Transcriptional Roles of CCAAT/Enhancer Binding Protein-β, Nuclear Factor-κB, and C-promoter Binding Factor 1 in Interleukin (IL)-1β-induced IL-6 Synthesis by Human Rheumatoid Fibroblast-like Synoviocytes

The involvement of interleukin (IL)-6 in the pathogenesis of rheumatoid arthritis (RA) has been recently demonstrated. IL-1β stimulated rheumatoid fibroblast-like synoviocytes (FLSs) to produce IL-6 in a concentration- and time-dependent manner. In the present study we investigated how the IL-6 promoter is transcriptionally regulated in rheumatoid FLSs in response to a physiologically relevant mediator of inflammation, IL-1β. Deletion analysis showed that the IL-6 promoter is regulated by two positive elements (located at −159 to −142 base pairs (bp) and −77 to −59 bp). Electrophoretic mobility shift assays revealed that CCAAT/enhancer binding protein-β (C/EBPβ) binding to nucleotides −159 to −142 bp was constitutively present. The probe corresponding to nucleotides −77 to −59 bp gave three positive bands. The two slower migrating bands were induced by IL-1β and comprised an nuclear factor (NF)-κB p50/p65 heterodimer and a p65/p65 homodimer. The faster migrating band was constitutively expressed and identified as Epstein-Barr virus C-promoter binding factor 1, CBF1. Site-specific mutagenesis analysis demonstrated that the NF-κB and CBF1 binding elements regulated inducible activity of the IL-6 promoter in response to IL-1β stimulation, whereas the C/EBPβ binding element mainly regulated basal activity. We also provide the first evidence that CBF1 functions as a positive regulator of human IL-6 gene transcription.

Rheumatoid arthritis (RA) 1 is a chronic inflammatory disease in which the synovial environment is characterized by intense immunological activity (1). In particular, the macrophage-like synoviocytes and fibroblast-like synoviocytes (FLSs) of the hyperplastic lining layer exhibit an activated phenotype (2)(3)(4). These cells are a major source of several inflammatory cytokines such as IL-1, tumor necrosis factor-␣, and IL-6, proteins which play crucial roles in the pathophysiology of RA (2,5,6).
IL-6 is a pleiotropic cytokine involved in T cell growth (7), B cell differentiation that might lead to the production of rheumatoid factors (8,9), and the induction of acute phase proteins (10). Immunological disorders often associated with RA include polyclonal plasmacytosis, production of autoantibodies, increased levels of acute phase proteins, and an increased number of peripheral blood platelets, all of which are related to the biological actions of IL-6 (11). In fact, IL-6 is produced mainly by FLSs in the synovium (5,9) and is found at high levels in the synovial fluids and the serum obtained from RA patients (12,13). Recently, administration of anti-IL-6 receptor antibody dramatically improved the symptoms and clinical markers of RA patients (14), indicating that IL-6 is one of the key cytokines for the development of RA.
So far, several regulatory elements such as AP-1, multiple responsive element (MRE), 5Ј-C/EBP␤, 3Ј-C/EBP␤, and NF-B have been identified within the human IL-6 promoter region (see Fig. 2) (11). These promoter elements are regulated in a cell type-specific manner. AP-1, MRE, 5Ј-C/EBP␤, and NF-B binding elements are required for a murine monocyte/macrophage cell line, the PU5-1.8 (15), whereas 5Ј-C/EBP␤, 3Ј-C/ EBP␤, and NF-B binding elements are required for a human monocytic leukemia cell line, the THP-1 cell (16). The NF-B site is sufficient for the full expression of this promoter by a human T cell leukemia virus type I-infected T cell line (17). More recently, it was reported that Epstein-Barr virus C-promoter binding factor 1 (CBF1) acts as a negative regulator of the IL-6 gene in a mouse embryonal carcinoma, the F9 cell, and in a mouse fibrosarcoma cell, the L929sA (18,19). CBF1 is a DNA-binding protein that is targeted by the viral transactivator Epstein-Barr virus nuclear antigen 2 in Epstein-Barr virusinfected human B lymphocytes (20). CBF1 is also known as recombination binding protein J involved in the rearrangement of the immunoglobulin V(D)J gene (21), although conflicting data were reported (22).
IL-1␤ is a potent activator of IL-6 synthesis by human rheumatoid FLSs (9). Although several transcriptional elements have been implicated in the induction of human IL-6 gene by IL-1␤, it is now apparent that the regulatory mechanisms of human IL-6 gene transcription are more complex and divergent among various cell types and phenotypes than initially described (11). There have been no previous reports regarding the molecular mechanisms of basal transcriptional activity of the IL-6 gene in the absence of inflammatory stimulation. In addition, transcriptional regulation of this gene in rheumatoid FLSs has not been elucidated. In the present study we inves-tigated how the IL-6 promoter is regulated in rheumatoid FLSs both, in the unstimulated state and in response to IL-1␤.
Cells-Synovial tissue samples were obtained from patients with RA or osteoarthritis (OA) undergoing total joint replacement. All RA and OA patients were evaluated by a certified rheumatologist and were diagnosed as having RA and OA, respectively, according to the criteria of the American College of Rheumatology (23,24). Written informed consent was obtained from each patient. Each tissue specimen was minced and then digested with 4 mg/ml collagenase for 2 h at 37°C. Cells were plated in RPMI 1640 (Nikken, Kyoto, Japan) with 10% fetal bovine serum (Life Technologies, Inc.). When the cells had grown to confluence, they were treated with trypsin/EDTA and split at a 1:4 ratio. For experiments, FLSs were used in passages 3-10. A FLS line, MH7A, 2 was established from synoviocyte cultures obtained from a 53-year-old female RA patient. The MH7A cells were plated in RPMI 1640 (Nikken, Kyoto, Japan) supplemented with 10% fetal bovine serum (Life Technologies, Inc.). When the cells had grown to confluence, they were treated with trypsin/EDTA and split at a 1:5 ratio.
Measurement of IL-6 -Cells at confluence were cultured in RPMI 1640 medium supplemented with 0.1% bovine serum albumin for the designated time periods. The culture supernatants were kept frozen until measured for IL-6, which was done by a specific sandwich enzymelinked immunosorbent assay. Briefly, supernatants or serial dilutions of recombinant IL-6 standards (Genzyme) were incubated overnight at 4°C in 96-well microtiter plates (Nunc, Roskilde, Denmark) previously coated overnight at 4°C with anti-human-IL-6 monoclonal antibody (2 g/ml; R&D Systems, Minneapolis, MN) and then for 2 h at room temperature for saturation. After the plates had been washed, biotinylated anti-human IL-6 polyclonal antibody (2 g/ml; R&D) was added. The incubation was carried out for 4 h at room temperature. After subsequent incubation (2 h, room temperature) with horseradish peroxidase-conjugated avidin (Zymed Laboratories Inc., South San Francisco, CA), 3,3Ј,5,5Ј-tetramethylbenzidine (Dojindo Labs., Kumamoto, Japan) was added to the wells. The absorbance at 450 nm was measured by a microplate reader (Bio-Rad). The minimum detection limit of the assays was 6.25 pg/ml.
Plasmid Construction-For generation of plasmid pIL6 -3BLuc, a SacI/XhoI fragment containing the 1169-bp (Ϫ1158 to ϩ11 relative to the transcription initiation site) BamHI/XhoI 5Ј-upstream sequence of the IL-6 gene was excised from the plasmid pGEMhIL-6 GT (Riken Gene Bank, Tsukuba, Japan) (29) and inserted into the compatible site (SacI/XhoI) of the luciferase reporter plasmid pGL3-Basic (Promega Corp., Madison, WI). A series of deletion mutants of the 5Ј-flanking region of the IL-6 gene were created as follows. For the construction of pIL6NX-3BLuc, pIL6 -3BLuc was digested with NheI and XhoI to generate the NheI/XhoI (Ϫ225 to ϩ11) fragment of the IL-6 upstream region, which was then cloned into pGL3-Basic. For the construction of pIL6BX-3BLuc, pIL6AX-3BLuc, pIL6MX-3BLuc, pIL6HX-3BLuc, and pIL6SsX-3BLuc, pIL6B-3BLuc was digested with BfaI, AatII, MseI, HaeIII, or SspI, respectively, made blunt-ended with Klenow enzyme or T4 polymerase, and released by XhoI digestion. The resulting fragments were subcloned into pBluescript® II KS(ϩ) phagemid vector (Stratagene, La Jolla, CA) which was then digested with SacI and XhoI. The SacI/XhoI fragment was next inserted into pGL3-Basic. For the construction of pIL6KX-3BLuc, a synthesized NheI/XhoI (Ϫ77 to ϩ11) fragment of the IL-6 gene was cloned into the pGL3-Basic.
The C/EBP␤ site at positions Ϫ158 to Ϫ145, the IL6-B site at positions Ϫ73 to Ϫ64, the CBF1 site at positions Ϫ67 to Ϫ60, and both IL6-B and CBF1 sites at positions Ϫ73 to Ϫ60 of the human IL-6 promoter were mutated by the insertion of synthesized DNA. The mutant sequence utilized for the C/EBP␤, IL6-B, CBF1, and both IL6-B and CBF1 sites were ACAgatatCAATCT (C/EBP␤-mt), aatATTTTCC (mt1), TTCCCtcG (mt3), and GGGATTTTagacTG (mt2), respectively. Lowercase letters represent the mutant nucleotides.
Transient Transfection and Luciferase Assay-Cells were grown to confluence in 75-cm 2 culture flasks containing RPMI 1640 medium supplemented with 10% fetal bovine serum, harvested, and resuspended in RPMI 1640 medium without phenol red to give a cell concentration of 2 ϫ 10 6 cells/ml. A 500-l volume of cell suspension, 5-10 g of the luciferase reporter plasmids, and 0.4 g of pCMV-␤-galactosidase were placed into a Bio-Rad 4-mm cuvette; electroporation (Bio-Rad Gene Pulser, 0.29 kV, 960 F) was then performed. The transfected cells were cultured for the designated time periods for each reporter plasmid in RPMI 1640 medium without phenol red. All assays were carried out in 96-well plates (Culture Plate TM ; Packard Instrument Company, Meriden, CT). The luminescence was measured with a Top Count (Packard) at the conditions of 0.15 min of counting time, a single photon counting mode, and a 15-min period of dark adaptation at 22°C (Top Count). The ␤-galactosidase activity was measured as described previously (26). Briefly, cells were washed with PBS(Ϫ), and then 100 l of 1.5 mM chlorophenol red-␤-D-galactopyranoside (Boehringer Mannheim, Germany) in PBS(Ϫ) containing 20 mM KCl, 2 mM MgSO 4 , 100 mM 2-mercaptoethanol, and 0.5% Nonidet P-40 (Wako) were added. The absorbance at 570 nm was thereafter measured by a microplate reader (Bio-Rad).
Preparation of Nuclear Extracts-Nuclear extracts were prepared by a method previously reported (27)  l of hypotonic buffer A (10 mM HEPES (pH 7.9), 10 mM KCl, 10 mM NaCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride) for 15 min on ice. The cells were then lysed by the addition of 0.6% Nonidet P-40 and vortexing for 10 s. Nuclei were separated from the cytosol by centrifugation at 12,000 ϫ g for 30 s, washed with 400 l of buffer A containing 0.6% Nonidet P-40, resuspended in buffer C (20 mM HEPES (pH 7.9), 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride), vigorously vortexed for 15 s, and incubated for 5 min on ice. This step was repeated three times. Nuclear extracts were then obtained by centrifugation at 12,000 ϫ g for 10 min. Protein concentration was measured essentially following the method of Bradford (28) using a protein dye reagent (Bio-Rad).
Electrophoretic Mobility Shift Assays (EMSAs)-Eight oligonucleotides as shown in Fig. 3 were synthesized with a DNA synthesizer (Sawady Technology, Tokyo, Japan). The 3Ј ends of the oligonucleotides were 32 P-labeled with Klenow DNA polymerase (Megaprime DNA labeling systems; Amersham International plc, Buckinghamshire, UK). Samples containing 5 g of nuclear extract were incubated with 10,000 cpm of labeled oligonucleotides and 1 g of poly(dI-dC) in a 10-l volume of 10 mM Tris (pH 7.5), 50 mM NaCl, 1 mM dithiothreitol, 1 mM EDTA, 5% glycerol in the presence or absence of competitor oligonucleotides for 15 min at room temperature and run on 4% polyacrylamide gels (acrylamide:bisacrylamide, 30:1, wt/wt) in 0.5ϫ TBE buffer (1ϫ TBE: 89 mmol/liter Tris-HCl, (pH 8.0), 89 mmol/liter boric acid, and 2 mmol/liter EDTA) at 150 V. The gel was subsequently dried and exposed to Rx-U film (Fuji Photo Film, Tokyo, Japan) at Ϫ70°C.

Induction of IL-6 by IL-1␤ in Rheumatoid FLSs and OA
FLSs-Initial studies were designed to confirm the induction of IL-6 synthesis by IL-1␤ in the human rheumatoid FLSs. Rheumatoid MH7A cells, primary rheumatoid FLSs, and primary OA FLSs were incubated with various concentrations of IL-1␤ for 24 h. For all cells there was a concentration-dependent increase in the amount of IL-6 released into the medium (Fig.  1A). Peak IL-6 production was noted with 1-10 ng/ml IL-1␤. After the addition of 1 ng/ml IL-1␤, IL-6 production was detectable within 2 h and reached a maximum at 24 h (Fig. 1B). IL-6 production by the FLSs of RA patients in response to IL-1␤ was significantly greater than that by the control FLSs from patients with OA ( Fig. 1).
Functional Analysis of 5Ј-Cis-regulatory Elements of the Human IL-6 Gene Operating in Rheumatoid MH7A Cells and OA FLSs-As illustrated in Fig. 2, approximately 1200 bp of the 5Ј-flanking region of the human IL-6 gene contains various putative responsive elements: AP-1 (Ϫ283 to Ϫ277), MRE (Ϫ168 to Ϫ153), 5Ј-C/EBP␤ (Ϫ155 to Ϫ148), 3Ј-C/EBP␤ (Ϫ83 to Ϫ75), IL6-B (Ϫ73 to Ϫ60), and TATA box (Ϫ53 to Ϫ47 and Ϫ30 to Ϫ23) (11,15,29,30). Functional analysis of the 5Јflanking region of the human IL-6 gene was carried out using the Ϫ1158 to ϩ11 fragment as well as a series of 5Ј-deletion mutants of the IL-6 promoter linked to the luciferase reporter gene (Fig. 2). For transient expression of the reporter gene, each plasmid was used for transfection of rheumatoid MH7A cells and OA FLSs by electroporation, and the luciferase activity of the cell lysates was then measured 18 h after the addition of IL-1␤ (1 ng/ml). IL-1␤ markedly induced the luciferase activity in the MH7A cells transfected with the luciferase reporter plasmid carrying the IL-6 promoter ( Fig. 2A). The luciferase activity significantly decreased when two regions (Ϫ159 to Ϫ142 bp and Ϫ77 to Ϫ59 bp) were deleted. The C/EBP␤ consensus element (Ϫ155 to Ϫ148 bp), NF-B consensus element (Ϫ73 to Ϫ64 bp), and CBF1 consensus element (Ϫ67 to Ϫ60 bp) were contained in these two gene segments. 5Ј-CAT- GGGAA-3Ј (Ϫ60 to Ϫ67) was homologous to the reported CBF1 consensus binding element 5Ј-A(G/C)CGTGGGAA-3Ј (31). IL-1␤ induced less luciferase activity in the OA FLSs transfected with the luciferase reporter plasmids carrying the IL-6 promoter compared with that induced in the rheumatoid FLSs (Fig. 2B).
EMSA Targeting Positive Regulatory Elements-For the further identification of these positive regulatory elements, we prepared eight double-stranded oligonucleotides as shown in Fig. 3. C/EBP␤ (Ϫ158 to Ϫ142) oligonucleotide contained the C/EBP␤ binding element, and IL6-B (Ϫ75 to Ϫ60) oligonucleotide contained the NF-B and CBF1 binding elements. C/EBP␤-mt was the mutant oligonucleotide for C/EBP␤. IL6-B-mt1 was the mutant for NF-B. IL6-B-mt2 was the double mutant for NF-B and CBF1. IL6-B-mt3 was the mutant for CBF1. Ig-B oligonucleotide was the NF-B consensus binding element of the mouse immunoglobulin chain gene (32). The m8 oligonucleotide contained the CBF1 consensus binding element of the m8 gene of Drosophila Enhancer of split (31). We carried out EMSA using the nuclear extracts of MH7A cells that had been either unstimulated or stimulated with IL-1␤ for 4 h. As shown in Fig. 4, when the C/EBP␤ oligonucleotide was used as a probe, a broad band was constitutively detected (lanes 1 and 2). This complex was specifically competed by the unlabeled C/EBP␤ oligonucleotide (lanes 3 and 4). The complex was scarcely observed when the C/EBP␤-mt oligonucleotide containing a 5-bp mutation of the C/EBP␤ element was used as a probe (lane 5). This band disappeared with the addition of anti-C/EBP␤ antibody (lane 8).
When the 32 P-labeled IL6-B oligonucleotide was incubated with the nuclear extracts obtained from MH7A cells, three complexes (C-1, C-2, and C-3) were detected (Fig. 5A). A specific faster migrating band, C-3, was constitutively expressed upon incubation of the IL6-B oligonucleotide with the noninduced nuclear extract (lane 1). In addition to this band, two new slower migrating bands, C-1 and C-2, were detected with the induced nuclear extract (lane 2). These three complexes were specifically abolished by the addition of unlabeled IL6-B oligonucleotide (lanes 3 and 4). Both C-1 and C-2 complexes were competed out by the unlabeled Ig-B oligonucleotide (lanes 6 and 7), whereas the C-3 complex was not affected. In contrast, the C-3 complex was abolished by the unlabeled m8 oligonucleotide (lanes 9 and 10), whereas the C-1 and C-2 complexes were not affected.
We next performed supershift experiments employing antibodies against p50, p52, p65, c-Rel, CBF1, HMG I(Y), C/EBP␤, and STAT-1 to identify individual proteins detected by EMSA. A representative result is shown in Fig. 5B. C-1 and C-2 complexes consisted of NF-B p65/p65 homodimer and p50/p65 heterodimer, respectively (lanes 3 and 5), as supershifts were observed for anti-p50 and anti-p65, but not for anti-p52 or anti-c-Rel antibodies (lanes 4 and 6). The C-3 complex contained the CBF1 molecule, as a supershift was observed with anti-CBF1 antibody (lane 7), but not with anti-HMG I(Y) antibody (lane 8). None of the three bands was affected by the addition of irrelevant antibodies (lanes 9 and 10).
Separation of NF-B and CBF1 Bindings by the Site-specific Mutagenesis of the IL6-B Element-EMSA using the IL6-B oligonucleotide showed the complex nature of the DNA-protein interactions at this motif involving proteins of the NF-B p50/ p65 heterodimer, p65/p65 homodimer, and CBF1 (Fig. 5). So we introduced site-specific mutations into this motif to separate CBF1 binding from NF-B binding. The effects of the mutations on DNA-protein interactions were then determined by EMSA. Fig. 6 shows that the IL6-B-mt1 oligonucleotide,  1, 2, and 6). Competition experiments were performed by preincubating with a 30or 100-fold molar excess of the unlabeled C/EBP␤ oligonucleotide for 20 min at room temperature (lanes 3 and 4). which contained a 3-bp mutation in the NF-B binding element (Fig. 3), enhanced CBF1 binding but eliminated NF-B binding (lane 3). Competition with the unlabeled IL6-B-mt1 oligonucleotide eliminated CBF1 binding complex of the unstimulated (data not shown) and the stimulated (lanes 8 and 9) nuclear extracts, but did not affect the binding of NF-B (lanes 8 and 9). The IL6-B-mt2 oligonucleotide containing a 4-bp mutation in both NF-B and CBF1 binding elements (Fig. 3) eliminated all specific binding (lane 4). Competition with the unlabeled IL6-B-mt2 oligonucleotide did not affect either NF-B or CBF1 binding (lanes 11 and 12). The IL6-B-mt3 oligonucleotide containing a 2-bp mutation in the CBF1 element (Fig. 3) permitted NF-B binding, but eliminated CBF1 binding (lane 5). Competition with the unlabeled IL6-B-mt3 oligonucleotide eliminated the binding of NF-B (lanes 14 and 15). The Ig-B oligonucleotide did not bind CBF1 protein (lane 6), as expected from the competition experiment using the unlabeled Ig-B oligonucleotide (Fig. 5B, lanes 6 and 7).
Cooperation of NF-B and CBF1 Is Required for IL-1␤-induced Gene Transcription-For analysis of the functional significance of NF-B and CBF1 bindings, luciferase reporter constructs containing four copies of the NF-B and/or CBF1 binding elements connected to the tk promoter gene were constructed (illustrated in Fig. 7A) and used to transfect rheumatoid MH7A cells, primary rheumatoid FLSs, and primary OA FLSs. Luciferase activity was measured 6 h after IL-1␤ stimulation. A representative result is shown in Fig. 7B. The reporter construct carrying both NF-B and CBF1 binding elements (IL6-B ϫ 4) was clearly transcribed in response to IL-1␤. In contrast, IL6-B-mt1 ϫ 4, IL6-B-mt3 ϫ 4, and Ig-B ϫ 4 reporter constructs were not significantly transcribed by rheumatoid MH7A cells or primary rheumatoid FLSs in response to IL-1␤. In this regard, the Ig-B ϫ 4 reporter construct was transcribed by Jurkat T cells stimulated with phytohemagglutinin and phorbol 12-myristate 13-acetate (data not shown). The IL6-B ϫ 4 reporter construct was significantly less transcribed by OA FLSs in response to IL-1␤ compared with that transcribed by rheumatoid FLSs. There was no significant difference in the luciferase activity among OA FLSs transfected with IL6-B ϫ 4, IL6-B-mt3 ϫ 4, or Ig-B ϫ 4 reporter constructs.  Fig. 8. In rheumatoid MH7A cells (Fig. 8A), the activity of the wild-type IL-6 promoter was increased by 4.6-fold upon stimulation with IL-1␤. A 5-bp mutation introduced into the C/EBP␤ binding element reduced the basal IL-6 promoter activity to 51% of the wild-type value, and reduced the induction of IL-6 promoter in response to IL-1␤ to 40%. A 3-bp mutation introduced into the NF-B element of the IL-6 promoter did not affect the basal activity of this promoter, but reduced the IL-1␤-inducible activity to 74%. A 2-bp mutation introduced into the CBF1 binding element also reduced the IL-1␤-inducible activity to 68%, whereas it did not affect the basal activity. Mutations in both NF-B and CBF1 binding elements did not affect the basal activity, but totally abolished the induction of this promoter activity by IL-1␤. These results demonstrate that the C/EBP␤ binding element is required for the full expression of the IL-6 promoter in rheumatoid MH7A cells, although both NF-B and CBF1 binding elements are critical for the transcriptional induction in response to IL-1␤. In contrast to rheumatoid FLSs, OA FLSs exhibited significantly lower luciferase activity upon stimulation with IL-1␤ (Fig. 8B). The C/EBP␤ mutation reduced the basal IL-6 promoter activity of OA FLSs to 46% of the wild-type value, and reduced the induction of IL-6 promoter in response to IL-1␤ to 39%. The NF-B mutation of the IL-6 promoter did not affect the basal activity, but completely abolished the IL-1␤-inducible activity. The CBF1 mutation did not affect either the basal or the IL-1␤-inducible activity of the IL-6 promoter in OA FLSs. DISCUSSION Our present study clearly demonstrated that IL-1␤ stimulated rheumatoid FLSs to produce IL-6 in a concentration-and time-dependent manner (Fig. 1). IL-6 production of rheumatoid FLSs was significantly enhanced compared with that of OA FLSs. IL-1␤-induced up-regulation of IL-6 promoter activity was localized to two DNA segments, positions Ϫ159 to Ϫ142 bp and Ϫ77 to Ϫ59 bp, by the deletion analysis (Fig. 2). The former positive motif was identified as the C/EBP␤ binding site, and the latter was the complex binding sites of the NF-B p50/p65 heterodimer, p65/p65 homodimer, and CBF1 (Figs. [3][4][5]. The NF-B and CBF1 binding elements regulated inducible IL-6 promoter activity in response to IL-1␤ in rheumatoid FLSs, whereas the C/EBP␤ binding element mainly influenced basal activity (Figs. 2, 7, and 8). In addition we have provided the first evidence that CBF1 functions as a positive regulator of human IL-6 gene transcription in rheumatoid FLSs (Figs. 7 and 8). In contrast to that by rheumatoid FLSs, induction of IL-6 promoter activity by OA FLSs was less pronounced (Fig.  2B). CBF1 seemed not to be involved in IL-1␤-induced upregulation of the IL-6 promoter by OA FLSs (Figs. 7 and 8).

Requirement of C/EBP␤, NF-B, and CBF1 for IL-6 Gene
RA is a chronic inflammatory disease characterized by the proliferation of the synovial membrane into a highly vascularized tissue known as a pannus. The pannus consists of several distinct cell types, which include resident FLSs and infiltrating mononuclear cells capable of producing imflammatory cytokines such as IL-1, tumor necrosis factor-␣, and IL-6 (33). Several studies including ours have demonstrated that rheumatoid FLSs produce large amounts of IL-6 upon stimulation with IL-1 or tumor necrosis factor-␣ (9). IL-6 induces the production of acute phase proteins by hepatocytes (34). It also facilitates differentiation of B cells and may contribute to the production of rheumatoid factors (11). Moreover, IL-6 is involved in Fc␥RI expression on monocytes through the induction of STAT family factors (35). The excessive production of IL-6 seems to be related to the immunological abnormalities associated with RA. Therefore, it is essential to delineate the mo-lecular and cellular mechanisms of IL-6 production by rheumatoid FLSs. Rheumatoid FLSs, however, pose several problems as useful in vitro models, including their limited source and growth potential, requirement for a high concentration of serum for growth, and lot-to-lot variability in functional assays. Recently, we established a rheumatoid FLS line, MH7A, 2 derived from an RA patient, and showed herein that the MH7A cells functionally retained IL-1␤-inducible IL-6 production ( Fig. 1) and transcriptional inducibility of the IL-6 promoter and 4 tandem repeats of the IL6-B element in response to IL-1␤ (Figs. 2, 7, and 8). With this system we sought to elucidate the transcriptional regulation of the human IL-6 gene by rheumatoid FLSs.
Transcriptional regulation of the IL-6 promoter involves the interaction of transcription factors such as NF-B/rel family, C/EBP of the bZIP family, cAMP-response element binding protein, and AP-1 (11). As different sets of transcription factors may regulate the IL-6 gene in a cell type-specific manner, it is still uncertain which element is functional in the rheumatoid FLSs. Therefore, we first performed deletion analysis of the IL-6 promoter and found that the IL6-B motif was critical for the transcriptional induction of the IL-6 promoter in IL-1␤stimulated rheumatoid MH7A cells and OA FLSs and that the 5Ј-C/EBP␤ element was further required for the full transcriptional activation of this promoter (Fig. 2). The IL6-B motif seemed to function more efficiently for the transcriptional induction in rheumatoid FLSs than in OA FLSs.
Next, we performed EMSAs to identify the transcription factors binding to these elements of the IL-6 promoter. NF-B binding activity was significantly up-regulated in rheumatoid MH7A cells stimulated with IL-1␤ (Fig. 5A, lane 2). Supershift experiments identified the two slower IL6-B binding complexes to contain NF-B p50/p65 heterodimer and p65/p65 homodimer (Fig. 5B). It was reported that the p65 subunit of NF-B is responsible for the transactivation due to the existence of a potent transactivation domain located in its C-terminal portion (36 -38). It is well assumed that IL-1␤-induced up-regulation of the p65 subunit may lead to transcriptional induction by IL-6 promoter in rheumatoid FLSs. In fact, several investigators identified the existence of NF-B p50 and p65 subunits in the nuclei of cells located in the lining and subsynovial regions of the rheumatoid synovium in situ (39,40).
CBF1 was identified as a constitutive component of the IL6-B binding complex (Fig. 5B). The mutation introduced in the CBF1 binding element significantly reduced the response to IL-1␤ (Fig. 8A), indicating the accessory role of this element in rheumatoid FLSs. A mutation introduced into the CBF1 binding element resulted in a profound loss of the transcriptional induction of 4 tandem repeats of IL6-B element in response to IL-1␤ in rheumatoid FLSs, but not in OA FLSs (Fig. 7). Our data clearly indicated that CBF1 acted as a positive regulator of the human IL-6 gene transcription by rheu-matoid FLSs. It has been reported that CBF1 functions as a negative regulator of the IL-6 gene in mouse cell lines (18,19). The differences in species and cell types may account for the discrepancy regarding the opposite CBF1 functions. In addition, the double mutant of NF-B and CBF1 binding elements totally abolished the transcriptional induction of IL-6 promoter (Fig. 8A), suggesting that the cooperation of the two elements of the IL-6 promoter plays an essential role in the efficient transcriptional induction of the IL-6 gene in response to IL-1␤ in rheumatoid FLSs. The finding is consistent with the report that the constitutively activated form of human Notch1 binds to CBF1 and activates transcription through the CBF1-responsive element (41). In this regard, HMG I(Y) was previously shown to be required for virus induction of the interferon-␤ gene (42). HMG I(Y) is unable to stimulate (or inhibit) promoter activities by itself but is able to interact with NF-B to modify promoter activity (43). CBF1 might behave by itself like HMG I(Y), whereas HMG I(Y) was not involved in the transcriptional regulation of the IL6-B motif.
In contrast to the inducible binding of NF-B, C/EBP␤ binding activity was constitutively present in rheumatoid MH7A cells (Fig. 4). Deletion of this element or 5-bp mutants revealed that C/EBP␤ was involved in the basal level of transcription rather than in the induction of the IL-6 promoter ( Figs. 2 and  8, A and B). It is interesting to note, however, that the induction of transcriptional activity was reduced in proportion to the basal activity when the C/EBP␤ element was deleted or mutated ( Figs. 2 and 8, A and B). The C/EBP␤ binding element seems essential for the full activation of this promoter in both rheumatoid and OA FLSs. Several investigators have suggested that physical interaction between C/EBP␤ and NF-B lead to synergistic transactivation of the IL-6 promoter through the 5Ј C/EBP␤, 3Ј C/EBP␤, and NF-B sites (16) in a way similarly observed for the human IL-8 promoter (44). Although the IL-6 promoter activity with the double mutation of NF-B and CBF1 binding elements was not induced by IL-1␤, the 5Ј-C/EBP␤ element along with either NF-B or CBF1 element was sufficient for the activation by IL-1␤ (Fig. 8, A and B), indicating that the physical interaction of C/EBP␤ with either NF-B or CBF1 synergistically transactivates the IL-6 promoter through 5Ј-C/EBP␤ and IL6-B motifs.
IL-1 signaling is initiated by its type I receptor, which triggers the mitogen-activated protein (MAP) kinase cascade, leading not only to IB phosphorylation followed by the translocation of NF-B complexes into the nucleus (45), but also to phosphorylation and transactivation of C/EBP␤ (46). The MAP kinase cascade is now classified into three different types of MAP kinase systems, i.e. p42/p44, p54/JNK, and p38/RK (47). More recently, it has been reported that a specific p38 MAP kinase inhibitor, SB203580, significantly inhibited the IL-6 production by MRC-5 fibroblasts (48); whereas a selective inhibitor of the p42/p44 MAP kinase pathway, PD098059 (49), had no effect on the IL-6 production by MH7A cells. 2 Our present study demonstrated that human rheumatoid FLSs produced larger amounts of IL-6 upon stimulation with IL-1␤ compared with OA FLSs. The mutation introduced in the NF-B binding element reduced the response to IL-1␤ by no more than 30% in rheumatoid FLSs, but totally abolished the response to IL-1␤ in OA FLSs (Fig. 8B) and other cell types (50), indicating that IL-6 gene regulation by the IL6-B element is altered in rheumatoid FLSs. The CBF1 element was shown to be involved in the transcriptional induction by rheumatoid FLSs but not in that by OA FLSs (Figs. 7 and 8, A and  B). Thorough elucidation of the transcription factors and elements involved in the IL-6 gene transcription in rheumatoid FLSs might facilitate the future development of a specific IL-6 synthesis inhibitor for RA therapy.