Insulin-like Growth Factor-I Augments Erythropoietin-induced Proliferation through Enhanced Tyrosine Phosphorylation of STAT5*

Insulin-like growth factor (IGF-I) is known to synergistically stimulate the proliferation of hematopoietic cells in combination with other hematopoietic growth factors. However, the precise mechanism underlying the cooperative effects of IGF-I is unknown. In a human interleukin-3 or erythropoietin (EPO)-dependent cell line, F-36P, IGF-I alone failed to stimulate DNA synthesis but did augment the EPO-dependent DNA synthesis of F-36P cells. The treatment of F-36P cells with a combination of EPO and IGF-I (EPO/IGF-I) was found to enhance EPO-induced tyrosine phosphorylation of STAT5, whereas IGF-I alone did not. Furthermore, c-CIS mRNA expression, one of the target molecules of STAT5, was more effectively induced by EPO/IGF-I than by EPO alone. To examine the mechanisms of the EPO- and EPO/IGF-I-induced proliferation of F-36P cells, we expressed dominant negative (dn) mutants of STAT5 and Ras in an inducible system. The EPO-induced DNA synthesis and the cooperative effect of EPO/IGF-I were significantly inhibited by the inducible expression of dn-STAT5 or dn-Ras. In addition, the inducible expression of dn-Ras abolished the IGF-I-enhanced tyrosine phosphorylation of STAT5. These results suggest that IGF-I may augment EPO-induced proliferation by enhancing tyrosine phosphorylation of STAT5 and raise the possibility that Ras may be involved in the augmentation of STAT5 tyrosyl phosphorylation.

Insulin-like growth factor (IGF-I), 1 a 70-amino acid peptide structurally related to insulin, is a metabolic hormone that mediates a number of the anabolic effects of growth hormones (for a review see Ref. 1). It has also been reported that IGF-I plays an important role in the control of cell growth and survival in a variety of cell types, including fibroblasts, smooth muscle cells, and hematopoietic cells (2)(3)(4). In a hematopoietic system, IGF-I has been shown to promote growth in a broad spectrum of hematopoietic cells. A number of studies have shown that IGF-I is able to cooperate with erythropoietin (EPO) in promoting the formation of erythroid burst-and colony-forming units in serum-free colony assays (5)(6)(7). It has also been reported that IGF-I enhances granulopoiesis and expansion of pro-B cells from normal hematopoietic progenitor cells (8,9). In addition, daily infusion of IGF-I into normal mice has been found to result in a dramatic increase in splenic B and thymic T cells (8,10). However, this cell growth cannot be sustained by IGF-I alone but requires synergistic stimulation by a combination with other growth factors such as EPO, platelet-derived growth factor, and IL-7 (5,10,11). Although the intracellular signaling pathways from IGF-I receptor (IGF-IR) have been the subject of intense investigation (for a review see Ref. 1), the mechanisms by which IGF-I cooperates with other growth factors remain unknown.
EPO is a glycoprotein hormone that is required for the survival, proliferation, and differentiation of committed erythroid progenitor cells (for a review see Ref. 12). In a recent report, mice that contained null mutations in both EPO and EPO receptor (EPOR) genes were shown to die at around embryonic day 13 because of the failure of definitive erythropoiesis in the fetal liver (13). EPOR belongs to the cytokine receptor superfamily, which includes receptors for interleukins, colony-stimulating factors, and thrombopoietin. These receptors do not appear to contain any recognized kinase domain or enzymatic motif in the cytoplasmic domain but are capable of inducing a series of biochemical events, including tyrosine phosphorylation and activation of Janus family of protein tyrosine kinases (JAKs), signal transducers and activators of transcription (STATs), phosphatidylinositol 3-kinase, and Shc (for a review see Ref. 14). EPOR, JAK2, and STAT5 are all known to be tyrosine phosphorylated and activated upon ligand binding. The activated STAT5 dimerizes, translocates to the nucleus, binds to specific DNA sequences, and participates in transcriptional regulation (15,16).
Among six members of the STAT proteins, STAT5 was originally identified as a mammary gland factor (MGF) that was regulated by prolactin (17). STAT5 is known to be activated by multiple cytokines such as IL-3, IL-5, granulocyte-macrophage colony-stimulating factor (GM-CSF), granulocyte colony-stimulating factor, EPO, and thrombopoietin (18 -22). Although the functional roles of STAT5 in cytokine-dependent proliferation and differentiation have been investigated by a number of investigators, the results of these studies were contradictory: one study demonstrated that a dominant-negative (dn) STAT5 suppressed IL-3-induced proliferation (23), and the other study showed that a mutant EPOR, which cannot activate STAT5, transmitted EPO-induced mitogenic signals as efficiently as the wild-type receptor (24). Furthermore, it has been reported that STAT5 is involved in EPO-induced erythroid differentiation of murine erythroleukemia cell lines, ELM-I-1 and SKT6, whereas an opposite result was shown in a human IL-3-dependent erythrocytic leukemia cell line TF-1 (25)(26)(27).
In this study, we investigated the molecular mechanism by which IGF-I exerts its effect on EPO-induced proliferation of a human IL-3-dependent erythroleukemia cell line, F-36P. By using dn forms of STAT5 and Ras, we demonstrate here that both STAT5 and Ras are involved in the EPO-induced DNA synthesis of F-36P cells and that IGF-I augments EPO-induced DNA synthesis through the enhanced tyrosine phosphorylation of STAT5.
Plasmid Construct-A dn mutant of human c-H-Ras (H-Ras S17N ) cDNA was kindly provided from Dr. T. Satoh (Tokyo Institute of Technology, Yokohama, Japan) (28). A dn HA-tagged STAT5 (HA-MGF 694F ) was described in previous papers (26,29). These cDNAs were subcloned into the NotI site of pOPRSVI-CAT (Stratagene, La Jolla, CA), instead of the chroramphenicol acetyltransferase gene by using NotI linkers.
[ 3 H]Thymidine Incorporation Assay-To quantitate DNA synthesis of the cells, a [ 3 H]thymidine incorporation assay was used as previously described (31). In brief, after 24-h serum and rhIL-3 starvation, triplicate aliquots of cells (3.0 ϫ 10 4 cells resuspended in 100 l of serum-free Cos 004 medium (Cosmobio, Tokyo, Japan)) were cultured in 96-well flat bottom microtiter plates for 48 h at 37°C in the presence or absence of the growth factor(s). Northern Blot Analysis-Isolation of total cellular RNA and the method for Northern blot were described previously (32). The probe for c-CIS mRNA was full-length cDNA, and that for c-myc mRNA corresponded to exon 3.
Immunoprecipitation and Immunoblotting-The isolation of cellular lysates, immunoprecipitation, gel electrophoresis, and immunoblotting were performed according to the methods described previously (33). Briefly, after serum and rhIL-3 starvation for 12 h, F-36P cells were treated with an appropriate growth factor(s). The cells were then lysed in lysis buffer, and insoluble material was removed by centrifugation. For immunoprecipitation, the precleared lysates obtained from 1 ϫ 10 7 cells were incubated with 1 g of anti-STAT5b Ab or 2 g of anti-HA mAb, followed by the addition of protein G-Sepharose beads. The immunoprecipitates or whole cell lysates (15 g/lane) were subjected to SDS-polyacrylamide gel electrophoresis. The proteins were electrophoretically transferred onto a polyvinylidene difluoride membrane (Immobilon, Millipore Corp., Bedford, MA). After blocking the residual binding sites on the filter, immunoblotting was performed with an appropriate Ab. Immunoreactive proteins were visualized with the enhanced chemiluminescence detection system (NEN Life Science Products, Boston, MA). In some experiments, the filters were then stripped and reprobed with anti-STAT5b Ab to confirm the amounts of the immunoprecipitated STAT5.
Inducible Expression of Dominant Forms of STAT5 and Ras-To inducibly express dn-STAT5 and dn-Ras in F-36P cells, we used a LacSwitch TM II inducible expression system (Stratagene), in which the target cDNA is inducibly expressed by the addition of IPTG. In short, F-36P cells were initially transfected with an expression vector of Lac repressor (Lac-R), pCMV-LacI, by electroporation (250 V, 960 microfarad) (Bio-Lad, Richmond, CA). The transfected cells were screened by the culture with hygromycin (Sigma, St. Louis, MO) at a concentration of 0.5 mg/ml. Several hygromycin-resistant cells were cloned, and expression levels of Lac-R mRNA were examined by Northern blot analysis. Of these clones, one clone (designated Lac-R/cl3) was selected, because it showed the most intensive expression of Lac-R mRNA and also retained characteristics that were nearly the same as the parental F-36P cells. Lac-R/cl3 cells were next transfected with pOPRSVI each containing dn-STAT5 and dn-Ras. The expression vector pOPRSVI contains Rous sarcoma virus promoter linked to the Escherichia coli lactose operon, and expression of the target cDNA is suppressed by Lac-R through the lactose operon. When IPTG was added to culture medium, Lac-R was released from lactose operon, and transcription of the target cDNA is initiated. The transfected cells were screened by the culture with G418 (Sigma) at a concentration of 1.5 mg/ml. G418resistant cells were cloned, and induction levels of the target protein were examined before and after the treatment with IPTG by Western blot analyses. After the selection, 694F cl1 and N17 cl2 were subjected to further analyses, because dn-STAT5 and dn-Ras proteins were most efficiently induced by treatment with IPTG in these clones.
Luciferase Assay-Two types of luciferase plasmids each containing potential STAT5-and AP1-binding sequence were used as reporter genes. The details of 3x AP1-Lu, which can be transactivated by Rasmediated AP1, have been described previously (34). To construct a reporter gene for STAT5 (3x ␤-Cas-Lu), three tandem repeats of GAS sequence from ␤-casein gene (5Ј-TCGAAGATTTCTAGGAATTCAAAT-CGTAC-3Ј; recognition site is underlined) were subcloned into a location just upstream of the murine minimal JunB promoter (Ϫ42 to ϩ136) linked to the firefly luciferase gene (17). The luciferase assay was performed by using the Dual-Luciferase Reporter System (Promega, Madison, WI), in which the transfection efficiency was monitored by cotransfected pRL-CMV-Rluc, an expression vector of renilla luciferase. The cultured cells were electroporated with 30 g of reporter gene, together with 30 g of pRL-CMV-Rluc. The transfected cells were serum-and IL-3-starved for 12 h and then stimulated with rhEPO (10 units/ml) and/or rhIGF-I (10 ng/ml) for 5 h. To examine the effects of dn mutants, the cells were pretreated with 1 mM IPTG for 24 h prior to electroporation and cultured with IPTG during the assay. The cells were lysed in lysis buffer supplied by manufacturer followed by measurement of the firefly and the renilla luciferase activities on luminometer LB96P (Berthold Japan, Tokyo, Japan). The relative firefly luciferase activities were calculated by normalizing transfection efficiency according to the renilla luciferase activities. The experiments were performed in triplicate, and similar results were obtained from at least three independent experiments.
Effects of rhIGF-1 on Cytokine-induced Tyrosine Phosphorylation of STAT5-To analyze the mechanism underlying the synergistic effect of rhEPO and rhIGF-I on DNA synthesis of F-36P cells, we examined the changes in surface expression of IGF-IR during the 48-h treatment with rhEPO by flow cytometric analysis. The expression level of IGF-IR was found to be considerably high even before the treatment and to be not affected by the EPO treatment, suggesting that the synergistic effect was not mediated by up-regulation of IGF-IR by rhEPO (data not shown).
Because STAT5 has been reported to be involved in the EPOand IL-3-dependent proliferation in previous studies (19,21), we examined changes in tyrosine phosphorylation of STAT5 after 15 min of stimulation with rhEPO or rhIGF-I or in combination (rhEPO/rhIGF-I). Consistent with the previous reports, the treatment with rhEPO resulted in tyrosine phosphorylation of STAT5 ( Fig. 2A, EPO lane). Although IGF-I alone failed to induce tyrosine phosphorylation of STAT5 ( Fig. 2A, IGF-I lane), the treatment with rhEPO/rhIGF-I led to a significantly increased level of tyrosine phosphorylation of STAT5 as compared with that induced by rhEPO alone (Fig. 2A, EPO/ IGF-I lane). In a time course analysis, rhIGF-I was shown to enhance rhEPO-induced tyrosine phosphorylation of STAT5 during the entire test period (Fig. 2B). However, an apparent difference was not observed in the kinetics of the tyrosine phosphorylation of STAT5 between the treatments with rhEPO and rhEPO/rhIGF-I (Fig. 2B). The stimulation with rhIL-3 or rhGM-CSF was more effective in inducing tyrosine phosphorylation of STAT5 than that with rhEPO (Fig. 2C, IL3 and GM-CSF lanes), and rhIGF-I was ineffective in augmenting rhIL-3-or rhGM-CSF-induced tyrosine phosphorylation of STAT5 (Fig. 2C, IL3/IGF-I and GM-CSF/IGF-I lanes). In contrast to the effect of IGF-I, SCF, whose receptor is a member of receptor tyrosine kinase family, did not enhance rhEPO-induced tyrosine phosphorylation of STAT5 (Fig. 2C, EPO lane versus EPO/SCF lane).
Treatment with rhEPO/rhIGF-I Enhances the Transactivating Activities of STAT5-To evaluate the changes in transactivating activity of STAT5 before and after treatment with rhEPO or rhEPO/rhIGF-I, we examined the induction of c-CIS mRNA by Northern blot analysis, which is one of the target molecules of STAT5 (35,36). The treatment with rhEPO for 30 min was found to induce c-CIS mRNA expression in F-36P cells (Fig. 3A). The treatment with rhIGF-I alone was not capable of inducing c-CIS mRNA expression, whereas the treatment with rhEPO/rhIGF-I led to a higher level of c-CIS mRNA expression than that with rhEPO alone (Fig. 3A). In addition, a time course analysis revealed that rhEPO/rhIGF-I was more effective in inducing c-CIS mRNA than rhEPO alone from 30 to 90 min (Fig. 3B). In contrast to c-CIS mRNA induction, the induction levels of c-myc mRNA were essentially the same for rhEPO alone and rhEPO/rhIGF-I (Fig. 3A). To quantitate the cooperative effects of rhEPO and rhIGF-I, we performed a luciferase assay with two types of reporter genes, 3x ␤-Cas-Lu and 3x AP1-Lu, that are transactivated by STAT5 and Ras-mediated AP1, respectively. The treatments with rhEPO and rhEPO/ rhIGF-I stimulated luciferase activities of 3x ␤-Cas-Lu by 5.1and 7.7-fold, respectively, whereas rhIGF-I alone had no affect on the ␤-Cas-driven luciferase activity (Fig. 4A). The activity of 3x AP1-Lu was stimulated 4.5-fold by treatments with rhEPO, 4.3-fold by rhIGF-I, and 4.7-fold with rhEPO/rhIGF-I, respectively (Fig. 4B). In contrast to the results obtained from 3x ␤-Cas-Lu, rhIGF-I showed no additive effect on EPO-induced luciferase activity in 3x AP1-Lu (4.7-fold versus 4.5-fold).
Inducible Expression of dn-STAT5 and dn-Ras-To characterize the mechanism of synergistic effect of rhIGF-I on rhEPOinduced DNA synthesis of F-36P cells, we prepared stable transfectants that can inducibly express HA-tagged dn-STAT5 (designated 694F cl1) and dn-Ras (designated N17 cl2). As shown in a representative blot (Fig. 5A), the addition of 1 mM IPTG to the culture medium led to the induction of dn-STAT5 as early as 4 h, and its expression peaked at ϳ24 h and was retained for up to 48 h in the HA-immunoprecipitated proteins from 694F cl1 cells. In addition, Western blot analysis on the whole cell lysates from N17 cl2 cells showed that treatment with IPTG resulted in the induction of dn-Ras (Fig. 5B). In the absence of IPTG pretreatment, the treatment with rhIGF-I enhanced rhEPO-induced tyrosine phosphorylation of STAT5 in 694F cl1 cells in a manner similar to that of parental F-36P cells (data not shown); by contrast, when dn-STAT5 was induced to express by pretreatment with IPTG, both rhEPO-and rhEPO/rhIGF-I-induced tyrosine phosphorylations of STAT5 were found to be severely reduced as compared with those in the absence of dn-STAT5 (data not shown). In addition, dn-STAT5, HA-MGF 694F , was not tyrosine-phosphorylated by treatment with rhEPO, rhIGF-I, or rhEPO/rhIGF-I (Fig. 5C). These results suggest that both rhEPO and rhEPO/rhIGF-I catalyze the phosphorylation of only a single tyrosine residue (tyrosine 694) of STAT5 and that the rhIGF-I-enhanced tyrosine phosphorylation occurred at the tyrosine 694 residue but not at other tyrosine residues.
Inhibitory Effects of dn-STAT5 and dn-Ras on the Signaling Pathway-To evaluate the inhibitory effects of dn-STAT5 and dn-Ras in each signaling pathway, we performed a luciferase assay with reporter plasmids, 3x ␤-Cas-Lu and 3x AP1-Lu. In 694F cl1 cells, both rhEPO and rhEPO/rhIGF-I induced luciferase activities of 3 ϫ ␤-Cas-Lu in about the same manner as in parental F-36P cells in the absence of dn-STAT5: a 5.3-fold induction by rhEPO and an 8.0-fold induction by rhEPO/ rhIGF-I (Fig. 6A). In contrast, when dn-STAT5 was inducibly expressed by pretreatment with IPTG, both rhEPO-and rhEPO/rhIGF-I-induced luciferase activities were severely reduced: 1.3-fold by rhEPO and 1.8-fold by rhEPO/rhIGF-I (Fig.  6A). The inhibitory effects of dn-Ras on 3x AP1-Lu were also observed in N17 cl2 cells. In the absence of dn-Ras, the treatment with rhEPO, rhIGF-I, and rhEPO/rhIGF-I stimulated  3. Effects of rhEPO and rhEPO/rhIGF-I on the expression of c-CIS and c-myc mRNA. A, after 6 h of serum and rhIL-3 starvation, F-36P cells were stimulated with rhEPO (10 units/ml) and/or rhIGF-I (10 ng/ml) for 30 min, and expression of c-CIS and c-myc mRNA was examined by Northern blot analysis. B, F-36P cells were stimulated with rhEPO or rhEPO/rhIGF-I, and expression of c-CIS mRNA at each condition was examined at the time indicated.
Effects of dn-STAT5 and dn-Ras on rhEPO-and rhEPO/ rhIGF-I-induced DNA Synthesis-We next examined the effects of dn-STAT5 and dn-Ras on rhEPO-and rhEPO/rhIGF-Iinduced DNA synthesis of F-36P cells. The 694F cl1 and N17 cl2 cells were IL-3-starved for 24 h in the presence or absence of IPTG, and the effects of rhEPO (10 units/ml) and/or rhIGF-I (10 ng/ml) on DNA synthesis were then examined after a 48-h culture period with or without IPTG. In the absence of IPTG pretreatment, the responses in DNA synthesis of Vector cl1 (a stable transformant with an empty vector), 694F cl1, and N17 cl2 cells to each growth factor were similar to those of parental F-36P cells (Fig. 7). The addition of IPTG had no effect on the responses of Vector cl1 cells to each growth factor. In contrast, the inducible expression of dn-STAT5 by IPTG led to a reduction of rhEPO-induced DNA synthesis by 34% and the rhEPO/ rhIGF-I-induced DNA synthesis by 61% in 694F cl1 cells, when DNA synthesis was compared between the presence and absence of IPTG treatment (Fig. 7). In addition, in N17 cl2 cells, dn-Ras suppressed the rhEPO-induced DNA synthesis by 32% and rhEPO/rhIGF-I-induced DNA synthesis by 54% (Fig. 7). Thus, rhEPO/rhIGF-I-induced DNA synthesis was more significantly inhibited by dn-STAT5 or dn-Ras than rhEPO-induced DNA synthesis, suggesting that both STAT5 and Ras are involved in the rhEPO-induced proliferation of F-36P cells and also, at least in part, in the synergistic effect of rhIGF-I on rhEPO-induced proliferation.

Involvement of Ras Pathway in the IGF-I-enhanced
Tyrosine Phosphorylation of STAT5-In an effort to characterize the mechanism by which IGF-I enhances the tyrosine phosphorylation of STAT5, we examined changes in tyrosine phosphorylation of JAK2 after the treatment with rhEPO or rhEPO/ rhIGF-I. As shown in Fig. 8A, no significant difference in intensity and kinetics of JAK2-tyrosyl phosphorylation was observed between the rhEPO and rhEPO/rhIGF-I treatments, suggesting that JAK2 tyrosine kinase is probably not involved in the IGF-I-enhanced tyrosine phosphorylation of STAT5. Because the synergistic effect of rhIGF-I and rhEPO on DNA synthesis was inhibited by dn-Ras as efficiently as by dn-STAT5 (Fig. 7), we next examined the effects of dn-Ras on the IGF-I-enhanced tyrosine phosphorylation of STAT5 (Fig. 8B). N17 cl2 cells were stimulated with rhEPO or rhEPO/rhIGF-I in the presence and absence of IPTG pretreatment. Without IPTG pretreatment, rhEPO/rhIGF-I was found to enhance rhEPOinduced tyrosine phosphorylation of STAT5 in N17 cl2 cells as well as F-36P cells. In contrast, the rhEPO/rhIGF-I-enhanced tyrosine phosphorylation of STAT5 was abolished by the pre-

FIG. 5. Inducible expression of dn-STAT5 and dn-Ras in F-36P cells.
694F cl1 and N17 cl2 cells were treated with 1 mM IPTG for the time indicated. HA-immunoprecipitated (IP) proteins from 694F cl1 cells (A) or total cell lysates from N17 cl2 cells (B) were subjected to SDS-polyacrylamide gel electrophoresis, and the blots were probed with anti-STAT5b Ab, and anti-Ras mAb, respectively. C, during 24-h pretreatment, 694F cl1 cells were serum-and rhIL-3-starved for 12 h and treated with 1 mM IPTG. The cells were then stimulated with rhEPO (10 units/ml) and/or rhIGF-I (10 ng/ml) for 15 min. HA-immunoprecipitated proteins (HA-MGF 694F ) were probed with anti-phosphotyrosine mAb.
FIG. 6. Evaluation of the inhibitory effects of dn-STAT5 and dn-Ras on each signaling pathway. 694F cl1 and N17 cl2 cells were electroporated with 30 g of each reporter gene (3x ␤-Cas-Lu and 3x AP1-Lu, respectively) together with 30 g of pRL-CMV-Rluc. The transfected cells were serum-and IL-3-starved for 12 h and then stimulated with rhEPO (10 units/ml) and/or rhIGF-I (10 ng/ml) for 5 h. To examine the effects of dn mutants, the cells were pretreated with 1 mM IPTG for 24 h before electroporation and cultured with IPTG during the assay. The relative firefly luciferase activities were calculated by normalizing transfection efficiency according to the renilla luciferase activities. The results are shown as the means Ϯ S.D. of triplicated experiments. treatment with IPTG. This result raises the possibility that Ras-mediated signaling may be involved in the enhanced tyrosine phosphorylation of STAT5 by rhIGF-I. DISCUSSION The IGF-IR is a tyrosine kinase receptor with 70% homology to the insulin receptor. On ligand binding, IGF-IR autophosphorylates and dimerizes, activating its kinase activity. The activated IGF-IR induces the tyrosine phosphorylation of insulin receptor substrate 1 and Shc, leading to activation of the phosphatidylinositol 3-kinase/Akt kinase pathway and a Ras/ MAPK cascade (for reviews see Refs. 37 and 38). The Ras/ MAPK cascade, which consists of Raf, MEK (MAPK kinase) and MAPK is known to transmit mitogenic signals from IGF-IR, resulting in gene expression such as c-fos, c-jun, c-myc, and cyclin D1 (39 -42). However, a number of previous studies have reported that IGF-I alone shows only a minimum effect on cell proliferation but synergizes with other growth factors to induce cell proliferation (5,10,11). Consistent with these findings, we here showed that rhIGF-I alone was unable to promote DNA synthesis of F-36P cells, but that it augmented the GM-CSF-, IL-3-, SCF-, and EPO-induced DNA synthesis. Because rhIGF-I activated Ras pathways as effectively as EPO in the luciferase assay, it was suggested that the activated Ras signaling from IGF-IR alone may not be sufficient to induce DNA synthesis.
In F-36P cells, both dn forms of STAT5 and Ras inhibited rhEPO-induced DNA synthesis. Although the precise role of STAT5 in growth and differentiation of hematopoietic cells is not fully understood, our results suggest that STAT5, in addition to Ras, may be involved in the EPO-induced proliferation of F-36P cells. This conclusion is supported by the recent finding that although STAT5A targeted mice underwent normal development without apparent hematopoietic abnormalities (43), the macrophages obtained from the bone marrow of these mice showed a significantly decreased proliferative response to rhGM-CSF of 33% as compared with those from normal mice (16). Furthermore, dn-STAT5 and dn-Ras were found to inhibit the synergistic effect of rhIGF-I on rhEPO-induced DNA synthesis of F-36P cells. These results suggest that both STAT5 and Ras may be involved in the cooperative effect of rhIGF-I. In N17 cl2 cells, dn-Ras showed a considerable but not complete inhibitory effect on the cooperation in DNA synthesis, whereas it completely abrogated IGF-I-enhanced tyrosine phosphorylation of STAT5. Recently, Zhou et al. (44) reported that IGF-I is capable of stabilizing STAT5 protein in a murine IL-3-dependent myeloblastic leukemia cell line, 32D, and that this stabilization might contribute to the protection from apoptosis in- duced by IL-3 withdrawal. Although we did not study the expression levels of STAT5 protein during the treatment with rhEPO, rhIGF-I, or rhEPO/rhIGF-I, the enhanced tyrosine phosphorylation and the stabilization of STAT5 may be involved in the synergistic effect of EPO and IGF-I on F-36P cells.
In addition to hematopoietin receptors, several receptor tyrosine kinases such as receptors for erythroid growth factor, platelet-derived growth factor, and insulin have recently been reported to induce tyrosine phosphorylation and transcriptional activation of STAT proteins (45)(46)(47). Furthermore, insulin has been shown to phosphorylate on serine residue of STAT3 and STAT5, which is required for the maximal activation of these molecules (47,48). In the present study, IGF-I alone was ineffective in inducing the tyrosine phosphorylation of STAT5 but enhanced the EPO-induced tyrosine phosphorylation of STAT5. The enhancing effect of IGF-I on tyrosine phosphorylation of STAT5 was abrogated by dn-Ras, suggesting that the Ras pathway(s) may be involved in the increased phosphorylation of STAT5 by IGF-I. Because the levels of tyrosine phosphorylation of JAK2 were not apparently augmented by the addition of rhIGF-I to rhEPO, JAK2 may not participate in the process. Because STAT5 is known to associate with EPOR and JAK2 (14,49), it is possible that IGF-I might strengthen the affinity of these associations. Tyrosinephosphorylated STATs are known to be rapidly dephosphorylated by an as-yet-unidentified tyrosine phosphatase. It was previously reported that the N-terminal (amino acids 1-61) domain of STAT1 was required for the interaction with this tyrosine phosphatase and that the N-terminal deletion mutant of STAT1 was constitutively phosphorylated on tyrosine (50). Therefore, it is also possible that IGF-I might modulate the activity of this tyrosine phosphatase through the Ras pathways. Because this domain is highly conserved between the STAT proteins, it would be interesting to investigate whether or not rhEPO and rhEPO/rhIGF-I shows similar effects on the N-terminal deletion mutant of STAT5.
In F-36P cells, rhEPO alone was less effective in inducing tyrosine phosphorylation of STAT5 than rhIL-3 or rhGM-CSF. As in the case of F-36P cells, it was recently reported that EPO-induced STAT5 activation was absent or severely suppressed in bone marrow cells from patients with myelodysplastic syndrome, whereas IL-3 was able to normally activate STAT5 in these cells (51). Because myelodysplastic syndrome is characterized by ineffective erythropoiesis despite erythroid hyperplasia on which EPO shows little effect in most cases, IGF-I might be a candidate for a new therapeutic agent for myelodysplastic syndrome patients.
In summary, we showed here that IGF-I cooperated with EPO, at least in part, through the enhanced tyrosine phosphorylation of STAT5 and that Ras was involved in this process. These results demonstrate a new type of cross-talk between the receptor tyrosine kinases and Jak-STAT pathways. Although the precise mechanism and significance of this cross-talk remains unknown, further studies will undoubtedly lead to a better understanding of the regulation of the growth of hematopoietic cells.