A Distinct Function of STAT Proteins in Erythropoietin Signal Transduction*

The Janus kinase (JAK)-signal transducers and activators of transcription (STAT) pathway is an important signaling pathway of interferons and cytokines. We examined the activation of STAT proteins induced by interleukin-3 (IL-3), granulocyte-macrophage colony-stimulating factor (GM-CSF), or erythropoietin (EPO) using the human leukemia cell line, UT-7, which requires these cytokines for growth. IL-3, GM-CSF, and EPO induced DNA-binding activity to the oligonucleotides corresponding to the sis-inducible elements (SIE) of c-fos, in addition to the β-casein promoter (β-CAP), SIE- and β-CAP-binding proteins were identical to Stat1α and Stat3 complex and to Stat5 protein, respectively. This indicates that IL-3, GM-CSF, and EPO commonly activated Stat1α, Stat3, and Stat5 proteins in UT-7. However, EPO hardly activated Stat1α and Stat3 in UT-7/GM, which is a subline of UT-7 that grows slightly in response to EPO. Transfection studies revealed that UT-7/GM cells constitutively expressing Stat1α, but not Stat3, can grow as well in response to EPO as GM-CSF, suggesting that Stat1α is involved in the EPO-induced proliferation of UT-7. Thus, although Stat1α, Stat3, and Stat5 proteins are activated by GM-CSF, IL-3, and EPO, our data suggest that each STAT protein has a distinctive role in the actions of cytokines.

Cytokines play an important role in the proliferation and differentiation of hematopoietic cells. The binding of a cytokine to its specific receptor on the cell surface induces the tyrosine phosphorylation and activation of several proteins, including the receptor itself, receptor-associated tyrosine kinase such as Janus kinases (JAKs), 1 and signal transducers and activators of transcription (STATs) (1,2). STATs were first identified in studies of interferon (IFN)-regulated gene expression (3,4). In IFN signal transduction, the activated JAK kinases induce the tyrosine phosphorylation and activation of STATs. Activated STATs then form homo-or heterodimers, translocate into the nucleus, and bind to specific DNA sequences, leading to induction of the specific genes for IFN (5). Cytokines may induce the activation of specific STAT proteins and the formation of specific complexes, as do IFNs. If so, it would explain the specificity of action of each cytokine. Therefore, it is important to identify the STATs activated by each cytokine.
So far six members of the STAT family have been identified (6 -11). One of them, Stat5 protein, is activated by several cytokines including interleukin (IL)-3, granulocyte macrophage-colony-stimulating factor (GM-CSF) (12)(13)(14), and erythropoietin (EPO) (14 -16). Stat5 recognizes the palindromic motif TTCXXXGAA corresponding to the ␤-casein promoter (␤-CAP). Although sis-inducible elements (SIE) of c-fos do not have this motif, SIE binding complexes are induced by GM-CSF in the human monocytic leukemia cell line, U937 (17). This suggested that GM-CSF induces the activation of Stat1␣ and Stat3 proteins, because the SIE consensus sequence is recognized by Stat1␣ and Stat3 complex. In addition, GM-CSF and IL-3 activate Stat1␣ protein in human eosinophils (18), and these cytokines also activate Stat3 or Stat1 proteins in COS cells reconstituted with the receptors for these cytokines (19,20). Both Stat1␣ and Stat3 are activated by EPO in the murine EPO-dependent cell line, HCD-57 (21). By contrast, EPO reportedly activates a Stat5-related factor but not Stat1␣, Stat2, Stat3, or Stat4 proteins in murine IL-3-dependent BAF-3 cells expressing exogenous EPO receptor or in UT-7 cells maintained in another laboratory (14). In addition, Stat1␣ and Stat5 activation is induced by EPO in immature erythroid cells (22).
Thus several independent investigators have raised the question as to whether IL-3, GM-CSF, and EPO all actually activate Stat1␣, Stat3, and Stat5 proteins. To address this issue, we performed electromobility shift assays with oligonucleotides corresponding to the ␤-CAP and SIE consensus sequences and nuclear proteins extracted from IL-3-, GM-CSF-, or EPO-stimulated UT-7 cells, because the growth and survival of these cells absolutely required GM-CSF, IL-3, or EPO (23). In addition, we examined the role of Stat1␣ and Stat3 in the EPO-induced cellular proliferation using UT-7 and its subline, UT-7/GM.
Hematopoietic Preparation of Nuclear and Cytoplasmic Extracts-After stimulation, cells were washed with ice-cold PBS containing 2 mM Na 3 VO 4 , resuspended in a hypotonic buffer (20 mM HEPES (pH 7.9), 10 mM KCl, 1 mM MgCl 2 , 10% glycerol, 0.5 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 15 g/ml aprotinin, 3 g/ml leupeptin, 3 g/ml pepstatin, and 2 mM/liter Na 3 VO 4 , with 0.2% Nonidet P-40, and homogenized. After centrifugation at 1,000 ϫ g for 5 min, the supernatant was separated from the nuclear pellet and then centrifuged at 14,000 ϫ g for 20 min at 4°C. The debris was removed, and the supernatants were collected as cytoplasmic extracts. The nuclear pellets were resuspended in hypotonic buffer with 300 mM NaCl; debris was removed by centrifugation (14,000 ϫ g for 20 min), and the supernatants were collected as nuclear extracts.
Colorimetric MTT Assay for Cell Proliferation-Cell growth was also examined by a colorimetric assay according to Mosmann (29) with some modification. Briefly, cells were incubated at a density of 1 ϫ 10 4 /0.1 ml in 96-well plates in IMDM containing 10% fetal bovine serum in the absence or presence of various concentrations of GM-CSF or EPO. After 72 h of culture at 37°C, 20 l of sterilized 5 mg/ml 3-(4,5-dimethyl thiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) (Sigma) was added to each well. Following a 2-h incubation at 37°C, 100 l of 10% sodium dodecyl sulfate (SDS) was added to each well to dissolve the dark-blue crystal product. The optical density was measured at a wavelength of 595 nm using a microplate reader (model 3550, Bio-Rad).
RNA Extraction and Northern Blotting-Total RNA was isolated from cells according to the method of Chomczynski and Sacchi (31). RNA was resolved by electrophoresis on agarose formaldehyde gels, transferred to nylon membranes (Zeta-probe, Bio-Rad) in 10 ϫ standard sodium citrate (SSC), and hybridized to human cDNA fragments for EPOR. The fragment was labeled with [␣-32 P]CTP by random priming. After an overnight incubation at 43°C in the presence of 50% formamide, blots were washed 3 times with 2 ϫ SSC, 0.5 ϫ SSC, or 0.1 ϫ SSC containing 0.1% SDS for 15 min each. The membranes were autoradiographed using Kodak XAR-5 film with an intensifying screen at Ϫ70°C.
Preparation of Cell Lysates, Immunoprecipitation, and Western Blotting-UT-7, UT-7/GM, and UT-7/GM cells transfected with Stat1␣ or Stat3 cDNA were lysed on ice in lysis buffer composed of 20 mM Tris (pH 7.4), 137 mM NaCl, 10% glycerol, 1% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 15 g/ml aprotinin, and 2 mM sodium orthovanadate. After a 30-min incubation on ice, insoluble materials were removed by centrifugation at 15,000 ϫ g for 10 min. The supernatants were boiled for 5 min in SDS-PAGE sample buffer. After a brief centrifugation, the supernatants were resolved by SDS-PAGE and then electroblotted onto a nitrocellulose membrane. The blots were blocked with 3% skim milk in PBS for 30 min at RT and then incubated with the appropriate concentration of primary antibodies including polyclonal antibodies against Stat1 or Stat3 overnight at 4°C. After washing with PBS containing Tween 20 (1:2,000), the blots were probed with a 1:1,000 dilution of goat anti-rabbit horseradish peroxidase-conjugated second antibodies for 90 min at RT. After a second wash, the blots were incubated with an enhanced chemiluminescence substrate (enhanced chemiluminescence detection kit; Amersham, Buckinghamshire, UK) and exposed to Hyperfilm enhanced chemiluminescence to visualize immunoreactive bands. In some experiments, UT-7/GM and UT-7/EPO cells were starved of growth factor for 24 h. After stimulation with cytokines at 37°C for 10 min, cells were lysed on ice in lysis buffer (50 mM HEPES (pH 7.5), 0.1% Tween 20, 250 mM NaCl, 1 mM EDTA, 2.5 mM EGTA, 1 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, pepstatin, and aprotinin, 1 mM sodium fluoride, and 0.1 mM sodium orthovanadate). The supernatants were immunoprecipitated with anti-Stat1, anti-Stat3, or anti-Stat5 antibody coated onto protein G-Sepharose for 4 h at 4°C in an Eppendorf shaker. Immunoprecipitates were collected by a brief centrifugation and washed three times with 1 ml of lysis buffer. The immunoprecipitated proteins were boiled for 5 min in SDS-PAGE sample buffer. After a brief centrifugation, the supernatants were resolved by SDS-PAGE and then electroblotted onto a nitrocellulose membrane. The blots were incubated with the appropriate concentration of primary antibodies including antiphosphotyrosine (Tyr(P)) monoclonal antibody 4G10 overnight at 4°C. After washing with PBS containing Tween 20 (1:2,000), the blots were probed with a 1:1,000 dilution of goat anti-mouse horseradish peroxidase-conjugated second antibodies for 90 min at RT and visualized using the enhanced chemiluminescence detection kit. The blots were then stripped with 62.5 mM Tris-HCl (pH 6.8), 2% SDS, 100 mM ␤-mercaptoethanol at 50°C for 30 min, washed, blocked, and reprobed.

RESULTS
Induction of DNA-binding Activity to SIE and ␤-CAP Consensus Sequences by Cytokines-We initially analyzed the STAT signaling pathway activated by GM-CSF, IL-3, or EPO using UT-7. After 24 h of starvation, UT-7 cells were stimulated with GM-CSF (10 ng/ml), IL-3 (100 ng/ml), EPO (10 units/ml), or IL-6 (100 ng/ml) for 15 min and then harvested and prepared for EMSA. As shown in Fig. 1, GM-CSF, IL-3, and EPO induced the three distinct SIE binding complexes (A, B, and C). In addition, IL-6 induced the formation of DNA binding complex A and, to a much lesser degree, complexes B and C. Moreover, the formation of ␤-CAP binding complex was induced by GM-CSF, IL-3, and EPO but not by IL-6 ( Fig. 1). The formation of these complexes was completely inhibited by a 150-fold molar excess of unlabeled probe, indicating that the proteins specifically bound to the SIE or ␤-CAP consensus sequence. The density of SIE binding complex but not that of ␤-CAP binding complex paralleled the growth activity against UT-7 cells (see Ref. 24 and data not shown).
Collectively, these results indicate that the SIE-binding proteins contain Stat1␣ and Stat3 and that the ␤-CAP-binding protein contains Stat5 or Stat5-related protein. Thus, GM-CSF, IL-3, and EPO all induced the activation of Stat1␣, Stat3, and Stat5 proteins in UT-7.
Time-dependent Activation of STAT Proteins-The activation of the STAT proteins by GM-CSF, IL-3, and EPO may be involved in cell proliferation because these cytokines act on UT-7 cells as growth factors. If so, the induction of SIE and ␤-CAP binding complexes in response to these three cytokines would have similar kinetics. To examine this notion, we ana-lyzed the time course of the activation of SIE and ␤-CAP binding complexes in UT-7 cells stimulated with GM-CSF, IL-3, or EPO. The SIE binding complexes A, B, and C were detectable after a 1-min exposure to GM-CSF. The binding activity peaked at 5 min and gradually diminished over 60 min (Fig. 3A). On the other hand, the ␤-CAP binding complex was detectable after 1 min. The binding activity peaked at 1-5 min and was sustained for up to 60 min (Fig. 3A). IL-3 had similar kinetics as GM-CSF in SIE-binding activity (Fig. 3B). The ␤-CAP-binding activity was detectable after a 1-min exposure to IL-3, and it peaked at 5-10 min and thereafter declined (Fig.  3B). EPO had similar kinetics as IL-3 in the ␤-CAP-binding activity (Fig. 3C). SIE binding complexes were detectable after a 5-min exposure to EPO. The maximal level was reached after 20 -30 min and sustained for up to 60 min (Fig. 3C). Thus, the time course of the induction of SIE-and ␤-CAP-binding activities significantly differed among GM-CSF, IL-3, and EPO.
Dose-dependent Activation of STAT Proteins-We examined the concentration of each cytokine required for the induction of SIE-and ␤-CAP-binding activities. Both DNA-binding activities were elevated with increasing concentrations of GM-CSF, IL-3, and EPO. Among the SIE binding complexes, complex C was detectable at 0.1 ng/ml GM-CSF, and plateau levels were obtained at 10 ng/ml GM-CSF. Complexes A and B were induced at 10 ng/ml GM-CSF (Fig. 4A). The ␤-CAP binding complex was detectable at 0.01 ng/ml GM-CSF, and a plateau was reached at 1 ng/ml (Fig. 4A). In contrast, 10 ng/ml IL-3 was  4 and 8). EMSA was then performed on the extracts using 32 P-labeled oligonucleotides containing SIE or ␤-CAP sequences. required to induce both SIE and ␤-CAP binding complexes (Fig.  4B). Complex C was detectable at 0.1 unit of EPO/ml, whereas complexes A and B were detectable at 1 unit of EPO/ml. The formation of these complexes reached the maximum at 10 units/ml EPO (Fig. 4C). On the other hand, the ␤-CAP binding complex was induced at 0.1 unit/ml EPO, and the level was maximal at 1 unit/ml EPO (Fig. 4C). (16,32). To exclude the possibility that the activation of Stat1␣ and Stat3 by EPO is restricted to UT-7, we examined whether or not EPO induced SIE-binding activity in the EPO-responsive cell lines, UT-7/GM, UT-7/EPO, F-36E, and TF-1, which have endogenously functional EPORs. Fig. 5A shows the dose-response curves of these cell lines to EPO. As reported, UT-7/EPO and F-36E cells proliferated very well in response to EPO (25,26), whereas UT-7/GM and TF-1 cells slightly proliferated with EPO (24,27). Using these cell lines, we performed EMSA with SIE and ␤-CAP probes. As shown in Fig. 5B, EPO induced SIE-binding activity in UT-7/EPO and F-36E but not in UT-7/GM and TF-1. In contrast, ␤-CAP-binding activity was detected in all four cell lines.

EPO-induced Activation of Stat1␣ and Stat3 Is Not Restricted to UT-7-Other reports have indicated that only Stat5 is activated by EPO in cell lines exogenously expressing EPOR
EPO Induces the Tyrosine Phosphorylation of Stat1␣, Stat3, and Stat5-To further confirm that EPO actually activates Stat1␣, Stat3, and Stat5, we examined whether or not EPO induces the tyrosine phosphorylation of these proteins in UT-7/EPO and UT-7/GM. As shown in Fig. 6, EPO induced the tyrosine phosphorylation of Stat1␣, Stat3, and Stat5 in UT-7/ EPO. However, none of them were tyrosine-phosphorylated by GM-CSF, because this cell line cannot respond to GM-CSF (25). On the other hand, Stat5 alone was tyrosine-phosphorylated by EPO in UT-7/GM (data not shown). These findings are consistent with the data from the EMSA.
Thus, Stat1␣ and Stat3 are activated by EPO in EPO-hypersensitive cell lines but not in EPO-hyposensitive cell lines. By analogy, Stat1␣ and/or Stat3 activation may be closely involved in EPO-induced cellular proliferation.
Close Relationship between Proliferative Response and the Formation of SIE Binding Complexes-To clarify this notion, we transfected Stat1␣ or Stat3 cDNA into UT-7/GM cells by electroporation and obtained three independent clones for each transfectant. We confirmed that each clone indeed abundantly expressed exogenous Stat1␣ or Stat3 protein by Western blotting (Fig. 7A). Moreover, EMSA revealed that Stat1␣ and Stat3 are constitutively activated without EPO in each transfectant (Fig. 7B). Using these transfectants, we examined the effect of EPO on their growth in liquid culture. As shown in Fig. 7C, all three clones expressing exogenous Stat1␣ proliferated in re- sponse to EPO better than the parent UT-7/GM. In contrast, all three clones expressing Stat3 responded slightly to EPO like parent UT-7/GM and cells transfected with vector alone (Fig.  7C and data not shown). Similar results were obtained by MTT reduction assay (data not shown). These results suggest that the activation of Stat1␣, but not Stat3, is required for the EPO-induced proliferation of UT-7.
As we reported, UT-7/GM expressed a small amount of EPOR at the mRNA level (33), presumably resulting in low sensitivity to EPO. The increased sensitivity of Stat1␣ transfectants may be due to the up-regulation of EPOR. To verify this notion, we examined the EPOR mRNA level of Stat1␣ transfectants by Northern blotting. As shown in Fig. 8, EPOR mRNA was faintly detectable in each clone like the parent UT-7/GM. This result suggests that exogenous Stat1␣ induced the increase of the sensitivity to EPO without affecting expression of the EPOR gene.
Overexpression of EPOR Restores the Proliferative Response to EPO and EPO-induced Activation of Stat1␣ and Stat3 in UT-7/GM Cells-To examine whether or not loss of the activation of Stat1␣ and Stat3 by EPO is due to the down-regulation of EPOR, we generated a UT-7/GM transfectant expressing a large amount of exogenous EPOR (Fig. 9A). As shown in Fig. 9B, the proliferative response to EPO is restored in the transfectant cells. In addition, EPO induced the activation of Stat1␣ and Stat3 in these cells but not in parent UT-7/GM cells (Fig. 9C). These findings suggest that the activation of Stat1␣ and Stat3 depends on the expression of EPOR. Thus, our data indicate that Stat1␣ and Stat3 lie downstream of the EPO receptor in EPO signaling pathway. DISCUSSION In this study we demonstrated by means of EMSA using oligonucleotides corresponding to the SIE consensus sequence that GM-CSF, IL-3, and EPO all activate the three distinct SIE binding complexes in UT-7 cells. A supershift study with specific antibodies against Stat1␣ and Stat3 revealed that the three SIE binding complexes A, B, and C were identical to the Stat3-Stat3 homodimers, Stat1␣-Stat3 heterodimers, and Stat1␣-Stat1␣ homodimers, respectively. In addition, we found that GM-CSF, EPO, and IL-3 all activate ␤-CAP binding complex. This complex was supershifted by anti-Stat5 antibody, suggesting that the ␤-CAP-binding protein is identical to Stat5 or Stat5-related protein. Collectively, these results indicate that GM-CSF, IL-3, and EPO commonly induced the activation of Stat1␣, Stat3, and Stat5 in UT-7.
It is notable that in addition to Stat5, the activation of Stat1␣ and Stat3 is induced by EPO in UT-7. This is in contrast to previous reports indicating that only Stat5 is activated by EPO in cell lines exogenously expressing EPOR (16, 32). Although we cannot exclude the possibility that this discrepancy is in part due to the cell lines used for these experiments, differences in sensitivity to EPO in cell lines appear to be closely associated with the activation of Stat1␣ and Stat3. Indeed, EPO induced the activation of Stat1␣ and Stat3 molecules in UT-7/EPO and F-36E having high sensitivity to EPO. In contrast, EPO did not induce the activation of these Stat proteins in TF-1 and UT-7/GM having low sensitivity to EPO. These results suggest that sensitivity to EPO is dependent on EPO-induced SIE-binding activity. This notion is supported by the finding that UT-7/GM cells transfected with Stat1␣ cDNA have a restored proliferative response to EPO.
Although Stat1␣ and Stat3 were not activated by EPO in UT-7/GM cells, these proteins were actually activated by GM-CSF in these cells, indicating that Stat1␣ and Stat3 molecules are functionally intact in UT-7/GM cells. In addition, Stat1␣ and Stat3 are expressed in UT-7/GM at the same level as they are in UT-7 (data not shown). The concept is speculative at present, but a large amount of EPOR appears to be required for the activation of Stat1␣ and Stat3 proteins, because UT-7/GM cells transfected with human EPOR cDNA restored the proliferative response to the EPO and EPO-induced activation of Stat1␣ and Stat3.
Stat1␣ knockout mice demonstrated that elimination of the Stat1␣ gene does not affect erythropoiesis (34,35). This seems to contradict our observation. This discrepancy may be explained by two possibilities. One is that other Stat protein(s) including Stat3 compensates for the lack of Stat1␣. The other is that Stat1␣ is involved in abnormal erythropoiesis, as shown by results obtained using leukemic cell line cells.
Very recently, Marra et al. (36) reported that the activation of Stat1␣ promotes the epidermal growth factor-or plateletderived growth factor-induced DNA synthesis. Consistent with this, our preliminary data showed that constitutive activation of Stat1␣ but not Stat3 actually promotes the cell cycle progression by EPO treatment (data not shown). Therefore, it is possible that Stat1␣ activates cell cycle-associated gene(s). Indeed, Chin et al. (37) reported that Stat1␣ can bind to the promoter region of the p21 gene, one of the cyclin-dependent kinase inhibitors. We found that GM-CSF but not EPO transiently induced the p21 gene at the mRNA level in UT-7/GM cells (data not shown). In addition, a recent study suggested that p21 did not inhibit but rather promoted the cell cycle progression in a human leukemia cell line (38). Taken together with these observations and our present data, it would not be surprising that activated Stat1␣ by EPO plays some role in the progression of the cell cycle mediated through activation of cell cycle-associated genes, one of which may be the p21 gene.
The function of Stat3 on EPO signaling pathway remains unknown. Overexpression of Stat3 had no effect on the EPOinduced proliferation of UT-7/GM cells. However, Stat3 is constitutively activated in cells transformed by the src oncogene (39,40). In addition, not only Stat1␣ but also Stat3 is autonomously activated in an erythroleukemia cell line, the growth of which was changed from EPO-dependent to EPO-independent by transfection with Friend virus (21). Taken together, Stat3 may play some role in the inhibition of erythroid differentiation.
Although EPO slightly stimulated the cellular proliferation of UT-7/GM, this cytokine induced the activation of Stat5 to the same degree as GM-CSF. Moreover, there was no significant difference in the degree of Stat5 activation by EPO between UT-7 and UT-7/GM cells (Figs. 3C and 5B), suggesting that a small amount of EPOR is sufficient for the activation of Stat5 protein. The activation of Stat5 is reportedly associated with EPO-induced cellular proliferation (41,42). In contrast, Quelle et al. (32) reported that the tyrosine residues in the EPO receptor required for activation of Stat5 are not critical for the mitogenic response. Thus, although the role of Stat5 in EPOinduced cellular proliferation remains controversial, our results indicate that Stat5 activation may be required but is not sufficient for EPO-induced maximal cell growth. It is notable that Stat5 or Stat5-related proteins activated by GM-CSF, IL-3, or EPO were supershifted to two distinct SS5 by anti- FIG. 9. Overexpression of EPOR restores proliferative response to EPO and EPO-induced activation of Stat1␣ and Stat3 in UT-7/GM cells. A, expression of EPOR mRNA in UT-7/GM cells transfected with human EPOR cDNA. Northern blot analysis was performed with a 32 P-labeled human EPOR cDNA probe. B, profile of proliferative response to EPO and GM-CSF. Cells were plated at a density of 10 4 /well in IMDM supplemented with 5% fetal bovine serum and cultured with increasing concentrations of EPO (0.01-100 units/ml) or GM-CSF (0.01-100 ng/ml). MTT reduction was measured after 3 days in culture. Values represent means of triplicate cultures. C, induction of SIE-binding activity by EPO stimulation. EMSA was performed on the extracts using 32 P-labeled oligonucleotides containing SIE sequence as a probe.
Stat5 antibody. One possibility is that these two bands consist of two Stat5 or Stat5-related proteins or a complex of Stat5 with unidentified cofactor(s). More detailed studies are necessary to address this issue.
Studies on interferon signal transduction have indicated that the activation of a specific combination of STAT proteins by each cytokine can contribute to the distinct responses induced by each cytokine. Although our results did not support this notion at least among GM-CSF, IL-3, and EPO, it is notable that there is a significant difference in the kinetics of Stat1␣ and Stat3 and Stat5 activation among three cytokines. Considering that the target gene(s) of Stat1␣ and Stat3 differs from that of Stat5, the difference in the time courses of these target gene activations may explain the specificity of each cytokine (43).
Not only JAK but also other kinases or phosphatases may be involved in the activation or inactivation of STATs (44 -46). Indeed, serine phosphorylation also participates in the activation of STAT proteins, including Stat1␣ and Stat3 (47)(48)(49)(50)(51)(52). Since STAT proteins have a consensus sequence that can be phosphorylated by mitogen-activated protein kinase (49), phosphorylation on serine residues of STAT proteins may be induced by activated mitogen-activated protein kinases (49,51). This notion is supported by evidence that IL-3, GM-CSF, or EPO activates not only JAK2 (53) but also the Ras-Raf-mitogen-activated protein kinase pathway (54). In addition, some STAT proteins may be tyrosine-phosphorylated by the Src family of tyrosine kinases such as Lyn or Fps, since these proteins are associated with the common ␤ chain of the receptors for IL-3 and GM-CSF (55)(56)(57)(58). Alternatively, since Shp 1 (HCP, PTP1C) is associated with the common ␤ chain (58) and the EPO receptor (44), this molecule may be involved in the dephosphorylation and inactivation of STAT proteins. Thus, although speculative at present, several kinases and phosphatases may modify the activity of STAT proteins. This notion would explain the finding that there is a significant difference in the kinetics of Stat1␣ and Stat3 and Stat5 activation.
In summary, we demonstrated that GM-CSF, IL-3, and EPO all activate Stat1␣, Stat3, and Stat5, although the biological function of each STAT remains uncertain. Among them, Stat1␣ may play an important role in EPO-induced cellular proliferation.