Autoregulation of the Stat3 Gene through Cooperation with a cAMP-responsive Element-binding Protein*

STAT3 (signal transducer and activator of transcription 3) is a key transcription factor mediating the signals for a variety of cytokines, including interleukin-6 (IL-6). The Stat3 gene itself is activated by IL-6 signals. We show that the region of the signal-transducing subunit, gp130, essential for STAT3 activation, is also required for activation of the Stat3 gene. To elucidate the mechanisms activating the Stat3 gene, we identified an IL-6 response element (IL-6RE) in the Stat3gene promoter containing both a low affinity STAT3-binding element and a cAMP-responsive element (CRE). Electrophoretic mobility shift assays showed that IL-6 induced a slowly migrating complex on the IL-6RE containing a STAT3 homodimer and an unidentified CRE-binding protein. With the combination of transient transfection assays using mutantStat3 promoter-reporter constructs and electrophoretic mobility shift assays, we found that the formation of a slowly migrating complex was required for full activation of theStat3 gene. Thus, STAT3 activates the Stat3gene in cooperation with an unidentified CRE-binding protein. This regulatory mechanism is similar to that of the junB gene, which is activated by IL-6 through the junB IL-6RE, which contains a low affinity STAT3-binding site and a CRE-like site.

STAT3 (signal transducer and activator of transcription 3) is a key transcription factor mediating the signals for a variety of cytokines, including interleukin-6 (IL-6). The Stat3 gene itself is activated by IL-6 signals. We show that the region of the signal-transducing subunit, gp130, essential for STAT3 activation, is also required for activation of the Stat3 gene. To elucidate the mechanisms activating the Stat3 gene, we identified an IL-6 response element (IL-6RE) in the Stat3 gene promoter containing both a low affinity STAT3-binding element and a cAMP-responsive element (CRE). Electrophoretic mobility shift assays showed that IL-6 induced a slowly migrating complex on the IL-6RE containing a STAT3 homodimer and an unidentified CRE-binding protein. With the combination of transient transfection assays using mutant Stat3 promoter-reporter constructs and electrophoretic mobility shift assays, we found that the formation of a slowly migrating complex was required for full activation of the Stat3 gene. Thus, STAT3 activates the Stat3 gene in cooperation with an unidentified CRE-binding protein. This regulatory mechanism is similar to that of the junB gene, which is activated by IL-6 through the junB IL-6RE, which contains a low affinity STAT3-binding site and a CRE-like site.
STAT (signal transducer and activator of transcription) proteins have been shown to play pivotal roles in cytokine signaling pathways, which are involved in regulating cell growth and differentiation in systems ranging from Drosophila to mammals (1)(2)(3). In mammals, the STAT family consists of at least six members, designated STAT1-6 (3). Different STAT family members are activated by a variety of cytokines, including interferon-␣/␤, interferon-␥, the interleukin-6 (IL-6) 1 family, growth hormone, erythropoietin, and leptin in a manner dependent on the Janus kinase family of tyrosine kinases (4 -12). Furthermore, several growth factors (such as epidermal growth factor, platelet-derived growth factor, and fibroblast growth factor) and non-receptor tyrosine kinases (including c-Src, c-Abl, and their v-oncogene products) have been shown to activate STAT proteins (13)(14)(15)(16)(17)(18)(19). Thus, three different types of tyrosine kinases, Janus kinases, receptor-type tyrosine kinases, and certain Src family tyrosine kinases, can all activate STAT proteins, probably by directly phosphorylating the tyrosine residue critical for causing the STAT molecules to form homo-or heterodimers through their SH2 domains (20 -25). Tyrosine-phosphorylated STAT dimers enter the nucleus and activate the target genes by binding to their specific target DNA sequences (26). The cis-acting DNA sequences recognized by STAT dimers, the STAT-binding element (SBE), have the general structure TTN 5 AA (1,27). The sequence and the size of the spacer region affect the binding of the respective STAT dimers (27). In certain cases, STAT proteins bind to the target DNA in combination with other DNA-binding proteins. For instance, the STAT3 homodimer forms a complex with p36 CRE-binding protein for an IL-6 response element in the junB gene (JRE-IL6) (28) or with c-Jun for the ␣ 2 -macroglobulin APRE (29). The STAT1 homodimer has been shown to make a complex with Sp1 for the interferon-␥ response element of the intercellular adhesion molecule-1 gene (30). Stocklin et al. (31) showed that the glucocorticoid receptor can act as a coactivator and enhance STAT5-dependent transcription by making complexes with STAT5. STAT dimers even form complexes with other STAT dimers through their amino-terminal regions on the repeated low affinity SBEs in the interferon-␥ gene intron (32). Therefore, in addition to the sequence of a SBE itself, the sequence outside the SBE is also critical for determining the target gene specificity because the complex formation with other proteins or other STAT dimers sometimes changes the binding specificity or increases the affinity of the complex for the DNA (21,28,32). Other important determinants for STAT function might be the intensity and duration of the STAT activity and the amounts of STAT proteins ready to be activated in the cells. All of these factors eventually determine the range of target genes and the duration and intensity of target gene activation and thereby often determine the outcome, such as differentiation, growth, and cell survival or death.
We have previously shown that STAT3 is critical for IL-6induced gene regulation, including the repression of c-myb and c-myc, the induction of junB and IRF1, and IL-6-induced growth arrest and terminal macrophage differentiation in M1 leukemic cells (33,34). Moreover, STAT3 activity is critical for the gp130-mediated anti-apoptotic signal in murine pro-B BaF/ B03 cells (35). To understand how STAT3-mediated signals affect growth, differentiation, and cell death or survival in different types of cells, it is important to study the regulatory mechanisms of Stat3 gene expression itself. In this study, we characterize the signals required for Stat3 gene activation and show that STAT3 activates the Stat3 gene promoter through a * This work was supported by grants-in-aid for scientific research from the Ministry of Education, Science, Sports, and Culture of Japan; the Special Coordination Fund of the Science and Technology Agency of the Japanese Government; the Yamanouchi Foundation for Research on Metabolic Disorders; the Osaka Foundation for Promotion of Clinical Immunology; and the Sankyo Foundation of Life Science. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence reported in this paper has been submitted to the DDBJ and GenBank™/EMBL Data Banks with accession number AB008160.
Northern Blot Analysis-Total RNA was extracted using the TRIzol reagent (Life Technologies, Inc.) according to the procedures recommended by the manufacturer. Total RNA (15 g/sample) was separated by electrophoresis in formaldehyde-containing 1% agarose gels and transferred to Hybond N ϩ nylon membranes (Amersham Corp.). Membranes were hybridized with 32 P-labeled cDNA fragments overnight, washed three times with 0.1ϫ SSC and 0.1% SDS at 56°C, and subjected to autoradiography. The amount of loaded RNA was verified by ethidium bromide staining or by measuring the expression level of CHO-B mRNA. The probes used were the 2.5-kb SalI-BamHI fragment of pBSHA-STAT3 containing a full-length STAT3 cDNA and the 0.6-kb EcoRI-BamHI fragment of CHO-B cDNA (a gift from J. E. Darnell, Jr.).
Isolation of the 5Ј-Flanking Region of the Mouse Stat3 Gene and Plasmid Construction-A mouse Stat3 5Ј-flanking region was obtained by PCR using the Promoter Finder DNA Walking kit (CLONTECH). Two gene-specific primers, 5Ј-CTCAGCGATCCGGTTAGGGCTCGT-T-3Ј (GSP1) and 5Ј-CAGGTTCCCCCTCCCTGTC TACACT-3Ј (GSP2), were made based on the sequence of the mouse Stat3 cDNA reported by Akira et al. (38). Primary PCR and secondary nested PCR were done using the GSP1 and GSP2 primers, the primers (AP1 and AP2) corresponding to the adaptor sequences, and adaptor-ligated genomic libraries. The resulting 2.2-kb PCR fragment was subcloned using the TA cloning vector pCR2.1 (Invitrogen). DNA sequencing was carried out using an automated sequencer, and data were obtained for both strands. A series of 5Ј-deletion mutations of the Stat3 5Ј-flanking region were made by subcloning the EcoRV-BamHI, SspI-BamHI, ScaI-BamHI, AflII-BamHI, StyI-BamHI, and PvuII-BamHI fragments into the reporter construct pSPLuc, containing a luciferase gene and SV40 poly(A) signal inserted into pSP72 between the PvuII and BamHI sites.
Site-directed Mutagenesis by PCR-Mutations were introduced at the STAT3-binding element or the CRE site of the Stat3 IL-6RE in the context of the intact Stat3 promoter with the 2.2-kb upstream region by the overlap extension technique using PCR. The primers used for these mutations were as follows: for mSTAT, 5Ј-ACTGCCAGGTCCTCAGCT-GAGTTTTCAG-3Ј and 5Ј-AGCTGAGGACCTGGCAGTGCGTGACGT-3Ј; for mCRE, 5Ј-GTGTCTTGTCATCACGCACTGCCAGGAACT-3Ј and 5Ј-AGTGCGTGATGACAAGACACTTTGAATGCCCT-3Ј; for the 5Јprimer, 5Ј-CCCAAATGCTTAAGTGGGGTGACACCT-3Ј; and for the 3Јprimer, 5Ј-AGCTCGGATCCACTAGTAACGGCCG-3Ј. The underlined bases are the mutated ones. After the PCR products were subcloned into pCR2.1, the sequences of the PCR products were verified by sequencing the products. The PCR products were digested with SacI and inserted at the proper position of p2166-Stat3-Luc-WT to make p2166-Stat3-Luc-mSTAT and p2166-Stat3-Luc-mCRE. The primers used to make the p478/229-Luc series were as follows: for the 5Ј-primer, 5Ј-AACGCTGCAGTTAAGTGGGGTGACACCTGG; and for the 3Ј-primer, 5Ј-CCAGGTACCCCAAGGGACGCGCAGAGGCC. The PCR products were digested with PstI and KpnI and inserted upstream of the minimal junB promoter at the PstI-KpnI sites of pSPBLuc.
Transient Transfection Assay-For transfection experiments, HepG2 cells were transfected with DNA using the calcium phosphate coprecipitation method (36). Typically, 1.2 g of one of the reporter plasmids containing the firefly luciferase gene and 1 g of pEF-lacZ, an expression vector containing the lacZ gene encoding ␤-galactosidase as an internal control for transfection efficiency, were used. Three g of pCAGGS-Neo (an expression vector without an insert; control) or pCAGGS-NeoHA-STAT1F or pCAGGS-NeoHA-STAT3F (expression vectors containing a cDNA encoding either HA-STAT1F or HA-STAT3F (33), respectively) was cotransfected in some experiments. Cells were incubated with DNA precipitates for 16 h, washed with phosphatebuffered saline, fed Dulbecco's modified Eagle's medium containing 0.1% fetal calf serum for 20 -24 h, and stimulated with 100 ng/ml IL-6 for the last 16 h. Approximately 40 -45 h after transfection, cells were collected in 120 l of lysis buffer and subjected to assays for luciferase and ␤-galactosidase activities as described (28).
Electrophoretic Mobility Shift Assay-This was performed according to the procedure published previously (28). Briefly, nuclear extracts (10 g) were incubated in a final volume of 20 l of 10 mM HEPES, pH 7.9, 80 mM NaCl, 10% glycerol, 1 mM dithiothreitol, 1 mM EDTA, and 100 g/ml poly(dI-dC)⅐poly(dI-dC) with each 32 P-labeled probe (10,000 cpm, 0.5-1 ng) for 20 min at room temperature. The protein-DNA complexes were resolved on a 4.5% nondenaturing polyacrylamide gel containing 2.5% glycerol in 0.25ϫ TBE (1ϫ TBE is 0.13 M Tris base, 0.12 M boric acid, and 2.0 mM EDTA, pH 8.8) at room temperature and autoradiographed. For competition analysis, extracts were preincubated with a 50-or 250-fold molar excess of unlabeled oligonucleotides for 5 min before the addition of labeled oligonucleotide. For antibody interaction studies, antisera or monoclonal antibody specific to several transcription factors was included in the binding reaction during a 30-min preincubation on ice. The antibodies used were as follows: anti-STAT1 N terminus (amino acids 1-194 of human STAT1; anti-ISGF3 G16920) monoclonal antibody (Transduction Laboratories, Lexington, KY); anti-STAT3 C terminus (amino acids 702-770 of human STAT3) polyclonal antibody (28); anti-CREB rabbit polyclonal antibody (9192; New England Biolabs Inc.); anti-CRE-BP1 (ATF2) rabbit polyclonal antibody (SC187X; Santa Cruz Biotechnology, Inc., Santa Cruz, CA); anti-c-Jun/ AP-1 goat polyclonal antibody recognizing the DNA-binding domains of c-Jun, JunB, and JunD (SC44X; Santa Cruz Biotechnology, Inc.); anti-ATF3 rabbit polyclonal antibody (SC188X; Santa Cruz Biotechnology, Inc.); and anti-ATF4 mouse monoclonal antibody (SC244X; Santa Cruz Biotechnology, Inc.). The effectiveness of the antibodies including anti-STAT1, anti-STAT3, anti-c-Jun, anti-ATF2, and anti-CREB was verified in electrophoretic mobility shift assays using the Stat3 IL-6RE probe and appropriate nuclear extracts containing the endogenous proteins or exogenously expressed ATF2, c-Jun, or CREB (data not shown). Anti-ATF3 and anti-ATF4 antibodies for gel shift assays were used following the manufacturer's instructions.

RESULTS
Increased Stat3 Gene Expression in Response to IL-6 -Akira et al. (38) reported that treatment with IL-6 increases the level of Stat3 mRNA severalfold in liver. Also, we showed that IL-6 treatment of M1 cells increases the protein level of STAT3 and that the increased level is sustained for Ͼ48 h (33). To begin a deeper analysis of STAT3 regulation, we first determined a detailed time course for the IL-6-induced mRNA expression of the Stat3 gene in M1, HepG2, and IL-6-dependent MH60 B cell hybridoma cells. In M1 cells, the level of Stat3 mRNA started to increase at 1 h in response to IL-6 and reached its maximum level at 3 h. After reaching its maximum, Stat3 mRNA levels showed a slight transient decrease at 6 h and then increased again for over 48 h (Fig. 1A, lanes 1-6). The induction of Stat3 mRNA was also tested in the presence of cycloheximide, a protein synthesis inhibitor. Treatment of M1 cells with cycloheximide failed to inhibit the IL-6-induced activation of the Stat3 gene (Fig. 1A, lanes 7-9), indicating that the induction of Stat3 by IL-6 did not require new protein synthesis. The rapid induction of Stat3 mRNA by IL-6 was also observed with HepG2 cells (Fig. 1B) and MH60 cells (data not shown).
The Cytoplasmic Region from Amino Acids 108 to 133, Especially the Third Tyrosine (Tyr-3) of gp130, Is Required for Regulating the Expression of Stat3-We next analyzed which part of gp130 was necessary for the induction of Stat3 gene expression. Using M1 transformants expressing chimeric receptors consisting of the extracellular domain of the GHR and the transmembrane and cytoplasmic domains of gp130 with progressive C-terminal truncations and point mutations at the tyrosine residues (34), we proceeded to identify the cytoplasmic region of gp130 required for Stat3 gene transactivation. Growth hormone enhanced Stat3 gene expression in the transformants expressing the chimeric receptors containing at least 133 amino acid residues of the gp130 cytoplasmic domain (Fig.  2, lanes 1-6). Chimeric receptors with further truncation of gp130 to 108 or 68 amino acid residues (GHR108 and GHR68) were unable to induce Stat3 gene expression (Fig. 2, lanes  7-12). GHR133F2, which contains a Tyr-to-Phe mutation at the second tyrosine (Tyr-2), could activate the Stat3 gene (Fig. 2,  lanes 13-15), but GHR133F3, which has a Tyr-to-Phe mutation at the third tyrosine (Tyr-3), and GHR133F2/3, which has Tyr-to-Phe mutations at both Tyr-2 and Tyr-3, did not activate the Stat3 gene (lanes 16 -21). These results show that the cytoplasmic region of gp130 from amino acids 108 to 133, with an intact YXXQ motif at Tyr-3, is required for activating the Stat3 gene. As this region corresponds to the region required for activating the STAT3 protein, it is likely that STAT3 participates in the transcriptional activation of its own promoter.
The Ϫ478/Ϫ229 Region of the Stat3 Promoter Shows IL-6 Responsiveness in a Manner Dependent on STAT3-To elucidate the molecular mechanisms activating the Stat3 gene promoter, we obtained a 2.2-kb fragment of the Stat3 gene 5Јflanking region by genomic PCR. The sequence of the proximal region up to Ϫ478 bp with the known DNA motifs is shown in Fig. 3A. The sequence published by Shi et al. (39) is missing a 0.25-kb region between positions Ϫ424 and Ϫ169. We first constructed a reporter gene containing the 2.2-kb Stat3 gene 5Ј-flanking region linked to the luciferase reporter gene and tested whether this 2.2-kb fragment showed IL-6 responsiveness by transfecting the reporter gene construct into HepG2 cells, followed by IL-6 stimulation for 16 h. Dominant-negative STAT3 and STAT1 expression vectors were also cotransfected with the promoter-reporter construct to test whether STAT3 is involved in the IL-6 activation of the Stat3 gene promoter. As shown in Fig. 3B, IL-6 increased the promoter activity of the 2.2-kb 5Ј-flanking region by ϳ5-fold, and this induction was effectively inhibited by dominant-negative STAT3 (DN-Stat3), but not by dominant-negative STAT1 (DN-Stat1), indicating that the IL-6 activation of the 2.2-kb Stat3 gene promoter was dependent on STAT3. Next, to localize the region containing the IL-6 response element, a series of 5Ј-deletion mutations of the Stat3 gene promoter constructs were made and tested for IL-6 responsiveness as described above. The promoter regions retained a similar level of IL-6 responsiveness when the promoter contained up to position Ϫ478 (Fig. 3C). However, deletion up to position Ϫ229 substantially decreased the IL-6 responsiveness, suggesting that an IL-6 response element(s) resides in the Ϫ478/Ϫ229 region (Fig. 3C).
Both STAT3 and CRE-binding Proteins Are Necessary for the IL-6-induced Transcriptional Activation of the Stat3 Gene-To test whether the Ϫ478/Ϫ/229 region alone has the ability to render a heterologous minimal promoter responsive to IL-6, we assayed the IL-6 responsiveness of a luciferase construct containing the Ϫ478/Ϫ229 DNA fragment inserted upstream of the minimal junB promoter-luciferase gene construct (p478/ 229-Luc). As shown in Fig. 4A, IL-6 increased the transcription of the reporter gene driven by the Ϫ478/Ϫ229 fragment by ϳ10-fold. As expected, dominant-negative STAT3 (DN-Stat3), but not STAT1 (DN-Stat1), effectively inhibited the IL-6 responsiveness of the Ϫ478/Ϫ229 region. These results indicate that STAT3 activates its own transcription through the Ϫ478/ Ϫ229 region. We next searched for STAT3-binding sequence(s) within the Ϫ478/Ϫ229 DNA region. Although we could not find a typical STAT3-binding sequence (a TT-AA motif with a spacing of 5 bp) within the Ϫ478/Ϫ229 DNA region, instead we found an atypical SBE (TGCCTGGAA) and a CRE (TGACGTCA) with a 5-bp spacing between the motifs, which is very similar to the junB response element for IL-6 (JRE-IL6) ( Fig. 3A) (28). To investigate whether both the SBE and CRE were responsible for IL-6 responsiveness, we made two different p478/229-Luc mutants containing mutations at the SBE (p478/229-Luc-mSTAT) or the CRE (p478/229-Luc-mCRE), as illustrated in Fig. 4B, and tested them for IL-6 responsiveness. Mutations at the putative SBE reduced the IL-6 responsiveness by Ͼ90%, and mutations at the CRE reduced it by Ͼ80% (Fig. 4B). These results indicated that both DNA motifs were required for IL-6 responsiveness. To assess the role of the two DNA motifs in the intact 2.2-kb Stat3 gene promoter, we introduced the same mutations as described above into the Stat3 IL-6RE in the intact 2.2-kb Stat3 promoter (p2166-Stat3-Luc-mSTAT and p2166-Stat3-Luc-mCRE). As shown in Fig. 4C, mutations at the putative STAT3-binding site in the 2.2-kb Stat3 promoter (p2166-Stat3-Luc-mSTAT) severely reduced the IL-6 responsiveness of the Stat3 promoter by Ͼ80%, and mutations at the CRE reduced the IL-6 responsiveness of the promoter by ϳ60%. The weaker inhibition by the mutations at the CRE in the otherwise intact promoter compared with that by the mutations at the CRE in the context of the Ϫ478/Ϫ229 region alone may be due to the existence of unidentified region(s) cooperatively working with STAT3 outside the Ϫ478/ Ϫ229 region. These results, in any case, indicate that the two DNA motifs in the Stat3 IL-6RE are the major determinants in the 2.2-kb Stat3 promoter for IL-6 responsiveness.
Characteristics of Stat3 IL-6RE-binding Proteins-We next examined the nature of the Stat3 IL-6RE-binding complexes in the nuclear extracts from IL-6-stimulated M1, HepG2, and MH60 cells by electrophoretic mobility shift assays using an oligonucleotide containing the Stat3 IL-6RE as a probe. We also used another oligonucleotide containing the ␣ 2 -macroglobulin APRE (40) as a control for STAT binding. Nuclear extracts from IL-6-stimulated MH60 cells showed a prominent IL-6-inducible Stat3 IL-6RE-binding complex on the Stat3 IL-6RE probe (indicated as Stat3IL6RE-BC), which appeared rapidly at 15 min, declined transiently at 3 h, and returned to the level of 15 min at 6 h (Fig. 5A, lanes 2-5). The mobility of the IL-6-induced Stat3 IL-6RE-binding complex is slower than that of the APREbinding complex containing a STAT3 homodimer, called APRF (40) (Fig. 5A, lane 7). More important, the level of the IL-6inducible complex is equivalent to that of APRF (Fig. 5A, compare lanes 2 and 7). This IL-6-inducible complex could be seen more clearly in MH60 nuclear extracts than in M1 and HepG2 nuclear extracts, although similar IL-6-inducible complexes were present in the nuclear extracts from all three cell lines (data not shown). Therefore, only results with MH60 nuclear extracts are shown. To investigate the binding sites and binding specificities of the Stat3 IL-6RE-binding complex, we used two mutant Stat3 IL-6RE oligonucleotides containing mutations either at the CRE site (Stat3-IL-6RE-mCRE) or at the STAT3-binding element (Stat3-IL-6RE-mSTAT) as probes and other oligonucleotides as competitors, including the APRE, JRE-IL6, and somatostatin CRE. The Stat3-IL-6RE-mSTAT probe did not show any IL-6-inducible Stat3 IL-6RE-binding complexes (Fig. 5A, lanes 8 and 9). On the other hand, the Stat3-IL-6RE-mCRE probe showed an IL-6-inducible complex with the same mobility as that of APRF (Fig. 5A, lanes 10 and  11), and this complex was likely to be a STAT3 homodimer since anti-STAT3 antibody shifted the complex (data not shown). Both the APRE and JRE-IL6 oligonucleotides effectively competed with the Stat3 IL-6RE probe for forming the IL-6-inducible Stat3 IL-6RE-binding complex (Fig. 5B, lanes  2-5), whereas the somatostatin CRE oligonucleotides competed for all of the CRE-binding complexes and the Stat3 IL-6REbinding complex, but made the appearance of a fast migrating IL-6RE-binding complex (lanes 6 and 7). The complex is likely to be a STAT3 homodimer since the complex migrated as APRF on the APRE probe, and anti-STAT3 antibody shifted the complex (data not shown). This STAT3 homodimer could barely be seen on the wild-type Stat3 IL-6RE probe (Fig. 5A). These results indicate that formation of the Stat3 IL-6RE-binding complex requires both the STAT-binding element and the CRE and suggest that the STAT-binding element has a low affinity for STAT3 homodimers alone. A, the reporter containing the Ϫ478/Ϫ229 fragment of the Stat3 promoter in front of the minimal junB promoter-luciferase gene (p478/229-Luc) was tested for IL-6 responsiveness as described for Fig. 3. A dominant-negative STAT3 (pCAGGS-NeoHA-STAT3F (DN-Stat3)) or STAT1 (pCAGGS-NeoHA-STAT1F (DN-Stat1)) expression vector was included to test the STAT dependence, and normalized luciferase activities are shown as described for Fig. 3B. The values are the averages of three independent experiments performed in duplicate. -Fold increases in response to IL-6 are indicated by the numbers above the bars. B, the two DNA motifs of the Stat3 IL-6RE are necessary and sufficient for the IL-6 responsiveness of the Stat3 promoter. Each of the Ϫ478/Ϫ229 fragments containing mutations at either the CRE or STAT-binding element was inserted upstream of the minimal junB promoter of pSPBLuc, as illustrated. The three p478/229-Luc reporter constructs, with wild-type IL-6RE (p478/229-Luc-WT), with a mutated STAT-binding element (p478/229-Luc-mSTAT), and with a mutated CRE (p478/229-Luc-mCRE), were tested for IL-6 responsiveness, and normalized luciferase activities are shown as described for Fig. 3B. Results are the averages of three independent experiments. C, the role of the two DNA motifs was assessed in the intact Stat3 promoter with a 2.2-kb upstream region. Two reporter gene constructs containing the Stat3 promoter including its 2.2-kb upstream region with mutations at either the SBE (p2166-Stat3-Luc-mSTAT) or CRE (p2166-Stat3-Luc-mCRE) were tested for IL-6 responsiveness as described for Fig. 3B. The results with the wild-type Stat3 promoter construct (p2166-Stat3-Luc-WT) are also shown. Relative luciferase activities from three independent experiments are shown.
Next, to test whether the IL-6-inducible Stat3 IL-6RE-binding complex contained STAT3, STAT1, or other known CREbinding proteins, we included the appropriate antibody against such molecules in the electrophoretic mobility shift assay (Fig.  5C). Anti-STAT3 antibody, not anti-STAT1 antibody, shifted the inducible complex, indicating that the IL-6-inducible Stat3 IL-6RE-binding complex contained STAT3 as a constituent. None of the other antibodies against transcription factors that are known to bind the CRE sites, such as c-Jun, CREB, ATF2, ATF3, and ATF4, shifted or decreased the IL-6-inducible Stat3 IL-6RE-binding complex (Fig. 5C). These results indicated that the Stat3 IL-6RE-binding complex contained both a STAT3 homodimer and unidentified CRE site-binding protein(s) distinct from CREB, ATF1, ATF2, ATF3, or ATF4. DISCUSSION In this study, we showed that STAT3 rapidly induced transcriptional activation of the Stat3 gene through an IL-6RE located at positions Ϫ335 to Ϫ314 in the Stat3 gene promoter. This IL-6RE was shown to consist of a low affinity SBE (TGC-CAGGAA) and a CRE (TGACGTCA). The functional role of the two DNA motifs in the Stat3 IL-6RE was confirmed in two ways. First, mutations at either the SBE or CRE effectively inhibited the IL-6 responsiveness of the Ϫ478/Ϫ229 fragment linked to a heterologous promoter. Second, the same mutations at the DNA motifs in the otherwise intact 2.2-kb Stat3 promoter also effectively inhibited the IL-6 responsiveness of the promoter. The functional roles of the two DNA motifs were further confirmed by the demonstration that IL-6 rapidly induced the formation of complexes with the Stat3 IL-6RE. A slowly migrating complex with the IL-6RE required both the SBE and CRE. We showed that this complex contained a STAT3 dimer and unidentified CRE-binding protein(s) distinct from CREB, ATF1, ATF2/CRE-BP1, ATF3, ATF4, or the c-Jun/ AP-1 family of proteins. Ternary complex formation on other IL-6 response elements containing a SBE and CRE has been shown by us (28), including the IL-6 response elements seen in the junB promoter, JRE-IL6 (36), and the IRF1 promoter, IR/IRF1 (41). Table I compares the sequences of such IL-6 response elements with that of the Stat3 IL-6RE. The size of the spacing between the two DNA motifs also varies from 1 to 5 bp. The combination of the DNA motifs varies: a low affinity SBE with an atypical CRE (JRE-IL6), a high affinity SBE with an atypical CRE (IR/IRF1), and a low affinity SBE with a typical CRE (Stat3 IL-6RE). Considering the binding specificity of the Stat3 IL-6RE-binding complex shown here and the JRE-  6 and 7), Stat3-IL-6RE-mSTAT (lanes 8 and 9), or Stat3-IL-6-mCRE (lanes 10 and 11). The position of the IL-6inducible Stat3 IL-6RE-binding complex (Stat3IL6RE-BC) is indicated by an arrow. The APRE-binding complex, APRF (also called SIF-A), is indicated as Stat3 homodimer with an arrow. B, binding specificity of the Stat3 IL-6RE-binding complex. Nuclear extracts prepared from MH60 cells stimulated with IL-6 for 15 min were preincubated with unlabeled oligonucleotide competitors (50-or 250-fold molar excess as indicated) for 5 min, followed by incubation with labeled Stat3 IL-6RE probes. The competitors used were APRE (lanes 2 and 3), JRE-IL6 (lanes 4 and 5), and somatostatin CRE (somCRE; lanes 6 and 7). C, characterization of the protein components of the IL-6-inducible binding complex. Nuclear extracts prepared from IL-6-stimulated MH60 cells were preincubated with the indicated antiserum or monoclonal antibody for 30 min on ice prior to the addition of 32 P-labeled Stat3 IL-6RE probes and were then subjected to an electrophoretic mobility shift assay. The effectiveness of these antibodies, except for anti-ATF3 and anti-ATF4, was verified using recombinant proteins overexpressed in transfected 293T cells (data not shown). The results of the kinetics study of the IL-6 induction of Stat3 mRNA in wild-type M1 cells (Fig. 1A) are fully consistent with the levels of STAT3 activity detected by tyrosine phosphorylation of STAT3 and its DNA binding activity shown previously by us (33), supporting the notion of the STAT3-mediated activation of the Stat3 gene. It is remarkable that three of the IL-6 target genes inducible in M1 cells, junB, IRF1, and Stat3, contain very similar IL-6 response elements in their promoters. Interestingly, these genes can be activated by a low dose of IL-6 (such as 1 ng/ml) in M1 cells. 2 In contrast, in another case of IL-6-induced gene regulation, the IL-6 repression of c-myc and c-myb is turned on only by higher doses of IL-6 in the same cell line. 2 The difference in sensitivity to IL-6 concentration may be explained by CRE or CRE-like site-binding proteins that may be crucial in recruiting small amounts of activated STAT3 to the IL-6 response element by making complexes with STAT3 and DNA. The identification and functional role of the unidentified CRE-binding protein would help to prove this model.
It has been noted that in IL-6-stimulated M1 cells, STAT3 activity persists at a high level for Ͼ48 h (33). This activation pattern is not detected in other cell lines. The sustained STAT3 activity at high levels in M1 cells may be important in causing growth arrest and terminal differentiation. Therefore, it is likely that the activation of the Stat3 gene by STAT3 in conjunction with the cooperative CRE-binding protein forms an autoregulatory loop, and this may be one of the mechanisms by which the persistent activation of STAT3 protein in M1 cells is induced. Recently, O'Brien and Manolagas (42) showed that transcription of the signal-transducing ␤-subunit of the IL-6 receptor complex (gp130) is also activated by STAT3 through a STAT-binding element in the promoter. Their findings and ours suggest the existence of autoregulatory mechanisms in the IL-6 signal-transducing system at the levels of both the receptor and the signal-transducing transcription factor.  -IL6) Ϫ125 GGCTCGCGTCAC----TGTCAGGAAGCG Ϫ148 Human IRF1 IL-6RE (IR/IRF1) Ϫ102 GCGTGCCGTCAT----TTCGGGGAAATC Ϫ125