Interleukin-6-induced Tethering of STAT3 to the LAP/C/EBPβ Promoter Suggests a New Mechanism of Transcriptional Regulation by STAT3*

LAP/C/EBPβ is a member of the C/EBP family of transcription factors and contributes to the regulation of the acute phase response in hepatocytes. Here we show that IL-6 controlsLAP/C/EBPβ gene transcription and identify an IL-6 responsive element in the LAP/C/EBPβ promoter, which contains no STAT3 DNA binding motif. However, luciferase reporter gene assays showed that STAT3 activation through the gp130 signal transducer molecule is involved in mediating IL-6-dependent LAP/C/EBPβ transcription. Southwestern analysis indicated that IL-6 induces binding of a 68-kDa protein to the recently characterized CRE-like elements in the LAP/C/EBPβ promoter. Transfection experiments using promoter constructs with mutated CRE-like elements revealed that these sites confer IL-6 responsiveness. Further analysis using STAT1/STAT3 chimeras identified specific domains of the protein that are required for the IL-6-dependent increase inLAP/C/EBPβ gene transcription. Overexpression of the amino-terminal domain of STAT3 blocked the IL-6-mediated response, suggesting that the STAT3 amino terminus has an important function in IL-6-mediated transcription of the LAP/C/EBPβ gene. These data lead to a model of how tethering STAT3 to a DNA-bound complex contributes to IL-6-dependent LAP/C/EBPβ gene transcription. Our analysis describes a new mechanism by which STAT3 controls gene transcription and which has direct implication for the acute phase response in liver cells.

leucine zipper transcription factors currently comprising six members involved in the regulation of various aspects of cellular differentiation and function in multiple tissues. Three closely related members of the C/EBP family, C/EBP␣, LAP/C/ EBP␤, and C/EBP␦, are known to contribute to liver-specific gene transcription. Specific functions of these transcription factors have been described in the regulation of the acute phase response and during liver regeneration, where they seem to be involved in the process of hepatocyte proliferation and differentiation (reviewed in Ref. 1).
Several inflammatory signals, including lipopolysaccharides (LPS), interleukin 6 (IL-6), interleukin 1 (IL-1), tumor necrosis factor ␣, and interferon gamma (IFN-␥) contribute to the regulation of the acute phase response in hepatocytes. The activities of the C/EBP family members are differentially modulated in response to the various inflammatory stimuli on the mRNA and protein level. Upon induction of the acute phase response C/EBP␣ mRNA levels decrease in the liver, whereas LAP/C/ EBP␤ and C/EBP␦ gene transcription is enhanced. For the regulation of LAP/C/EBP␤, however, post-translational mechanisms were described in addition to transcriptional activation (reviewed in Ref. 2). Several phosphorylation sites in the LAP/ C/EBP␤ protein have been shown to be functionally relevant (3)(4)(5)(6). Phosphorylation occurs in vitro in response to an intracellular Ca 2ϩ increase, through activation of the protein kinase C pathway and through activation of the mitogen-activated protein (MAP) kinase following induction of the Ras pathway. Although all these events lead to increased transactivation of LAP/C/EBP␤-dependent genes, only the MAP kinase pathway can be linked to the acute phase response. Moreover, stimulation with lipopolysaccharides, IL-6, IL-1, dexamethasone, and glucagon strongly induces LAP/C/EBP␤ expression (7)(8)(9)(10). Initially, LAP/C/EBP␤ has also been cloned because it binds to IL-6-responsive elements in the promoters of acute phase response genes (11,12) and thus was originally named NF-IL6 (nuclear factor involved in the IL-6 gene expression).
Of the numerous cytokines and growth factors that are involved in regulating the acute phase response, IL-6 plays a major role (reviewed in Ref. 13). Binding of IL-6 to its receptor induces homodimerization of the signal-transducing component gp130 (14), which results in the activation of constitutive associated JAK kinases and phosphorylation of gp130 at different tyrosine residues (15,16). Tyrosine phosphorylation creates specific docking sites for signaling molecules containing SH2 domains (17). In the case of gp130 at least three signaling cascades, the STAT1, the STAT3, and the MAP kinase pathway are activated (18 -21). Activation of STAT3 is the most prominent pathway during this process (15,22,23). Phosphorylation of STAT3 through JAK kinases results in its homo-or het-erodimerization and nuclear translocation. STAT3 then binds to target sequences in different promoters and enhances gene transcription. Binding sites for STAT3 are located in most promoters of acute phase genes, and originally STAT3 was identified as acute phase response factor (24 -26).
Analysis of the C/EBP␦ promoter revealed that STAT3 contributes to higher gene transcription after IL-6 stimulation (27,28). Besides DNA binding, cooperative interaction of STAT3 with Sp1 is essential for the regulation of the C/EBP␦ promoter. In contrast, no information is available about the molecular mechanisms involved in the IL-6-dependent increase of LAP/C/EBP␤ gene transcription. Therefore, we were interested in investigating how IL-6 controls the activation of this gene.
In this analysis we show that IL-6 activates LAP/C/EBP␤ gene transcription using the two already characterized CRElike sites in the LAP/C/EBP␤ promoter (29). STAT3 activation through the gp130 signal transducer is essential for increased LAP/C/EBP␤ transcription despite the lack of sequence-specific STAT DNA binding sites in this promoter region. Enhanced IL-6-dependent transcription of the LAP/C/EBP␤ gene correlates with the binding of a new IL-6-inducible factor with a molecular mass of 68 kDa to the CRE-like sites in the LAP/C/ EBP␤ promoter. Functional dissection of the gp130 signal transducer molecule and the use of STAT1/STAT3 chimeras further demonstrated the essential role of STAT3 and identified specific regions of the protein that are required for the IL-6-dependent increase of LAP/C/EBP␤ gene transcription. These results lead to a model of how tethering STAT3 to the promoter region contributes to IL-6-dependent gene transcription without direct DNA binding and thus indicate a new mechanism for STAT3-dependent gene transcription.

Stimulation of Mice and Preparation of Total RNA and of Liver
Nuclear Extracts-C3H mice were stimulated intraperitoneally with 40 g of human recombinant IL-6. At different time points after injection, the livers were removed and preparation of RNA and of liver nuclear extracts was performed. At each time point at least three animals were used in parallel. RNA was isolated by the guanidinium isothiocyanate method (30). For preparation of nuclear extracts the pooled livers were rinsed in ice-cold phosphate-buffered saline, and liver nuclear proteins were prepared as described previously (30). All steps were performed at 4°C. Nuclear proteins were aliquoted and frozen immediately in liquid nitrogen.
Northern Blot Analysis-Northern blot analysis was performed as described before, according to standard procedures (30). 30 g of total RNA was analyzed through a 1% agarose formaldehyde gel followed by transfer to Hybond N ϩ membranes (Amersham Pharmacia Biotech, Braunschweig, Germany). The LAP/C/EBP␤ and glyceraldehyde-3phosphate dehydrogenase (GAPDH) cDNA probes were labeled with [␣-32 P]CTP according to the instructions for Rediprime (Amersham Pharmacia Biotech). The hybridization procedure was performed as described previously (30). Blots were exposed for autoradiography and exposed to an imaging plate (Fuji) for quantification. LAP/C/EBP␤ signals were normalized to the GAPDH signals and set to 1 for untreated animals. The values for IL-6 treatment were shown as -fold stimulation.
Expression Vectors for STAT1, STAT3, and Chimeric STAT Proteins-A series of expression vectors for truncated STAT3 proteins and various STAT3-STAT1 chimeric proteins were constructed as described elsewhere (22,31). For the construction of cDNAs coding for STAT3-STAT1 chimeric proteins, additional unique restriction sites were introduced into the STAT cDNAs by site-directed mutagenesis or authen-tic restriction sites were used. The chimeric STAT cDNAs were constructed by exchanging the respective DNA fragments within the pBluescript vector context. The cDNAs were then subcloned into the pRc/CMV vector (Invitrogen, Groningen, The Netherlands). The STAT3D (EE 3 AA) construct was kindly provided by Toshio Hirano (32).
Other Plasmids-The pCMV␤Gal vector was used as internal standard. The pRcCMV-PIAS3 expression vector was kindly provided by Chan Chung (34). The Epo-receptor/gp130 chimeras were described before (19). They consisted of the extracellular domain of the murine Epo-receptor fused to the human transmembrane domain and different cytoplasmic tyrosine modules of the IL-6 signal transducer gp130. The EpoR/MAP site corresponds to the Tyr-759 module and the EpoR/ STAT3 site corresponds to the Tyr-767 module. The EpoR/STAT1 site comprised only the membrane-proximal part of gp130, fused to a peptide covering the Tyr-440 tyrosine motif of the IFN-␥R.
Cell Culture, Transfection Experiments, and Luciferase Assays-HepG2 cells (ATCC) were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. DNA transfection was performed using a modified calcium phosphate precipitation method as described previously (35). HepG2 cells were grown on 60-mm dishes to about 50% confluence when used for transfection experiments. The amount of reporter and expression vectors is indicated in the figure legends. The total amount of DNA was kept constant in each transfection experiment by adding pBSK ϩ DNA (Stratagene, La Jolla, CA) to 5 g. All transfections contained 0.1 g of the ␤-galactosidase reporter pCMV␤Gal as an internal standard. For stimulation experiments, cells were starved with 1% fetal calf serum for 24 h after transfection. Stimulation was performed with human recombinant IL-6 (Strathmann Biotech, Hannover, Germany) in the amounts and for the time points indicated or with 7 units/ml human recombinant erythropoietin (Epo) (Roche Molecular Biochemicals, Mannheim, Germany) for 4 h. 1 unit of IL-6 is equivalent to 0.005 ng (Strathmann Biotech).
To measure luciferase activity, cells were washed twice with phosphate-buffered saline and lysed by adding 350 l of extraction buffer (25 mM Tris-H 3 PO 4 , pH 7.8, 2 mM EDTA, 10% (v/v) glycerol, 1% (v/v) Triton X-100, and 2 mM DTT) for 10 min. The lysates were cleared by centrifugation. 50 l of the supernatant was assayed by addition of 300 l of measuring buffer (25 mM glycylglycine, 15 mM MgSO 4 , and 5 mM ATP). The light emission was measured in duplicate for 10 s in a Lumat LB 9501 (Berthold) by injecting 100 l of 250 M luciferin. Data are represented as the mean Ϯ S.D. of triplicate experiments and are representative for three independent experiments.
Gel Retardation Assays-For gel retardation assays nuclear extracts were used as indicated in the figure legends. Binding buffer consisted of 25 mM HEPES, pH 7.6, 5 mM MgCl 2 , 34 mM KCl, 2 mM DTT, 0,2 mM PMSF, 50 ng of poly(dI-dC)/l, and 100 ng of bovine serum albumin/l. The binding reaction was performed for 30 min on ice. Free DNA and DNA⅐protein complexes were resolved on a 6% polyacrylamide gel.
The oligonucleotides were purchased from Eurogentec (Seraing, Belgium) and used as 32 P-labeled probes. Oligonucleotides A to M were derived from the LAP/C/EBP␤ promoter. Oligonucleotide A corresponds to Ϫ123 to Ϫ95 (GCG GCC GGG CAA TGA CGC GCA CCG ACC CG), oligonucleotide B to Ϫ102 to Ϫ75 (CCG ACC CGG CGG CGG GGC GGC GGG AGG G), oligonucleotide C to Ϫ90 to Ϫ68 (CGG GGC GGC GGG AGG GGC CCC GG), oligonucleotide D to Ϫ75 to Ϫ43 (GGC CCC GGC GTG ACG CAG CCC GTT GCC AGG CGC), oligonucleotide E to Ϫ123 to Ϫ107 (GCG GCC GGG CAA TGA CG), oligonucleotide F to Ϫ110 to Ϫ87 (GAC GCG CAC CGA CCC GGC GGC GGG), and oligonucleotide M to Ϫ123 to Ϫ99, mutated in Ϫ109 to Ϫ107 (GCG GCC GGG CAA TGG TTC GCA). The SIE consensus oligonucleotide corresponds to the sequence: 5Ј-GTG CAT TTC CCG TAA ATC TTG TCT ACA-3Ј. Supershift experiments were performed with antibodies against STAT3 or CREB1 (Santa Cruz Biotechnology, Santa Cruz, CA). Nuclear extracts were modified in Fig. 4 as indicated. Prior to the incubation with the 32 Plabeled probe, nuclear extracts were incubated with 0.08% desoxycholic acid (DOC) for 20 min at 4°C, then 0.15% Nonidet P-40 was added. For the depletion assay 2.5 l of antibody to CREB1 (Santa Cruz Biotechnology) was coupled to 10 l of protein A-Sepharose, saturated with 1 g/l bovine serum albumin, in TNE buffer (10 mM Tris, pH 8.0, 100 mM NaCl, 1 mM EDTA, 2 mM DTT, 2 mM Pefabloc, 1% Nonidet P-40) for 4 h at 4°C, washed twice in TNE buffer, and twice in gel shift sample buffer. Incubation with 3 g of nuclear extract was performed at 4°C overnight. Protein A-Sepharose without antibody was treated under the same conditions and served as control. The supernatants were analyzed in gel shift experiments.
Southwestern Analysis-Southwestern analysis was performed as described with some modifications (36,37). 20 g of liver nuclear extracts prepared before and 1 h after stimulation of C3H mice with 40 g of IL-6 was heated for 30 min at 60°C in SDS sample buffer and separated on a 10% SDS-polyacrylamide gel. After electrophoresis the gel was incubated in renaturing buffer (10 mM Tris, pH 7.5, 1 mM EDTA, 50 mM NaCl, 1 mM DTT, 4 M urea, adjusted to pH 7.5) for 4 h at room temperature and then blotted onto a nitrocellulose membrane (Millipore, Bedford, MA) in a solution containing 25 mM Tris and 192 mM glycine (adjusted to pH 8.3) at 4°C for 2 h at a constant current of 200 mA. The membranes were prehybridized in binding buffer (10 mM Tris, pH 7.5, 0.1 mM EDTA, 1 mM DTT, 50 mM NaCl, 2ϫ Denhardt's solution, adjusted to pH 7.2) at 4°C overnight and then incubated with 2.5 ϫ 10 6 cpm 32 P-labeled oligonucleotide A for 1 h at room temperature. After incubation the membranes were washed several times in binding buffer and exposures were made to x-ray films.
Western Blot Analysis-Nuclear extracts were separated on a 10% SDS-polyacrylamide gel (38) and blotted onto nitrocellulose membrane (Millipore) in 1% SDS, 20% methanol, 400 mM glycine, 50 mM Tris, pH 8.3, at 4°C for 2 h at 200 mA. The antibody directed against Gal4-(1-147) was purchased from Santa Cruz Biotechnology, the antibody directed against phospho-STAT3 was purchased from New England Bio-Labs (Beverly, MA). The antigen⅐antibody complexes were visualized using the ECL detection system as recommended by the manufacturer (Amersham Pharmacia Biotech, Braunschweig, Germany).

IL-6-induced LAP/C/EBP␤ mRNA Expression in the Liver
Correlates with STAT3 Activation-LAP/C/EBP␤ was originally identified, because an increase in DNA binding was evident after IL-6 treatment in human hepatoma cells (11,12). IL-6 stimulates LAP/C/EBP␤ at the transcriptional and posttranslational level (2,7). However, the molecular mechanisms responsible for IL-6-mediated LAP/C/EBP␤ gene transcription are unknown. In an initial experiment, C3H mice were stimulated with IL-6 and LAP/C/EBP␤ mRNA expression was studied by Northern blot analysis (Fig. 1A). LAP/C/EBP␤ signals were normalized to the GAPDH signals and set to 1 for untreated animals. Quantification revealed an increase in LAP/C/EBP␤ mRNA expression already after 30 min while maximum levels (3.8-fold) were found 6 h after stimulation ( Fig. 1A). At later time points LAP/C/EBP␤ mRNA expression decreased again. After binding of IL-6 to its receptor, several pathways are activated. However, STAT3 activation has been shown to stimulate transcription of many acute-phase genes. Therefore, activation of STAT3, as already described before (39), was studied. IL-6 induced an increase in nuclear expression of tyrosine-phosphorylated STAT3 30 min after injection as shown by Western blot analysis (Fig. 1B). The nuclear expression of phospho-STAT3 remained high for the first 4 h and was no more detectable after 12 h. Gel shift experiments with an SIE element as STAT3 target sequence showed activation of DNA binding 30 min after IL-6 injection (Fig. 1C). DNA binding remained high for up to 3 h and then subsequently decreased. No complex formation was found after 12 h. The specificity of STAT3 in the new appearing complex was verified by FIG. 1. IL-6-induced LAP/C/EBP␤ mRNA expression in the liver correlates with STAT3 activation. A, C3H mice were stimulated with 40 g of IL-6, total RNA was isolated from the liver before and at different time points after stimulation (as indicated), and Northern blot analysis with a 32 P-labeled DNA for LAP/C/EBP␤ (upper panel) and for GAPDH were performed. The ratio (lower panel) between the LAP/C/EBP␤ signal and the GAPDH signal was calculated and set to 1 for the untreated probe. Changes after IL-6 treatment are shown as -fold stimulation. B, Western blot analysis was performed with 10 g of liver nuclear extract prepared before and at different time points after stimulation (as indicated) of C3H mice with 40 g of IL-6. Nuclear expression of tyrosine-phosphorylated STAT3 (P-STAT3) was detected with an antibody directed against tyrosine-phosphorylated STAT3. C, gel shift experiments were performed with 3 g of liver nuclear extract prepared before and at different time points after stimulation (as indicated) of C3H mice with 40 g of IL-6. Complex formation was performed with the 32 P-labeled STAT3-specific oligonucleotide comprising the SIE site (Santa Cruz Biotechnology). supershift experiments (data not shown) and has been shown before using a STAT3 DNA target sequence derived from the ␣2-macrolgobulin promoter (39). Thus, these experiments show a close correlation between an increase in LAP/C/EBP␤ mRNA expression and STAT3 activation after IL-6 stimulation in the liver in vivo.
IL-6-induced LAP/C/EBP␤ Transcription Is Mediated by a Region in Proximity to the TATA Box-As LAP/C/EBP␤ mRNA expression increased after IL-6 stimulation in vivo, we performed experiments to identify regulatory sequences in the LAP/C/EBP␤ promoter that mediate IL-6 induction. In transfection experiments luciferase reporter constructs with increasing deletions in the 5Ј-flanking region, located upstream of the start site of transcription in the LAP/C/EBP␤ gene, were analyzed in HepG2 cells. The constructs were cotransfected with 50 ng of STAT3 expression vector, and cells were treated with 1000 units/ml IL-6 for 4 h (Fig. 2A). The relative luciferase activity of the respective LAPPRO reporter construct without stimulation was set to 1, and the changes after IL-6 treatment were shown as -fold stimulation. Strong induction (up to 20fold) of the luciferase reporter gene was observed when the whole 1.4-kb 5Ј-region of the promoter was used (LAPPRO 1). IL-6-dependent stimulation of the 5Ј-flanking region was still found when deletions up to nucleotide Ϫ121 were analyzed. No increase in reporter gene activity after IL-6 treatment was evident with the LAPPRO 9 construct (truncated to position Ϫ71, see Fig. 2A). These data indicate that the region located between nucleotides Ϫ121 and Ϫ71 in close proximity to the TATA box is involved in mediating IL-6-dependent transcrip-tion of the LAP/C/EBP␤ gene. The IL-6-dependent effect on the LAP/C/EBP␤ promoter was further investigated by dose and time-kinetic experiments (Fig. 2, B and C). This analysis showed that a maximal effect on LAP/C/EBP␤ promoter activity was found with 1000 units/ml IL-6 for 4 h.
STAT3 Activation Is Essential for IL-6-mediated LAP/C/ EBP␤ Transcription-Our next interest was to analyze the relevance of the STAT3 signaling cascade for IL-6-mediated LAP/C/EBP␤ transcription despite the finding that there is no STAT consensus sequence located in the LAP/C/EBP␤ promoter region, which confers IL-6 inducibility. Therefore cotransfection experiments were performed with the LAPPRO 8 reporter construct and an empty control vector or expression vectors for STAT1 or STAT3. These studies showed that IL-6 stimulated the promoter construct 2-fold, whereas cotransfection of the STAT3 expression vector enhanced this effect (Fig.  3A). In contrast, cotransfection of a STAT1 expression vector had no influence on the IL-6 inducibility compared with cotransfection of the empty control vector (Fig. 3A). In further experiments, the STAT3-specific inhibitor PIAS3 (protein inhibitor of activated STAT3) (14) was used to characterize the STAT3-dependent mechanism. Cotransfection experiments of the LAPPRO 8 constructs and increasing amounts of an expression vector for PIAS3 were performed with or without an expression vector for STAT3. Under both conditions PIAS3 inhibited the IL-6-dependent increase in LAP/C/EBP␤ gene transcription in a concentration-dependent manner (Fig. 3B).
After binding of IL-6 to gp130 at least three intracellular signaling cascades, STAT1, STAT3, and the Map kinase path-  Fig. 2 and then stimulation was performed with 1000 units of IL-6/ml for 4 h. The relative luciferase activity of the respective transfection without stimulation was set to 1, and the changes after IL-6 treatment are shown as -fold stimulation. B, HepG2 cells were cotransfected with 2 g of the LAPPRO 8 luciferase reporter construct, without (Ϫ) or with (ϩ) 100 ng of a STAT3 expression vector (Stat3) and with increasing amounts of a PIAS3 expression plasmid (PIAS) as indicated. Cells were starved as indicated in Fig. 2 and then stimulation was performed with 1000 units of IL-6/ml for 4 h. The relative luciferase activity of the transfection without stimulation and without PIAS3 cotransfection was set to 1 for the samples without and with STAT3 cotransfection, respectively, and the changes after IL-6 treatment are shown as -fold stimulation. C, EpoR/gp130 chimeras. The receptor chimeras consisted of the extracellular domain of murine EpoR fused to the transmembrane domain and different cytoplasmic tyrosine modules of the IL-6 signal transducer gp130 (EpoR/Map site, EpoR/STAT3 site). The EpoR/Stat1 site comprised only the membrane-proximal part of gp130, fused to a peptide covering the Tyr-440 tyrosine motif of the IFN-␥R. D, HepG2 cells were cotransfected with 2 g of the LAPPRO 8 luciferase reporter construct, 2 g of the respective expression vector coding for the receptor chimeras as indicated, and with 50 ng of a STAT3 expression vector. Cells were starved as indicated in Fig. 2 and then stimulation was performed with 7 units of Epo/ml for 4 h. The relative luciferase activity of the respective transfection without stimulation was set to 1, and the changes after Epo treatment are shown as -fold stimulation. activating STAT3 (gp130 Y767, EpoR/STAT3 site), the MAP kinase pathway (gp130 Tyr-759, EpoR/Map site), or STAT1 (IFN ␥R Tyr-440/EpoR/STAT1 site) (Fig. 3C). Stimulation of HepG2 cells with erythropoietin (Epo) had no effect on LAP-PRO 8 activity (data not shown). Epo stimulation after cotransfection of HepG2 cells with the LAPPRO 8 construct and the EpoR/Map site or the EpoR/STAT1 site fusion receptors also resulted in no increase of reporter gene activity (Fig. 3D). However, when the EpoR/STAT3 site chimera was used, a 3-fold induction of LAP/C/EBP␤ gene transcription was found. Therefore these data indicate that of the three pathways studied only STAT3 can mediate IL-6-dependent transcription of the LAP/C/EBP␤ gene.
IL-6 Induces Binding of a 68-kDa Protein to the CRE-like Sites in the LAP/C/EBP␤ Promoter-The cotransfection experiments indicated that STAT3 is involved in mediating IL-6-dependent LAP/C/EBP␤ transcription. However, even computer-assisted analysis showed no potential STAT DNA binding site (41) between nucleotide Ϫ121 and the TATA box in the LAP/C/EBP␤ promoter (for the whole sequence see Ref. 29).
To exclude the possibility that STAT3 may bind to a yet unknown target sequence in this region, gel shift experiments were performed using mouse liver nuclear extracts prepared before and 2 h after IL-6 treatment when STAT3 binding to its consensus sequence occurred. Overlapping oligonucleotides derived from the LAP/C/EBP␤ promoter (A-M, Fig. 4A) were used as 32 P-labeled probes. As shown in Fig. 4B, complex formation of two complexes was only found with oligonucleotide A and D containing the two CRE-like sites of the LAP/C/EBP␤ promoter. Complex formation was blocked when the mutant oligonucleotide M (mutation of the first CRE-like site) was used as a 32 P-labeled probe. In contrast to unstimulated extracts, a third, new complex was detected in IL-6-stimulated liver nuclear extracts, when the intact CRE-like site was included in the oligonucleotides.
In time course experiments, formation of the IL-6-stimulated complex was studied. Oligonucleotide A (see Fig. 4A) represents the sequence between nucleotide Ϫ123 and Ϫ95 of the LAP/C/EBP␤ promoter that was used as the 32 P-labeled probe. After IL-6 stimulation the intensity of the two complexes found already in untreated animals did not change (Fig. 4C). Formation of the IL-6-inducible third complex was found 1 h after IL-6 injection. Maximal DNA binding was evident after 2 h. At later time points complex formation decreased again (Fig. 4C). Supershift experiments were performed to study whether STAT3 binds to the CRE-like sequence in the LAP/C/EBP␤ promoter after IL-6 stimulation (Fig. 4D). Antibodies directed against CREB and STAT3 were used as described before (30,39). However, DNA binding of the IL-6-dependent complex was not changed by using a STAT3 antibody, whereas DNA binding of one of the two unstimulated complexes could be inhibited with a CREB antibody (Fig. 4D). The protein of the second unstimulated complex is not known (38).
Because binding of this IL-6-inducible factor occurred at the CRE-like site in the LAP/C/EBP␤ promoter, we were interested in examining whether binding of the IL-6-induced factor to the CRE-like binding site occurs only in the presence of CREB or if the IL-6-induced factor is able to bind independently of CREB. Therefore, gel shift experiments with modified liver nuclear extracts as described under "Experimental Procedures" were performed. We used detergent incubation with desoxycholic acid and Nonidet P-40 (DOC) to inhibit protein⅐protein interaction. However, DNA binding of the IL-6-induced factor was unchanged after DOC treatment (Fig. 4E). Complete depletion of liver nuclear extracts with an anti-CREB antibody coupled to protein A-Sepharose (Prot.A/CREB) also had no effect on DNA binding of the IL-6-induced factor (Fig. 4E). Thus, these data indicate that the IL-6-induced factor binds to the CRE-like site in the LAP/C/EBP␤ promoter independently of CREB.
Next, Southwestern analysis was performed to characterize the molecular weight of the IL-6-induced factor (Fig. 4F). In control as well as in IL-6-stimulated liver nuclear extracts we found p30, corresponding to the lower gel shift band, and p43, corresponding to CREB. An additional p68 protein corresponding to the upper band was only found in IL-6-stimulated extracts (Fig. 4F). In parallel UV cross-linking, experiments showing very weak signals (data not shown) confirmed our results indicating a molecular mass of 68 kDa for the IL-6inducible factor binding to both CRE-like sites in the LAP/C/ EBP␤ promoter. The Southwestern analysis revealed weak binding of an additional IL-6-stimulated factor with a molecular mass of ϳ45 kDa.
Both CRE-like Sites in the LAP/C/EBP␤ Promoter Are Essential for IL-6-dependent Induction-To further characterize the significance of the two CRE-like sites for IL-6 induction, we analyzed mutant LAP/C/EBP␤ promoter constructs in HepG2 cells (Fig. 5A). Mutations in both CRE-like sites (LAPPRO 8 MUT IϩII) led to a strong decrease in IL-6-dependent transcription (Fig. 5B). No stimulation was observed after transfection of the LAPPRO 9 construct that contains only one CRElike site. Recently Cantwell et al. (27) described the cooperative function of STAT3 and Sp1 in IL-6-mediated C/EBP␦ induction. We therefore analyzed the deletion construct LAPPRO 8 WT ⌬, which carries a deletion of 27 nucleotides between the two CRE-like sites in the GC-rich sequence with possible Sp1 binding sites. This deletion caused a marked decrease in IL-6-dependent transcription (Fig. 5B). The data show that both CRElike sites contribute to IL-6-mediated LAP/C/EBP␤ induction and indicate that the spacer sequence between them might be of functional relevance.
The Amino-terminal Domain of STAT3 Is Essential for IL-6mediated LAP/C/EBP␤ Transcription-Our results indicate that STAT3 contributes to IL-6-dependent LAP/C/EBP␤ transcription without direct DNA binding to the promoter region. To better understand the role of different domains of STAT3 in this context, cotransfection experiments were performed with the LAPPRO8 reporter construct and expression vectors for different STAT3/STAT1 domain swap mutants and analyzed after IL-6 stimulation (Fig. 6). The expression level and the characterization of the domain swap mutants have been shown before (15).
After receptor stimulation, the SH2 domain of the STAT proteins interacts with specific phosphorylated tyrosines in the intracellular receptor domain (19,22). Substitution of the STAT3 by the STAT1 SH2 domain (Stat 3/1 (SH2)) prevented IL-6-mediated LAP/C/EBP␤ transcription (Fig. 6). Therefore, the SH2 region of STAT3 is essential for the IL-6 effect, as already implicated by the EpoR transfection experiments (Fig.  3D). The exchange of the tyrosine region (Stat 3/1 (Y)), responsible for dimerization of STAT proteins after phosphorylation by members of the JAK kinase family, maintained the IL-6 effect (Fig. 6). The amino-terminal part (NT) of the STAT proteins contributes to several protein⅐protein interactions (reviewed in Ref. 41). Substitution to STAT1 NT (Stat 3/1 (NT)) abolished the IL-6 effect (Fig. 6). Therefore, our experiments indicate that the amino-terminal region of STAT3 is required for IL-6-mediated LAP/C/EBP␤ transcription. The carboxylterminal part (CT) of STAT proteins mediates the transcriptional activation of target genes (42,43). The exchange of the carboxyl-terminal region (Stat 3/1 (CT)) or a mutation of serine 727 to alanine (Stat 3S/A) maintained IL-6 stimulation (Fig. 6). However, no IL-6-dependent transcription was found The two CRE-like binding sites in the LAP/C/EBP␤ promoter (TGACGCGC, Ϫ111 to Ϫ104 and TGACGCAG, Ϫ65 to Ϫ58) are marked as circles. B, gel shift experiments were performed with 3 g of liver nuclear extract prepared before (Ϫ) and 2 h after stimulation (ϩ) of C3H mice with 40 g of IL-6. Complex formation was performed with the respective oligonucleotide as 32 P-labeled probe. IL-6-induced complex formation is marked with an arrowhead. C, gel shift experiments were performed with 3 g of liver nuclear extracts prepared before and at different time points after stimulation (as indicated) of C3H mice with 40 g of IL-6. Complex formation was performed with oligonucleotide A (Ϫ123 to Ϫ95) derived from the LAP/C/EBP␤ promoter as 32 P-labeled probe. IL-6-induced complex formation is marked with an arrowhead. D, for supershift/binding inhibition analysis antibodies against STAT3 or CREB were added to the 2-h value after IL-6 stimulation (w/o ϭ without antibody). IL-6-induced complex formation is marked with an arrowhead. E, gel shift experiments were performed with 3 g of liver nuclear extract prepared before (Ϫ) and 2 h after stimulation (ϩ) of C3H mice with 40 g of IL-6 and oligonucleotide A (Ϫ123 to Ϫ95) as 32 P-labeled probe. Before complex formation was performed, the extracts were modified as follows: Doc ϭ incubation with detergent (desoxycholic acid and Nonidet P-40), Prot. A ϭ incubation with protein A-Sepharose beads, Prot. A/CREB ϭ when cotransfection experiments were performed with carboxyl-terminal deletion mutants (Stat3 ⌬715, Stat3 ⌬711). A mutation in the DNA binding region (Stat 3D) as well as substitution of the STAT3 DNA binding region to STAT1 (Stat 3/1 (D,Linker)) abolished IL-6 induction (Fig. 6).
These data indicate that interaction with the basal transcriptional machinery and, thereby, activation of LAP/C/EBP␤ transcription is mediated by the STAT3 carboxyl-terminal region and that this function is exchangeable toward STAT1. In contrast, the amino-terminal part and the DNA binding region of STAT3 are absolutely required to mediate IL-6 induction, which shows that in addition to the SH2 site these two domains of STAT3 mediate specificity during this process.

Overexpression of the Amino-terminal Domain of STAT3 Prevents IL-6-mediated Transcription of the LAP/C/EBP␤ Promoter-
The experiments using the chimeric STAT3/STAT1 constructs suggested that interactions at the amino terminus of STAT3 are crucial to mediate IL-6-dependent transcription of the LAP/C/EBP␤ promoter. To confirm this hypothesis we fused the amino-terminal part of STAT3 or STAT1 to the Gal4 DNA binding domain to provide this region with a nuclear translocation signal and investigated whether overexpression of the STAT3 construct might act as a dominant-negative inhibitor of IL-6-dependent LAP/C/EBP␤ induction (Fig. 7A). Fusion to the Gal4 DNA binding domain was performed with the amino-terminal (5ЈGal4-Stat1NT, 5ЈGal4-Stat3NT) as well as with the carboxyl-terminal domain (Stat1NT-3ЈGal4, Stat3NT-3ЈGal4) of the respective STAT amino terminus (Fig. 7A). Both approaches were chosen, because it would minimize the possible risk of misfolding in the tertiary structure.
Comparable expression of the respective fusion proteins after transfection of HepG2 cells was found by Western blot analysis of nuclear extracts (Fig. 7B). In luciferase reporter gene experiments cotransfection of 5ЈGal4-STAT3NT with LAPPRO 8 inhibited the IL-6-mediated LAP/C/EBP␤ induction in a concentration-dependent manner, whereas the 5ЈGal4-STAT1NT and the Gal4 alone showed no effect (Fig. 7C). The same results were found with the 3Ј fusion constructs. STAT3NT-3ЈGal4 prevented IL-6 induction, whereas STAT1NT-3ЈGal4 had no influence (Fig. 7D). However, the STAT3NT-3ЈGal4 constructs were significantly more effective (concentration range 10 -100 ng) compared with the 5ЈGal4-STAT3NT constructs (concentration range 100 -1000 ng) (Fig.  7, C and D). These experiments clearly show that the aminoterminal domain of STAT3 is able to act as a dominant-negative inhibitor, and therefore, these results suggest that the STAT3 amino terminus has an important function in IL-6mediated transcription of the LAP/C/EBP␤ gene.

DISCUSSION
Most of the acute phase genes are activated on a transcriptional level through IL-6-responsive elements in their promoter regions. The best studied example for a transcription factor involved in this mechanism is STAT3 (23,25); however, C/EBP incubation with CREB antibody coupled protein A-Sepharose beads. CREB ϭ no modification, supershift reaction with CREB antibody. IL-6-induced complex formation is marked with an arrowhead. F, Southwestern analysis. 20 g of liver nuclear extract prepared before (Ϫ) and 1 h after stimulation (ϩ) of C3H mice with 40 g of IL-6 were separated by SDS-polyacrylamide gel electrophoresis, blotted on nitrocellulose, incubated with 32 P-labeled oligonucleotide A (Ϫ123 to Ϫ95), and exposed to autoradiography. The arrows mark the proteins in the stimulated extracts (p68, p43, p30) corresponding to the three gel shift bands. Molecular mass markers are shown in kDa on the left.

FIG. 5. Both CRE-like sites in the LAP/C/EBP␤ promoter are essential for the IL-6-dependent induction.
HepG2 cells were cotransfected with 2 g of the respective LAPPRO luciferase reporter constructs and with 100 ng of a STAT3 expression plasmid. Cells were starved as indicated in Fig. 2 and then stimulation was performed with 1000 units of IL-6/ml for 4 h. The relative luciferase activity of the respective transfection without stimulation was set to 1, and the changes after IL-6 treatment are shown as -fold stimulation. A, the following constructs were used: the LAPPRO 8 WT construct; the LAPPRO 8 MUT IϩII construct, where the two CRE-like sites were mutated; the LAPPRO 8 WT⌬ construct, where a part of the spacer between the two CRE-like sites is deleted; and the LAPPRO 9 construct, which contains only the second CRE-like site. B, following IL-6 stimulation the relative changes in luciferase activity compared with the unstimulated construct are shown as -fold stimulation.
family members also contribute to this regulation (2,24). Induction of the acute phase response in hepatocytes enhances the expression of LAP/C/EBP␤ and C/EBP␦, which then activate transcription of several acute phase response genes. Earlier results showed that the C/EBP␦ promoter contains two Sp1 binding sites, an STAT3 binding site, and a CRE-like binding site in close proximity to the TATA-box. For the IL-6-dependent increase in C/EBP␦ gene transcription, the STAT3 and the Sp1 DNA binding sites function cooperatively, whereas the CRElike site does not participate in this regulation (27,28). STAT3 is also supposed to contribute to IL-6-dependent transcription of the LAP/C/EBP␤ gene (2,44).
The STAT3 Signaling Cascade Is Involved in IL-6-mediated LAP/C/EBP␤ Transcription-Our in vivo and in vitro experiments demonstrated that the IL6/gp130/STAT3 pathway induces higher LAP/C/EBP␤ gene transcription. An increase in LAP/C/EBP␤ gene transcription was also reported after lipopolysaccharide (LPS) stimulation and hepatocyte growth factor treatment. This is mediated by an autoregulatory loop in which phosphorylation of LAP/C/EBP␤ increases transactivation and thus higher LAP/C/EBP␤ gene transcription. The LAP/C/EBP␤ binding site(s), which contribute to this regulation, were mapped outside the IL-6-responsive region characterized in our study (45,46). IL-6 also stimulates the MAP kinase pathway, which has been shown before to lead to posttranslational activation of LAP/C/EBP␤ by its phosphorylation (4). This mechanism could be further excluded, because the EpoR/Map site chimera was unable to induce LAP/C/EBP␤ transcription, indicating that an autoregulatory loop does not explain our observations.
The IL-6-responsive Element in the LAP/C/EBP␤ Promoter Contains No STAT3 Consensus Sequence and Is Regulated through Two CRE-like Binding Sites-Analysis of the whole 1.4-kb LAP/C/EBP␤ promoter by transfection experiments revealed that the three potential STAT DNA binding sites are not involved in mediating the increase in IL6-dependent gene transcription. In contrast, the recently described CRE-like sites (Ϫ111 to Ϫ104 and Ϫ65 to Ϫ58) in the promoter control this effect, and IL-6 stimulates DNA binding of a 68-kDa protein to these sequences. Induction of CRE-like site binding proteins after IL-6 stimulation was already reported for the STAT3 (47) and JunB promoter (48). However, these proteins were not further characterized. Meanwhile, the ATF/CREB family comprises a growing number of proteins that recognize the CREconsensus or CRE-like binding sites (reviewed in Ref. 49). The mechanism for the induction of the DNA binding activity of the CRE-like site binding proteins after IL-6 stimulation is not known.
Mutations in the CRE-like sites of the LAP/C/EBP␤ promoter resulted in a lack of p68 binding and in a decrease in IL-6-dependent transcription. Mutation of the first site had a stronger effect on IL-6 inducibility than that of the second site (data not shown), but strongest inhibition was observed with the double mutant construct. Additionally, our results indicate that the spacer region between the two CRE sites is important in mediating IL-6-dependent gene transcription. To exclude the possibility that other transcription factors are involved in this regulation, we subcloned the two CRE sites with the wild type or an unrelated spacer region in a heterologous promoter. These results showed that IL-6 inducibility is not dependent on the sequence of the spacer region (data not shown), indicating that only the spacing between the CRE sites and no factors binding to this region are crucial to confer IL-6 inducibility of the LAP/C/EBP␤ promoter.
A Model for Tethering STAT3 to the LAP/C/EBP␤ Promoter-Our results suggested that STAT3 contributes to higher IL-6-dependent LAP/C/EBP␤ gene transcription without DNA binding. Further analyses with STAT3/STAT1 domain swap mutants suggested that the amino-terminal part, the SH2 domain, and the DNA binding domain of STAT3 are crucial for the IL-6-dependent increase in LAP/C/EBP␤ gene transcription. The essential role of the STAT3 DNA binding domain for this regulation was unexpected. Our mutation analysis in the LAP/C/EBP␤ promoter clearly showed that only the CRE-like sites are relevant for IL-6 inducibility. Consequently, STAT3 would have to bind to this region in a nonspecific manner. Thus we cannot completely rule out the possibility that STAT3 contributes to p68 DNA binding through a yet undefined mechanism, which is not detected by gel shift experiments. To prove this possibility we also depleted IL-6-induced nuclear extracts from STAT3 as shown for CREB. However, STAT3-depleted extracts had no influence on DNA binding of p68 (data not shown). Therefore we favor the possibility that the DNA binding domain of STAT3 contributes to protein⅐protein interaction. In a recent report (50), parts of the STAT3 DNA binding domain are shown to be involved in mediating the interaction between STAT3 and c-Jun.
The STAT3/STAT1 amino terminus (NT) domain swap mutants and the Gal4-STAT3NT fusion constructs indicated that the amino-terminal part of STAT3 is very crucial for mediating the link between the basal machinery and the two CRE-like sites. This observation implies that the amino terminus most likely does not bind to a more general factor, which would be a part of the general RNA polymerase machinery, but to one of the proteins, e.g. p68, binding to the CRE-like sites in the LAP/C/EBP␤ promoter. Thus this interaction seems to mediate specificity during the regulation. The Gal4 part was fused to both the carboxyl-terminal and the amino-terminal end of the STAT3 amino terminus to minimize steric effects. Fusion to the carboxyl-terminal end (STAT3NT-3ЈGal4) resulted in a 10-fold more potent dominant-negative inhibitor than fusion to the amino-terminal end (5ЈGal4-STAT3NT). The difference be- with the DNA binding domain of the Gal4 protein (nuclear localization signal) are shown. The Gal4 domain was either cloned 5Ј of the respective STAT amino terminus (5ЈGal4-Stat1NT and 5ЈGal4-Stat3NT) or 3Ј of the respective STAT amino terminus (Stat1NT-3ЈGal4 and Stat3NT-3ЈGal4). B, expression of the respective fusion proteins was shown by Western blot analysis. HepG2 cells were transfected with pBS, 2 g of 5ЈGal4-STAT1NT, 2 g of 5ЈGal4-STAT3NT, 1 g of STAT1NT-3ЈGal4, or 1 g of STAT3NT-3ЈGal4 as indicated. Cells were harvested 48 h after transfection; nuclear extracts were prepared and analyzed by Western blotting with a Gal4-specific antibody. C, HepG2 cells were cotransfected with 2 g of the LAPPRO 8 luciferase reporter construct, without (Ϫ) or with (ϩ) 100 ng of a STAT3 expression vector (Stat3), and with increasing amounts (100, 500, 1000, and 2000 ng) of the 5ЈGal4-STAT3NT, the 5ЈGal4-STAT1NT, or the Gal4 construct. Cells were starved as indicated in Fig. 2 and then stimulation was performed with 1000 units of IL-6/ml for 4 h. The relative luciferase activity without stimulation for the respective transfection was set to 1, and the changes after IL-6 treatment are shown as -fold stimulation. D, HepG2 cells were cotransfected with 2 g of the LAPPRO 8 luciferase reporter construct, without (Ϫ) or with (ϩ) 100 ng of a STAT3 expression vector (Stat3) and with increasing amounts (10,25,50, and 100 ng) of the STAT3NT-3ЈGal4 or the STAT1NT-3ЈGal4 construct. Cells were starved as indicated in Fig. 2 and then stimulation was performed with 1000 units of IL-6/ml for 4 h. The relative luciferase activity without stimulation for the respective transfection was set to 1, and the changes after IL-6 treatment are shown as -fold stimulation. tween the two constructs might be best explained by the localization of the interacting region.
In recent reports there is evidence that several transcription factors activate transcription without DNA binding. In specific situations a pre-existing complex of transcription factors bound to DNA allows the association with an additional transcription factor without DNA binding. This mechanism has also been called tethering, and meanwhile several examples are known. ATF6 can be tethered to the c-Fos and to the atrial natriuretic factor promoter by serum response factor bound to the serum response element, without direct binding of ATF6 to the serum response element (51,52). Ubeda et al. (53) reported the recruitment of CHOP to an AP1⅐DNA complex without DNA binding of CHOP. Glucocorticoid receptors can also act as transcriptional coactivators for STAT5 without binding to a glucocorticoid response element (54,55). The involvement of STAT1 in tumor necrosis factor-␣-mediated apoptosis, where SH2 mutants can restore wild-type function in STAT1 knockout cells, lead to the model that STAT1 might be recruited to a promoter through protein⅐protein interaction with a DNAbound partner (56,57). STAT3 has also been reported as a transcriptional coactivator without association with its DNA binding motif. IL-6-activated STAT3 associates with ligandbound glucocorticoid receptor to form a transactivating complex bound to a glucocorticoid response element (58). Besides the model of the tethered coactivator function for STAT3, there is evidence for its role during other mechanisms not related to DNA binding. An adapter function for STAT3 has been implicated in the recruitment of phosphatidylinositol 3-kinase to the IFN-␣ receptor (59).
The close correlation between DNA binding of p68 to the CRE-like elements and the relevance of STAT3 for the increase in LAP/C/EBP␤ transcription indicates that these mechanisms could be directly linked to each other. Therefore, based on our results we would like to propose a hypothetical model explaining the role of STAT3 during IL-6-dependent activation of LAP/C/EBP␤ gene transcription. We suggest that IL-6 activates STAT3 and DNA binding of p68. STAT3 is tethered to the p68-containing complex at the CRE-like site in the LAP/C/ EBP␤ promoter. In this regulation, STAT3 would act as transcriptional coactivator whereby the carboxyl-terminal region interacts with the basal transcription machinery and activates LAP/C/EBP␤ transcription while the amino terminus may interact with p68 or other related factors. It is conceivable that certain steric conditions are necessary for tethering STAT3 by p68 bound to both CRE-like sites. This would explain the critical distance between the two sites for IL-6 induction. DNA binding can affect the quaternary structure of transcriptional regulators and thereby determine heterodimerization partners (60). This conformational change would allow tethering of STAT3 to p68 after binding to the CRE-like site in the LAP/C/ EBP␤ promoter.
Importance for an Interplay of Different Regulatory Sites and for Interaction of Different Activators with the Same Sites in the LAP/C/EBP␤ Promoter-Among mouse, rat, and humans the two CRE-like sites and the GC-rich linker sequence are highly conserved regulatory elements in the LAP/C/EBP␤ promoter (61). Recently, we have reported that basal and protein kinase A-inducible LAP/C/EBP␤ promoter activity in cells of hepatic and neuronal origin requires the synergistic activity of the two CRE-like sites (29). Since then additional examples for the relevance of a cAMP-dependent regulation of LAP/C/EBP␤ through these sites were described during memory formation in hippocampal neurons (62), in human endometrial stroma cells as part of prolactin induction (63), in sertoli cells (64), for the gonadotropin response in granulosa cells (65,66), and dur-ing adipocyte differentiation (67).
Besides the DNA binding elements in the promoter region other more cell type-specific mechanisms, e.g. starvation, proliferation, determine gene regulation in a certain cellular background. Analysis of the LAP/C/EBP␤ promoter in monocytic cells showed the relevance of an interplay between an Sp1 site and the second CRE-like site for basal activity and for PMA or LPS stimulation, whereas the first CRE-like site was not involved in this regulation (61). In hepatocytes both CRE-like sites are required and recognized by different activators and coactivators, which contribute to the regulation of LAP/C/ EBP␤ gene transcription during the acute phase response and during liver regeneration. The increase in promoter activity varied after protein kinase A stimulation (29) and as shown here after IL-6 treatment between 6-and 20-fold. The mechanisms, which contribute to these differences, are currently unknown. However, for example, after IL-6 stimulation, binding of an additional factor p45 as shown by Southwestern analysis could be detected, which might be a possible heterodimerization partner for p68 modulating its activity. Additionally, the proliferative state of the cell may directly modulate the basal activity of the LAP/C/EBP␤ promoter. Therefore, to better understand all the different transcriptional control mechanisms under various physiological conditions, it might be necessary in the future to further characterize the transcription factors binding to the CRE-like sites and to identify other possible interacting coactivators.