Functional Analysis of Interleukin 6 Response Elements (IL-6REs) on the Human γ-Fibrinogen Promoter

Several families of transcription factors play important roles in modulating liver-specific gene expression during an acute phase response (APR). Stat3/APR factor is the main transactivator of gene expression by the interleukin (IL)-6 family of cytokines signaling through gp130. During an APR, fibrinogen (FBG) genes are coordinately up-regulated by IL-6 and glucocorticoids. Except for rat γFBG, attempts to demonstrate direct binding of IL-6-activated Stat3 to FBG CTGGGAA promoter elements have not been successful. Herein we show the presence of three functional type II IL-6 response elements (IL-6REs) on the human γFBG promoter and that the magnitude of Stat3 binding to these elements correlates negatively with their functional activity in reporter gene assays. Stat3-specific binding to γFBG IL-6REs was confirmed by cross-competition with α2-macroglobulin IL-6RE and specific interactions with anti-Stat3 in electrophoretic mobility shift assays. All type II IL-6REs contributed to full promoter activity; however, transactivation from Site II at –306 to –301 was strongest. In contrast to a previous report, IL-6 failed to induce activation of serum amyloid A-activating factor-1/c-Myc-associated zinc finger protein (SAF-1/MAZ), and mutation of the SAF-1RE had little effect on IL-6 induction of γFBG promoter activity. In the absence of a functional glucocorticoid receptor response element, dexamethasone potentiated IL-6-induced γFBG promoter activity 2-fold, requiring promoter-proximal Site I and Site II; the promoter-distal Site III had no effect on dexamethasone potentiation of IL-6-induced promoter activity. Notably the propensity for Stat3 binding to human γFBG IL-6REs was low compared with Stat3 binding to the α2-macroglobulin IL-6RE. Together these data suggest that Stat3 transactivation via IL-6REs on FBG promoters likely involves participation of additional transcription factors and/or coactivators to achieve optimal coordinated up-regulation during an APR.

Fibrinogen (FBG) 1 and fibrin are the targets of two complex and opposing biochemical pathways, the coagulation and fi-brinolytic cascades, respectively, that together preserve vascular integrity and maintain the hemostatic balance (1,2). In addition, the fibrin clot provides a critical provisional matrix at sites of injury, inflammation, or infection in which cells can proliferate, organize, and carry out specialized functions (3)(4)(5). FBG is a dimeric molecule with each half composed of three polypeptide chains, A␣, B␤, and ␥. Three separate single copy genes encode the FBG polypeptide chains, and each gene is separately transcribed and translated. The FBG chains are rapidly assembled in a sequential series of steps by the independent attachment of the preformed A␣ and ␥ chains, drawn from intracellular pools, to the nascent B␤ chain (6). These chains, and each half-molecule, are linked by a network of disulfide bonds to form the circulating 340-kDa protein. Expression of the FBG genes is transcriptionally regulated, and in response to inflammation, hepatic transcription of the three genes is coordinately up-regulated (7).
The acute phase response is the mechanism by which the host responds to disruption of homeostasis by infection, tissue injury, or neoplasia. The major regulatory cytokines of the acute phase response are tumor necrosis factor-␣, interleukin (IL)-1, and IL-6, which act on the liver to change the expression of a number of plasma proteins collectively known as the acute phase proteins (APPs). Change in APP gene expression is either positive or negative depending on the relative -fold up-or down-regulation of each gene, respectively. The positive APPs are further divided into three groups depending on the magnitude of their induction. Group 1 APPs such as C3 and C4 are increased by 50%, Group 2 APPs such as FBG and haptoglobin are increased by 2-10-fold, and Group 3 APPs including C-reactive protein and serum amyloid A (SAA) are increased by 100 -1000-fold in plasma. In addition, APP genes are classified into two categories based on their cytokine responsiveness. Type I APP genes are up-regulated by IL-1, IL-6, tumor necrosis factor-␣, and glucocorticoids (GCs) or various combinations thereof; whereas type II genes respond to IL-6 and GCs (8). In contrast, neither tumor necrosis factor-␣ nor IL-1␤ induce expression of type II APP genes but instead frequently downregulate their expression even in the presence of high levels of IL-6 (9, 10). The FBG genes are classified as type II in that GCs act synergistically with IL-6 in up-regulation of FBG gene expression (11,12).
Several families of transcription factors play important roles in modulating liver-specific APP gene expression including the CCAAT/enhancer-binding protein (C/EBP), nuclear factor (NF)-B, and the signal transducer and activator of transcription (Stat). NF-B and C/EBP transcription factors are important in regulation of type I APP genes. IL-1␤ and IL-6 induce expression of C/EBP-␤ and C/EBP-␦, while IL-1␤ and tumor necrosis factor-␣ activate NF-B. These transcription factors in turn up-regulate IL-6 production, which activates the Stat family of transcription factors; activated Stat3/acute phase response factor binds to IL-6 response elements (REs) on promoters of type II APP genes. Basal promoter elements and enhancer sequences for constitutive and IL-6-regulated expression of the human and rat FBG genes have been described (13). Unlike the rat ␥FBG promoter, which contains three functional type II IL-6REs (14), only one such IL-6RE was found on the human ␥FBG promoter (13). Furthermore, unlike the IL-6REs on the human A␣ and B␤ chain promoters, IL-6 induction of human ␥FBG promoter activity is not responsive to the presence of elevated levels of either ␤ or ␦ isoforms of C/EBP (e.g. type I IL-6REs). Recently another transcription factor, SAA-activating factor-1 (SAF-1/MAZ), was shown to contribute to IL-6 regulation of ␥FBG gene expression in human hepatoma cells (15).
Although rat and human FBG genes are clearly responsive to IL-6, the identity of the cognate transcription factors that bind to IL-6 enhancer elements on promoter regions of the FBG genes has remained elusive. IL-6-activated Stat3 binds to the type II cis-acting elements of the rat ␥FBG promoter (14); however, no such complexes have been identified on the human ␥FBG promoter (13,15) or the rat and human A␣ and B␤ chain promoters (7,16,17). Because IL-6-mediated Stat3 signaling plays critical roles in essential biological functions including the immune response, inflammation, hematopoiesis, oncogenesis, and embryogenesis by regulating cell growth, survival, and differentiation (for reviews, see Refs. 18 and 19), we investigated the role of Stat3 binding to IL-6REs and transactivation of the human ␥FBG promoter as described in this report.

EXPERIMENTAL PROCEDURES
Cell Culture, Reagents, and Antibodies-HepG2 cells (HB-8065) from the ATCC were grown in Eagle's minimal essential medium containing 1 mM sodium pyruvate, 0.1 mM nonessential amino acids, penicillin, streptomycin, and 10% fetal bovine serum; cell culture reagents were from Invitrogen. The passage number of cells used in this report ranged from 12-18 from the ATCC stock culture. T 4 polynucleotide kinase was obtained from Promega (Madison, WI), [␥-32 P]ATP and [ 14 C]chloramphenicol were obtained from PerkinElmer Life Sciences, poly(dI-dC) was from Amersham Biosciences, and oligonucleotides were synthesized by Invitrogen. The following antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA): goat polyclonal anti-actin (sc-1615), which recognizes multiple actin isoforms; goat anti-Stat1 (sc-7988-X); and mouse monoclonal anti-Stat3 (sc-8019-X). Dexamethasone (DEX), ethyl acetate, and o-nitrophenyl ␤-D-galactopyranoside were purchased from Sigma, and recombinant human IL-6 was from Research Diagnostics (Flanders, NJ).
Western Blot Immunodetection-HepG2 cells were grown to near confluence and then incubated in medium containing various concentrations of IL-6 for 24 h; the cells were washed with phosphate-buffered saline and lysed as described previously (12). After determining the protein concentration by Bradford assay, equivalent amounts of protein from the cell lysates were resolved after reduction by SDS-10% PAGE. After transfer to nitrocellulose membranes, the ␥ chain was detected by Western blotting with monoclonal antibody J88B, which specifically recognizes the ␥ chain of human FBG (20). The blots were stripped and reprobed with antibody to actin to control for variability in protein loaded on the gel. Immune complexes were visualized by enhanced chemiluminescence according to the manufacturer's instructions (PerkinElmer Life Sciences).
Superfect Transfections and Luciferase Assays-Transfection of HepG2 cells was performed using Superfect (Qiagen, Valencia, CA) according to the manufacturer's protocol. Cells were grown to 40 -50% confluence in 35-mm culture dishes and transfected with reporter construct, 5 ng of pRL-SV40 as the internal control for transfection efficiency, and pSG5 plasmid DNA to a total of 2 g in each transfection reaction. The amount of reporter and expression vectors used in each experiment is indicated in the figure legends. After incubation with Superfect-DNA complexes for 16 -18 h, fresh medium with various concentrations of IL-6 Ϯ 0.1 M DEX was added to the cells and incubated for 24 h. Luciferase activity in cell lysates was determined by the Dual-Luciferase ® reporter assay system according to the manufacturer's protocol (Promega). The wtStat3 construct and Stat3⌬55C mutant, which functions as a transdominant inhibitor, were generous gifts of Dr. H. Baumann (Roswell Park Cancer Institute, Buffalo, NY).
Preparation of Nuclear Extracts-HepG2 nuclear extracts were prepared as described previously (23). After the desired treatment, the cells were washed twice with ice-cold phosphate-buffered saline, scraped into 15-ml conical tubes, and then centrifuged for 5 min at 1000 ϫ g at 4°C. Cell pellets were resuspended in hypotonic buffer (10 mM HEPES-KOH, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride), allowed to swell on ice for 10 min, and then vortexed for 10 s. After centrifugation for 3 min at 1300 ϫ g at 4°C, the supernatant was discarded, and the pellet was resuspended in cold hypertonic buffer (20 mM HEPES-KOH, pH 7.9, 25% glycerol, 420 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride) and incubated on ice for 20 min. Cellular debris were removed by centrifugation at 14,000 ϫ g at 4°C for 15 min. The supernatant fractions containing nuclear proteins were stored at Ϫ70°C until use.
Electrophoretic Mobility Shift Assays (EMSAs)-The sense strand used to prepare each double-stranded probe corresponding to the human ␥FBG IL-6REs for EMSAs is given in Fig. 3A. The rat ␣ 2 -macroglobulin (␣ 2 M) type II IL-6RE that binds with high affinity to Stat3 served as a positive control (9). Double-stranded probes were endlabeled with [␥-32 P]ATP by T 4 polynucleotide kinase. DNA-protein binding reactions were performed in 10 mM HEPES, pH 7.8, 50 mM KCl, 1 mM EDTA, 5 mM MgCl 2 , 10% glycerol, 5 mM dithiothreitol, 0.7 mM phenylmethylsulfonyl fluoride, 100 ng of poly(dI-dC)/l, and 1 g of bovine serum albumin/l as described previously (24). Two to 10 g of HepG2 nuclear extract and 20 fmol of probe were incubated at room temperature for 10 min and then the reaction was stopped by adding loading buffer. For supershift, the reaction was incubated at room temperature for 10 min followed by adding 1 l of specific antibody and incubating an additional 10 min. Nondenaturing 6% polyacrylamide gels were precleared by electrophoresis at 20 mA for 30 min at 4°C in 1ϫ running buffer (7 mM Tris-HCl, pH 7.5, 3 mM sodium acetate, 1 mM EDTA); free DNA and DNA-protein complexes were resolved at 30 mA for 150 min at 4°C. The dried gels were exposed to phosphorimaging screens, and the signal was detected using a Storm 820 Phosphor-Imager and ImageQuant software (Amersham Biosciences).
Chloramphenicol Acetyltransferase (CAT) Reporter Plasmid Constructions, Calcium Phosphate Transfections, and CAT Assays-To resolve the conflict in localization of the IL-6REs on the human ␥FBG promoter obtained in this report using the luciferase reporter constructs with the data obtained by Ray (15) using the pCAT3 basic vector (Promega) and calcium phosphate transfection method, we essentially repeated our studies using identical ␥FBG promoter-CAT reporter constructs following the exact methods of Ray (15). Transient transfections were conducted by the calcium phosphate method using a mixture of DNA containing 3 g of pCAT3-␥FBG reporter plasmid, 1 g of pSV-␤-gal plasmid (Promega) as a control for transfection efficiency, and carrier DNA so that the total amount of DNA in each transfection remained constant. After incubation with DNA calcium phosphate precipitates for 17-19 h, the medium was changed, and cells were incubated with or without 50 ng/ml IL-6 for an additional 24 h. CAT activity was determined from cell extracts as described previously (15). Furthermore EMSAs were performed on nuclear extracts prepared from control and IL-6-treated HepG2 cells using the human ␥FBG SAF-1RE (sequence as shown in Fig. 3A) as probe and, as positive control, the SAF-1 consensus sequence from the SAA promoter as described previously (25) following the EMSA reaction conditions described by Ray et al. (26).

Dose-dependent IL-6 Up-regulation of ␥FBG Production-
Previously we reported that IL-6 up-regulates FBG expression in both HepG2 and A549 cell lines (27)(28)(29). To confirm this expression and determine whether increasing concentrations of IL-6 would result in elevated production of ␥FBG protein, Western blotting of IL-6-treated HepG2 cell lysates was performed using monoclonal antibody J88B to specifically detect the ␥ chain. After 24 h of treatment, IL-6 induced ␥FBG production in HepG2 cells in a dose-dependent manner. The magnitude of ␥FBG induction with 50 ng/ml IL-6 was Ͼ10-fold over control in HepG2 cells (Fig. 1A). Functional assays using the Dual-Luciferase reporter assay system were performed to test the response of the human ␥FBG promoter to IL-6 treatment. Increasing concentrations of IL-6 induced luciferase activity driven by the human ␥FBG promoter construct pGL3-954 in transiently transfected HepG2 cells (Fig. 1B). At 50 ng/ml IL-6, the ␥FBG promoter was transactivated ϳ30-fold over control. These data confirm that cis-element(s) within the 954-bp promoter region efficiently mediate IL-6-induced transactivation of the human ␥FBG promoter in HepG2 cells.
Additional Type II IL-6RE Identified on the Human ␥FBG Promoter-While Mizuguchi et al. (13) demonstrated the lack of type I (C/EBP) IL-6RE on the human ␥FBG promoter region, a functional type II IL-6RE was localized to Ϫ306 to Ϫ301. The type II consensus motif serves as a docking site for IL-6-activated Stat3 to transactivate expression of Group 2 APP genes (i.e. positive APP up-regulated by 2-10-fold in response to IL-6 and also possessing the type II IL-6RE) such as the rat ␣ 2 M (24) and rat ␥FBG (14) genes. However, past attempts to show Stat3 binding to the functional IL-6REs of the human ␥FBG promoter were unsuccessful (13,15). IL-6 is a definitive inducer of expression of all three FBG genes across species (7,13,(15)(16)(17); however, the role of Stat3 binding to the IL-6REs on FBG promoter elements and functional transactivation remains unresolved. Therefore, we examined whether there might be additional IL-6-responsive elements on the human ␥FBG promoter by analysis of the promoter region up to Ϫ1000 bp from the ϩ1 site of transcription initiation. By computerassisted analysis, we looked for the type II IL-6 consensus motif CTGG(G)AA found in most Group 2 APP genes (30) and the SAF-1 consensus sequence RGGGRAGGRR where R represents any purine (31). In addition to the functional type II IL-6REs previously identified (13), two more putative type II IL-6REs were found. We designated these sites from promoterproximal to promoter-distal as Sites I, II, and III, corresponding to the nucleotide positions on the human ␥FBG promoter: Site I, Ϫ157 to Ϫ151; Site II, Ϫ306 to Ϫ301; and Site III, Ϫ531 to Ϫ525 ( Fig. 2A). No additional SAF-1 binding sites were found on the human ␥FBG promoter other than the one at position Ϫ271 to Ϫ262 (15).
Functional Analysis by Luciferase Reporter Assays of the 5Ј-Flanking Region of the Human ␥FBG Promoter in Response to IL-6 -To test whether each putative IL-6RE plays a role in mediating IL-6 induction of the human ␥FBG gene, progressive deletions of the 5Ј-flanking promoter region were cloned in front of the luciferase reporter gene in pGL3-basic (Fig. 2B). These constructs were transiently expressed in HepG2 cells, and the luciferase activity with IL-6 was compared with that without IL-6 to get the relative -fold induction, which represents the capacity of each promoter construct to respond to IL-6 treatment. For purposes of comparison, the promoter activity of pGL3-600, which contains three putative type II IL-6REs and one SAF-1 site, is defined as 100% activity (Fig. 2B). The loss of the distal IL-6RE in pGL3-400 resulted in a 25% loss of promoter responsiveness to IL-6 treatment (p ϭ 0.0004). Deletion of an additional 100 bp, which, in addition to Site III, eliminated the middle IL-6RE corresponding to Site II, led to greater than 85% reduction in promoter activity in response to IL-6 (p Ͻ 0.0001). Although the deletion of the next 100 bp resulted in loss of the SAF-1 site as well as Sites II and III, the promoter activity of pGL3-200 in response to IL-6 treatment was similar to that of pGL3-300 (p ϭ 0.8501) (Fig. 2B). Only 4% promoter activity remained in response to IL-6 in pGL3-100 in which all putative IL-6-responsive elements were deleted. These results indicate that the three putative type II IL-6REs on the human ␥FBG 600-bp promoter show differing degrees of responsiveness to IL-6 induction; Site II plays the most significant role in IL-6-induced gene expression, which is consistent with previous data (13). About 25% of the transactivation potential of the ␥FBG promoter construct resides in the Ϫ600 to Ϫ400 region, which contains the distal IL-6RE (Fig. 2A). These results sug- gest that Sites II and III account for the majority of the wild type promoter activity defined by pGL3-600. While Site I has minimal transactivation potential in response to IL-6, its loss (pGL3-200 to pGL3-100) resulted in a statistically significant reduction in luciferase activity (p ϭ 0.0110). Deletion of Ϫ300 to Ϫ200, which includes the SAF-1 IL-6RE, did not affect luciferase expression in response to IL-6 treatment (Fig. 2B).

Functional Analysis by Luciferase Reporter Assays of Sitedirected Internal Deletions of the Human ␥FBG Promoter in
Response to IL-6 -Although the results from functional analysis of the deletion constructs suggest that Sites II and III account for the majority of the IL-6 enhancer activity of the human ␥FBG promoter, we cannot rule out the possibility that the combination of any two sites may be functionally important in response to IL-6. In addition, the IL-6 responsiveness of one or a combination of the putative IL-6REs may have been compromised by the 100-or 200-bp deletion of flanking sequences within the 600-bp promoter region. Therefore, the functional activity of site-directed internal deletions, a mutation of each element, or a combination thereof was tested in transient transfection assays. The luciferase activity of the wild type promoter (pGL3-600) in response to IL-6 was considered 100%. Deletion of the promoter-proximal IL-6RE led to a 37.2% reduction in promoter activity (p ϭ 0.0011) (Fig. 3B), whereas using the series deletion constructs in transient expression assays showed that Site I had minimal IL-6 enhancer activity (Fig. 2B). Site-directed mutation of the SAF-1RE caused a 30% reduction in promoter activity. Similarly deletion of Site III resulted in about a 25% reduction in full promoter activity (p ϭ 0.029). Thus, the loss of Site I, Site III, or SAF-1 functional IL-6REs reduced wild type promoter activity by 25-37%. In contrast, in all constructs in which Site II was deleted, the promoter activity in response to IL-6 was reduced by 75-85% compared with pGL3-600 (p Ͻ 0.001). Notably, when all IL-6REs are lost, the promoter retained almost 13-15% of the transactivation potential of the wild type promoter in response to IL-6 (p Ͻ 0.001). These data indicate that Site II drives the majority of IL-6 enhancer activity on the human ␥FBG promoter but that full activity requires all IL-6REs (Fig. 3B).
Human ␥FBG Type II IL-6RE-Stat3 Complex Formation Detected by EMSA-To determine whether IL-6-activated Stat3 can bind to one or more of the functional type II IL-6REs on the human ␥FBG promoter, EMSAs were performed using IL-6treated HepG2 nuclear extracts. Nuclear extracts from HepG2 cells treated with 2.5 ng/ml IL-6 for 20, 40, 80, and 160 min were incubated with end-labeled probes corresponding to IL-6RE Sites I, II, or III; the ␣ 2 M probe was used as the positive control. EMSA results show that Sites I and III formed protein-DNA complexes that migrated to the same position in the gel as the ␣ 2 M complex (Fig. 4A); however, Site II did not form a detectable complex with nuclear extracts from HepG2 cells when treated with 2.5 ng/ml IL-6. In comparison to the ␣ 2 M complex, faint complexes were formed using the ␥FBG probes for Sites I and III, suggesting much weaker affinity of these sites for activated Stat3. To confirm the specificity of the reaction, EMSA was performed using 100ϫ cold ␣ 2 M probe as competitor for DNA-protein binding and supershift of the complex with anti-Stat3 or anti-Stat1 antibodies. The results confirmed that the protein complex formed with the ␣ 2 M probe was specific for IL-6-activated Stat3 (Fig. 4B). Cross-competition experiments using 100ϫ cold probe indicated that Sites I and III were reasonably efficient competitors for complex formation of IL-6-activated Stat3 with the labeled ␣ 2 M probe; however, Site II showed only partial inhibition (Fig. 4C). These results indicate that binding affinities for activated Stat3 differ among the type II IL-6REs. Using the intensity of band formation as a relative indicator of probe affinity for Stat3, the data show the following rank order from strongest to weakest: ␣ 2 M Ͼ Ͼ Site I Ͼ Site III Ͼ Ͼ Site II (Fig. 4, C and D). Only very faint complexes could be formed with the Site II probe over 15-30 min and diminished by 90 min when cells were treated with higher concentrations of IL-6. The data indicate that complex formation with ␥FBG Sites I-III respond in a dose-dependent manner to IL-6 treatment as well (Fig. 4D). Furthermore anti-Stat3 but not anti-Stat1 antibodies ablated complex formation with Sites I and III (Fig. 4D). Taken together, the results demonstrate that, at 50 ng/ml IL-6, at least Sites I and III form demonstrable complexes with activated Stat3 from HepG2 cells.
Overexpressed Wild Type Stat3 Potentiates IL-6-induced ␥FBG Promoter Activity-The functional data indicate that all three type II IL-6REs contribute to IL-6-induced expression of the ␥FBG 600-bp promoter in HepG2 cells (Figs. 2 and 3). In addition, the EMSA analyses suggest that IL-6-activated Stat3 likely plays a role in transactivation of luciferase gene expression driven by the ␥FBG promoter (Fig. 4). Because the relative affinity of Stat3 for the ␥FBG IL-6REs is low, we determined whether overexpression of wtStat3 in HepG2 cells would lead to IL-6-enhanced transactivation. In response to 50 ng/ml IL-6, the relative luciferase activity using the pGL3-600 reporter construct was increased 25-fold, whereas, using pGL3-600-⌬I,II,III, a residual induction by IL-6 was observed. In the presence of increasing concentrations of wtStat3 transiently expressed in IL-6-treated HepG2 cells, the relative -fold luciferase activity was further enhanced 37-fold above the not IL-6-treated pGL3-600, resulting in 1.5-fold potentiation of ␥FBG promoter activity over IL-6-treated pGL3-600 due to Stat3 overexpression (Fig. 5). However, overexpression of wtStat3 did not potentiate IL-6-induced luciferase expression of pGL3-600-⌬I,II,III suggesting that no additional or cryptic Stat3 binding sites remain in the this 600-bp promoter region. Although, due to the high levels of endogenous Stat3 in HepG2 cells, the use of the transdominant Stat3⌬55C as a competitive inhibitor of IL-6-induced Stat3 transactivation was limited (32), in the presence of increasing concentrations of Stat3⌬55C, the 1.5-fold potentiation in the ␥FBG promoter activity of pGL3-600 induced by wtStat3 was inhibited (not shown). Taken together, these data indicate that type II IL-6REs on the human ␥FBG promoter are required for IL-6-mediated wtStat3 potentiation of gene expression in HepG2 cells.
Induction of Human ␥FBG Promoter Activity Is IL-6 Concentration-dependent-EMSA analysis indicated that the degree of Stat3 complex formation with each of the ␥FBG IL-6REs was dependent on the concentration of IL-6 used to stimulate HepG2 cells (Fig. 4). Therefore, we investigated whether the transactivation potential of each of the type II ␥FBG IL-6REs was affected by increasing concentrations of IL-6. Using the wild type human ␥FBG 600-bp promoter as control, we tested the promoter responsiveness of single-site (Fig. 6A), two-site (Fig. 6B), or three-site (Fig. 6C) internal deletions to different concentrations of IL-6. The results showed that the magnitude of luciferase activity promoted by each single or multiple site-deleted construct was responsive to increasing concentrations of IL-6; however, each IL-6RE on the ␥FBG promoter showed the same rank order of transactivation potential (Site II Ͼ Ͼ Site I Ͼ Site III) as determined by the data shown in Fig. 3B regardless of the IL-6 concentration.
Dexamethasone Potentiates IL-6-induced Activation of the Human ␥FBG Promoter in the Absence of a Functional GRE-GCs synergistically enhance IL-6 induction of Group 2 APP genes even in the absence of a functional GRE (7). The human ␥FBG 600-bp promoter region lacks the functional GRE at Ϫ1116 to Ϫ1102 as described by Asselta et al. (21). Therefore, we compared the functional activity of the wild type pGL3-600 promoter construct to pGL3-600-⌬I,II,III, in which all Stat3 binding sites were deleted, in response to 50 ng/ml IL-6, 0.1 M DEX, or both. Whereas IL-6 induced the wild type promoter activity 30-fold as expected, DEX alone induced pGL3-600 luciferase activity by 25-fold over pGL3-600 activity in the absence of IL-6 treatment (Fig. 7A, white bars) (p Ͻ 0.001). The 2.25-fold induction in wild type promoter activity by 0.1 M DEX in the presence of 50 ng/ml IL-6 compared with IL-6 treatment of pGL3-600 alone was statistically significant (p ϭ 0.0063) (Fig. 7A, white bars), consistent with a previous report (33). In contrast, when all three type II IL-6REs were deleted, 0.1 M DEX failed to potentiate the residual promoter activity in the presence of IL-6 ( Fig. 7A, black bars), suggesting that the SAF-1RE does not contribute to DEX enhancement of ␥FBG   FIG. 3. Functional analysis of IL-6REs by site-directed deletions or mutations using luciferase reporter assays. A, sequences of type II and SAF-1 IL-6REs on the human ␥FBG promoter region. Upper schematic, the wild type promoter with the four putative IL-6 binding sites is displayed. The type II IL-6 consensus sequences (CTGG(G)AA) are depicted inside the ovals; the SAF-1RE is enclosed in a box. The flanking regions for each of the IL-6REs are shown. The sequences for Sites I-III and SAF-1 also represent the specific ␥FBG oligos used in EMSAs. The numbers indicate the map position of each region within the human ␥FBG promoter. Lower schematic, ␥FBG promoter fragment with type II IL-6 sequences deleted (X in box) and SAF-1 sitedirected mutations (X in diamond). B, transient transfection using these mutant constructs fused to the luciferase (luc) gene was performed as described under "Experimental Procedures." For purposes of comparison, the promoter activity of the wild type ␥FBG promoter (pGL3-600) is defined as 100%.
FIG. 4. The ␥FBG type II IL-6REs form protein-DNA complexes with HepG2 nuclear extracts treated with IL-6. A, HepG2 cells were treated with IL-6 (2.5 ng/ml) for 0, 20, 40, 80, and 160 min, and cells were collected for nuclear extract preparation. The sequences of oligos used for EMSA probes are as depicted in Fig. 3A. The rat ␣ 2 M IL-6RE that binds strongly to activated Stat3 was used as a positive control. Nine g of IL-6-treated HepG2 nuclear extracts were used in each binding reaction. B, the specificity of Stat3 binding to ␣ 2 M IL-6RE was tested by competition with 100ϫ nonlabeled oligo and by supershift analysis with antibody against Stat3 or Stat1. HepG2 cells were treated with IL-6 (2.5 ng/ml) for 40 min, and cells were collected for nuclear extract preparation. Two g of nuclear extracts were used for each binding reaction. C, each nonlabeled ␥FBG IL-6RE-containing oligo was used at 100ϫ concentration for cross-competition with labeled ␣ 2 M probe in binding to Stat3 from 2 g of nuclear extract from IL-6-treated HepG2 cells. D, HepG2 cells were treated with increasing concentrations of IL-6 (2.5, 10, and 50 ng/ml) for 0, 15, 30, and 90 min, and cells were collected for nuclear extract preparation. The promoter activity in the presence or absence of IL-6. Together these data indicate that the DEX effect requires the presence of one or more type II IL-6-responsive elements in the ␥FBG 600-bp promoter element and that DEX induction of pGL3-600 occurs in the absence of a known GRE.
To further evaluate how each individual IL-6RE contributes to DEX enhancement of IL-6 induction, we compared the transactivation potential of 0.1 M DEX in the presence of 50 ng/ml IL-6 on each of the type II IL-6RE site-deleted constructs to the wild type pGL3-600 promoter. The data show that DEX treatment enhanced IL-6-induced wild type promoter activity by ϳ2-fold (p ϭ 0.0214), whereas deletion of Site I or II individually or in combination with Site III resulted in complete loss of the DEX-enhanced promoter activity (Fig. 7B, black bars). Furthermore deletion of Site II resulted in inhibition of DEX potentiation as well as IL-6 induction of ␥FBG promoter activity (Fig. 7B, black bars). Deletion of the promoter-distal IL-6RE (Site III) resulted in an equivalent potentiating effect by DEX as that observed for the wild type promoter. Whereas all three type II ␥FBG IL-6REs contributed to the full activity of the 600-bp promoter in response to IL-6, the data suggest that together Sites I and II but not Site III are required for DEXenhanced expression of IL-6-induced ␥FBG promoter activity.

Elucidation of Human ␥FBG IL-6 Promoter Elements Using CAT Reporter Constructs Compared with Luciferase Reporter
Constructs-Because the IL-6-responsive sites identified on the human ␥FBG promoter using the luciferase reporter constructs described in this report differ considerably from the site identified by A. Ray (15) using CAT gene reporter constructs, we repeated the reporter gene assays following the methods used by Ray (15). This involved subcloning our human ␥FBG promoter constructs into the pCAT3-basic vector, performing transient transfections by the calcium phosphate method, measuring relative changes in promoter activity induced by IL-6 using the [ 14 C]chloramphenicol acetyltransferase assay, and performing EMSAs with the ␥FBG SAF-1 sequence compared with the SAF-1RE of SAA as a positive control. A representative thin layer chromatogram of [ 14 C]chloramphenicol acetylated by CAT expressed in response to IL-6 transactivation of the various human ␥FBG pCAT3 constructs is shown in Fig.  8A; the relative CAT activity reported as the mean and S.E. of three independent experiments is shown in Fig. 8B. A comparison in expression efficiency of the human ␥FBG 600-bp element promoter element in pCAT3-basic, designated pCAT3-600, shows that IL-6 induced CAT activity by 6.4-fold using calcium phosphate, whereas IL-6 induced CAT activity by 18.5fold using Superfect for transient transfection (Fig. 8B, compare data set 2 to data set 10) (p Ͻ 0.0001). This 3-fold differ-ence in transfection efficiency would account for the overall higher levels in ␥FBG promoter activity induced by IL-6 using the pGL3-luciferase expression system. Furthermore the luciferase reporter system is more sensitive than the CAT reporter system as shown by comparing the 18.5-fold IL-6-induced expression of pCAT3-600 (Fig. 8B, data set 10) to the 25-30-fold IL-6-induced expression of pGL3-600 (Figs. 5-7) by Superfectmediated transfection. Therefore, the increased sensitivity of the luciferase reporter system and the more efficient expression of reporter constructs by the Superfect method of transient transfection in HepG2 cells allowed us to measure smaller changes in reporter gene activity.
Aside from the lower sensitivity of the CAT assay and efficiency of expression by the calcium phosphate precipitation method of transfection, the results we obtained with the human ␥FBG 600-bp promoter-CAT reporter constructs were essentially the same as the results we observed with the luciferase FIG. 5. Overexpression of wild type Stat3 potentiates ␥FBG promoter activity in response to IL-6. HepG2 cells were cotransfected with increasing amounts of wtStat3 (as indicated) plus 500 ng of the wild type ␥FBG promoter (pGL3-600, open squares) construct or the ␥FBG promoter construct with deletions in all three type II IL-6REs (pGL3-⌬I,II,III, closed diamonds) as reporter. The luciferase activity of the reporter with no IL-6 treatment was set to 1 (n ϭ 5).
reporter constructs (Fig. 8B and Table I). We showed by EMSA that protein from HepG2 cell nuclear extracts binds to the positive control probe derived from the SAA promoter as well as to the ␥FBG probe containing the SAF-1RE at nucleotides Ϫ271 to Ϫ262 as described by Ray (15). The specificity of SAF-1 binding was confirmed by competition with excess nonlabeled oligo (Fig. 8C). However, IL-6 treatment did not alter the amount of SAF-1 that bound to either the SAA or ␥FBG probes (Fig. 8, C and D). Furthermore, in contrast to the results of Ray (15), the HepG2 cells used in this study express high constitutive levels of SAF-1 (Fig. 8C). Using increasing concentrations of nonlabeled probes and 2 g of nuclear extracts, cross-competition of protein binding to labeled SAA probe with cold ␥FBG SAF-1 oligo and vice versa was performed. Although the ␥FBG SAF-1 sequence competitively inhibited binding of nuclear protein to the labeled SAA probe, it failed to completely inhibit this binding. In contrast, nonlabeled SAA probe completely abolished the binding of protein from HepG2 nuclear extracts to the ␥FBG SAF-1 probe, suggesting that SAF-1 binds with lower affinity to the ␥FBG SAF-1 promoter element (Fig.  8D). Deletion or mutation of the ␥FBG SAF-1 site had minimal effect on IL-6 induction of the human ␥FBG promoter containing the type II IL-6-responsive sites, in particular Site II (Figs.  2B, 3B, and 8B). Taken together, the data in this report suggest that the ␥SAF-1 site contributes little to the human ␥FBG promoter activity in response to IL-6.

DISCUSSION
In addition to the IL-6 family of cytokines, Stat3 is activated by leptin, granulocyte colony-stimulating factor, and epidermal growth factor. Targeted disruption of the stat3 gene results in lethality between embryonic days 6.5 and 7.5, demonstrating that Stat3 is essential early in embryonic development (34). Analyses of conditional cell type-or tissue-specific Stat3-deficient mice indicate that Stat3 plays a crucial role in a variety of biological functions including cell growth, cell motility, and suppression and induction of apoptosis (18,35,36). In addition, targeted disruption of genes involved in Stat3 activation such as IL-6 and gp130, as well as Stat3 target genes including members of the C/EBP transcription factor family, leads to dysregulation of innate immunity, metabolism, hematopoiesis, or embryogenesis (37)(38)(39)(40)(41). Stat3 is the main mediator of APP gene induction downstream of IL-6 family cytokines and gp130 signaling (42). IL-6 activation of Stat3 is essential for inducing APP gene expression including the FBG A␣, B␤, and ␥ chain genes in response to lipopolysaccharide, IL-6, or localized tissue damage in conditionally Stat3-deficient mice (38,39). Furthermore the functional importance of the Stat3 cognate binding sites in promoting expression of rat and human A␣, B␤, and ␥ chain genes is well established. However, the inability to demonstrate direct binding of Stat3 to wild type promoter elements on each of these genes (13,(15)(16)(17)43), except the rat ␥ chain IL-6REs (14), remains unexplained. Similarly attempts to demonstrate binding of the mouse haptoglobin IL-6RE to Stat3 by EMSA failed for unknown reasons, although functional assays demonstrated it as the cis-acting element required for regulation by Stat3 (32).
We demonstrate for the first time that, like the rat promoter, the human ␥FBG promoter contains three functional type II IL-6REs within 600 bp upstream of the transcription initiation site. The relative position of promoter elements on the human 600-bp promoter is compared with corresponding elements on the rat ␥FBG promoter region in Fig. 9. We confirmed that the functional IL-6RE identified at Ϫ306 to Ϫ301 (13) (designated Site II in Fig. 9) is the major IL-6-responsive site on the human ␥FBG promoter and provide evidence that Sites I (Ϫ157 to Ϫ151) and III (Ϫ531 to Ϫ525) contribute to full promoter activity in response to IL-6. In contrast to the report by Ray (15), our results do not show that the SAF-1 site is the major IL-6responsive element on the human ␥FBG promoter; therefore, we repeated a number of experiments using the CAT reporter gene expression system to verify the positions of the IL-6REs. The results we obtained with the human ␥FBG promoter-CAT reporter constructs were essentially the same as the results we observed with the luciferase reporter constructs and did not replicate the work of Ray (15). Although we cannot fully explain the reasons for this discrepancy, one major difference between our report and that of Ray is that we see high levels of constitutive expression of SAF-1 in our HepG2 cells, whereas Dr. Ray did not (Fig. 5 of Ref. 15). Furthermore EMSA results showed that IL-6 did not induce binding of a HepG2 nuclear protein to labeled probes corresponding to the SAF-1 consensus motif of either SAA or ␥FBG. These data suggest that high constitutive levels of SAF-1 may abrogate or mask IL-6 induction of low levels of SAF-1. SAF-1 is a zinc finger transcription factor originally cloned from a rabbit brain cDNA library (25) and is Ͼ95% identical to the deduced primary structure of mouse Pur-1 (44) and human MAZ (c-Myc-associated zinc finger protein) (45) orthologs, which are ubiquitously expressed housekeeping genes. Interestingly SAF-1 mRNA is abundantly expressed in mouse heart, liver, lung, brain, skeletal muscle, testis, and kidney (25), suggesting that SAF-1 protein is ubiquitously expressed as well. In light of these data, we cannot  1-10) based on the promoter construct tested (as identified in B) without (Ϫ) or with (ϩ) 50 ng/ml IL-6. B, relative -fold change in CAT activity without (white bars) or with (black bars) 50 ng/ml IL-6 for 24 h (n ϭ 3). pCAT3-⌬,⌬ represents pCAT3-⌬I,II,III,⌬SAF-1. Data set 10 represents transient transfection of HepG2 cells using Superfect to compare the transfection efficiency to calcium phosphate-mediated transfection (data set 2) with pCAT3-600. C, EMSA using the SAF-1 cis-element from SAA and ␥FBG as probes. HepG2 cells were incubated with medium with or without IL-6 (50 ng/ml) for 24 h. Labeled SAF-1 site probes were incubated with 10 g of nuclear extracts for EMSA. The specificity of binding was determined by incubation of 10 g of HepG2 nuclear extracts with 100ϫ nonlabeled "self " probe. D, the specificity of the DNA-protein binding interactions was further tested by cross-competition using 2 g of nuclear extracts, labeled SAA or ␥FBG probe, and increasing concentrations of the heterologous nonlabeled probe as competitor. explain why the HepG2 cells used by Ray (15) did not express constitutive levels of the SAF-1/MAZ transcription factor.
Herein we show, also for the first time, that all three type II IL-6-responsive elements identified on the human ␥FBG 600-bp promoter form demonstrable protein complexes with IL-6-activated Stat3 from HepG2 nuclear extracts. Based on the relative intensity of each complex, the strength of Stat3 binding to Sites I and III appears stronger than that with Site II. In addition, the degree of complex formation was IL-6 dosedependent. However, compared with the magnitude of Stat3 binding to the ␣ 2 M type II core element, Stat3 complex formation with the human ␥FBG type II IL-6RE probes of the same specific activity occurred with much lower intensity. The specificity of Stat3 binding to ␣ 2 M and ␥FBG IL-6REs was confirmed by cold competition with excess probe and either supershift (␣ 2 M) or ablation of complex formation (␥FBG Sites I and III) with anti-Stat3 but not anti-Stat1 antibodies. In crosscompetition studies, 100-fold excess nonlabeled probes for ␥FBG Sites I and III successfully competed off most of the Stat3 binding to the labeled ␣ 2 M probe, whereas excess ␥FBG Site II probe only partially inhibited Stat3-␣ 2 M complex formation. In addition, overexpression of wild type Stat3 potentiated the IL-6-induced activity of the wild type 600-bp promoter 1.5-fold, which could be inhibited by expression of the transdominant Stat3⌬55C construct, whereas site-directed deletion of all three type II ␥FBG IL-6REs abrogated this potentiation. Furthermore we show that Sites I and II on the ␥FBG 600-bp promoter, which lacks a GRE, are critical for DEX potentiation of IL-6induced promoter activity. The promoter-distal Site III, which lies upstream of a negative response region identified by Mizuguchi et al. (Fig. 9), had no effect on DEX potentiation of IL-6 activity.
Taken together, the data in this report show the following rank order of Stat3-DNA complex formation from strongest to FIG. 9. Comparison of human and rat ␥FBG 600-bp promoter elements. The basal promoter and enhancer sequences identified to date are shown on the Ϫ600 bp portion of human (H, top strand) and rat (R, bottom strand) ␥FBG promoter regions. Both human and rat ␥FBG promoters contain a TATA-like sequence, one CAAT box, and one upstream stimulatory factor (USF) binding site to constitute the minimal promoter. An Sp1 site is found in the rat promoter that is lost in the human ␥FBG basal promoter region. The previously identified type II IL-6-responsive element is located at Ϫ306 to Ϫ301 on the human ␥FBG promoter and is designated as Site II; the positions of the two additional type II IL-6REs identified as Sites I and III in this report are shown; the three type II IL-6REs previously identified on the rat ␥FBG promoter are indicated (references cited in text). No SAF-1 site has been identified in the rat ␥FBG promoter. The NF-B site that overlaps the rat Stat3-II site is not conserved in the human ␥FBG promoter. Negative response regions previously identified are indicated. hStat3, human Stat3; rStat3, rat Stat3. weakest: ␣ 2 M Ͼ Ͼ Site I Ͼ Site III Ͼ Ͼ Site II. In contrast, the functional studies in transiently transfected HepG2 cells indicate a reverse in rank order for ␥FBG IL-6RE transactivation potential with Site II Ͼ Ͼ Site I Ͼ Site III; however, full promoter activity in response to IL-6 requires all three type II sites. The weak binding of Stat3 to the wild type ␥FBG IL-6REs explains, in part, the inability to demonstrate complex formation with wild type IL-6 promoter regions of several APP genes (13,15,17,32,43). Indeed the results of EMSA reported by Zhang et al. (14) demonstrate that complex formation of Stat3 with the cis-acting elements on the rat ␥ chain promoter is also much weaker than Stat3 binding to the rat ␣ 2 M probe. Collectively the data of others together with the data in this report indicate that the magnitude of Stat3 binding to CTGG(G)AA promoter elements is not the sole determinant in transactivation of gene expression in response to IL-6. The presence of three IL-6REs with weak binding affinity for Stat3, as found in the rat and human ␥FBG promoters, may be as efficient at driving gene expression as one IL-6RE with strong affinity for Stat3. This is consistent with the report showing that oligomerization of the mouse haptoglobin IL-6RE in the absence of C/EBP-responsive sites is required to obtain the same magnitude of Stat3 induction of the haptoglobin promoter in which C/EBP sites flanking the single IL-6RE enhance Stat3 transactivation (32). Stat family members, including Stat3, function as transactivators by recruiting p300/CBP (46) and other nuclear cofactors (47) that aid in complex formation of binding partners to their cognate enhancer elements as well as with the basal transcription machinery. By integrating multiple signaling events with the transcription apparatus, p300/CBP cofactors regulate appropriate levels of gene activity in response to diverse physiological cues. In addition, the acetyltransferase activity of p300/ CBP modifies transcription factors and chromatin through acetylation to modulate promoter activity (48). Cooperative interaction of Stat3 with other transcription factors is demonstrated by the DEX potentiation of IL-6 induction of the ␥FBG promoter. Zhang et al. (33) have shown that IL-6-activated Stat3 associates with ligand-bound GR to form a transactivating complex, which can function through either an IL-6RE or a GRE. Thus, ligand-bound GR acts as a transcriptional coactivator without direct association with the GR DNA binding motif (33). Similarly, in the absence of a functional GRE, DEX coactivates IL-6 induction of mouse haptoglobin gene expression driven by the promoter-proximal region. The potentiating effect of GC and IL-6 relies on the functional interaction between Stat3 and ligand-bound GR (7,32). The proximity of the IL-6RE to the basal transcription machinery may also play a role in the efficiency of DEX-enhanced gene expression through direct interaction with Stat3 and other transcriptional coactivators such as p300/CBP (49). This concept is supported by our results demonstrating that the promoter-distal IL-6RE at Site III is not involved in DEX potentiation of IL-6 transactivation of the human ␥FBG promoter. The binding of Stat3 complexed with ligand-activated GR to the IL-6RE at Ϫ531 to Ϫ525 may place this complex too far from the promoter-proximal Sites II and I, both of which are required for DEX potentiation of IL-6-induced promoter activity. Such an interaction would effectively remove the distal site from interactions with the putative enhancer complex formed with Stat3 and GR on Sites II and I as well as with the basal transcription apparatus including coactivators such as p300/CBP. Fuller and Zhang (7) have proposed further that chromatin remodeling by nuclear cofactor acetylation of histones upstream of the ␥FBG gene, which is positioned at the 5Ј proximal end on both the human and rat FBG loci, may facilitate the coordinated transcriptional activa-tion of the A␣ and B␤ chain genes during an acute phase response. Because elevated levels of circulating FBG correlate with increased risk for cardiovascular disease and stroke, elucidation of the molecular mechanisms that regulate FBG expression becomes important for control of diseases associated with high levels of FBG (50,51). Thus, in addition to furthering our understanding of how FBG is up-regulated in response to inflammatory stimuli, there is considerable interest in elucidation of the mechanisms that down-regulate FBG gene expression with the hopes of identifying agents that will normalize elevated plasma FBG in patients at risk for cardiovascular disease.