Stimulation of Stat5 by Granulocyte Colony-Stimulating Factor (G-CSF) Is Modulated by Two Distinct Cytoplasmic Regions of the G-CSF Receptor

Granulocyte CSF (G-CSF)³ plays a critical role in regulating the proliferation and differentiation of myeloid progenitor cells and maintaining neutrophil levels in peripheral blood (1, 2). G-CSF exerts its biological activities via binding to a cell surface receptor that is a member of the cytokine receptor superfamily (1, 2). Incubation of hemopoietic cells with G-CSF leads to the activation of the Jak/STAT pathway and the Ras/Raf/MAP kinase pathway (2). Activation of Jak1 and Jak2 requires the membrane-proximal cytoplasmic region of the G-CSF receptor (3, 4, 5, 6, 7). Jaks permit tyrosine phosphorylation of the STAT transcription factors, which then translocate to the nucleus, bind enhancer elements, and stimulate the transcription of cellular genes (8). In contrast, tyrosine phosphorylation of Shc and subsequent initiation of Ras/Raf/MAP kinase signaling appear to require the carboxyl-terminal region of the G-CSF receptor (9, 10, 11, 12). Activated MAP kinases translocate to the nucleus and phosphorylate several transcription factors that induce the expression of immediate early genes that are distinct from those activated by STATs (13).

Several recent studies have identified distinct cytoplasmic regions of the G-CSF receptor that are involved in transducing signals for cell proliferation and differentiation (14, 15). The membrane-proximal region is required and sufficient for mitogenic signaling, whereas the distal carboxyl tail of the receptor mediates growth-suppressing signals and is involved in induction of terminal granulocytic maturation. Truncations of the carboxyl-terminal region of the G-CSF receptor as a result of point mutations have been reported in patients with severe congenital neutropenia and acute myeloid leukemia (16, 17, 18, 19). However, little is known about the mechanisms whereby the carboxyl-terminal region of the G-CSF receptor mediates granulocytic maturation and growth-suppressing signals. In other related receptors such as the erythropoietin receptor, proliferation is also down-regulated by the receptor carboxyl terminus, which appears to provide the binding site for the PTPase SHP-1 that inactivates Jak2 (20).

Although Stat5 was originally identified as a mammary gland factor that is regulated by prolactin (21), recent studies have shown that Stat5 is activated by other cytokines as well, including IL-2, IL-3, IL-5, granulocyte-macrophage CSF, growth hormone, thrombopoietin, and G-CSF (22, 23, 24, 25, 26). Activation of Stat5 by IL-3 appears to be associated with cell proliferation (27). In this study, we show that Stat5 activation by G-CSF is regulated by distinct cytoplasmic regions of the G-CSF receptor. We further demonstrate that Stat5 signaling pathway is critical for G-CSF-mediated cell proliferation and survival.

Murine BAF3 and 32D cells, stably transfected with cDNAs encoding either the wild-type or the truncated forms of the human G-CSF receptor, have been described (15, 28). Cells were grown in RPMI 1640 medium supplemented with 10% FCS, 50 μg/ml gentamicin, and 10% WEHI-3B cell-conditioned media. COS-7 cells were maintained in DMEM medium containing 10% FCS.

BAF3 or 32D cells were starved overnight in the absence of conditioned media before being incubated with 100 ng/ml G-CSF for the times indicated. Cells (2 × 10⁷) were collected by centrifugation, washed with PBS, and resuspended in ice-cold extraction buffer (20 mM HEPES, pH 7.5, 10 mM KCl, 300 mM NaCl, 0.5 mM DTT, 1% Triton X-100, 1 mM PMSF, and 1 mM vanadate). The suspension was gently vortexed for 10 s and allowed to incubate at 4°C for 20 min. The mixture was centrifuged at 12,000 × g for 20 min at 4°C, and the supernatant was collected.

The EMSA was performed as previously described using whole cell extracts (29, 30, 31). The β-casein GAS of the promoter of the bovine β-casein gene (5′-AGATTTCTAGGAATTCAAATC-3′) was end labeled using polynucleotide kinase and [γ-³²P]ATP, and used in all EMSAs.

Whole cell extracts were prepared as described above and incubated with rabbit anti-Jak2 antiserum (Upstate Biotechnology, Lake Placid, NY) or rabbit anti-Stat5a and Stat5b antisera for 2–4 h at 4°C. Anti-Stat5a and anti-Stat5b antisera were raised against synthetic peptides corresponding to amino acids 573–593 of Stat5a and amino acids 576–586 of Stat5b of the murine proteins (32). The immunoprecipitates were analyzed by 8% SDS-PAGE, followed by transfer to Immobilon-P. The membranes were then probed with biotin-labeled anti-phosphotyrosine Ab 4G10 (Upstate Biotechnology) or with specific Abs against Jak2, Stat5a, or Stat5b, and detected by using ECL.

Cells were starved in serum-free medium for 14 h and then stimulated with G-CSF for times as indicated. Whole cell extracts were prepared as above and incubated with anti-Erk2 Ab (TR10; kindly provided by Michael Weber, University of Virginia, Charlottesville). Immunocomplexes were washed three times in lysis buffer and once in kinase buffer (20 mM HEPES, pH 7.5, 10 mM MgCl₂, 2 mM EGTA). Kinase reaction was conducted in 20 μl kinase buffer containing 20 μM unlabeled ATP, 1.5 mg/ml myelin basic protein, and 10 μCi [γ-³²P]ATP (6000 μCi/mmol). After incubation at room temperature for 15 min, the reaction was terminated by adding 6 μl of 4× sample buffer. Samples were heated at 95°C for 5 min and separated by SDS-PAGE. The proteins were then transferred to Immobilon-P, followed by autoradiography.

Cos-7 cells were transfected using the DEAE-dextran method. The expression vectors encoding the wild-type (pLNCX-WT), D715 (pLNCX-DA) human G-CSF receptor, mouse Stat5a (pXM-Stat5a), and mouse Stat5b (pXM-Stat5b) have been described (15, 32). One microgram of pLNCX-WT or pLNCX-DA, 0.5 μg of pXM-Stat5a, and 0.5 μg of pXM-Stat5b were used in each transfection. Twenty hours after transfection, cells were deprived of serum for 4 h and stimulated with G-CSF for times as indicated.

For transfection of BAF3 cells, 10 μg of Stat5a cDNA or a carboxyl-terminally truncated Stat5a (amino acids 713) cDNA with FLAG epitope in Prk vector (33) (kindly provided by James Ihle, St. Jude Children’s Research Hospital, Memphis, TN) was coelectroporated into cells with 5 μg of an expression vector for green fluorescent protein (GFP). Fourteen hours after transfection, cells were selected for expression of GFP by FACS, and cells expressing GFP were used in subsequent experiments.

A total of 10⁵ cells was incubated in triplicate in 100 μl of RPMI medium supplemented with 10% FCS in 96-well plates in the presence or absence of G-CSF (50 ng/ml) for 20 h. Cells were then pulsed with 1 μCi [³H]thymidine for 4 h, and [³H]thymidine incorporation was measured by liquid scintillation counting.

BAF3 cells used in this study did not express detectable level of endogenous G-CSF receptor (15). To study G-CSF-mediated Stat5 signaling, BAF3 cells transfected with the human wild-type G-CSF receptor were incubated without (lane 1) or with G-CSF (lanes 2–10) before preparation of cellular extracts to examine whether Stat5 proteins were activated by G-CSF (Fig. 1,A). Extracts were incubated with a ³²P-labeled oligonucleotide probe corresponding to the GAS element present in the promoter of the β-casein gene. This element has been shown previously to bind tyrosine-phosphorylated Stat5 (34). A complex that binds to the β-casein GAS element was detected in extracts prepared from G-CSF-treated BAF3 cells (lane 2), which was displaced by the addition of a 50 molar excess of unlabeled oligonucleotides corresponding to the labeled probe, or to the IFN regulatory factor-1 (IRF-1) or the FcγR1 GAS element (lanes 3–5). However, the G-CSF-induced complex was not displaced by the addition of unlabeled oligonucleotide corresponding to the IFN stimulated response element (ISRE) or AP-1 enhancers (lanes 6 and 7). Addition of antisera that recognize Stat5 (lane 8) inhibited the formation of the G-CSF-stimulated complex, while antisera that recognize Stat1 or Stat3 had no effect (lanes 9 and 10). These data indicated that G-CSF treatment of cells resulted in activation of Stat5. Although Stat1 and Stat3 are tyrosine phosphorylated as a result of incubation of cells with G-CSF, the β-casein probe has a low affinity for binding of these activated STAT proteins (34). To directly demonstrate that Stat5 was tyrosine phosphorylated as a consequence of G-CSF treatment of BAF3 cells, extracts from untreated or treated cells were subjected to immunoprecipitation with antisera that recognized both Stat5a and Stat5b (Fig. 1 B). Immunoprecipitated protein was resolved by SDS-PAGE and transferred to Immobilon, and the resulting membrane was probed with anti-phosphotyrosine Ab. Incubation of cells with G-CSF for 5 min stimulated tyrosine phosphorylation of Stat5, and the proteins remained phosphorylated for 30 min.

It has been shown that the membrane-proximal cytoplasmic region of the G-CSF receptor is sufficient for Stat1 activation by G-CSF, whereas activation of Stat3 requires an additional region of 30 amino acids (5, 12). To examine which cytoplasmic regions of the G-CSF receptor were necessary for Stat5 activation, extracts were prepared from cells expressing the wild-type, D715, or D685 receptor (Fig. 2,A). EMSAs demonstrated that the G-CSF-stimulated complex was present in all three cell lines (Fig. 2,B). Immunoprecipitations of Stat5, followed by immunoblotting with anti-phosphotyrosine Abs, confirmed tyrosine phosphorylation of Stat5 induced by G-CSF in the three cell lines (Fig. 2,C). Notably, activation of Stat5 was consistently more robust in cells that expressed the D715 receptor (Fig. 2, B and C). The differences in the magnitude of Stat5 activation appeared not to be related to the expression levels of the different receptor forms. The wild-type and D715 forms of the receptor have been shown to be expressed at about the same levels, although the expression of the D685 receptor was approximately 10-fold greater (15).

Interestingly, when cells were washed and cultured in medium containing no cytokines following initial G-CSF stimulation for 10 min, the duration of Stat5 activation showed dramatic differences in the three cell lines (Fig. 2, B and C). While the G-CSF-activated Stat5 disappeared within 60 min in cells expressing the wild-type receptor in the absence of G-CSF (Fig. 2, B and C), cells expressing the D715 or D685 receptor displayed significantly prolonged activation of Stat5, which remained unchanged for at least 2 h and was approximately 18% of the original intensities at 3 h after removal of G-CSF, as determined by phosphor imager analysis. The attenuation of IL-3-mediated Stat5 activation was comparable in cells expressing the wild-type or D715 receptor (Fig. 2 B, lower panel). These results suggested that the carboxyl terminus of the G-CSF receptor is involved in down-regulating Stat5 activation induced by G-CSF.

The tyrosine kinases Jak1 and Jak2, which are signaling molecules upstream of Stat5, have been shown to be activated as a consequence of being tyrosine phosphorylated by G-CSF stimulation (3, 4). To determine whether tyrosine phosphorylation of these Jaks was altered in BAF3 cells expressing the truncated forms of the receptor, cells were incubated with G-CSF for 10 min before being cultured in medium containing no cytokines. G-CSF-induced tyrosine phosphorylation of Jak2 was of approximately equal intensity in the three cell lines (Fig. 3). The attenuation of Jak2 tyrosine phosphorylation following G-CSF withdrawal was also comparable. Thus, the kinetics of Jak2 activation did not appear to correlate with that of Stat5 activation in the three cell lines. No significant induction of Jak1 tyrosine phosphorylation was observed after G-CSF stimulation of the three cell lines (data not shown).

In addition to activation of Jak/STAT signaling pathway, G-CSF stimulation has been shown to activate the Ras/Raf/MAP kinase pathway (9, 10, 11, 12). To examine whether the carboxyl terminus of the G-CSF receptor may have a role in the regulation of the Ras/MAP kinase signaling pathway, cells were treated with G-CSF, as described above, and the activity of MAP kinase was determined in in vitro kinase assays. Consistent with previous reports (10, 12), G-CSF induced rapid activation of MAP kinase in cells expressing either the wild-type or the D715 receptor (Fig. 4). The extent of MAP kinase activation by G-CSF and the attenuation of MAP kinase activity upon G-CSF withdrawal both were comparable in the two cell lines. Only a very weak activation of MAP kinase was observed following G-CSF stimulation of BAF3 cells expressing the D685 receptor (data not shown).

BAF3 cell line was derived from murine Pro-B cells. To address whether the carboxyl terminus of the G-CSF receptor also down-regulates Stat5 activation in myeloid cells, we took advantage of murine myeloid 32D cells that did not express endogenous G-CSF receptor and were transfected with the different forms of the human G-CSF receptor (15). Similar to what were observed in BAF3 cells, the G-CSF-activated Stat5 disappeared rapidly and was undetectable 1 h after G-CSF withdrawal in 32D cells expressing the wild-type G-CSF receptor, whereas in 32D cells transfected with the D715 receptor Stat5 remained activated for at least 2 h (Fig. 5). Interestingly, the attenuation of G-CSF-activated Stat5 in cells expressing the D755 receptor was slower than in cells expressing the wild-type receptor, but was faster than in cells expressing the D715 receptor. It is also of note that the truncated forms of the G-CSF receptor consistently induced stronger activation of Stat5 than the wild-type receptor (Fig. 5). These results provide further evidence for the negative effect of the G-CSF receptor carboxyl terminus on Stat5 activation.

We next addressed the question of whether the carboxyl terminus of the G-CSF receptor is functional in nonhemopoietic cells in terms of down-regulating Stat5 activation. COS-7 cells were transiently transfected with the wild-type or the D715 G-CSF receptor, and Stat5 activation was examined by EMSA. In contrast to hemopoietic cells, G-CSF stimulation resulted in sustained Stat5 activation in COS-7 cells transfected with either the wild-type or the D715 receptor (Fig. 6). Significantly, the activated Stat5 in COS-7 cells persisted for at least 8 h without any detectable decay.

Recent studies have suggested that down-regulation of STAT signaling pathway following its activation by cytokine stimulation involves dephosphorylation of active STATs by PTPase and/or proteasome-mediated proteolysis (35, 36, 37, 38). To determine what might be the possible mechanism by which the carboxyl terminus of the G-CSF receptor exerts the negative effect on Stat5 activation in hemopoietic cells, we examined the effects of a potent proteasome inhibitor (MG132) and a tyrosine phosphatase inhibitor (vanadate) on Stat5 activation. BAF3 cells expressing the wild-type G-CSF receptor were pretreated with DMSO, MG132, or vanadate, followed by stimulation with G-CSF for 10 min. Cells were then incubated in the absence of cytokines, and the attenuation of activated Stat5 was monitored by EMSA. G-CSF-induced activation of Stat5 was markedly prolonged by addition of vanadate (Fig. 7). Although MG132 has been shown to stabilize the DNA-binding activities of Stat1 and Stat5 (36, 37, 38), no significant effect of MG132 on G-CSF-activated Stat5 was seen (Fig. 7). Cycloheximide (CHX), an inhibitor of protein synthesis, also had no effect on Stat5 activation. Interestingly, vanadate did not affect the decay of activated Stat5 in BAF3 cells expressing the D715 receptor (data not shown). These results suggest that a PTPase may play a critical role in the down-regulation of G-CSF-induced Stat5 activation.

To examine whether Stat5 signaling plays a role in G-CSF-dependent cell proliferation, BAF3 cells expressing the wild-type G-CSF receptor were transiently transfected with the carboxyl-terminally truncated Stat5a that has been shown to suppress Stat5-dependent gene expression (33). The expression level of the mutant Stat5a protein in BAF3 cells was at least 5 times higher than that of the endogenous Stat5a protein, as indicated by Western blot analysis (data not shown). G-CSF-dependent cell growth was significantly suppressed by the mutant Stat5a protein (Fig. 8). In contrast, overexpression of the wild-type Stat5a protein, which was expressed at approximately the same levels as that of the mutant protein in BAF3 cells (data not shown), had no detectable effect on cell growth.

We also investigated the effect of sustained Stat5 activation mediated by the D715 receptor on cell survival. BAF3 cells expressing the wild-type or the D715 receptor were cultured in G-CSF-containing medium for 8 h. They were then incubated in serum-free medium, and the cell viability was determined from 12 to 60 h later. As shown in Fig. 9,A, BAF3 cells expressing the wild-type receptor lost viability at a rate that was significantly faster than cells expressing the D715 receptor. Notably, the rates at which the two cell lines lost viability upon IL-3 withdrawal were comparable. Moreover, the prolonged survival of BAF3 cells expressing the D715 receptor was suppressed by expression of the dominant-negative Stat5a mutant (Fig. 9 B). 32D cells expressing the D715 receptor also showed significantly prolonged survival compared with cells expressing the wild-type receptor in the absence of G-CSF (data not shown).

Cytokines regulate many aspects of cell proliferation, differentiation, and survival. In the case of incubation of myeloid progenitor cells with G-CSF, a program of both cell proliferation and differentiation to granulocytic lineage is initiated. Transduction of signals for cell proliferation by the G-CSF receptor requires the cytoplasmic region proximal to the transmembrane domain, while the carboxyl-terminal region of the receptor is required for induction of cell differentiation (14, 15). The mechanisms by which the different domains of the G-CSF receptor regulate cell proliferation and differentiation are poorly understood. In this work, we demonstrate that activation of Stat5 by G-CSF stimulation requires only the membrane-proximal region of the receptor. We further show that a dominant-negative Stat5 mutant significantly inhibits G-CSF-induced cell proliferation. These data indicate that the Stat5 signaling pathway may have a role in the regulation of cell proliferation, consistent with recent studies showing that Stat5 is involved in IL-3- and granulocyte-macrophage CSF-mediated cell proliferation (27, 39). However, the Stat5 mutant does not completely eliminate G-CSF-dependent cell proliferation, suggesting additional signaling pathways that may contribute to cell proliferation. It is noteworthy that phosphatidylinositol 3′-kinase pathway has recently been shown to be activated by the membrane-proximal region of the G-CSF receptor (40).

In contrast to the membrane-proximal region of the G-CSF receptor, the distal carboxyl terminus of the receptor functions to down-regulate Stat5 signaling. Deletion of the carboxyl terminus results in increased extent and prolonged duration of Stat5 activation. Notably, the carboxyl terminus has also been shown to suppress cell growth (15, 28), further supporting the conclusion that Stat5 is a positive regulator of cell growth. In addition, our data demonstrate that sustained Stat5 activation mediated by the D715 form of the G-CSF receptor, which lacks the carboxyl terminus, is associated with the prolonged cell survival in the absence of G-CSF. Moreover, prolonged cell survival is markedly suppressed by overexpression of the dominant-negative mutant of Stat5a. Taken together, these results indicate that Stat5 signaling pathway plays an important role in the maintenance of cell survival.

The D715 and D685 forms of the G-CSF receptor, both of which mediate sustained Stat5 activation, are defective in inducing differentiation signals. It remains unknown whether the prolonged activation of Stat5 may contribute to the inability of the truncated G-CSF receptors to mediate differentiation signals (15). It is noteworthy that erythropoietin-induced erythroid differentiation of TF-1 cells is associated with an inability of erythropoietin to activate Stat5 (41).

In BAF3 cells expressing the wild-type G-CSF receptor, activation of Stat5 was prolonged significantly by incubating cells with the tyrosine phosphatase inhibitor vanadate, but not by the proteasome inhibitor MG132 (Fig. 7), suggesting that a PTPase may play a critical role in terminating G-CSF-stimulated Stat5 signaling. The expression of such a PTPase appears to be cell specific, because the full-length G-CSF receptor expressed in COS-7 induced prolonged activation of Stat5 similar to that seen when the truncated receptor was expressed in COS-7 cells (see Fig. 6). It is unclear which component in the Stat5 signaling pathway is being down-regulated as a consequence of expression of the carboxyl terminus of the receptor. Notably, the D685 receptor, which induces sustained Stat5 activation in BAF3 cells, contains no tyrosine in the cytoplasmic domain, indicating that G-CSF-mediated Stat5 signaling does not require receptor tyrosine phosphorylation. It is significant that G-CSF-induced activation of Jak2 and MAP kinase is comparable in BAF3 cells expressing either the wild-type or the truncated G-CSF receptor proteins (Figs. 3 and 4). Thus, it is less likely that the prolonged activation of Stat5 in cells expressing the truncated G-CSF receptor proteins resulted from delayed inactivation of the receptors or Jak2. It is possible that other protein tyrosine kinases involved in the regulation of Stat5 activation could be the target of the PTPase. Several such protein tyrosine kinases, including src and abl, have been shown under certain conditions to induce STAT binding (42, 43), and lyn and syk have been shown to be activated by G-CSF stimulation (44).

An alternative mechanism would be that the carboxyl terminus of the G-CSF receptor may indirectly regulate the activity of a potential nuclear phosphatase that could dephosphorylate Stat5. Such PTPase activities have been described with respect to inactivation of STATs after prolonged treatment of cells with IFNs (35). Interestingly, it has been shown that the Stat1 amino-terminal domain, which is highly conserved among all STAT family members, is required for tyrosine dephosphorylation of this protein (45).

Another likely PTPase that may have a role in the regulation of G-CSF signaling would be the SH2 domain-containing SHP-1 that has been implicated as a negative regulator of Jak/STAT signaling cascade by dephosphorylating Jak1 and Jak2 (20, 46, 47). However, the carboxyl terminus of the G-CSF receptor does not appear to regulate the Jak2 activity (Fig. 3). In addition, bone marrow cells from moth-eaten mice, which express no SHP-1 (48, 49), show no alterations in G-CSF-induced STAT activation as compared with cells from wild-type mice (data not shown). Using a variety of techniques, we have not been able to detect PTPase activity or SHP-1 protein associated with the receptor signaling complex. Taken together, these results indicate that the carboxyl terminus of the G-CSF receptor is not mediating its actions via SHP-1.

The region of 30 amino acids proximal to the carboxyl-terminal region of the G-CSF receptor dramatically up-regulates G-CSF-induced Stat5 activation (Fig. 2). The PTPase SHP-2 has been implicated as a positive regulator of both IFN-α and prolactin activation of the Jak/STAT signaling cascade (46, 50). Previous studies have shown that G-CSF treatment of cells results in tyrosine phosphorylation of SHP-2 and its association with GRB2 (10), which makes this phosphatase an attractive candidate to function as a positive regulator of G-CSF-stimulated Jak/STAT pathway. Experiments are being initiated to address this possibility. Although many of the detailed mechanisms of regulation of the Jak/STAT pathway by G-CSF remain to be elucidated, the experiments presented in this work indicate that the cytoplasmic domain of the G-CSF receptor contains distinct functional regions that are responsible for modulating the duration and the intensity that this cascade is being activated by G-CSF.