Signal Transducers and Activators of Transcription 3 Augments the Transcriptional Activity of CCAAT/Enhancer-binding Protein α in Granulocyte Colony-stimulating Factor Signaling Pathway*

The Janus kinase (Jak)-Stat pathway plays an essential role in cytokine signaling. Granulocyte colony-stimulating factor (G-CSF) promotes granulopoiesis and granulocytic differentiation, and Stat3 is the principle Stat protein activated by G-CSF. Upon treatment with G-CSF, the interleukin-3-dependent cell line 32D clone 3(32Dcl3) differentiates into neutrophils, and 32Dcl3 cells expressing dominant-negative Stat3 (32Dcl3/DNStat3) proliferate in G-CSF without differentiation. Gene expression profile and quantitative PCR analysis of G-CSF-stimulated cell lines revealed that the expression of C/EBPα was up-regulated by the activation of Stat3. In addition, activated Stat3 bound to CCAAT/enhancer-binding protein (C/EBP)α, leading to the enhancement of the transcription activity of C/EBPα. Conditional expression of C/EBPα in 32Dcl3/DNStat3 cells after G-CSF stimulation abolishes the G-CSF-dependent cell proliferation and induces granulocytic differentiation. Although granulocyte-specific genes, such as the G-CSF receptor, lysozyme M, and neutrophil gelatinase-associated lipocalin precursor (NGAL) are regulated by Stat3, only NGAL was induced by the restoration of C/EBPα after stimulation with G-CSF in 32Dcl3/DNStat3 cells. These results show that one of the major roles of Stat3 in the G-CSF signaling pathway is to augment the function of C/EBPα, which is essential for myeloid differentiation. Additionally, cooperation of C/EBPα with other Stat3-activated proteins are required for the induction of some G-CSF responsive genes including lysozyme M and the G-CSF receptor.

The proliferation and differentiation of hematopoietic progenitor cells are regulated by cytokines (1). Among these, gran-ulocyte colony-stimulating factor (G-CSF) 1 specifically stimulates cells that are committed to the myeloid lineage (2). The importance of G-CSF to the regulation of granulopoiesis has been confirmed by the observation of severe neutropenia in mice carrying homozygous deletions of their G-CSF or G-CSF receptor genes (3,4). Cytokines activate several intracellular signaling pathways, and the Janus kinase (Jak) signal transducers and activators of transcription (Stat) pathway is essential for cytokine function (5,6). The binding of G-CSF to cell surface G-CSF receptors activates Jak1, Jak2, and Tyk2 followed by the activation of Stat1, Stat3, and Stat5 (7)(8)(9). Stat3 is the principle protein activated by G-CSF (8,10). Phosphorylated Stats translocate from the cytoplasm into the nucleus and induce transcription of their target genes within a short period of time. 32Dcl3 cells differentiate to neutrophils following treatment with G-CSF. In contrast to their parental cells, 32Dcl3 cells expressing dominant-negative Stat3 (32Dcl3/ DNStat3) proliferate in the presence of G-CSF, but they maintain immature morphologic characteristics without evidence of differentiation (11). Additionally, transgenic mice with a targeted mutation of their G-CSF receptor that abolishes G-CSFdependent Stat3 activation show severe neutropenia with an accumulation of immature myeloid precursors in their bone marrows (12). To clarify the role of Stat3 in the G-CSF signaling pathway, we wished to identify target genes of Stat3.
We found that the levels of CCAAT/enhancer-binding protein (C/EBP) ␣ mRNA were up-regulated following G-CSF stimulation in 32Dcl3 but were unchanged in 32Dcl3/DNStat3. In addition, the activation of Stat3 augmented the function of C/EBP␣, which is the essential transcriptional factor for myeloid differentiation. G-CSF-induced granulocytic differentiation was restored in 32Dcl3/DNStat3 cells by the conditional expression of C/EBP␣. These results show that one of the major roles of Stat3 in the G-CSF signaling pathway is to enhance the function of C/EBP␣.
Microarray Analysis-32D cl3 and 32Dcl3/DNStat3 cells maintained in IL-3 were washed twice with PBS and starved of cytokine in RPMI 1640 containing 10% fetal bovine serum for 8 h and then stimulated with 10 ng/ml G-CSF. Total RNA was extracted, by the acid guanidinium method, from 32Dcl3 and 32Dcl3/DNStat3 cells before or after the stimulation for 2 h with G-CSF. Double-stranded cDNA synthesized from the total RNA (20 g/sample) was then used to prepare biotinlabeled cRNA for the hybridization with GeneChip MGU74Avs2 microarrays (Affymetrix, Santa Clara, CA) harboring oligonucleotides corresponding to ϳ6000 known genes as well as ϳ6000 expressed sequence tag sequences. Hybridization, washing, and detection of signals on the arrays were performed with the GeneChip system (Affymetrix).
Quantitative Real-time Reverse Transcription-PCR Assay-32Dcl3 and 32Dcl3/DNStat3 cells maintained in IL-3, were washed twice with PBS and starved of cytokines for 8 h and then stimulated with 10 ng/ml G-CSF. Cells were harvested at the indicated times, and total RNA was isolated using Isogen (Nippon gene, Tokyo, Japan) according to the manufacturer's instructions. One microgram of extracted RNA was transcribed in a 20-l cDNA synthesis reaction using an RNA PCR kit (AMV) (Takara, Tokyo, Japan). Real-time PCR for C/EBP␣, G-CSF receptor, lysozyme M, neutrophil gelatinase-associated lipocalin precursor (NGAL), and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was performed by a TaqMan assay on an ABI 7000 system. PCR primers and probes were designed as follows: murine C/EBP␣, sense, 5Ј-CCA TGT GGT AGG AGA CAG AGA CCT A-3Ј, and antisense,  5Ј-CTC TGG GAT GGA TCG ATT GTG-3Ј; probe FAM-5Ј-CGG CTG  GCG ACA TAC AGT ACA CAC AAG-3Ј-TAMRA, and sense, 5Ј-CCA  AGA AGT CGG TGG ACA AGA-3Ј, and antisense, 5Ј-CGG TCA TTG  TCA CTG GTC AAC T-3Ј; probe FAM-5Ј-AGC ACC TTC TGT TGC GTC  TCC ACG TT-3Ј-TAMRA; murine G-CSF receptor, sense, 5Ј-CTA AAC  ATC TCC CTC CAT GAC TT-3Ј, and antisense, 5Ј-GGC CAT GAG GTA  GAC ATG ATA CAA-3Ј; probe FAM-5Ј-CAT CTT CTC TGT CCC CAC  CGA CCA A-3Ј-TAMRA; murine lysozyme M, sense, 5Ј-TGC CTG TGG  GAT CAA TTG C-3Ј, and antisense, 5Ј-ATG CCA CCC ATG CTC GAA  T-3Ј; probe 5Ј-FAM-CAG TGA TGT CAT CCT GCA GAC CA-TAMRA-3Ј; murine NGAL, sense, 5Ј-GGC CTC AAG GAC GAC AAC A-3Ј, and  antisense, 5Ј-CAC CAC CCA TTC AGT TGT CAA T-3Ј; probe 5Ј-FAM-CAT CTT CTC TGT CCC CAC CGA CCA A-TAMRA-3Ј, and  Each amplification reaction also contained 100 nM appropriate detection probe. Each PCR amplification was performed in duplicate, using conditions of 50°C for 2 min preceding 95°C for 10 min followed by 40 cycles of amplification (95°C for 15 s, 60°C for 1 min). In each reaction, GAPDH was amplified as a housekeeping gene to calculate a standard curve and allow for the correction for variations in target sample quantities. Relative copy numbers were calculated for each sample from the standard curve after normalization to GAPDH by the instrument software.
Conditional C/EBP␣ Expression-pMY-IRES-GFP/C/EBP␣-ER was transfected into 32Dcl3 and 32Dcl3/DNStat3 cells by electroporation. 5 ϫ 10 6 cells were transfected with 20 g of expression vector, and GFP-positive cells were sorted by FACS Vantage (BD Biosciences). Expression of C/EBP␣ was determined by Western blotting analysis (see below).
Luciferase Assay-293T cells were transfected by the calcium phosphate precipitation method in 6-well plates, and luciferase activity was assayed using a luminometer Lumat LB9507 (Berthold Technologies, Bad Wildbad, Germany) according to the manufacturer's protocol. Each expression plasmid amount was 50 -100 ng/well, and the same amount of empty expression vector was used as control, respectively. Results of reporter assays represent the average values for relative luciferase activity generated from five independent experiments.
Flow Cytometry-1 ϫ 10 7 cells were incubated with 5 l of recombinant phosphatidylethanolamine-conjugated rabbit anti-murine Gr1 monoclonal antibody (BD Biosciences) for 30 min at 4°C, washed twice in PBS, and analyzed on a FACS Calibur (BD Biosciences).
Immunoprecipitation and Immunoblotting-Cells were lysed with lysis buffer, and lysates were immunoprecipitated with anti C/EBP␤ (Santa Cruz Biotechnology, Santa Cruz, CA) as described previously (8). Total cell lysates or the immunoprecipitates were resolved by 10% SDS-PAGE and transferred to a nitrocellulose membrane. Membranes , and G-CSF receptor. Twelve hours after transfection, cells were stimulated with 10 ng/ml G-CSF. Promoter activity was measured as luciferase activity 36 h after transfection. The vertical axis number is the fold induction when compared with control. B, 32Dcl3 cells or 32Dcl3/DNStat3 cells were cultured with IL-3 and then deprived of IL-3 for 12 h. Cells were treated with the G-CSF for 30 min and lysed. Post-nuclear supernatants were resolved by 10% SDS-PAGE and transferred to nitrocellulose membranes. Membranes were probed using the indicated antibodies. p-ERK1/2, phosphorylated ERK1/2.
were probed using the indicated antibodies followed by an IgG-horseradish peroxidase-conjugated secondary antibody (Amersham Biosciences) and visualized with the ECL detection system (Amersham Biosciences). Anti-phospho-ERK1/2 antibodies were purchased from Cell Signaling (Beverly, MA). Anti-phospho-Stat1 and -Stat5 antibodies were obtained from New England Biolabs (Beverly, MA), and anti-Stat1, -Stat3, and -C/EBP␣ antibodies were purchased from Santa Cruz Biotechnology. Membranes were probed using and visualizes with the ECL detection system (Amersham Biosciences).
Proliferation Assay-32D cl3 and 32Dcl3/DNStat3 cells maintained in IL-3 were washed twice with PBS and starved of cytokine for 8 h and then stimulated with 10 ng/ml G-CSF. The number of viable cells was determined by trypan blue dye exclusion using a hemocytometer.  32Dcl3 and 32Dcl3/DNStat3 cells were starved of cytokines for 8 h and then stimulated or left unstimulated with 10 ng/ml G-CSF. Total RNA was extracted from each fraction and was subjected to the hybridization with high-density oligonucleotide microarrays (MGU74Av2). Fold induction means a rate of increase in gene expression level by G-CSF stimulation. Candidate genes were identified as transcripts that were up-regulated in 32Dcl3 cells and down-regulated or unchanged in 32Dcl3/DNStat3 cells after G-CSF stimulation. dominant-negative Stat3 (32Dcl3/DNStat3) proliferate following G-CSF treatment. These cells maintain immature morphologic characteristics without evidence of differentiation (11). First, we examined the effect of dominant-negative Stat3, carboxyl-truncated Stat3 that lacked 55 amino acids including the transactivation domain. We transfected reporter construct of STATt3-LUC, in which the ␣2-macroglobulin promoter (16) drives expression of the luciferase (LUC) reporter gene and G-CSF receptor, together with empty vector (pcDNA3) or DNStat3 to 293T cells. After 12 h of transfection, cells were stimulated with 10 ng/ml G-CSF. Cells were cultured for more 24 h, and luciferase assay was performed. As shown in Fig. 1A, G-CSF induced the transcriptional activity of Stat3 by 150-fold, and DNStat3 inhibited this G-CSF-induced Stat3 activation in a dose-dependent manner. G-CSF mainly induces the phosphorylation of Stat3, but it also phosphorylates Stat1 and Stat5 in some cells among the Stats family (8) and induces the activation of MAP kinases. In both 32Dcl3 cells and 32Dcl3/DNStat3 cells, neither Stat1 nor Stat5 was phosphorylated in response to G-CSF (data not shown). As for the MAP kinase activation, the degree of the phosphorylation of ERK1/2 by G-CSF stimulation in 32Dcl3/ DNStat3 cells was stronger than that in 32Dcl3 cells (Fig. 1B).
Identification of Genes Regulated by Stat3 in the G-CSF Signaling Pathway by Oligonucleotide Array Analysis-To identify Stat3-regulated genes involved in granulocytic differentiation, we compared gene expression change in both cell lines using microarray analysis. 32D cl3 and 32Dcl3/DNStat3 cells maintained in IL-3 were washed twice with PBS and starved in RPMI 1640 containing 10% fetal bovine serum lacking cytokine for 8 h and then stimulated with 10 ng/ml G-CSF.
Total RNA was isolated from 32Dcl3 cells and 32Dcl3/DNStat3 cells treated with G-CSF after 0 and 2 h, transcribed to biotinlabeled cRNA, and hybridized to GeneChip MGU74Av2 arrays to compare the expression profile of ϳ12,000 murine genes. The fold induction in the expression level of each gene was calculated as the ratio of GAPDH-normalized fluorescence intensity value of G-CSF-stimulated cells when compared with those before G-CSF stimulation. As shown in Table I, we could identify a set of candidate genes for Stat3 targets, expression of which was up-regulated in 32D cl3 cells but down-regulated or unchanged in 32Dcl3/DNStat3 cells. Such Stat3-dependent expression profiles were confirmed in triplicate experiments.
C/EBP␣ Is a Target Gene for Stat3 in G-CSF Signaling Pathway-Among the identified genes, it was decided to focus further efforts on C/EBP␣. C/EBP␣ has been shown to be critical for early granulocytic differentiation (17)(18)(19), and the factors regulating its activity are unclear. The expression of C/EBP␣ was examined by real-time quantitative reverse transcription-PCR. C/EBP␣ mRNA levels are rapidly up-regulated in 32Dcl3 cells, being elevated 2.39-fold after 6 h and 4.20-fold after 48 h ( Fig. 2A). In contrast to 32Dcl3 cells, the C/EBP␣ mRNA levels were not changed in 32Dcl3/DNStat3 cells after G-CSF stimulation ( Fig. 2A). A similar expression pattern was seen in separate experiments with independently designed primers and probes (data not shown). Levels of C/EBP␣ mRNA were unaffected by cycloheximide treatment (Fig. 2B). The expression level of the sum of Stat3 plus dominant-negative Stat3 in 32Dcl3/DNStat3 cells is a little larger than that of Stat3 in 32Dcl3 cells (Fig. 2C).
Activated Stat3 Binds to C/EBP␣ and Enhances the Transcription Activity of C/EBPa-We next examined the effect of Stat3 abrogation on the balance of intracellular signals in other cytokine pathways. Although Stat1 was not phosphorylated by leukemia inhibitory factor stimulation in neither 32Dcl3 cells nor 32Dcl3/DNStat3 cells, its activation in response to IFN-␥ occurred at the same degree in both 32cl3 cells and 32Dcl3/ DNStat3 cells (Fig. 3A). As for the Stat5 activation, the phosphorylation of Stat5 by IL-3 stimulation in 32Dcl3 cells was stronger than that in 32Dcl3/DNStat3 cells (Fig. 3B). These data indicated that there was the possibility that abrogation of Stat3 signaling can alter the balance of intracellular signals in other cytokine signaling pathways. The transcription of C/EBP␣ is regulated by C/EBP␣ itself (20,21). Then we examined whether activated Stat3 in G-CSF signaling enhance C/EBP␣ activity or not.
We transfected a reporter construct of a minimal TK promoter with CEBP-binding sites (p(C/EBP)2TK), C/EBP␣, and G-CSF receptor to 293T cells. After 12 h of transfection, cells were stimulated with 10 ng/ml G-CSF. Cells were cultured for more 24 h, and a luciferase assay was performed. C/EBP␣ up-regulated the C/EBP␣-dependent gene expression, and the G-CSF stimulation enhanced this C/EBP␣-dependent gene expression (Fig. 4A). Next we examined the effect of constitutive active Stat3 (Stat3C) on the augmentation of C/EBP␣ transcriptional activity instead of the G-CSF stimulation. We transfected reporter construct p(C/EBP)2TK, C/EBP␣, and Stat3C to 293T cells. After 24 h of transfection, luciferase assay was performed. Stat3C augmented the C/EBP␣-dependent gene expression, although Stat3C alone had no influence on the luciferase activity (Fig. 4, B and C).
As p(C/EBP)2TK contains only a C/EBP␣-binding site and does not contain a Stat3-binding sequence, the possibility that Stat3C makes a complex with C/EBP␣ and augments the function of C/EBP␣ is raised. Then we transfected C/EBP␣, Stat3, and G-CSF receptor to 293T cells and stimulated cells with G-CSF for 6 h. There is no detectable level of endogenous C/EBP␣ or C/EBP␤ protein in 293T cells. Cells were lysed and immunoprecipitated with C/EBP␤ antibody (this antibody cross-reacts with C/EBP␣). As shown in Fig. 4D, immunoprecipitants with anti-C/EBP␤ contain Stat3. In addition, the complex formation between C/EBP␣ and Stat3 is augmented by G-CSF stimulation, indicating that activated Stat3 makes the complex with C/EBP␣. regulated C/EBP␣ function in the G-CSF signaling pathway, we transfected a C/EBP␣-tamoxifen receptor fusion protein (C/EBP␣-ER) into 32Dcl3 and 32Dcl3/DNStat3 cells (32Dcl3/ CEBPA cells, 32Dcl3/DNStat3/CEBPA cells, respectively). The expression of C/EBP␣-ER in these cells was verified by Western blotting (Fig. 5A). C/EBP␣-ER localizes to the cytoplasm and is in an inactive form in the absence of tamoxifen. Upon treatment with tamoxifen, it translocates from cytoplasm to nucleus and becomes active. 32Dcl3, 32Dcl3/CEBPA, 32Dcl3/DNStat3, and 32Dcl3/DNStat3/CEBPA cells were cultured with G-CSF in the presence or absence of tamoxifen, and cell proliferation was examined by both counting viable cells and [ 3 H]thymidine incorporation. 32Dcl3/DNStat3 proliferated in response to G-CSF, and proliferation was not affected by the presence of tamoxifen. Conversely, G-CSF-induced proliferation of 32Dcl3/ DNStat3/CEBPA cells in the presence of tamoxifen was dramatically reduced (Fig. 5, B and C).
32Dcl3/DNStat3 cells maintain morphologically immature characteristics and proliferate without granulocytic differentiation after G-CSF stimulation. We examined the morphological changes in 32Dcl3 and 32Dcl3/DNStat3 cells induced by G-CSF after translocation of C/EBP␣ from the cytoplasm to the nucleus. When tamoxifen was added to medium containing G-CSF, 32Dcl3/DNStat3/CEBPA cells rapidly began to differentiate into granulocytes, and 5 days later, about 40% of the cells were morphologically similar to mature neutrophils. In contrast, 32Dcl3/DNStat3/CEBPA cells cultured in G-CSF-con-taining medium without tamoxifen appeared immature with blast-like morphologic features (Fig. 6, Table II). To quantitatively analyze the difference in granulocyte maturation in 32Dcl3/DNStat3/CEBPA cells stimulated by G-CSF in the presence of tamoxifen, the mature granulocyte marker Gr-1 was monitored by FACS analysis. 32Dcl3 cells differentiate into Gr-1-positive neutrophils in response to G-CSF (Fig. 7A). As shown in Fig. 7D, Gr-1-positive cells were increased by the addition of tamoxifen in 32Dcl3/DNStat3/CEBPA cells treated with G-CSF, although low levels were detected in the absence of tamoxifen.
C/EBP␣ Up-regulates Genes That Are Related to Granulocytic Differentiation-In a conditional expression system, induction of C/EBP␣ leads to expression of granulocyte-specific genes, such as neutrophil primary granule genes (lysozyme M, NGAL) and the G-CSF receptor gene (17). In 32Dcl3/DNStat3 cells, the expression of these genes following G-CSF stimulation was inhibited (Fig. 8, A, C, and E). Interestingly, only NGAL was up-regulated by G-CSF in 32Dcl3/DNStat3/CEBPA cells following the restoration of C/EBP␣ (Fig. 8B). Conversely, the expression of lysozyme M and the G-CSF receptor were not changed by the restoration of C/EBP␣ (Fig. 8, D and F). These data suggest that regulatory factors in addition to C/EBP␣ may be involved in the induction of expression of granulocyte-specific genes by G-CSF. DISCUSSION G-CSF plays a pivotal role in granulopoiesis and granulocytic differentiation. The binding of G-CSF to its receptor leads to the activation of the Jak-Stat pathway, phosphatidylinositol-3 kinase pathway, and Ras-MAP kinase cascade (22). In the Jak-Stat pathway, G-CSF activates Jak1, Jak2, and Tyk2 followed by the activation of Stat1, Stat3, and Stat5 (7,8).
Dominant-negative Stat3 inhibits G-CSF-induced transcriptional activity of Stat3 (Fig. 1A), as does G-CSF-induced granulocytic differentiation in vitro (11). Also, more transgenic mice with a targeted mutation of their G-CSF receptor that abolishes G-CSF-dependent Stat3 activation show severe neutropenia with an accumulation of immature myeloid precursors in their bone marrows (12). Consequently, Stat3 is thought to play an essential role in G-CSF-induced granulocytic differentiation.
32Dcl3 cells differentiate into neutrophils after treatment with G-CSF, and 32Dcl3/DNStat3 cells (32Dcl3 cells expressing dominant-negative Stat3) proliferate in G-CSF without differentiation. The degree of the phosphorylation of ERK1/2 by G-CSF stimulation in 32Dcl3/DNStat3 cells was stronger than that in 32Dcl3 cells (Fig. 1B). We reported that Stat3 null bone marrow cells displayed a significant activation of ERK1/2 after G-CSF stimulation than wild-type bone marrow cells did using Stat3 conditional deficient mice (23). Then the augmented phosphorylation of ERK1/2 in response to G-CSF in 32Dcl3/ DNStat3 cells might be caused by the functional abrogation of Stat3 in 32Dcl3/DNStat3 cells. We compared gene profiles between two cell lines, 32Dcl3 and 32Dcl3/DNStat3 cells, to identify target genes of Stat3 in G-CSF signaling. We found that C/EBP␣ mRNA levels are rapidly up-regulated in 32Dcl3 cells following G-CSF treatment; these levels are increased 2.39-fold after 6 h and 4.20fold after 48 h of treatment. In contrast to 32Dcl3 cells, C/EBP␣ mRNA levels are not changed in 32Dcl3/DNStat3 cells after G-CSF stimulation ( Fig. 2A). The observation that cycloheximide does not inhibit G-CSF-induced increases in C/EBP␣ transcript levels (Fig. 2B) suggests that C/EBP␣ is induced by G-CSF directly downstream of Stat3. Dahl et al. (24) also reported that G-CSF induced the expression of C/EBP␣ in IL-3dependent progenitors. SOCS3 is one of the major target genes of Stat3. We previously reported that the expression level of SOCS3 protein in Stat3-deficient bone marrow cells is a trace, and it is not augmented by G-CSF stimulation (23). Contrary to this suppression of SOCS3 in Stat3-deficient cells, the induction of SOCS3 by G-CSF is not abolished in 32Dcl3/DNStat3 cells (data not shown).
The phosphorylation of ERK1/2 by G-CSF is stronger and the phosphorylation of Stat5 by IL-3 is weaker in 32Dcl3/DNStat3 cells when compared with those in 32D/Cl3 cells, although Stat1 phosphorylation by IFN-␥ was not changed between these two cells (Figs. 1B and 3). Then there is the possibility that the transfection of dominant-negative Stat3 affects other signaling pathways in 32Dcl3/DNStat3 cells, resulting in the change of C/EBP␣ regulation. To clarify whether Stat3 directly up-regulates C/EBP␣ in the G-CSF signaling pathway in 32Dcl3 cells or not, we examined the effect of Stat3C on the transcription of C/EBP␣. C/EBP␣ up-regulated the C/EBP␣-dependent gene expression, and the G-CSF stimulation enhanced this C/EBP␣-dependent gene expression (Fig. 4A). Strikingly, Stat3C augmented the C/EBP␣-dependent gene expression as G-CSF stimulation did (Fig. 4, B and C). This means that G-CSF-induced up-regulation of C/EBP␣-dependent gene expression is, at least partly, due to the activation of Stat3.
Two possibilities arise for the mechanism of the induction of C/EBP␣ transcription by activated Stat3 in the G-CSF signaling pathway. One is that activated Stat3 binds to the promoter region of C/EBP␣ and induces the transcription of C/EBP␣. Analysis of the reported murine C/EBP␣ promoter sequence (20) identified no Stat-responsive elements (TTN5AA) (25,26), but we found six Stat-responsive elements between 6 and 4 kb upstream of the C/EBP␣ transcription initiation site. Activated Stat3 might bind these Stat-responsive elements between 6 and 4 kb upstream of the C/EBP␣ transcription initiation site. The other possibility is that activated Stat3 might form the complex with C/EBP␣ and augment the transcriptional activity of C/EBP␣ because C/EBP␣ itself is the only protein reported to activate the murine C/EBP␣ promoter (20,21). When a minimal TK promoter with CEBP-binding sites (p(C/EBP)2TK) together with C/EBP␣ was transfected to 293T cells, C/EBP␣ up-regulated C/EBP␣-dependent gene expression. Activated Stat3 (Stat3C) enhanced this C/EBP␣-dependent gene expression in collaboration with C/EBP␣, although only Stat3C has no transcriptional activity on p(C/EBP)2TK (Fig. 4, B and C). In addition, the stimulation of G-CSF allows Stat3 to make the complex with C/EBP␣ (Fig. 4D). Then activated Stat3 by G-CSF makes the complex with C/EBP␣ and augments the transcriptional activity of C/EBP␣. This is one of the reasons why induction of C/EBP␣ transcript through Stat3 activation by G-CSF occurred in 32Dcl3 cells. Several reports have described factors that repress C/EBP␣ promoter activity, such as SP1 (27), AP2A (28), or MYC (29). We show here for the first time that Stat3 augments the C/EBP␣ promoter activity. Intracellular transcript levels of several genes were changed following G-CSF treatment downstream of Stat3 activation (Table I). To better identify the role of C/EBP␣ in Stat3-mediated signaling in G-CSF-induced granulocyte differentiation, C/EBP␣-ER (C/EBP␣-tamoxifen receptor fusion protein) was stably expressed in 32Dcl3 and 32Dcl3/DNStat3 cells. C/EBP␣-ER translocates from the cytoplasm to the nucleus and becomes activated upon treatment with tamoxifen. Strikingly, transfection of C/EBP␣-ER into 32Dcl3/DNStat3 cells abolished proliferation and induced myeloid differentiation by G-CSF without Stat3 activation (Figs. 5, B and C, and 6). These data indicate that C/EBP␣ activation induced by G-CSF through Stat3 plays an essential role in stopping the cell proliferation and inducing the differentiation to the myeloid lineage. The CEBP family of transcription factor is expressed in multiple cell types, including hepatocytes, adipocytes, keratinocytes, enterocytes, and cells of the lung (30,31). C/EBP␣ transactivates the promoters of hepatocyte-and adipocyte-specific genes, which are important for energy homeostasis (32,33), and C/EBP␣-deficient mice lack hepatic glycogen stores and die from hypoglycemia within 8 h of birth (34). In the hematopoietic system, C/EBP␣ is exclusively expressed in myelomonocytic cells (35,36). C/EBP␣ expression is prominent in mature myeloid cells, and previous investigations found that C/EBP␣ is critical for early granulocytic differentiation. Mice with a targeted disruption of the C/EBP␣ gene demonstrate an early block in granulocytic differentiation, but they develop normal monocytes (19). Conditional expression of C/EBP␣ is sufficient to induce granulocytic differentiation (17). In contrast to the essential role of C/EBP␣ in granulocytic differentiation, the role of Stat in granulopoiesis is controversial. Stat3 is the principle Stat protein activated by G-CSF, with Stat5 and Stat1 also activated to a lesser degree (8,10). In mice lacking Stat5a and Stat5b, the number of colonies produced in response to G-CSF was reduced 2-fold despite normal circulating numbers of neutrophils (9). Myeloid cell lines expressing dominant-negative forms of Stat3 (11,37,38) and transgenic mice with a targeted mutation of the G-CSF receptor that abolishes G-CSF-dependent Stat3 activation (12) demonstrate that Stat3-activation is required for G-CSF-dependent granulocytic proliferation and differentiation.
In the present study, we clearly demonstrate that the expression of C/EBP␣ mRNA is up-regulated through the activation of Stat3 in response to G-CSF, and the Stat3-C/EBP␣ signaling cascade plays an important role in G-CSF-induced differentiation. Contrary to these data, however, we and others showed that mice conditionally lacking Stat3 in their hematopoietic progenitors developed neutrophilia, and bone marrow cells were hyper-responsive to G-CSF stimulation (23,39). Additionally, mice with tissue-specific disruption of Stat3 in bone marrow cells die within 4 -6 weeks after birth with Crohn's disease-like pathogenesis (40). These mice exhibit phenotypes with dramatic expansion of myeloid cells, leading to massive infiltration of the intestine with neutrophils, macrophages, and eosinophils. Cells of the myeloid lineage also demonstrate autonomous proliferation. These apparently disparate results may be explained by the need for molecules in addition to Stat3 to regulate C/EBP␣ expression in vivo, the in vivo functional redundancy among C/EBP␣ regulators, or the absence of the abrogation of SOCS3 induction by G-CSF in 32Dcl3/DNStat3 cells. In 32Dcl3 cells, the Stat3-C/EBP␣ pathway might be favored, and other pathways may contribute little to granulocytic differentiation in response to G-CSF.
Among C/EBP family, C/EBP⑀ is important for late phase of granulocytic differentiation, and its expression is up-regulated by G-CSF independent of Stat3 (11). A previous report showed that C/EBP⑀ is a transcriptional target of C/EBP␣ in 32Dcl3 cells (41). From these reports and our results, we speculated that a small amount of C/EBP␣ is enough for the induction of the transcription of C/EBP⑀ by G-CSF or that there are multiple signaling steps except for Stat3-C/EBP␣ to induce the transcription of C/EBP⑀ by G-CSF. Induction of C/EBP␣ led to not only morphologic differentiation but also expression of granulocyte-specific genes (17). In 32Dcl3/DNStat3 cells, the induction of the G-CSF receptor, lysozyme M, and NGAL in response to G-CSF was abrogated (Fig. 8). Restoration of C/EBP␣ in these cells led to expression of only the NGAL gene, and thus, 32Dcl3/DNStat3 cells differentiated by the induction of C/EBP␣ may not be functional as mature neutrophils. In these cells, therefore, activation of C/EBP␣ is not sufficient for the induction of lysozyme M or G-CSF receptor genes, and the presence of other molecules appears to be required for their expression.