Cytokine Receptor Common (cid:1) Subunit-mediated STAT5 Activation Confers NF- (cid:2) B Activation in Murine proB Cell Line Ba/F3 Cells*

The cytokine receptor common (cid:1) subunit ( (cid:1) c ) trans-mits intracellular signals upon binding ligand such as granulocyte-macrophage colony-stimulating factor or interleukin-3 (IL-3); however, transcriptional regulation under the control of signaling events downstream of the (cid:1) c is not fully understood. Using murine Ba/F3 cells, here we demonstrate that the (cid:1) c -mediated signals stimulate NF- (cid:2) B-driven gene expression of not only the reporter construct but also endogenous target genes such as IL-6 . Analyzing the effects of several inhibitors or mutant receptors revealed that this NF- (cid:2) B activation is mediated neither by MEK/ERK/MAPK nor by the phos-phatidylinositol 3-kinase pathway but by STAT5. Over-expression experiments of the wild-type or constitutive active form of STAT5 further confirmed this notion. In addition, STAT5-dependent NF- (cid:2) B activation is mediated not through an inducible nuclear translocation but

Granulocyte-macrophage colony-stimulating factor (GM-CSF) 1 and interleukin-3 (IL-3) play important roles in regulating multiple cellular functions such as proliferation, differentiation, and survival in various hematopoietic cell lineages and their precursors (1). These cytokines bind to their cognate receptors and trigger a cascade of signaling events leading to various biological responses. The receptors for GM-CSF (GMR) and IL-3 (IL-3R) are composed of two subunits, the cytokine-specific ␣ and the common ␤ subunit (␤ c ), the latter of which is also shared by the receptor for IL-5 (2). Both subunits belong to the type I cytokine receptor superfamily, and the ␤ c , having a relatively large cytoplasmic domain, plays a pivotal role in downstream signal transduction (2). These receptors lack intrinsic kinase activity but interact with and activate Janus kinase 2 (JAK2) in response to binding ligand (3). Subsequently, the cytoplasmic domain of the ␤ c becomes tyrosine-phosphorylated and then cue multiple signaling pathways (4,5) including the mitogenactivated protein kinase (MAPK) pathway and the phosphatidylinositol 3-kinase (PI3K) pathway (6,7). In addition, at the transcriptional level, not only expression of some immediate early genes such as c-fos, c-jun, c-myc, egr-1, cis, and pim-1 (6,8,9) but also the functional regulation of transcription factors such as STAT5, AP-1, and CREB (9 -12) is known to be achieved by these cytokine receptors.
The STAT proteins are activated upon various cytokine stimulations and play a central role in following the transcriptional regulation of gene expression (13,14). Among seven known members of STAT family, both STAT5A and STAT5B, closely related isoforms, are the most predominant members activated by the ␤ c (10,11). Upon cytokine stimulation, JAK2 phosphorylates STAT5 on a tyrosine residue near the C-terminal transactivation domain in the cytoplasm (15). The phosphorylated STAT5 proteins dimerize and translocate into the nucleus, where they bind to specific DNA elements and thereby activate target gene expression (10,11). Using murine IL-3-dependent Ba/F3 cells, we have reported previously that STAT5 participates in growth-promoting signals downstream of the ␤ c (16 -18). Moreover, we have shown that STAT5 is a key regulator of not only proliferation but also differentiation or apoptosis in these cells (19). In particular, pim-1, JAB/SOCS-1/SSI-1, and p21 WAF1/Cip1 , all of which are transcriptionally regulated by STAT5, are suggested to be involved in a variety of biological outcomes induced by STAT5 (19). Our knowledge of the gene expression profile regulated by STAT5, however, has been extremely limited. A recent study using microarray analysis revealed that more than 300 genes are regulated at the transcriptional level by IL-3 signaling in murine pro-B cells (20), raising the possibility that more complicated mechanisms than described previously are operated in these cytokine-or STAT5dependent biological responses.
It was recently reported that several growth factors, including GM-CSF (21) and IL-3 (22), could induce the activation of another transcription factor NF-B in cells of hematopoietic lineages, although underlying mechanisms have remained unclear. NF-B was originally described as a regulator of immune and inflammatory responses. NF-B consists of a dimer from five related proteins, most typically a heterodimer composed of p65/RelA and p50 subunits (23)(24)(25). The regulation of NF-B is achieved through interaction with a family of inhibitory protein known as IB that binds to NF-B and sequesters it in the cytoplasm (26). Once cells are stimulated with inducers such as tumor necrosis factor-␣ (TNF␣) and IL-1, two serine residues of the IB protein are phosphorylated by IB kinases (27,28). Phosphorylation of IB targets it for ubiquitination and subsequent degradation by the 26S proteasome and renders the nuclear localization signal of NF-B unmasked (29,30). Then NF-B translocates from the cytoplasm into the nucleus and regulates the transcription of target genes. In addition to this "classical" milieu, recent reports have suggested that alternative pathways lead to NF-B activation through several nuclear mechanisms (31)(32)(33)(34)(35). As well as its well established role in activating the transcription of genes involved in immunological responses, it is indicated that NF-B is one of the central mediators of hematopoiesis (36). Therefore, it appears to be of importance to elucidate the linkage between the GM-CSF or IL-3 signaling and NF-B activation for a better understanding of biological responses under the control of these cytokines.
In the present study, we demonstrated that in Ba/F3 cells, GM-CSF as well as IL-3 activates NF-B-dependent transcription, and STAT5 is essential for this process. Of note, STAT5 affected neither IB degradation nor nuclear translocation of NF-B. In contrast, we indicated that STAT5 increases not only the DNA binding activity but also the transactivational potential of NF-B, possibly through a mechanism involving STAT5dependent synthesis of as yet undetermined humoral factor(s).
Plasmids-The reporter plasmids NF-B-Luc (formerly pNFBHL) and AP-1-Luc (formerly pAP-1HL), in which expression of the luciferase gene is under the control of transcription factors NF-B and AP-1, respectively, were described elsewhere (38). A ␤-casein promoter-inducible luciferase reporter construct, ␤-casein-Luc (pZZ1) (39), and an expression plasmid, pXM-STAT5⌬750VP16JAK2, encoding a fusion protein described in Ref. 40 (see the legend for Fig. 7), were kindly gifts of Dr B. Groner (Tumor Biology Center, Freiburg, Germany). To construct the wild-type (WT) STAT5A expression vector pCMX-WT STAT5A, an EcoRI-NotI fragment containing cDNA for mSTAT5A was excised from a retrovirus vector pMX-STAT5A (18), filled with the Klenow fragment, and inserted into EcoRV site of pCMX (41). A pCMX-STAT5A1*6, an expression vector encoding a constitutive active mutant form of STAT5A, was constructed according to the same strategy as above using pMX-STAT5A1*6 (18). A pCMX-STAT5A⌬749 encoding a mutant of STAT5A in which C-terminal 44 amino acids are deleted was derived by exchanging the NsiI and NheI fragments of pCMX-WT STAT5A with the PCR-generated fragment: primers used in the PCR reaction were 5Ј-GAGTTCGTCAATGCATCCAC-3Ј (sense primer) and 5Ј-AGTCAGTCGCTAGCTGAGCCATCTTGGTCAAGGAC-3Ј (antisense primer). Underlined sequences indicate recognition sites for NsiI for the sense primer or NheI for the antisense primer. Bold sequences indicate the introduced stop codon. A pCMX-STAT5AVVV encoding a DNA-binding defective mutant of STAT5A (42) (see the legend for Fig.  7) was generated by the use of sequential PCR reactions. Primers used here were 5Ј-AGAAGCAGCCTCCTCAGGTC-3Ј (sense primer-1), 5Ј-ATGGACGATAGCGGCCGCAGGGAGGGACAG-3Ј (antisense primer-1), 5Ј-TCCCTCCCTGCGGCCGCTATCGTCCATGGC-3Ј (sense primer-2), 5Ј-GTTGTAGTCCTCGAGGTGGT-3Ј (antisense primer-2), and two independent PCR reactions were carried out with either set of primers, respectively, using a pCMX-WT STAT5A as a template. The second step of the PCR reaction was performed with the outer primers, sense primer-1 and antisense primer-2, using combined products of the first round PCR as templates. The Bsu36I and XhoI fragment within the STAT5A coding region of a pCMX-WT STAT5A was replaced with the PCR-amplified fragments. Underlined sequences are sites for Bsu36I (sense primer-1) and XhoI (antisense primer-2), respectively, and the bold letters indicate introduced mutations at codons 466, 467, and 468. A Gal4-responsive luciferase reporter tk-Gal4px3-Luc was described elsewhere (43). To construct the chimeric plasmid for NF-B p65 and the DNA binding domain of Gal4 (amino acids 1-147), PCR-generated fragments containing cDNA encoding either the N-terminal half (amino acids 1-285) or C-terminal half (amino acids 286 -549) of m-p65 were cloned into EcoRI and EcoRV sites of pCMX-Gal4 (44) in-frame. Primers used here were 5Ј-AGTCAGTCGAATTCATGGACGAACTGTTCCCC-3Ј(sense primer) and 5Ј-AGTCAGTCGATATCTGATTCCATGG-GCTCACTGAA-3Ј(antisense primer) for pCMX-Gal4-p65 1-285 and 5Ј-GTGAGCCCATGGAATTCCAG-3Ј(sense primer) and 5Ј-AGTCAGTC-GATATCTGAGGAGCTGATCTGACTCAG-3Ј (antisense primer) for pCMX-Gal4-p65 286 -549 , respectively, and the PCR reaction was performed using pCAGGS-m-p65 (a gift from Dr. H. Handa, Tokyo Institute of Technology, Tokyo, Japan) as a template. Underlined sequences indicate recognition sites of EcoRI for sense primers or EcoRV for antisense primers. Bold sequences indicate the introduced stop codon. All plasmids constructed as above were verified by sequencing.
Transient Transfection and Reporter Assays-Plasmids were transiently transfected into Ba/F3 cells or WEHI 3B cells by electroporation as described previously (16). After transfection, cells were cultured in appropriate conditions, and cellular extracts were isolated as described below. When indicated, either PD98059 (Biomol Research Laboratories, Plymouth Meeting, PA) or wortmannin (Sigma) was added in the culture medium 1 h prior to the addition of cytokines. Anti-mouse TNF␣ antibodies (R&D systems, Minneapolis, MN) were also added in the culture medium 2 h prior to the cellular stimulation. For reporter gene assay, cells were harvested and lysed by three cycles of freezing and thawing after 12 h (Ba/F3 cells) or 18 h (WEHI 3B cells) of cellular stimulation. Cell lysates were subjected to luciferase assays, and relative light units (RLU) were normalized to the protein amount determined with protein assay reagent (Pierce) according to the manufacturer's instructions. Total amounts of plasmid DNA were adjusted to 10 g/transfection using pGEM 7Z (Promega, Madison, WI).
RNA Extraction and Northern Blotting-Total RNA was extracted by the acid guanidine method using TRIzol Reagent (Invitrogen Corp., Carlsbad, CA). Twenty g of total RNA was subjected to 1% agarose/ formaldehyde gel electrophoresis and subsequently transferred onto a MAGNACHARGE nylon membrane filter (Osmonics, Westborough, MA). Equal amounts of loading and the transfer efficiency of RNA samples were verified by ethidium bromide staining of 28 and 18 S rRNAs on the gel and the membrane. The cDNA probe corresponding to nucleotides 16 -761 of the m IL-6 cDNA was generated by reverse transcription-PCR from an RNA sample of parental Ba/F3 cells cultured under usual conditions. The probe was labeled with [␣Ϫ 32 P]dCTP by random priming using RediPrime II (Amersham Biosciences, Inc.) according to the manufacturer's instructions. Hybridization was carried out under ULTRAHyb conditions (Ambion Inc., Austin, TX) at 42°C overnight.
Preparation of Cell Extracts-For preparation of whole cell extracts, 1 ϫ 10 6 cells were washed with phosphate-buffered saline twice and incubated on ice for 15 min in lysis buffer (1% Nonidet P-40, 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride (PMSF), 10 g/ml aprotinin, 1 M pepstatin, 50 mM NaF, 1 mM Na 3 VO 4 ). After centrifugation at 12,000 ϫ g for 20 min in a microcentrifuge, the supernatant was used as whole cell extracts. Cy-tosol and nuclear extracts were prepared as described by Dignam et al. (45) with minor modifications. In brief, after two cycles of washing in ice-cold phosphate-buffered saline, 2 ϫ 10 7 cells were harvested and centrifuged at 1,000 ϫ g for 5 min in a microcentrifuge. The cell pellet was resuspended in 80 l of buffer A (10 mM Tris-HCl, pH 7.3, 1.5 mM MgCl 2 , 10 mM KCl, 1 mM dithiothreitol, 0.4% Nonidet P-40, 1 mM PMSF, 10 g/ml aprotinin, 1 M pepstatin, 50 mM NaF, and 1 mM Na 3 VO 4 ). After incubation on ice for 5 min, cell lysates were centrifuged, and the supernatant was used as cytosol. The nuclear pellet was then resuspended in 75 l of buffer C (20 mM Tris-HCl, pH 7.3, 1.5 mM MgCl 2 , 484 mM KCl, 1 mM dithiothreitol, 0.2 mM EDTA, 25% glycerol, 1 mM PMSF, 10 g/ml aprotinin, 1 M pepstatin, 50 mM NaF, and 1 mM Na 3 VO 4 ) and incubated for 30 min at 4°C. Nuclear debris were pelleted with centrifugation at 12,000 ϫ g for 15 min, and the supernatant was used as nuclear extracts. Protein concentration was determined using protein assay reagent (Pierce).
Immunoblotting-Ten g of protein was separated in 10% SDSpolyacrylamide gels and then transferred to polyvinylidene difluoride membranes. The membranes were blocked in Tris-buffered saline (TBS)-T (50 mM Tris-HCl, pH 7.6, 200 mM NaCl, 0.1% Tween 20) with 5% non-fat dried skim milk. The membranes were probed with either anti-p65 (sc-372), anti-p50 (sc-1190), or anti-IB␤ (sc-945, Santa Cruz Biotechnology) antibodies at 1:1000 dilution and then incubated with a secondary antibody conjugated to horseradish peroxidase (Santa Cruz Biotechnology). For analysis of IB␣ protein, the membranes were first probed with anti-phospho-IB␣ (Ser-32) antibody (New England Biolabs, Beverly, MA) at 1:1000 dilution followed by incubation with a secondary antibody conjugated to horseradish peroxidase. Subsequently, after detection of proteins, the same membranes were stripped, reprobed for anti-IB␣ antibody (sc-371, Santa Cruz Biotechnology) at 1:500 dilution, and probed with a secondary antibody. In all experiments, proteins were visualized with ECL detection systems (Amersham Biosciences, Inc.) according to the manufacturer's instructions.
Immunocytochemical Analysis-For immunocytochemical analysis, Ba/F3-wild cells were collected and washed once with phosphate-buffered saline, and then 2 ϫ 10 5 cells were cytospined to a silane-coated glass slide. Cells were fixed with 3.7% paraformaldehyde in TBS for 20 min at room temperature, permeabilized for 30 min using TBS containing 0.1% Triton X-100 and 3% bovine serum albumin, and washed twice with TBS. Slides were then treated with 0.1 mg/ml RNase for 1 h at 37°C. Incubation with 1:1000 dilution of anti-p65 antibody was performed for 1 h at 37°C followed by washing two times with TBS, and then incubation was carried out for 1 h with anti-rabbit IgG conjugated to fluorescein (Santa Cruz Biotechnology) at 1:200 dilution. The slides were rinsed and exposed to propidium iodide for 2 h at room temperature, and coverslips were mounted for viewing by a laser scanning confocal microscopy (Fluoview FV500, OLYMPUS, Tokyo, Japan). . Cells were cultured in depletion medium alone or in medium containing either 5 ng/ml GM-CSF or 4 ng/ml IL-3. After 12 h, cells were lysed and assayed for reporter activity. Luciferase activities were normalized to RLU per g of protein. Ba/F3wild cells were transiently transfected with NF-B-Luc and cultured in depletion medium for 1 h with the indicated doses of PD98059 or wortmannin or left untreated (B). Cells were then stimulated with 5 ng/ml GM-CSF or left unstimulated for another 12 h, and reporter activities were assayed. Luciferase activities were normalized to RLU per g of protein and indicated as fold activation as compared with those of nontreated cells. The structures of wild-type, Y6, and Fall mutants of the human GMR are shown (C). ␣ and ␤ c are the ␣ and common ␤ subunits of the GMR, respectively. Y and F indicate tyrosines and introduced phenylalanines, respectively, and their amino acid positions on the ␤ c are indicated. The position of each amino acid residue is also numbered in parentheses as 1 through 8 from the membrane proximal site in order. Ba/F3-wild, -Y6, and -Fall cells were transiently transfected with NF-B-Luc and cultured as described in panel A (D). Luciferase activities were normalized and indicated as fold activation as compared with those of nontreated Ba/F3-wild cells. All results are the mean Ϯ S.D. of 3 independent experiments. Parental Ba/F3, Ba/F3wild, -Y6, or -Fall cells were cytokine-deprived for 5 h and then stimulated with 4 ng/ml IL-3 (parental Ba/F3) or 5 ng/ml GM-CSF (Ba/F3wild, -Y6, or -Fall) for the indicated time periods, and the total RNA was extracted (E). Twenty g of each RNA sample was subjected to 1% agarose/formaldehyde gel electrophoresis, and then Northern blot analysis of IL-6 genes was performed (top) using the [␣Ϫ 32 p]dCTP-labeled probe corresponding to the nucleotides 16 -761 of the mouse IL-6 cDNA. Equal amounts of loading were verified by ethidium bromide staining of 28 and 18 S rRNAs on the gel (bottom). of the h GMR (Ba/F3-wild cells, Fig. 1C) (16) because signals evoked by the hGMR in these cells are able to substitute for those of endogenous IL-3R. To test whether NF-B-dependent gene expression is induced by IL-3 or GM-CSF, both Ba/F3-wild and parental Ba/F3 cells were transiently transfected with an NF-B-inducible reporter plasmid NF-B-Luc and cultured in the absence or the presence of these cytokines, and then cellular lysates were assayed for luciferase activities. As shown in Fig. 1A, treatment with IL-3 induced nearly 3-fold induction of reporter gene expression in both of these cells. In contrast, treatment with GM-CSF resulted in 5-fold increase of reporter gene expression only in Ba/F3-wild cells. These results suggested that not only the endogenous IL-3R but also transfected hGMR could activate NF-B-dependent gene expression in Ba/F3 cells and that cells expressing hGMR provide a suitable tool for analyzing this ␤ c -dependent NF-B activation.

GM-CSF as
As described previously, the ␤ c , upon ligand binding, activates multiple distinct signals such as the MEK/ERK/MAPK cascade, the PI3K pathway, and STAT5 (6,7,11). Because it has been revealed that both the MEK/ERK/MAPK cascade and PI3K pathway activate NF-B in certain conditions (31, 46 -52), the involvement of these pathways in NF-B activation in our systems was tested using specific inhibitors for each pathway. As shown in Fig. 1B, neither the MEK1 inhibitor PD98059 nor PI3K inhibitor wortmannin did affect GM-CSF-dependent induction of NF-B-responsive reporter gene expression. The bioavailability of PD98059 was verified by its 50 to 60% of inhibitory effect on AP-1-inducible reporter gene activation in response to GM-CSF (data not shown). Similarly, both PI3K activity and phosphorylation of Akt, a downstream target of PI3K, were completely abrogated by wortmannin at the concentrations used in the present study in Ba/F3 cells (53). These results indicate that NF-B activation is mediated through neither MEK/ERK/MAPK nor PI3K pathway. To further characterize the signal(s) responsible for NF-B activation, we next utilized the sublines of Ba/F3 cells expressing the mutated ␤ c with the intact ␣ subunit, Ba/F3-Y6 and -Fall cells (Fig. 1C). Accumulating data have revealed the essential role of the ␤ c in transmitting downstream signals, most of which are distinctively regulated through the phosphorylation of eight tyrosine residues on the ␤ c . Ba/F3-Y6 cells, in which only Tyr-750 out of eight tyrosine residues on the ␤ c remains intact, are a representative clone that activates STAT5 without eliciting MAPK or PI3K activation in the presence of GM-CSF (16) (Fig. 1C). Ba/F3-Fall mutant cells, in which all eight tyrosine residues are mutated to phenylalanines, hardly activate either MAPK, the PI3K pathway, or STAT5 (16) (Fig. 1C). As shown in Fig.  1D, treatment with IL-3 activated expression of the NF-Bresponsive reporter gene not only in Ba/F3-wild but in Ba/ F3-Y6 and -Fall cells as well. In contrast, treatment with GM-CSF stimulated reporter gene activation in Ba/F3-wild and -Y6 cells but not in Ba/F3-Fall cells (Fig. 1D). These results suggest that a certain signal involving neither MAPK nor PI3K pathway but involving phosphorylation of at least Tyr-750 on the ␤ c is essential for the ␤ c -dependent NF-B activation. Together with the potent ability of Ba/F3-Y6 cells but not Ba/F3-Fall cells to activate STAT5 (16), the involvement of STAT5 in this cytokine-dependent NF-B activation was strongly suggested.
We next asked whether induction of NF-B activity may contribute to expression of endogenous target genes in Ba/F3 cells. For this purpose, we examined mRNA expression of proinflammatory cytokine IL-6, gene expression of which is regulated by NF-B in various cell types and known to be induced by IL-3 in Ba/F3 cells (54). Moreover, analysis of the IL-6 gene seems appropriate for monitoring NF-B-driven gene expression in the present study because the promoter region of the IL-6 gene is lacking STAT5 response element (55). As shown in Fig. 1E, the mRNA level of IL-6 was strongly induced within 3 h in response to IL-3 treatment in parental Ba/F3 cells. Furthermore, similar induction of IL-6 mRNA expression was observed after a 3-h stimulation of GM-CSF in Ba/F3-wild and -Y6 cells. In contrast, treatment with GM-CSF did not induce IL-6 mRNA expression in Ba/F3-Fall cells, reflecting the incapability of this Fall mutant receptor to induce the activation of NF-B. Taken together, these results clearly indicate that the ␤ c -dependent NF-B activation leads to expression of endogenous target genes including IL-6.
STAT5 Activates NF-B in Ba/F3 Cells-Next, we explored the possibility that STAT5 activation is involved in cytokinemediated NF-B activation. For this purpose, we cotransfected into Ba/F3-wild cells an expression plasmid for WT STAT5A together with the reporter plasmids for either NF-B or STAT5. As shown in Fig. 2A, WT STAT5A amplified GM-CSFdependent induction of NF-B activity, suggesting the primary role of STAT5 in NF-B activation in these cells (left). The . Another clone of Ba/F3 cells stably expressing STAT5A1*6 (Ba/F3-STAT5A1*6) was also transfected with the same amount of NF-B-Luc or ␤-casein-Luc. Cells were cultured in the depletion medium for 12 h and subjected to assays for reporter activities. In the case of AP-1-Luc, cells transfected with an empty vector were also cultured in GM-CSFcontaining medium. Luciferase activities were assayed and indicated as fold activation as compared with those of cells transfected with empty vector and left nonstimulated. Results are the mean Ϯ S.D. of 3 independent experiments. cotransfection efficiency of these plasmids was confirmed by the enhancement of ␤-casein promoter activity induced by the presence of WT STAT5A in a dose-dependent fashion ( Fig. 2A,  right). The effect of STAT5B, an isoform of STAT5A, was also examined, and the obtained results were quite similar to that of STAT5A (data not shown).
To bypass the effects of the other signaling cascades triggered by the ␤ c rather than STAT5 activation, we also analyzed the effect of a constitutive active mutant of STAT5, STAT5A1*6, on NF-B activation in the absence of cytokine treatment. STAT5A1*6 has two amino acid substitutions in the WT STAT5A backbone, one in the DNA binding domain (H298R) and the other in the transactivation domain (S710F) (18). This mutant is constitutively phosphorylated on its tyrosine residue, localizes in the nucleus, and is transcriptionally active in the absence of growth factor stimulation (18). Efficient introduction of this strong transcriptional activator was verified by the fact that ␤-casein promoter activity was induced in a dose-dependent manner of STAT5A1*6 even in the absence of cytokine stimulation (Fig. 2B, top right). Of note, overexpression of STAT5A1*6 significantly induced NF-B activation despite the absence of cytokine stimulation (Fig. 2B, top left). We also investigated another subline of Ba/F3 cells stably expressing STAT5A1*6 (Ba/F3-STAT5A1*6), which are able to survive and proliferate without upstream cytokine signals (18). As expected, Ba/F3-STAT5A1*6 cells exhibited strong NF-B-dependent transcription in parallel with the activation of STAT5dependent reporter gene expression (Fig. 2B, top). Analysis of AP-1-driven reporter gene expression revealed that transcriptional activity of AP-1 was induced by GM-CSF stimulation but not by STAT5A1*6 (Fig. 2B, bottom). Taken together, we may conclude that STAT5 is essential for the ␤ c -mediated NF-B activation in Ba/F3 cells.
STAT5 Enhances the DNA Binding Activity of NF-B without Influencing Its Subcellular Localization-To investigate the DNA binding activity of NF-B and STAT5 in Ba/F3 cells, EMSA was performed. As shown in Fig. 3A, two complexes were observed between NF-B oligonucleotide probe and the nuclear extracts prepared from GM-CSF-treated Ba/F3-wild cells (lane 1). These complexes were sequence-specific because not nonspecific (lane 5) but specific competitor DNA (lanes 2-4) diminished the complex formation. Supershift assays revealed that anti-p65 antibody shifted the upper band (lane 6) and anti-p50 antibody shifted both upper and lower bands (lane 7). Antibodies against other Rel family proteins or Bcl-3, a member of the IB family proteins, did not affect these complex formations (lanes 8 -11), indicating that the upper complex is a heterodimer of p65/p50 and the lower complex is a p50/p50 homodimer. We obtained similar results when using the nuclear extracts from Ba/F3 cells after IL-3 stimulation (Fig. 3B,  lanes 5, 10, and 15, and data not shown). Note that the anti-STAT5 antibody, which recognizes both STAT5A and STAT5B, could not supershift either of these NF-B complexes (Fig. 3A,  lane 12). We also demonstrated the formation of protein-DNA complex between STAT5 oligonucleotide probe and the nuclear extracts from GM-CSF-treated Ba/F3-wild cells (Fig. 3A, lanes  14 -20).
Based on these experimental settings, we examined the timedependent alteration of NF-B DNA binding in response to GM-CSF stimulation in Ba/F3-wild, -Y6, and -Fall cells. As shown in Fig. 4B, the DNA binding activity of STAT5 was hardly detectable at the time point of stimulation in any of these cells (top panels, lanes 1, 6, and 11), indicating the rapid attenuation of STAT5 DNA binding activity after the removal of IL-3. When stimulated with GM-CSF, the DNA binding activity of STAT5 reached the maximum level within 15 min and was sustained for at least 3 h in both Ba/F3-wild (lanes 1-4) and -Y6 (lanes 6 -9) cells, whereas induction of STAT5 DNA binding was hardly observed in Ba/F3-Fall cells (lanes [11][12][13][14]. When assayed for NF-B, all of these cells showed a substantial amount of NF-B DNA binding activity in the nuclear extracts even after a 5-h deprivation of IL-3 (bottom panels, lanes 1, 6, and 11). After GM-CSF stimulation, the DNA binding activity of NF-B was induced ϳ2-2.5-fold at 1 and 3 h after stimulation in Ba/F3-wild (bottom panel, lanes 1-4) and -Y6 cells (lanes 6 -9). In Ba/F3-Fall cells, however, the DNA binding activity of NF-B seemed weaker at 3 h than that at the time point of stimulation (lanes [11][12][13][14]. Treatment with IL-3 resulted in an increase in DNA binding of NF-B in any of these cells (lanes 5, 10, and 15). These observations indicate that cytokine receptor-mediated STAT5 activation enhances the DNA binding activity of NF-B.
Regulation of NF-B activity in most cell types involves the targeted phosphorylation and degradation of IB proteins, allowing subsequent nuclear translocation of NF-B (see the Introduction). To test this in Ba/F3 cells, Ba/F3-wild cells were deprived of IL-3 for 5 h, stimulated with GM-CSF, and processed for immunocytochemical staining of p65 subunit of NF-B. As a control experiment, cells were treated with TNF␣ and examined in parallel. In TNF␣-treated cells, drastic nuclear translocation of p65 was observed after 15-30 min of stimula-  -10), and -Fall (lanes 11-15) cells were deprived of cytokine for 5 h and restimulated with either 5 ng/ml GM-CSF or 4 ng/ml IL-3 (B). After the indicated time periods, nuclear extracts were isolated and subjected to EMSA using the oligonucleotide probes for STAT5 (top panels) and NF-B (bottom panels). tion (Fig. 4A, right, and data not shown). In contrast, GM-CSFtreated Ba/F3-wild cells exhibited no detectable nuclear translocation of p65 upon stimulation (Fig. 4A, left). These striking differences were further confirmed by Western blot analysis of NF-B p65. Ba/F3-wild cells were stimulated with GM-CSF or TNF␣ as above, and cytosol and nuclear extracts were prepared. Fig. 4B shows that p65 proteins in the nuclear extracts of TNF␣-treated cells were apparently increased after 15 min of stimulation and then decreased within 180 min (Fig. 4B,  right). In clear contrast, p65 protein levels remained almost constant even after GM-CSF stimulation (left). Verification of our fractionation procedure was confirmed by Western blot using anti-Sp1 antibody for nuclear extracts and anti-␤ c antibody of the GMR for the cytosol (data not shown). Note that the presence of p65 protein in the nuclear extracts even before cytokine stimulation is consistent with the previous results showing substantial fluorescence signal of p65 staining (Fig.  4A) and the apparent DNA binding activity of NF-B (Fig. 3B). These results suggest that the ␤ c -dependent NF-B activation is not mediated through the classical activation pathway involving the nuclear translocation of NF-B. This was further supported by the following experiments in which the phosphorylation and degradation of IB proteins were analyzed; both rapid phosphorylation of IB␣ and subsequent decrease of both IB␣ and IB␤ proteins were detected in whole cell extracts from TNF␣-stimulated cells (Fig. 4C, right). However, phosphorylation of IB␣ was not observed, and the protein levels of both IB␣ and IB␤ remained unchanged after GM-CSF stimulation (Fig. 4C, left). There remains the possible mechanism that both inducible but undetectable phosphorylation and rapid resynthesis of IB proteins may occur, promoting the nuclear translocation of NF-B, whereas the gross protein amount of IBs remains constant. To test this, we also analyzed the time-dependent alteration of IB␣ and IB␤ in the presence of the protein synthesis inhibitor cycloheximide in IL-3-or GM-CSF-treated Ba/F3-wild cells; however, the elimination of interference by newly synthesized IB proteins did not explore any degradative process of these proteins (data not shown).
These somewhat curious findings that the ␤ c -dependent NF-B activation involves enhancement of DNA binding activity without affecting its subcellular localization prompted us to precisely investigate the effects of STAT5 on NF-B DNA binding activity. Previous reports described that when Ba/F3 cells are deprived of cytokine, the DNA binding activity of NF-B slowly declines (22). Thus, we also tested whether the decrease in NF-B DNA binding following the cytokine deprivation was affected by STAT5 activation. As shown in Fig. 5A, both STAT5 and NF-B are recovered as DNA-bound complexes in both Ba/F3-wild and -Fall cells cultured in the presence of IL-3 (top part, lanes 1 and 5). The DNA binding activity of NF-B was hardly discernible after a 9-h deprivation of IL-3 (lanes 4 and  8). Longer deprivation resulted in total loss of the DNA binding, although cell viability was seriously impaired (data not shown). In the case of STAT5, deprivation of IL-3 rapidly decreased DNA binding in both of these cells (lanes 4 and 8 and data not shown). Interestingly, once these cells were deprived of IL-3 but complemented with GM-CSF just after IL-3 re-

FIG. 4. Nuclear translocation of NF-B is not involved in GM-CSF-dependent NF-B activation.
Ba/F3-wild cells were stimulated with 5 ng/ml GM-CSF (left) or 20 ng/ml TNF␣ (right) after 5 h of cytokine deprivation (A). After the indicated time periods, cells were fixed and processed for immunocytochemical analysis. Subcellular localization of p65 was analyzed by indirect immunocytochemical staining using anti-p65 antibody as described under "Experimental Procedures." FITC, subcellular localization of p65 with fluorescein isothiocyanate; PI, nuclear staining with propidium iodide; Merged, the merged image of both. Ba/F3-wild cells were stimulated with GM-CSF (left) or TNF␣ (right) as described above (B). After the indicated time periods, cells were harvested, and both cytosol and nuclear extracts were isolated. Ten g of each was separated on a 10% SDS-polyacrylamide gel and immunoblotted with anti-p65 antibody. Ba/F3wild cells were stimulated with GM-CSF (left) or TNF␣ (right) and harvested at the indicated time points, and whole cell extracts were prepared (C). Ten g of each was separated on a 10% SDS-polyacrylamide gel and immunoblotted with antiphosphorylated IB␣ (p-IB␣) antibody, and the protein blot was reprobed with anti-IB␣ antibody. As a control, whole cell extracts from cells treated with TNF␣ for 5 min were loaded on the same gel (lane 8). Immunoblotting with anti-IB␤ antibody was also performed. N.S., non-specific.  2 and 6). In clear contrast, although GM-CSF was complemented, preformed NF-B-DNA complexes in the absence of IL-3 were not sustained in Ba/F3-Fall cells (lane 7), suggesting the involvement of STAT5 in the preservation of the ␤ c -dependent interaction between NF-B and DNA. Because protein levels of p65 and p50 appeared to be constant during the experiments (Fig. 5A, bottom part), it was suggested that this alteration in NF-B DNA binding activity is not due to a decrease of p65/p50 proteins. Moreover, subcellular localization of p65 in either of these cells was unaffected after a 9-h deprivation of IL-3 as assessed by Western blot analysis (Fig. 5B), indicating that STAT5-dependent signals downstream of the hGMR positively modulate the DNA binding activity of NF-B without influencing either the subcellular sequestration or protein amounts of p65/p50 complexes. To examine the effect of STAT5 more directly, we analyzed the effect of STAT5A1*6 on the DNA binding activity of NF-B. When the nuclear extract was prepared just after electroporation of empty vector alone, the baseline NF-B DNA binding was significantly detectable (Fig. 5C, bottom panel, lane 1). Relative loss of DNA binding of STAT5 in the same nuclear extract is reasonable because the cells had to stay cytokine-free for nearly 1 h during the procedures of cytokine removal before electroporation (top panel, lane 1). The DNA binding of NF-B then gradually decreased to the minimum level through 9 h of culture in the depletion medium (lanes 1-4). In contrast, when STAT5A1*6 was overexpressed, not only STAT5 but also NF-B DNA binding was observed until 9 h after electropora-tion even in the absence of IL-3 (top and bottom panels, lanes [5][6][7][8]. Thus, we may conclude that STAT5 is able to preserve the DNA binding activity of NF-B within the nucleus. We next tested the dose effect of STAT5 on the DNA binding activity of NF-B. For this purpose, vector alone or indicated amounts of the expression plasmid for WT STAT5A were transfected into Ba/F3-wild cells. Cells were cultured in the presence or the absence of GM-CSF for 9 h, and then EMSA was performed. The top panel of Fig. 5D represents that the DNA binding activity of STAT5 increased in proportion to increasing amounts of the expression plasmids for WT STAT5A. In contrast, the DNA binding activity of NF-B did not show a correspondent increase with the amounts of transfected plasmids (bottom panel). Note that these doses of the expression plasmids enhanced expression of NF-B-responsive reporter gene in a dose-dependent manner (Fig. 2A). These results, therefore, may argue the presence of an additional mechanism that enhances the transcriptional function of nuclear NF-B after DNA binding.
STAT5 Up-regulates Transactivational Function of p65-To test this hypothesis, we performed one hybrid assay to evaluate whether the transactivational function of p65 is also affected by STAT5. We constructed the indicated plasmids expressing GAL4 fusion proteins with either the N-or C-terminal half of p65 (Fig. 6A). We then transfected each of these plasmids, the GAL4-reporter plasmid, and either expression plasmid for WT STAT5A or STAT5A⌬749 into Ba/F3-wild cells, and cells were treated with GM-CSF as indicated (Fig. 6B). As in the case of GAL4 alone, the N-terminal half of p65 did not transactivate reporter gene expression in the presence of either GM-CSF or WT STAT5A (Fig. 6B). However, reporter gene expression in  4 and 8). Cells were harvested after 9 h of complementation as well as before substitution of cytokines (lanes 1 and 5). Cells were collected, and then cytosol and nuclear extracts or whole cell extracts were isolated. Nuclear extracts were subjected to EMSA using the oligonucleotide probes for STAT5 and NF-B (A, top part), and whole cell extracts were separated on a 10% SDS-polyacrylamide gel and immunoblotted with anti-p65 or anti-p50 antibody (A, bottom part). Cytosol and nuclear extracts of Ba/F3-wild (B, top panel) and -Fall (B, bottom panel) cells were also separated on a 10% SDS-polyacrylamide gel and immunoblotted with anti-p65 antibody. C, parental Ba/F3 cells, grown in the culture medium containing IL-3 were deprived of IL-3 and transiently transfected with 1 g of either an empty vector (lanes 1-4) or the expression plasmid encoding STAT5A1*6 (lanes 5-8). After transfection, cells were cultured in the depletion medium and harvested at the indicated time points. Nuclear extracts were prepared and subjected to EMSA for STAT5 (top panel) and NF-B (bottom panel). D, Ba/F3-wild cells, grown in the culture medium containing IL-3, were deprived of IL-3 and transiently transfected with either 1 g of empty vector (lanes 1 and 2) or the indicated amounts of expression plasmid encoding WT STAT5A (lanes [3][4][5]. Cells transfected with empty vector were cultured in the absence (lane 1) or presence (lane 2) of 5 ng/ml GM-CSF, and those transfected with WT STAT5A expression plasmids were cultured in the presence of GM-CSF (lanes 3-5). After 9 h of culture, cells were harvested, and nuclear extracts were prepared and subjected to EMSA for STAT5 (top panel) and NF-B (bottom panel).
the presence of the C-terminal half of p65 was significantly increased when WT STAT5A was overexpressed (Fig. 6B), indicating that the transactivational function of NF-B p65 is augmented by STAT5A. Moreover, this effect of STAT5 was supposed to be ascribed to the C-terminal transactivation domain because coexpression of STAT5A⌬749 exhibited no detectable enhancement of reporter gene expression (Fig. 6B).
Both DNA Binding Activity and Transactivational Functions of STAT5 Are Required for NF-B Activation-To further elucidate the mechanism for STAT5-dependent NF-B activation, we used several mutants of STAT5 and performed cotransfection experiments. STAT5AVVV mutant, in which amino acid residues Val-Val-Val (amino acids 466 -468) within the DNA binding domain of STAT5A are replaced by Ala-Ala-Ala, lacks DNA binding activity and is transcriptionally inactive (42). STAT5A⌬749, in which the C-terminal 44 amino acids containing the transactivation domain are deleted, exerts dominantnegative effects on wild-type STAT5-induced transcription despite retaining its DNA binding ability (56). As shown in Fig. 7, neither of these mutants was able to augment NF-B activation in response to GM-CSF, suggesting that NF-B activation requires both STAT5 DNA binding and transactivation domains. These data raised the possibility that STAT5 induces transcription of a certain gene, the product of which in turn enhances NF-B function. We tried to confirm this possibility by using a protein synthesis inhibitor cycloheximide on GM-CSF-induced NF-B activation but failed because cycloheximide itself was a potent inducer of NF-B DNA binding as estimated by EMSA in Ba/F3 cells (data not shown). Therefore, we took another approach using STAT5⌬750VP16JAK2, which consists of the JAK2 kinase domain and mammary gland factor, the sheep homologue of STAT5, the transactivation domain of which is replaced by that of VP16 (40). This chimeric protein constitutively activates STAT5-dependent transcription in the absence of upstream cytokines (40). Interestingly, STAT5⌬ 750VP16JAK2 could induce NF-B activation even in the absence of cytokine stimulation to the level comparable with that of STAT5A1*6 (Fig. 7). This exchangeability of the transactivation domain of STAT5 may support the indirect mechanism for NF-B activation, i.e. via induction of as yet unidentified target gene expression.
STAT5-dependent Production of a Humoral Factor(s) Leads to NF-B Activation in Ba/F3 Cells-Cytokine-mediated STAT5 activation results in expression of several target pro-teins such as Pim-1, CIS, and OSM (8,19,54). One recent report described that OSM stimulated tissue factor expression in vascular smooth muscle cells and that this transcriptional regulation is primarily mediated through an IB-independent mechanism of NF-B activation (57). Therefore, we hypothesized that STAT5-dependent gene expression and secretion of certain soluble factors such as OSM may participate in NF-B activation in Ba/F3 cells. To investigate such a possibility, the culture supernatant of Ba/F3 cells was examined for the capability to activate NF-B-dependent transcription. We used the culture supernatant of Ba/F3-STAT5A1*6 cells as a conditioned medium (CM STAT5A1*6) because these cells are able to survive and proliferate in cytokine-deprived medium, and therefore, we could eliminate the carry-over of IL-3, GM-CSF, and any factors secreted via other signaling events downstream of the ␤ c . Interestingly, as shown in Fig. 8A, CM STAT5A1*6 alone could stimulate NF-B activity to a level comparable with that observed after IL-3 stimulation in paren- tal Ba/F3 cells (top). This biological activity of CM STAT5A1*6 is supposed to be mediated through a protein factor(s), because heat inactivation of CM STAT5A1*6 at 65°C for 1 h nearly abolished the activity. Similar properties of inducing NF-Bdependent transcription were also observed in the CM of several different clones of Ba/F3-STAT5A1*6 or Ba/F3 cells in which another form of constitutive active mutant STAT5A-N642H (58) is stably transfected (data not shown). An autocrine mode of NF-B activation through secretion of IL-3 or GM-CSF was not likely because CM STAT5A1*6 induced neither STAT5-nor AP-1-dependent transcription as determined by reporter gene expressions (Fig. 8A, middle and bottom panels). We next tested the possibilities of several other humoral protein factors to be involved in this NF-B activation. As shown in Fig. 8B, however, neither IL-6 nor OSM could induce NF-B activation. In addition, IL-1␤, widely known as a potent inducer of NF-B in multiple cell types, also had no effect (top). TNF␣ could stimulate NF-B-dependent reporter activities to a  (5, 20, or 80 ng/ml) of either IL-6, OSM, IL-1␤, or TNF␣ was added to the depletion medium, or cells were cultured in the CM Ba/F3-STAT5A1*6 (top). The effects of anti-mouse TNF␣ antibodies (Ab) were examined by adding the indicated amounts of neutralizing antibodies or control IgG (asterisks) 2 h prior to cellular stimulation with 20 ng/ml TNF␣ or CM Ba/F3 STAT5A1*6 (bottom). Luciferase activities were assayed and indicated as fold activation as compared with those of cells cultured in the depletion medium. Parental Ba/F3 cells were deprived of IL-3 for 5 h, collected by centrifugation, and then resuspended in CM Ba/F3-STAT5A1*6 (C and D). Total RNA and whole cell extract were prepared after the indicated time periods, and Northern blot analysis of IL-6 mRNA (B) and immunoblot analysis of IB␣ proteins (D) were performed. N.S., non-specific. level comparable with that induced by IL-3 or CM STAT5A1*6 (top); however, involvement of this cytokine in STAT5-dependent NF-B activation seemed unlikely because the mechanism of NF-B activation induced by TNF␣ was distinct from that induced by the ␤ c -mediated signals in Ba/F3 cells (Fig. 5). In agreement with this, CM STAT5A1*6-dependent NF-B activation was not blocked by the addition of neutralizing antibodies against TNF␣ (Fig. 8B, bottom), and furthermore, enzymelinked immunosorbent assay revealed no detectable amount of TNF␣ in the CM STAT5A1*6 (data not shown). Strikingly, we found that this putative protein factor(s) in CM STAT5A1*6 could activate not only reporter gene expression but also transcription of endogenous NF-B target genes such as IL-6 ( Fig.  8C). It should be emphasized that the time-dependent increase of IL-6 mRNA in response to CM STAT5A1*6 was earlier than that observed in the case of IL-3 or GM-CSF stimulation (compare Fig. 8C with Fig. 1E). These results strongly support our hypothesis that the ␤ c -mediated NF-B activation is in some extent dependent on a secondary mechanism involving STAT5dependent transcription and de novo protein synthesis. In Western blot analysis, we again found that phosphorylation and subsequent degradation of IB proteins are not affected in this soluble factor-dependent NF-B activation process (Fig.  8D). We, therefore, may conclude that NF-B activation downstream of the ␤ c is mediated at least in part through STAT5dependent transcription and the production of a(n) unknown humoral factor(s) that in turn evokes a quite unique mode of NF-B activation.
Humoral Factor(s) Produced by Ba/F3-STAT5A1*6 Could Activate NF-B-dependent Transcription in WEHI 3B Cells-Finally, we examined whether STAT5 may induce NF-B activation in different cell types other than Ba/F3 cells. When myelomonocytic leukemic WEHI 3B cells were transiently transfected with an expression vector encoding STAT5A1*6 along with reporter plasmid ␤-casein-Luc, significant induction of reporter gene expression was observed (Fig. 9A, left). In contrast, transfection of the same amount of STAT5A1*6 expression plasmid did not augment but rather suppressed NF-B-dependent reporter gene activation (right). These results indicate that the mechanisms by which STAT5 up-regulates NF-B activation in Ba/F3 cells may not be generally conserved in a variety of cell types. Indeed, it was recently reported that prolactin-activated STAT5 inhibits NF-B-mediated signaling through a mechanism of competition for limiting transcriptional coactivators such as p300 (59). However, when WEHI 3B cells were stimulated with CM Ba/F3-STAT5A1*6 and assayed for reporter gene activity, an ϳ4-fold induction of NF-Bdriven reporter gene expression was observed (Fig. 9B). Thus, it is strongly proposed that this as yet undetermined soluble factor(s) secreted from certain cell types such as Ba/F3 cells may serve as an inducer(s) of NF-B-dependent gene expression in not only Ba/F3 cells but also other cells including WEHI 3B cells. DISCUSSION Here we show that signals downstream of the ␤ c lead to the activation of the transcription factor NF-B. To date, it is well characterized that the ␤ c , upon ligand binding, activates JAK2 and then provokes several distinct and experimentally separable signals such as the MEK/ERK/MAPK cascade, the PI3K pathway, and STAT5 (6,7,11). Therefore, we first attempted to determine the responsible signaling cascade that links the ␤ c to the NF-B activation pathway because several previous studies revealed that both ERK/MAPK and PI3K pathways functionally up-regulate the activity of NF-B in various conditions (31, 46 -52). In addition, one recent report showed that erythropoietin receptor-mediated JAK2 activation leads to NF-B activa-tion in neuronal cells (60). In our case with Ba/F3 cells, however, involvement of either the MEK/ERK/MAPK cascade or the PI3K pathway seemed unlikely because specific inhibitors for each pathway had no effect on NF-B activation. In addition, our observation that Ba/F3-Y6 cells did induce NF-B activation further supports this because this mutant receptor was previously shown to lack the potential to elicit MAPK or PI3K activation in the presence of GM-CSF (16). The possibility that JAK2 directly affect the NF-B signaling was also excluded by the observation that the Fall mutant receptor could not activate NF-B despite its retaining the potential to activate JAK2 (16). Instead, our results strongly suggested the involvement of STAT5, which was further supported by the observations that this NF-B activation is augmented by overexpression of wild-type STAT5. Furthermore, our findings that the constitutively active mutant STAT5A1*6 induced NF-B activation even in the absence of cytokines lead us to conclude the essential role of STAT5 in activating NF-B. Thus, we are able to indicate that whether signaling events such as JAK2, MEK/ERK/MAPK, or PI3K activation interact with NF-B activation pathway may vary depending on the cell type or the type of upstream stimulus.
Recent studies have revealed that STAT5 functionally interacts with other transcription factors. These include several distinct mechanisms by which STAT5 suppresses the function of other transcription factors such as glucocorticoid receptor (GR) (61) and peroxisome proliferator-activated receptor-␣ (62,63): the direct protein-protein interaction-mediated mechanism for the former and the yet unidentified indirect mechanism for the latter. In addition, it has also been revealed, in contrast to our present study, that prolactin-activated STAT5 inhibits NF-B-mediated signaling through a mechanism of competition for limiting transcriptional coactivators such as p300 (59). In the present study, we were unable to detect a physical association between STAT5 and NF-B, as assessed by EMSA supershift analysis and pull-down assays using in vitro translation products ( Fig. 3A and data not shown), suggesting a mechanism distinct from the previously described STAT5-GR association. In contrast, our data strongly indicate an indirect mechanism involving as yet unidentified humoral factor(s) in STAT5-dependent NF-B activation. At this moment, therefore, it remains unclear whether the reported STAT5dependent NF-B suppression mechanism via, for example, the squelching of transcriptional coactivators may be present in Ba/F3 cells. Because we reproduced the suppressive effects of STAT5A1*6 on NF-B-dependent reporter gene expression in WEHI 3B cells (Fig. 9), competition for limiting coactivators between STAT5 and NF-B may play a role in STAT5-mediated inhibition of NF-B-mediated signaling. Alternatively, enhancement of NF-B function in our systems may be considered as the net effect of the opposite influences of STAT5 on NF-B activation pathway: the inhibitory effect caused by protein-protein interaction between coactivators and these transcription factors and the promoting effect possibly mediated through intervening secondary humoral factor(s). In this regard, identification of this factor(s) responsible for activating NF-B in Ba/F3 cells would provide insight into the mechanism of cross-talk between STAT5 and NF-B signal transduction pathways in various cells.
In this report, we also addressed which step of the NF-B activation pathway is influenced by the ␤ c -mediated STAT5 activation. Ba/F3 cells exhibit a substantial amount of nuclear NF-B composed of p65/p50 under usual culture conditions as seen in other cells of B cell lineage (64 -66) or certain malignant cells (67). We showed that STAT5 activation leads to enhancement of the DNA binding activity of NF-B without its nuclear translocation. In addition, our results also suggested that STAT5 up-regulates the transactivational potential of p65 even after the DNA binding of NF-B is saturated. This mode of NF-B regulation within the nucleus displays a striking contrast with that observed when Ba/F3 cells are stimulated with TNF␣, which induces the rapid phosphorylation and subsequent degradation of IB proteins. Consistently, a growing body of evidence has suggested that NF-B-dependent gene expression is regulated not only via this well characterized IB-dependent mechanism but also at several regulatory steps within the nucleus (31)(32)(33)(34)(35)68). One possible mechanism of these nuclear regulations is modulatory phosphorylation of NF-B p65 because both DNA binding (69) and the transactivation potential of NF-B (33,51,70,71) are positively regulated by phosphorylation of p65 by such kinases as cAMP-dependent protein kinase (33), CK2 (72), protein kinase C (71), and IB kinases (73). Thus, it is suggested that STAT5-dependent signals may consequently target the catalytic activity of these kinases. The second possible mechanism is that STAT5dependent signals may up-regulate NF-B via a redox-dependent mechanism because it has been shown that DNA binding activity of NF-B is modulated by oxidation-reduction in vitro (74). Furthermore, we demonstrated that a redox-related factor, Ref-1, could enhance the DNA binding activity of NF-B without affecting the degradation of IB, nuclear translocation of NF-B, or phosphorylation status of NF-B proteins (35,38). Thirdly, STAT5-dependent signals may regulate the function of transcriptional coactivators because NF-B-dependent gene expression involves a growing number of such coactivators (34,75,76), and recent investigations have revealed that not only recruitment but also inducible activation of coactivators plays an important role in the efficient induction of transcription (77)(78)(79). Although the precise mechanisms of our described nuclear regulation of NF-B remain undetermined, it should be emphasized that the activation of NF-B is regulated through different modes of signaling pathways within the same cell. In this regard, it would be essential to identify and characterize the putative soluble factor for a better understanding of not only a new biological function exerted by STAT5 but also a unique mode of the NF-B activation pathway distinctively utilized from the classical IB-dependent pathway depending on the type of extracellular stimuli in vivo.
In summary, we demonstrated that signals downstream of the ␤ c induce NF-B activation in murine proB cell line Ba/F3 cells, which is mediated neither by MEK/ERK/MAPK nor by the PI3K pathway but instead exclusively mediated by STAT5. This NF-B activation is suggested to be induced in part through as yet unidentified humoral factor(s) that is expressed depending on the activation of STAT5. Furthermore, STAT5dependent signals confer a unique mode of NF-B activation that is distinct from well characterized mechanisms.