STAT3-mediated differentiation and survival of myeloid cells in response to granulocyte colony-stimulating factor: role for the cyclin-dependent kinase inhibitor p27Kip1

Article metrics


The signal transducer and activator of transcription (STAT) proteins have been implicated in cytokine-regulated proliferation, differentiation and cell survival. Granulocyte colony-stimulating factor (G-CSF), a regulator of granulocytic differentiation, induces a robust and sustained activation of STAT3. Here, we show that introduction of dominant negative (DN) forms of STAT3 interferes with G-CSF-induced differentiation and survival in murine 32D cells. G-CSF induces expression of the cyclin-dependent kinase (cdk) inhibitor p27Kip1 (but not p21Cip1), which is completely blocked by DN-STAT3. The ability of tyrosine-to-phenylalanine substitution mutants of the G-CSF receptor to activate STAT3 strongly correlated with their capacity to induce p27 expression and their ability to mediate differentiation and survival, suggesting a causal relationship between STAT3 activation, p27 expression and the observed cellular responses. We identified a putative STAT binding site in the promoter region of p27 that showed both STAT3 binding in electrophoretic mobility shift assays and functional activity in luciferase reporter assays. Finally, we studied G-CSF-induced responses in primary bone marrow and spleen cells of p27-deficient mice. Compared with wild-type, myeloid progenitors from p27-deficient mice showed significantly increased proliferation and reduced differentiation in response to G-CSF. These findings indicate that STAT3 controls myeloid differentiation, at least partly, via upregulation of p27Kip1.


Cytokine receptors lack intrinsic tyrosine kinase activity but activate cytoplasmic tyrosine kinases, in particular those of the Janus kinase (Jak) family (Demetri and Griffin, 1991; Isfort and Ihle, 1990). Jaks associate with the membrane-proximal cytoplasmic region of the receptors and become activated upon ligand binding (Dong et al., 1995; Quelle et al., 1994). Jak activation leads to phosphorylation of STAT (signal transducer and activator of transcription) proteins on a conserved tyrosine residue, just C-terminal to the Src homology 2 (SH2) domain (Shuai et al., 1994). Subsequently, STATs dimerize by phosphotyrosine/SH2 interactions, translocate to the nucleus and activate target genes by interaction with specific DNA sequences. Recently, a role for STATs in the control of cell growth and development has been demonstrated in a variety of cell systems. For instance, it was shown that STAT5 is involved in both granulocyte colony-stimulating factor (G-CSF)- and interleukin-3 (IL-3)-induced proliferation of Ba/F3 cells (Dong et al., 1998; Mui et al., 1996). In addition, activation of STAT3 is required for IL-6-induced growth arrest and macrophage differentiation of M1 cells (Nakajima et al., 1996), IL-10-induced inhibition of macrophage proliferation (O'Farrell et al., 1998), and G-CSF-induced neutrophilic differentiation (Shimozaki et al., 1997). However, the underlying mechanism(s) of STAT-mediated control of growth and differentiation are still largely unclear.

G-CSF stimulates proliferation, survival, and differentiation of myeloid progenitor cells towards neutrophilic granuylocytes (Demetri and Griffin, 1991; Nicola, 1987). G-CSF-deficient mice show chronic neutropenia and a reduced granulopoietic response to bacterial infections, indicating that G-CSF plays an essential role in the regulation of granulopoiesis in both steady-state and stress conditions (Lieschke et al., 1994). The biological effects of G-CSF are mediated through a cell-surface receptor (G-CSF-R) of the hematopoietin or class I cytokine receptor superfamily (Bazan, 1990; Fukunaga et al., 1990). G-CSF activates STAT1, STAT3, and STAT5 (Tian et al., 1994, 1996). Whereas the membrane-proximal cytoplasmic region of the G-CSF-R is sufficient for activation of STAT1 and STAT5, activation of STAT3 requires the membrane-distal C-terminal part of the receptor (de Koning et al., 1996a; Tian et al., 1996). The G-CSF-R C-terminus contains four conserved tyrosine residues (Y704, Y729, Y744, and Y764) and comprises a region that has specifically been implicated in the control of neutrophilic differentiation (Dong et al., 1993; Fukunaga et al., 1993). We and others have recently reported that Y704 and Y744 are involved in recruitment and activation of STAT3 (Chakraborty et al., 1999; de Koning et al., 1996a; Ward et al., 1999a). These tyrosines are also important for differentiation and survival signals from the G-CSF-R (Ward et al., 1999b).

Previously, it was shown that interleukin-6 and oncostatin M-induced growth inhibition of A375 melanoma cells is dependent on STAT3 activation and correlates with increased transcript levels of the cdk inhibitor p27Kip1 (Kortylewski et al., 1999). Here, we report that STAT3 is involved in G-CSF-mediated differentiation and survival and regulates the expression of p27Kip1. In addition, we show that the proliferation/differentiation balance of myeloid progenitor cells of p27-deficient mice in response to G-CSF is perturbed. Based on these data, we propose that STAT3-mediated expression of p27 represents one of the mechanisms by which G-CSF controls differentiation and survival of myeloid progenitor cells.


G-CSF induces sustained STAT3 activation in 32D cells undergoing proliferation followed by neutrophilic differentiation

IL-3-dependent 32D cells have an immature myeloblastic morphology. 32D transfectants expressing wild-type G-CSF-R respond to G-CSF by transient proliferation, followed by terminal neutrophilic differentiation after 7–10 days of culture (de Koning et al., 1998). To investigate STAT activation during the course of differentiation, cells were switched from IL-3- to G-CSF-containing medium and nuclear extracts were prepared daily. EMSAs with m67 oligonucleotides showed that STAT3- and/or STAT1-containing complexes were not activated in 32D cells proliferating in IL-3-containing medium. In contrast, G-CSF induced sustained and robust activation of STAT3 homodimers (Figure 1). In addition, some STAT1-STAT3 heterodimers, but no STAT1 homodimers, were formed. Interestingly, the mobility of the STAT3 homodimeric complexes slightly decreased after 3–4 days of G-CSF stimulation.

Figure 1

EMSA of nuclear extracts from 32D cells expressing the G-CSF-R. Cells were switched from IL-3- to G-CSF-containing medium after extensive washing to remove residual IL-3 and incubated at 37°C for the times indicated. Nuclear extracts were prepared and incubated with 32P-labeled double-stranded m67 oligonucleotides

Establishment of 32D transfectants stably expressing dominant-negative STAT3

To examine the role of STAT3 in G-CSF-mediated responses, we introduced wild-type (WT) hemagglutinin (HA)-tagged STAT3 (HA-STAT3WT) or two different dominant-negative (DN) STAT3 mutants (HA-STAT3F and HA-STAT3D) into the G-CSF-R expressing 32D cells. Expression levels of the STAT3 variants were determined in multiple independent transfectants by immunoprecipitation with anti-STAT3 antibodies, followed by Western blotting with anti-HA antibodies (Figure 2a). All experiments were performed and repeated on at least three independent clones of each mutant with approximately equivalent levels of HA-STAT3.

Figure 2

Dominant-negative STAT3 mutants in 32D transfectants. (a) expression of STAT3 variants in 32D/G-CSF-R cells. In STAT3F, the tyrosine residue at position 705 essential for STAT dimer formation was substituted for phenylalanine. In STAT3D, glutamic acids 434 and 435 important for DNA binding were replaced by alanines. STAT3 immunoprecipitates were subjected to Western blot analysis using anti-HA antibodies. Blots were reprobed with anti-STAT3 antibodies (against amino acids 750–769). Ig, immunoglobulins. (b) EMSA of nuclear extracts from 32D transfectants expressing STAT3 variants. Serum- and growth factor-deprived cells were incubated at 37°C with G-CSF (100 ng/ml) for the times indicated. Nuclear extracts were prepared and incubated with 32P-labeled double-stranded m67 or β-casein oligonucleotides

To determine whether the levels of DN-STAT3 were sufficient to inhibit G-CSF-induced STAT3 activation, STAT3F, STAT3D, and vector control cells were serum- and growth factor-deprived and subsequently incubated with G-CSF. EMSAs with m67 oligonucleotides showed that stimulation of both nontransfected and vector control cells with G-CSF resulted in activation of a large amount of STAT3 homodimers as well as some STAT1-STAT3 heterodimers and STAT1 homodimers, as described previously (de Koning et al., 1996a). In contrast, G-CSF-induced STAT3 activation was greatly diminished in STAT3F and STAT3D cells (Figure 2b), which continued during prolonged G-CSF treatment of these cells (data not shown). To determine the specificity of the DN-STAT3 mutants, G-CSF-induced STAT5 activation was investigated in STAT3F and STAT3D cells by performing EMSAs with β-casein oligonucleotides. In cells that were serum- and growth factor-deprived before stimulation, G-CSF induced a strong but rapidly declining activation of STAT5 in vector control cells that was not influenced by the presence of DN-STAT3 (Figure 2b).

Dominant-negative STAT3 mutants abrogate G-CSF-induced growth arrest and neutrophilic differentiation

After switching from IL-3- to G-CSF-containing medium, vector control cells proliferated in response to G-CSF for 5–7 days (Figure 3a). The cells then gradually stopped proliferating and developed into terminally differentiated neutrophils between days 8 and 11, showing the characteristic enlarged cytoplasm-to-nucleus ratio, neutrophilic cytoplasm, segmented nuclei, and granules (Figure 3b). In contrast, cells overexpressing DN-STAT3 proliferated continuously in response to G-CSF. Both STAT3D and STAT3F expressing clones maintained an immature myeloblastic morphology and could be cultured in G-CSF-containing medium for at least 4 weeks, in agreement with studies in L-GM-1 cells (Shimozaki et al., 1997). 32D transfectants overexpressing STAT3WT displayed growth and differentiation characteristics that were similar to vector control cells (Figure 3). This excludes that the effects of DN-STAT3 were merely caused by overexpression of STAT3 proteins, for instance by competing with other SH2-containing signaling proteins for docking to the G-CSF-R.

Figure 3

Effect of STAT3 variants on G-CSF responses of 32D cells. (a) proliferation of 32D transfectants expressing STAT3 variants in response to G-CSF. The numbers of viable cells were determined on the basis of trypan blue exclusion at the indicated times. In the presence of IL-3, all clones proliferated with similar kinetics, whereas upon growth factor deprivation, cells died within 1 day without showing signs of neutrophilic differentiation. (b) Neutrophilic differentiation of 32D transfectants expressing STAT3 variants. Morphology of cells cultured for 9 days in the presence of G-CSF (May-Grünwald-Giemsa staining; original magnification: ×630)

We have previously established conditions in which addition of the cell cycle inhibitors cytosine arabinoside (Ara-C) or hydroxyurea results in accumulation of 32D cells in the G1 phase of the cell cycle, while leaving their ability to differentiate in response to G-CSF unaffected (de Koning et al., 1998). G1-arrested 32D cells cultured with IL-3 remain myeloblastic, indicating that enforced cell cycle arrest per se does not induce differentiation of 32D cells in the absence of G-CSF (de Koning et al., 1998). We applied these conditions to determine whether DN-STAT3 also prevents neutrophilic differentiation in G1-arrested 32D cells. In the presence of Ara-C, G-CSF induced terminal neutrophilic differentiation in STAT3WT and vector control cells as well as in STAT3D and STAT3F cells with comparable efficiencies (data not shown). These results suggest that activation of STAT3 might be involved in controlling G-CSF-induced growth arrest that is an essential component of the differentiation program.

Dominant-negative STAT3 impairs survival at low ligand concentration

We have shown that 32D cells expressing dominant-negative STAT3 proliferate robustly at 100 ng/ml G-CSF, without undergoing differentiation. In contrast, however, survival of these clones at G-CSF concentrations of 0.1 ng/ml or lower was impaired relative to 32D cells expressing STAT3WT (Figure 4). Annexin V staining indicated that this is associated with increased apoptosis, which becomes prominent on day 3–5 of culture. A representative example of this analysis is shown in Figure 5.

Figure 4

Effect of dominant-negative STAT3 on G-CSF-mediated survival signaling. Cell survival analysis of 32D clones expressing either STAT3 WT (□) or STAT3 D () performed as described in Figure 3a, except at the concentrations of G-CSF indicated. Essentially, the numbers of viable cells were determined on the basis of trypan blue exclusion at the indicated times

Figure 5

Accumulation of apoptotic cells determined by Annexin V staining. 32D/WT cells expressing either STAT3D (left panels) or STAT3WT (right panels) were cultured for three days at G-CSF concentrations indicated. L: live cells; A: apoptotic cells. Percentages of total gated cells are indicated for each window

STAT3 is essential for G-CSF-induced expression of p27Kip1

To further investigate this growth arrest, we subsequently studied whether STAT3 is involved in control of the cell cycle via the cdk inhibitors p27Kip1 and p21Cip1. Northern blot analysis showed that in vector control cells p27 mRNA levels increased after 2–3 days of stimulation with G-CSF (Figure 6a). In contrast, p27 mRNA was not induced by G-CSF in 32D cells overexpressing DN-STAT3 (data not shown). In vector control cells, p27 protein also appeared upon 3 days of stimulation with G-CSF, and reached a maximum after 5–7 days (Figure 6b). This timing of p27 protein expression coincides with the observed growth arrest in these cells (Figure 3a). As expected, G-CSF did not induce p27 protein in 32D cells overexpressing DN-STAT3 (Figure 6b). Reprobing of the blots with anti-p21 antibodies revealed no alteration in the (low) levels of expression of p21 protein (data not shown), suggesting that p21 is not involved in G-CSF-induced neutrophilic differentiation.

Figure 6

G-CSF-induced p27 expression in 32D transfectants. Cells were switched from IL-3- to G-CSF-containing medium after extensive washing to remove residual IL-3 and incubated at 37°C for the times indicated. (a) Total RNA (10 μg) of 32D/G-CSF-R cells was analysed by Northern blot hybridisation using a 32P-labeled p27 probe. The blot was reprobed with GAPDH to confirm equal loading. The graph shows quantitative analyses of p27 mRNA induction, expressed as fold induction by G-CSF compared to IL-3. The means±s.e. of three independent clones are shown. (b) Lysates were prepared and analysed by Western blotting with anti-p27 antibodies. Ponceau staining of the blot indicated equal loading of the samples (not shown)

STAT3 activation by G-CSF-R mutants strongly correlates with induction of p27 expression

To determine the role of the four conserved cytoplasmic tyrosine residues of the G-CSF-R, we recently constructed several tyrosine-to-phenylalanine (Y-to-F) substitution mutants, including a quadruple Y-to-F or ‘null’ mutant, with no cytoplasmic tyrosines (mO), and a series of triple Y-to-F or ‘add-back’ mutants, which each retain a single tyrosine (mA, mB, mC, and mD). Expression in 32D cells revealed that signals from the G-CSF-R for differentiation and survival are mediated most strongly by Y704 (mA) and Y744 (mC) (Ward et al., 1999b). Strikingly, the differentiation and survival-inducing capacity of these mutants correlated strongly with their ability to activate STAT3 upon G-CSF stimulation, as measured by both tyrosine and serine phosphorylation, consistent with the observation that overexpression of DN-STAT3 totally blocked neutrophilic differentiation and impaired survival of 32D cells (Figures 3,4,5). Since the Y-to-F mutants of the G-CSF-R show varying levels of STAT3 activation, they provide a useful independent setting to investigate the involvement of STAT3 activation in induction of p27 expression. Therefore, we analysed both STAT3 activation status and p27 protein levels after 3 days G-CSF treatment for each mutant. Mutant mO, which hardly activated STAT3, also failed to induce p27 expression (Figure 7). Importantly, mO supported neither differentiation nor proliferation (Ward et al., 1999b), showing that the inability to induce proliferation does not automatically result in p27 expression. In clones expressing the ‘add-back’ mutants, G-CSF-induced expression of p27 protein was greatest with mA (Y704) and mC (Y744). Mutant mB (Y729) and mD (Y764) showed little STAT3 activation and no significant p27 induction. Thus, the ability of G-CSF-R mutants to induce p27 expression showed a striking correlation with their capacity to activate STAT3 (Figure 7), suggesting a positive role for STAT3 in the control of p27 expression.

Figure 7

STAT3 activation and p27 induction by G-CSF-R tyrosine mutants. 32D transfectants expressing wild-type or mutant G-CSF-Rs were switched from IL-3- to G-CSF-containing medium and incubated at 37°C for 3 days. Lysates were prepared and analysed by Western blotting with anti-STAT3pY, anti-p27 and anti-STAT3 (a.a. 750–769) antibodies

The p27Kip1 promoter contains a functional STAT3 binding site

Subsequently, we searched for STAT responsive elements in the p27 promoter sequence. We identified a potential STAT-binding site in the p27 promoter at position −1585 (Kwon et al., 1996). EMSAs were performed to verify whether G-CSF-activated STAT proteins indeed bind to this site. G-CSF induced the formation of three distinct nuclear complexes that bound to p27 promoter oligonucleotides. Supershift analysis showed that the slowest migrating complex consists of STAT3 homodimers, the complex with intermediate mobility of STAT1-STAT3 heterodimers, and the fastest migrating complex of STAT1 homodimers (Figure 8a). Overexpression of DN-STAT3 reduced the levels of STAT3 binding to p27 oligonucleotides, further suggesting that STAT3 is involved in the regulation of p27 expression.

Figure 8

The p27 promoter contains a functional STAT3 binding site. (a) Identification of STAT proteins that bind to a specific p27 promoter sequence by EMSA. Serum- and growth factor-deprived 32D transfectants were incubated at 37°C with G-CSF (100 ng/ml) for the times indicated. Nuclear extracts were prepared and incubated with 32P-labeled double-stranded p27 oligonucleotides. For supershift analysis, nuclear extracts from vector control cells stimulated for 15 min with G-CSF (100 ng/ml) were preincubated without (−) or with the indicated antibodies before the addition of 32P-labeled p27 oligonucleotides. (b) Effect of STAT3 variants on transactivation of p27 promoter in HeLa cells. HeLa cells were transfected with a p27-promoter luciferase reporter construct, a human G-CSF-R expression plasmid, and different amounts of constructs encoding STAT3WT or STAT3F as indicated. After transfection, cells were cultured for 48 h without factor, with cholera toxin (200 ng/ml; CT), and/or with G-CSF (100 ng/ml). Cell lysates were prepared and assayed for luciferase activity. Extracts from untransfected cells were used to determine basal levels. Data represent the mean of four independent experiments. Incubation with forskolin or dibutyryl-cAMP resulted in responses similar to those obtained with cholera toxin. (c) Activity of wild-type and mutant p27 promoter-derived STAT3 binding sequence in luciferase assay. Hela cells were transfected with reported constructs and 250 ng of STAT3WT and cultured as outlined under (b). Data are the mean and s.d. from three experiments

To investigate whether STAT3 can induce p27 promoter activity, we performed reporter assays in cells transiently transfected with a p27-promoter luciferase construct, a G-CSF-R expression plasmid, and different ratios of WT- and DN-STAT3. In the presence of WT-STAT3, G-CSF induced a sevenfold increase in p27 promoter activity (Figure 8b). The p27 promoter also contains a cAMP responsive element (CRE) at position −286 (Kwon et al., 1996). Because cAMP is known to induce p27 expression (Ward et al., 1996), we also treated cells with cholera toxin to elevate intracellular cAMP levels, as a control for the specific action of DN-STAT3. The p27 promoter activity was augmented approximately 2.4-fold by cholera toxin and 18-fold by G-CSF plus cholera toxin. Importantly, while increasing amounts of DN-STAT3 progressively inhibited transactivation induced by G-CSF, CRE-driven luciferase activity was not affected (Figure 8b). Luciferase experiments with constructs containing the putative p27 promoter-derived STAT3 binding site confirmed its functional involvement in G-CSF-induced transcription. As expected, this activity is lost upon mutation of nucleotides known to be critical for STAT3-binding (Figure 8c).

Myeloid progenitors of p27−/− mice show enhanced G-CSF-induced proliferation

Finally, to determine the involvement of p27 in G-CSF-induced myeloid cell development, we performed in vitro colony assays with bone marrow and spleen mononuclear cells of p27−/− and p27+/+ mice in the presence of G-CSF, GM-CSF, or IL-3. The generation of p27-deficient mice has been previously described (Fero et al., 1996). In agreement with the results of Fero et al. (1996), we observed elevated numbers of GM-CSF-responsive progenitor cells in p27−/− mice compared with control littermates (femur: 1.5-fold; spleen: twofold). Equivalent results were obtained for IL-3-responsive colony forming cells (femur: 1.4-fold; spleen: twofold). Interestingly, the difference in G-CSF-responsive progenitors between p27 knockout and wild-type mice was even greater (femur: twofold; spleen: 2.9-fold) (Figure 9). This selective increase in colony numbers might result from a more pronounced expansion of the G-CSF-responsive progenitor compartment in p27−/− mice, from an enhanced proliferative capacity of the G-CSF-responsive progenitor cells, or from both.

Figure 9

In vitro colony assays with bone marrow mononuclear cells of p27−/− and p27+/+ mice. Cells were plated in methylcellulose-containing media supplemented with G-CSF, GM-CSF or IL-3. Hematopoietic colonies containing 30 cells or more were scored after 7 days. Data represent the means±s.e. of three male mice of each genotype

If p27 is involved in G-CSF-induced growth arrest and differentiation of the progenitor cells, granulocyte colonies from p27−/− mice predictively contain less differentiated cells and are larger in size. Indeed, on day 7 of culture, colonies derived from p27−/− bone marrow or spleen cells contained a lower percentage of terminally differentiated neutrophils than from p27+/+ mice (Table 1). On days 7 and 14, the average numbers of cells per G-CSF-induced colony were approximately twice as high in p27−/− colonies than in p27+/+ colonies (Table 1). In contrast, the size and composition of colonies induced by GM-CSF or IL-3 were similar in p27−/− and p27+/+ mice. These results indicate that, although p27−/− mice contain more progenitors of several lineages, the lack of p27 specifically affects the G-CSF-mediated growth arrest and neutrophilic differentiation.

Table 1 Cellular composition and size of G-CSF-, GM-CSF- and IL-3-induced colonies from wild type and p27-deficient bone marrow cells


Previous studies have implicated STAT1 in interferon-induced G1 arrest via p21Cip1 (Chin et al., 1996) and STAT5 in thrombopoietin-induced megakaryocytic growth arrest and differentiation (Matsumura et al., 1997). Potential STAT responsive elements in the p21 promoter were recognized, which showed binding of STAT1 and STAT5 in band shift assays, suggesting that p21 is a candidate target for these STATs (Chin et al., 1996; Matsumura et al., 1997). In contrast, STAT3 did not bind to these elements in the p21 promoter. In this study, we showed that STAT3, but not STAT1 and STAT5, is activated robustly over a period of days by G-CSF in 32D cells undergoing neutrophilic differentiation. In addition, levels of p27, but not p21, were found to increase during this process. We further demonstrated that STAT3 regulates p27 transcription and is instrumental to both differentiation and survival in response to G-CSF. Together, these data suggest that G-CSF induces a novel program of cell cycle control in the process of neutrophil development.

Thus far, the control of p27 expression had been mainly attributed to translational mechanisms (Alessandrini et al., 1997; Polyak et al., 1994; Toyoshima and Hunter, 1994). However, our data strongly suggest that STAT3 is involved in the transcriptional control of p27, although it is not yet clear to what extent this mechanism contributes to the regulation of p27. For instance, it remains to be resolved why STAT3-DNA binding is already maximal 1 day after G-CSF induction, while the increase in p27 mRNA levels is seen only after day 2–3. There are several possible explanations for this time lag. For instance, conformational changes of the STAT3 complex or the involvement of other transcription modulatory proteins may be required for transcriptional activity. Suggestive of such an alteration, it is noteworthy that the electrophoretic mobility of STAT3 complexes consistently decreased after 3 days of culture (Figure 1). Recently, it was shown that the N-Myc interactor (Nmi) protein associates with different STAT proteins, including STAT3, and augments STAT-induced transcription by enhancing recruitment of the coactivator proteins CBP/p300 (Zhu et al., 1999). The fact that Nmi expression is induced by multiple cytokines provides a novel mechanism for specific enhancement of STAT transcriptional activity by growth factors, including G-CSF, which may be relevant in this context. This possibility is currently under investigation. Significantly, the appearance of these slower migrating complexes correlated more tightly with the kinetics of p27 induction than the initially formed STAT3 complexes.

An alternative explanation for the lag between STAT3 activation and p27 transcription is that full activation of the p27 promoter requires synergy between STAT3 and other transcriptional regulator(s) acting on a different site in the promoter. As shown in Figure 8b, the p27 promoter also contains a functional cAMP responsive element (CRE). However, neither the addition of dibutyryl-cAMP nor cAMP inducing agents (choleratoxin, forskolin) significantly altered the kinetics of G-CSF-induced p27 expression, which argues against a major contribution of the CRE in this setting. In agreement with this, elevation of intracellular cAMP levels did not influence the proliferation and differentiation characteristics of 32D cells in response to G-CSF (JP de Koning and IP Touw, unpublished results). Other transcription factors known to bind to the p27 promoter are Sp1, NF-κB, and Myb (Kwon et al., 1996). Furthermore, it was shown that Sp1 and STAT1 synergize in interferon-γ-induced responses (Look et al., 1995), while synergy between Sp1 and STAT3 was demonstrated in interleukin 6-induced transcription of the CCAAT/enhancer binding protein δ (Cantwell et al., 1998). These factors therefore represent additional candidates for facilitating the observed transcriptional effects of STAT3.

There is also a lag between p27 mRNA induction and the appearance of p27 protein. A mechanism that counteracts the accumulation of p27 protein in proliferating cells is ubiquitin-mediated degradation (Pagano et al., 1995). Recent data indicate that mitogen-stimulated Ras activity not only results in induction of cell cycle progression via cyclin D/cdk complexes, but also controls ubiquitination and degradation of p27 (Weber et al., 1997). Because G-CSF activates Ras in 32D cells during the proliferative phase (de Koning et al., 1998; Okuda et al., 1994), this may explain why p27 protein levels remain low during the first 3 days of culture, despite active transcription of the p27 gene.

What might be the function of p27 in the control of myeloid development? It is likely that p27 contributes to the G1 arrest, which is an essential step in the myeloid differentiation program (de Koning et al., 1998; Sherr and Roberts, 1995). The fact that cell cycle inhibitors could override the inhibitory effects of DN-STAT3 mutants on G-CSF-induced differentiation argues in that direction. However, it has recently been suggested that p27 might exert additional functions, such as prevention of apoptosis in mesangial cells and fibroblasts (Hiromura et al., 1999). The data presented here and elsewhere (Fukada et al., 1996; Ward et al., 1996b) linking activation of STAT3 to both differentiation and cell survival suggest that p27 has a similar survival promoting activity in differentiating myeloid cells.

Materials and methods

Cell culture, constructs, and transfectants

32D.C8.6, a subline of the IL-3-dependent murine myeloid cell line 32D (Greenberger et al., 1983), was maintained in RPMI 1640 medium supplemented with 10% fetal calf serum (FCS) and 10 ng/ml of murine IL-3. The pBabe expression vector (Morgenstern and Land, 1990) encoding wild-type human G-CSF-R cDNA was introduced into 32D.C8.6 cells by electroporation. Subsequently, cells were selected with puromycin (Sigma, Zwijndrecht, The Netherlands) at a concentration of 1 μg/ml. The G-CSF-R expression levels in multiple clones were determined by flow cytometry as previously described (de Koning et al., 1998). Various STAT3 cDNAs cloned into the pCAGGS-Neo expression vector were transfected into a 32D.C8.6 clone overexpressing wild-type G-CSF-R. The pCAGGS-Neo constructs encoding wild-type (WT) hemagglutinin peptide (HA)-tagged murine STAT3 cDNA or the dominant-negative mutants HA-STAT3F and HA-STAT3D were used (kindly provided by Drs K Nakajima and T Hirano). The tyrosine residue at position 705 was substituted for phenylalanine in HA-STAT3F, whereas glutamic acids 434 and 435 were replaced with alanines in HA-STAT3D (Nakajima et al., 1996). After transfection, cells were selected with G418 (GIBCO-BRL, Breda, The Netherlands) at a concentration of 0.8 mg/ml. Several independent clones were expanded for further analysis. Establishment of 32D.8.6 transfectants expressing different tyrosine-to-phenylalanine (Y-to-F) substitution mutants of the G-CSF-R has been described previously (Ward et al., 1999b). Clones expressing a quadruple Y-to-F or ‘null’ mutant, with no cytoplasmic tyrosines (mO), or a series of triple Y-to-F or ‘add-back’ mutants, which each retain a single cytoplasmic tyrosine (mA, mB, mC, mD) were used.

Electrophoretic mobility shift assay (EMSA)

Nuclear extracts were prepared as previously described (de Koning et al., 1996a). Extracts of 2×106 cells were incubated for 20 min at room temperature with 0.2 ng of 32P-labeled double-stranded oligonucleotide (5–10 000 c.p.m.) and 2 μg of poly(dI-dC) in 20 μl of binding buffer (13 mM HEPES [pH 7.8], 80 mM NaCl, 3 mM NaF, 3 mM NaMoO4, 1 mM dithiothreitol, 0.15 mM EDTA, 0.15 mM EGTA, 8% glycerol). The following oligonucleotides were used: m67 (5′-CATTTCCCGTAAATC-3′), a high-affinity mutant of the sis-inducible element (SIE) of the human c-fos gene (Wagner et al., 1990); β-casein (5′-AGATTTCTAGGAATTCAATCC-3′), the STAT5-binding site of the bovine β-casein promoter (Wakao et al., 1994); and p27 (5′-AATTTCCTGTAACAT-3′), a potential STAT-binding site in the p27Kip1 promoter located at position −1585 (Kwon et al., 1996). The DNA-protein complexes were separated by electrophoresis on 5% polyacrylamide gels containing 5% glycerol in 0.25×Tris-borate/EDTA electrophoresis buffer (TBE). The gels were dried and subsequently analysed by autoradiography. For supershift analysis, nuclear extracts were preincubated for 2 h with 2 μg of anti-STAT3 (raised against amino acids 750–769; Santa Cruz Biotechnology Inc, Santa Cruz, CA, USA), anti-STAT1α (raised against amino acids 716–739; Santa Cruz), or anti-STAT1αβ antibodies (raised against amino acids 688–710 mapping within the C-terminal sequence common to STAT1α and STAT1β; Santa Cruz), before the addition of 32P-labeled oligonucleotide.

Immunoprecipitation and Western blotting

Preparation of cell lysates, immunoprecipitation, and Western blotting were performed as described (de Koning et al., 1996b). Anti-STAT3 (raised against amino acids 750–769; Santa Cruz), anti-STAT3 (raised against amino acids 1–175; Transduction Laboratories Inc, Lexington, KY, USA), anti-phospho-STAT3[Y705] (9130; New England Biolabs Inc, Beverly, MA, USA), anti-HA (Boehringer, Mannheim, Germany), and anti-p27 antibodies (Transduction Laboratories) were used.

Cell proliferation, morphological analysis, and apoptosis assay

To determine proliferation, cells were incubated at an initial density of 3×105 cells/ml in 10% FCS/RPMI medium supplemented with 100 ng/ml of human G-CSF, 10 ng/ml of murine IL-3, or without growth factors. The medium was replenished every 2–4 days, and the cell densities were adjusted to 3×105 cells/ml. Viable cells were counted on the basis of trypan blue exclusion. To analyse the morphological features, cells were spun onto glass slides and examined after May-Grünwald-Giemsa staining. Apoptotic cells were analysed by flow cytometry (FACS Calibur, Becton-Dickinson, Sunnyvale, CA, USA) after combined Annexin V and propidium iodide staining according to established protocols. In brief, cells were incubated with Annexin V-biotin (Boehringer Mannheim, Germany) in incubation buffer (10 mM HEPES/NaOH, pH 7.4, 140 mM NaCl, 5 mM CaCl2) for 15 min at room temperature, washed and subsequently stained with streptavidin allophycocyanin (APC) conjugate (Caltag Laboratories, Burlingham, CA, USA) for 15 min at 4°C. Following washing, cells were resuspended in incubation buffer supplemented with 1 μg/ml propidium iodide and analysed by FACS.

Northern blotting

RNA was extracted from cells using the Ultraspec-II RNA isolation system (Biotecx Laboratories Inc, Houston, TX, USA). Agarose-formaldehyde gel electrophoresis and transfer to filters (Hybond; Amersham Life Sciences, Amersham, UK) was performed using standard procedures. As probes, murine p27Kip1 (888-bp NotI fragment) and murine GAPDH (777-bp HindIII-EcoRI fragment) were 32P-labeled by random priming (Boehringer, Mannheim, Germany). For quantification, filters were exposed to phosphorimager screens and analysed with ImageQuant software (Molecular Dynamics).

Luciferase assay

HeLa cells were maintained in Dulbecco's modified Eagle's medium (DMEM; GIBCO) supplemented with 10% FCS. For transient transfection, cells were plated into 12-well plates at a density of 9×104 cells/well and cultured for 24 h. Subsequently, cells were transfected by the DEAE-dextran method with 250 ng of p27Kip1-pGL-2, a reporter construct containing the p27 promoter (−1609 to +178) upstream of the luciferase gene (Kwon et al., 1996), 250 ng of pLNCX-G-CSF-R, an expression vector containing wild-type G-CSF-R cDNA (Dong et al., 1993), and 250 ng of pCAGGS constructs encoding various STAT3 cDNAs. Although the STAT3 variants were tested in several ratios, the total amount of pCAGGS plasmid was kept constant at 250 ng in each transfection. In a separate experiment, the putative STAT3-binding site within the p27 promoter (5′-TTAATTTCCTGTAACATC-3′) and a mutant form thereof (5′-TTAATTGTCTGCGACATC; altered nucleotides underscored) were introduced in triplicate into the pGL3 minimal promoter driven luciferase reporter plasmid (Promega, Leiden, The Netherlands). Cells were incubated with DNA/DEAE-dextran precipitates for 30 min, washed and cultured in 10% FCS/DMEM medium supplemented with 100 ng/ml of human G-CSF, 200 ng/ml of cholera toxin, 500 nM forskolin, 500 μM dibutyryl-cAMP, or without factors. After 48 h, cells were lysed in luciferase lysis buffer (25 mM Trisphosphate [pH 7.8], 8 mM MgCl2, 1 mM dithiothreitol, 1% Triton X-100, 15% glycerol) and assayed for luciferase activity on a Biocounter M2500 luminometer (Lumac, Landgraaf, The Netherlands) using an equal volume of luciferin solution (1 mM luciferin, 1 mM ATP, 8 mM MgCl2) as a substrate.

In vitro colony assays

p27-deficient mice (Fero et al., 1996) were kindly provided by Dr J Roberts. Femurs, tibias, and spleens of 6–12 month old male p27−/− and wild-type (p27+/+) mice were removed aseptically. To obtain bone marrow cell suspensions, femurs and tibias were crushed in a mortar in HBSS/10% FCS. Spleen and bone marrow cells were passed through a 70 μm sieve, spun down, and resuspended, resulting in mononuclear suspensions containing 98 to 99% viable cells as determined by trypan blue exclusion. Subsequently, cells were incubated in a cell culture flask in a humidified atmosphere containing 5% CO2 at 37°C for 1 h. Nonadherent cells were harvested, and 3×104 bone marrow or 3×105 spleen mononuclear cells were plated in triplicate in 35 mm Petri dishes in 1 ml of methylcellulose medium containing 30% fetal bovine serum, 1% bovine serum albumin, 0.1 mM β-mercaptoethanol, 2 mM L-glutamine, without additional growth factors, or with increasing concentrations of G-CSF, granulocyte-macrophage colony-stimulating factor (GM-CSF), or IL-3. Colonies containing at least 30 cells were counted on day 7 of culture. To analyse the morphological features and to determine the average cell number per colony, cultures were harvested and washed extensively to remove methylcellulose. Cells were counted, spun onto glass slides, and examined after May-Grünwald-Giemsa staining.


  1. Alessandrini A, Chiaur DS and Pagano M. . 1997 Leukemia 11: 342–345.

  2. Bazan JF. . 1990 Proc. Natl. Acad. Sci. USA 87: 6934–6938.

  3. Cantwell CA, Sterneck E and Johsnon PF. . 1998 Mol. Cell. Biol. 18: 2108–2117.

  4. Chakraborty A, Dyer KF, Cascio M, Mietzner TA and Tweardy DJ. . 1999 Blood 93: 15–24.

  5. Chin YE, Kitagawa M, Su WC, You ZH, Iwamoto Y and Fu XY. . 1996 Science 272: 719–722.

  6. de Koning JP, Dong F, Smith L, Schelen AM, Barge RM, van der Plas DC, Hoefsloot LH, Löwenberg B and Touw IP. . 1996a Blood 87: 1335–1342.

  7. de Koning JP, Schelen AM, Dong F, van Buitenen C, Burgering BM, Bos JL, Löwenberg B and Touw IP. . 1996b Blood 87: 132–140.

  8. de Koning JP, Soede-Bobok AA, Schelen AM, Smith L, van Leeuwen D, Santini V, Burgering BM, Bos JL, Löwenberg B and Touw IP. . 1998 Blood 9: 1924–1933.

  9. Demetri GD and Griffin JD. . 1991 Blood 78: 2791–2808.

  10. Dong F, Liu X, de Koning JP, Touw IP, Henninghausen L, Larner A and Grimley PM. . 1998 J. Immunol. 161: 6503–6509.

  11. Dong F, van Buitenen C, Pouwels K, Hoefsloot LH, Löwenberg B and Touw IP. . 1993 Mol. Cell. Biol. 13: 7774–7781.

  12. Dong F, van Paassen M, van Buitenen C, Hoefsloot LH, Löwenberg B and Touw IP. . 1995 Blood 85: 902–911.

  13. Fero ML, Rivkin M, Tasch M, Porter P, Carow CE, Firpo E, Polyak K, Tsai LH, Broudy V, Perlmutter RM, Kaushansky K and Roberts JM. . 1996 Cell 85: 733–744.

  14. Fukada T, Hibi M, Yamanaka Y, Takahashi-Tezuka M, Fujitani Y, Yamaguchi T, Nakajima K and Hirano T. . 1996 Immunity 5: 449–460.

  15. Fukunaga R, Ishizaka-Ikeda E and Nagata S. . 1993 Cell 74: 1079–1087.

  16. Fukunaga R, Seto Y, Mizushima S and Nagata S. . 1990 Proc. Natl. Acad. Sci. USA 87: 8702–8706.

  17. Greenberger JS, Sakakeeny MA, Humphries RK, Eaves CJ and Eckner RJ. . 1983 Proc. Natl. Acad. Sci. USA 80: 2931–2935.

  18. Hiromura K, Pippin JW, Fero ML, Roberts JM and Shankland SJ. . 1999 J. Clin. Invest. 103: 597–604.

  19. Isfort RJ and Ihle JN. . 1990 Growth Factors 2: 213–220.

  20. Kortylewski M, Heinrich PC, Mackiewicz A, Schniertshauer U, Klingmuller U, Nakajima K, Hirano T, Horn F and Behrmann I. . 1999 Oncogene 18: 3742–3753.

  21. Kwon TK, Nagel JE, Buchholz MA and Nordin AA. . 1996 Gene 180: 113–120.

  22. Lieschke GJ, Grail D, Hodgson G, Metcalf D, Stanley E, Cheers C, Fowler KJ, Basu S, Zhan YF and Dunn AR. . 1994 Blood 84: 1737–1746.

  23. Look DC, Pelletier MR, Tidwell RM, Roswit WT and Holtzman MJ. . 1995 J. Biol. Chem. 270: 30264–30267.

  24. Matsumura I, Ishikawa J, Nakajima K, Oritani K, Tomiyama Y, Miyagawa J, Kato T, Miyazaki H, Matsuzawa Y and Kanakura Y. . 1997 Mol. Cell. Biol. 17: 2933–2943.

  25. Morgenstern JP and Land H. . 1990 Nucleic Acids Res. 18: 3587–3596.

  26. Mui AL, Wakao H, Kinoshita T, Kitamura T and Miyajima A. . 1996 EMBO J. 15: 2425–2433.

  27. Nakajima K, Yamanaka Y, Nakae K, Kojima H, Ichiba M, Kiuchi N, Kitaoka T, Fukada T, Hibi M and Hirano T. . 1996 EMBO J. 15: 3651–3658.

  28. Nicola NA. . 1987 Int. J. Cell Cloning 5: 1–15.

  29. O'Farrell AM, Liu Y, Moore KW and Mui AL. . 1998 EMBO J. 17: 1006–1018.

  30. Okuda K, Ernst TJ and Griffin JD. . 1994 J. Biol. Chem. 269: 24602–24607.

  31. Pagano M, Tam SW, Theodoras AM, Beer-Romero P, Del Sal G, Chau V, Yew PR, Draetta GF and Rolfe M. . 1995 Science 269: 682–685.

  32. Polyak K, Lee MH, Erdjument-Bromage H, Koff A, Roberts JM, Tempst P and Masague J. . 1994 Cell 78: 59–66.

  33. Quelle FW, Sato N, Witthuhn BA, Inhorn RC, Eder M, Miyajima A, Griffin JD and Ihle JN. . 1994 Mol. Cell. Biol. 14: 4335–4341.

  34. Sherr CJ and Roberts JM. . 1995 Genes Dev. 9: 1149–1163.

  35. Shimozaki K, Nakajima K, Hirano T and Nagata S. . 1997 J. Biol. Chem. 272: 25184–25189.

  36. Shuai K, Horvath CM, Huang LH, Qureshi SA, Cowburn D and Darnell Jr JE. . 1994 Cell 76: 821–828.

  37. Tian SS, Lamb P, Seidel HM, Stein RB and Rosen J. . 1994 Blood 84: 1760–1764.

  38. Tian SS, Tapley P, Sincich C, Stein RB, Rosen J and Lamb P. . 1996 Blood 88: 4435–4444.

  39. Toyoshima H and Hunter T. . 1994 Cell 78: 67–74.

  40. Wagner BJ, Hayes TE, Hoban CJ and Cochran BH. . 1990 EMBO J. 9: 4477–4484.

  41. Wakao H, Gouilleux F and Groner B. . 1994 EMBO J. 13: 2182–2191.

  42. Ward AC, Csar XF, Hoffmann BW and Hamilton JA. . 1996 Biochem. Biophys. Res. Commun. 224: 10–16.

  43. Ward AC, Hermans MH, Smith L, van Aesch YM, Schelen AM, Antonissen C and Touw IP. . 1999a Blood 93: 113–124.

  44. Ward AC, Smith L, de Koning JP, van Aesch Y and Touw IP. . 1999b J. Biol. Chem. 274: 14956–14962.

  45. Weber JD, Hu W, Jefcoat Jr SC, Raben DM and Baldassare JJ. . 1997 J. Biol. Chem. 272: 32966–32971.

  46. Zhu M, John S, Berg M and Leonard WJ. . 1999 Cell 96: 121–130.

Download references


We thank L Smith, J Gits, C Erpelinck, S Oomen and M Hermans for excellent technical advice and assistance, M von Lindern for critical reading of the manuscript, K Nakajima and T Hirano for STAT3 plasmids, A Nordin for the p27Kip1-pGL-2 construct, T Hunter for murine p27Kip1 cDNA, G Gil Gomez and J Roberts for p27-deficient mice, R Bernards for anti-p27Kip1 antibodies, and K van Rooyen for exquisite graphical work. This work was supported by grants from the Dutch Cancer Society ‘Koningin Wilhelmina Fonds’, the Netherlands Organization for Scientific Research (NWO) and an EMBO Long Term Fellowship.

Author information

Correspondence to Ivo P Touw.

Rights and permissions

Reprints and Permissions

About this article


  • G-CSF
  • STAT3
  • p27Kip1
  • neutrophilic differentiation
  • survival

Further reading