Oncostatin M Stimulates c-Fos to Bind a Transcriptionally Responsive AP-1 Element within the Tissue Inhibitor of Metalloproteinase-1 Promoter*

,

mia inhibitory factor (LIF), oncostatin M (OSM), IL-11, ciliary neurotropic factor, and cardiotrophin-1 possess both unique and shared biological activities and utilize a common signal transducing receptor subunit gp130 (1,2). In addition, the polypeptide primary structures of OSM, LIF, and, to a lesser extent IL-6, ciliary neurotropic factor, and granulocyte colonystimulating factor display sequence homology, suggesting a functional relationship for these cytokines (3). Biochemical studies on signaling by gp130 have demonstrated tyrosine phosphorylation of a variety of components of the Ras-MAP kinase cascade (4,5) and activation of the recently characterized family of STATs (signal transducers and activators of transcription) (6,7). Upon activation, cytosolic STAT proteins homo-or heterodimerize and translocate to the nucleus where they bind DNA elements termed ␥-interferon activation sequence (GAS). One target gene of STAT proteins is c-fos, which harbors a GAS-like DNA element, the Sis-inducible element (SIE), within its promoter.
OSM is secreted as a 28-kDa polypeptide by mitogen-activated T cells and endotoxin-stimulated macrophages (8). It was first characterized by its ability to specifically inhibit the growth of the A375 melanoma cell line (9). Subsequently, OSM was demonstrated to stimulate growth of fibroblasts (10), aortic endothelial cells (11), and hematopoietic cells (12), as well as promote leukemic cell differentiation (3), while suppressing embryonic stem cell differentiation (13) and several tumor cell lines including the HTB10 lung carcinoma (14). OSM elicits acute phase protein production by hepatocytes and HepG2 hepatoma cells (15). In addition, unlike other IL-6-cytokines, OSM also specifically up-regulates low density lipoprotein receptor (16) in HepG2 cells. The transduction of signals that lead to biological activities unique to OSM but not other IL-6 family members is not yet defined.
Extracellular zinc-dependent endopeptidases (the matrix metalloproteinases) can be inhibited by a family of proteins called tissue inhibitors of metalloproteinases (includes TIMP-1, -2, -3, and -4), which act to modulate extracellular matrix metabolism by matrix metalloproteinases (17)(18)(19)(20)(21). TIMP-1 expression is up-regulated in fibroblasts by a variety of soluble factors including IL-1 (22), tumor necrosis factor (23), epidermal growth factor (23), transforming growth factor-␤ (23), phorbol esters (24), and retinoic acid (25). We have shown that OSM potently induces mRNA expression of the TIMP-1 gene in hepatocyte cell lines and primary fibroblasts (26). Work by Edwards et al. (27) has suggested that the Ϫ95 to ϩ47 region of the murine TIMP-1 promoter is sufficient to confer serum responsiveness in mouse cells and this region contains several putative regulatory motifs including SP1, AP-1, and Ets DNA elements. Here, we examine OSM regulation of the Ϫ95 to ϩ47 TIMP-1 promoter in HepG2 cells and characterize the partici-pation of c-Fos in the binding of nuclear factors to an AP-1 site between Ϫ62 and Ϫ53 that is necessary for marked OSM (but not IL-6) up-regulation of promoter/CAT reporter gene expression. We examined IL-6-and phorbol 13-myristate 12-acetate (PMA)-induced responses to identify OSM-specific effects on this promoter region and the participation of SP1, Ets, and other promoter sequences in regulation of TIMP-1 expression.
Northern Blots-Total RNA was prepared from HepG2 cells according to Chomczynski and Sacchi (28). Subconfluent HepG2 cultures were washed and replenished in medium containing 2% FBS. Cytokines (at indicated concentrations) were then added, and cultures were incubated for the indicated times before RNA isolation. Northern blots were prepared by standard techniques and probed with human TIMP-1 cDNA (gift of Dr. A. J. P. Docherty, Celltech, Slough, United Kingdom) and rat c-Fos cDNA (kindly provided by Dr. Tony Cruz, Mount Sinai, Toronto, Canada). The intensity of ethidium bromide-stained 18 S and 28 S bands on the blots was used to estimate loading of RNA.
Polymerase Chain Reaction (PCR) Deletion of the TIMP-1 Promoter and Cloning-Deletions of the TIMP-1 promoter were carried out by generating truncated PCR products within Ϫ95 to ϩ47. PCR reactions consisted of Vent polymerase (5 units; New England Biolabs), 100 ng of TIMP-1 Ϫ223 to ϩ47/pBLCAT3 (27) template DNA, 150 M sense and antisense oligonucleotide primers, 5 mM dNTPs, and 1 ϫ Vent polymerase New England Biolabs buffer in a 50-l final reaction volume. Reaction mixtures were overlaid with 30 l of mineral oil and denatured for 1 min at 95°C, followed by annealing of primers at 55°C for 2 min and primer extension at 72°C for 2 min in a Perkin-Elmer PCR thermal cycler for 35 cycles.
Chloramphenicol Acetyltransferase (CAT) Assays-HepG2 cells in 100-mm dishes were transfected with 10 g of CAT reporter plasmid DNA (co-transfected with 1.4 g of pSV-␤Gal plasmid, Promega) by the calcium phosphate coprecipitation method. Cells were allowed to recover overnight and then replated into six-well tissue culture plates. Prior to cytokine stimulation, cells were serum-starved in serum-free ␣-minimal essential medium for 6 h. Cytokines were then added for 18 h. CAT assays were carried out using cell lysates according to standard protocols and 14 C-labeled chloramphenicol products quanti-fied using a Molecular Dynamics PhosphorImager and ImageQuant software. Values were normalized to ␤-galactosidase activity in lysates (29).
Preparation of Nuclear Extracts-HepG2 cells were stimulated with the indicated cytokines for various time periods. Nuclear extracts were prepared according to Andrews et al. (30) with the following modifications. Buffers A and C contained 0.5 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotonin, and 2 g/ml pepstatin A and leupeptin. Cells were resuspended in 400 l of buffer A, and nuclear proteins were extracted in 100 l of buffer C, frozen in liquid nitrogen, and stored at Ϫ70°C.
Electrophoretic Mobility Shift Assays (EMSAs)-Nuclear extract (15 g) was incubated with 2 g of poly(dI⅐dC) and 5 g of calf thymus DNA in binding buffer (50 mM Tris-Cl (pH 7.5), 50 mM NaCl, 2 mM EDTA, 2 mM dithiothreitol, 1 mM spermidine, and 5% glycerol) for 15 min on ice. A total of 10 5 cpm of 32 P-labeled probe was then added, and the binding reaction (20 l) was incubated at room temperature for 20 min. Excess of unlabeled oligonucleotide (50 -100-fold) was added to the reaction for competition assays. Following the reaction, samples were electrophoresed on 5% polyacrylamide gels (40:1) containing 1.25% glycerol in 0.25 ϫ TBE (1 ϫ TBE: 89 mM Tris borate, 2 mM EDTA) at 95 V for 3.5 h and dried prior to autoradiography. For supershift analysis, antibodies were added following the binding reaction for 1 h at 4°C. Sequences of oligonucleotides used for mobility shift assays are shown in Table I, and their location in the TIMP-1 promoter is illustrated in Fig. 2. AP-1 and SP1 consensus oligonucleotides were purchased from Santa Cruz Biotechnology Inc. Oligonucleotides were annealed by heating to 100°C for 5 min in 100 mM MgCl 2 and 400 mM Tris-Cl (pH 8), followed by gradual cooling to 20°C. TIMP-1 DNA probes were end-labeled using polynucleotide kinase and the high affinity Sis-inducible element (hSIE) labeled by the fill-in reaction using [ 32 P]dCTP and the Klenow enzyme. Probes were then gel-purified.
Western Blots-Nuclear extracts were electrophoresed on a 8% sodium dodecyl sulfate (SDS)-polyacrylamide gel prior to transfer of proteins onto an Immobilon nitrocellulose membrane. Membranes were then blocked overnight in 1 ϫ PBS and 5% milk. Following a 10-min incubation in wash buffer (1 ϫ PBS, 0.5% Tween 20), membranes were incubated with primary rabbit polyclonal antibody (anti-c-Fos, antipan-Fos, anti-c-Jun, anti-JunB, or anti-JunD; Santa Cruz Biotechnology Inc.) at a 1:2500 dilution for 1 h at room temperature. To assess specificity of antibody binding, 2 g of peptide was preincubated with antibody for 1 h at 4°C (as recommended by Santa Cruz Biotechnology Inc.). Membranes were then washed three times for 10 min each. Goat anti-rabbit horseradish peroxidase (Sigma) was added for an additional (1:7500 dilution) 1 h and Immobilon filters were developed by Enhanced Chemiluminescence Renaissance (NEN Life Science Products).

Oncostatin M Induces TIMP-1 and c-Fos Expression-We
have found previously that OSM elevates TIMP-1 mRNA levels after overnight stimulation in a variety of cell types including HepG2 (26). Examination of TIMP-1 mRNA expression in HepG2 cells over a time course of OSM stimulation (Fig. 1) revealed a 5-fold increase in TIMP-1 message by 1 h, near maximal at 2 h (40-fold), and maximal mRNA levels at 6 h (60-fold). OSM also transiently stimulated mRNA levels of the AP-1 immediate-early gene c-fos (Fig. 1). As demonstrated previously in response to other stimuli (31), the up-regulation of c-Fos mRNA occurred early (maximal induction at 30 min (20-fold)) and decreased thereafter. Thus, c-Fos mRNA levels peaked slightly before the marked increase in TIMP-1 mRNA.
OSM Regulates the TIMP-1 Proximal Promoter-The proximal promoter of the TIMP-1 gene (Ϫ95/ϩ47 sequence) can regulate downstream CAT gene expression in pBLCAT3 chimeric reporter constructs (27). When transiently transfected into HepG2 cells, the expression of Ϫ95/ϩ47CAT could be elevated upon OSM (5.2-fold) or IL-6 (3.8-fold) stimulation (Fig. 2, construct A). Several putative DNA elements can be identified within Ϫ95 to ϩ47 region of TIMP-1, including those for SP1, AP-1 (Fos/Jun), and Ets proteins. To examine the potential role of these sites in transcription, we generated 5Ј deletions of this region and cotransfected HepG2 cells with constructs A-F (Fig.  2) and pSV-␤Gal to normalize for transfection efficiency. Basal levels of CAT transcription were reduced in the plasmid lack-ing Ϫ95 to Ϫ63 sequences (construct B), and further reduced in the plasmid that also lacked the Ϫ62 to Ϫ53 sequences, which contained the AP-1 site (construct C). The Ϫ62/ϩ47CAT chimera (construct B) demonstrated maximal responsiveness to OSM with a 11.4-fold increase in CAT activity over unstimulated cells, whereas IL-6 induced 4.1-fold increases. Deletion of the AP-1 site within this region (construct C) markedly reduced responsiveness to OSM (from 11.4-to 2.7-fold), whereas responses to IL-6 decreased from 4.1-to 2-fold. Thus, a promoter region from Ϫ62 to Ϫ53 of TIMP-1, containing a putative AP-1 binding site (at Ϫ59/Ϫ53), contributes to basal transcription and to induction of transcription by OSM. Deletion of this sequence removed any significant difference between OSM and IL-6 activity in this assay. Interestingly, deletion of sequences from ϩ1 to ϩ47 (construct E versus B) dramatically abrogated the responsiveness to OSM (from 11.4-to 3.7-fold) and also reduced IL-6 responses (from 4.1-to 2.3-fold). Thus, sequences within ϩ1/ϩ47 may cooperate with the TIMP-1 AP-1 element within Ϫ62/Ϫ53 for maximal responsiveness to OSM. SP1 elements are common within TATA-less promoters and have been demonstrated to participate in basal transcription (32). A chimeric plasmid with mutation of the SP1 site (construct F) showed reduced basal transcription (4-fold) but was still inducible by OSM (construct E versus F).
Nuclear Factors Bound to the TIMP-1 Promoter-We have examined a variety of overlapping oligonucleotide probes (probes 1-6; location shown in Fig. 2) spanning Ϫ95 to ϩ1 of the TIMP-1 promoter in EMSA analysis to identify elements that bind nuclear factors in HepG2 cells. A specific bandshift with either AP-1 or AP-1-Ets (probes 2 and 3) was seen in unstimulated cell nuclear extracts that we termed "complex 1" (Fig. 3, A and B). At 12 h of OSM treatment, much higher amounts of complex 1 are apparent. OSM induced the assembly TABLE I Oligonucleotides used in EMSA Oligonucleotides were synthesized, annealed, and purified as indicated under "Materials and Methods." The sequences correspond to the TIMP-1 promoter regions as indicated. Defined nuclear factor binding sites for SP1, AP-1, and Ets have been underlined. Probe 3A contains a mutated AP-1 site (uAP-1). Probe 4A contains a G at position 49 corresponding to the mouse TIMP-1 sequence, whereas probe 4 has A in that position corresponding to the human TIMP-1 sequence. Deletions of the TIMP-1 promoter spanning Ϫ95 to ϩ47 were generated by PCR and cloned into pBLCAT3 to examine their capacity to regulate CAT reporter gene expression. HepG2 cells were co-transfected with each of the constructs (schematically illustrated at left) and pSV-␤Gal and treated with either OSM (50 ng/ml) or IL-6 (100 ng/ml) for 18 h. Cellular extracts were then prepared, and CAT activity was measured by standard methods and normalized to ␤-galactosidase activity. Results are shown as percent conversion and -fold induction upon cytokine stimulation. Values represent the mean of three separate experiments (standard deviation in parentheses), each experiment done in duplicate.
of a second complex after 30 min of OSM stimulation with decreased mobility ("complex 2") that persisted for several hours. The mobility of complex 2 appeared to increase following 2 h of OSM stimulation. Promoter sequences at Ϫ49 to Ϫ40 displayed an Ets core binding element as well as homology to STAT DNA binding sites (consensus TTCCNNNAA). Using a probe spanning this region (probe 4), a very weak binding activity was detected, which was not altered upon OSM stimulation (Fig. 3C), and similar results were noted using the mouse TIMP-1 sequence (probe 4A), which differs by one base pair (data not shown). The mobility of this complex appeared similar to that of the AP-1-complex 1 and no other complexes were detectable. Nuclear factors binding to probe 4 were specifically competed by cold Ets probe but supershift experiments using an anti-Ets-1/Ets-2 polyclonal antibody reactive against a highly conserved DNA binding domain of Ets proteins did not identify Ets-1 or Ets-2 binding to this probe (data not shown). STAT-1 and 3 activation can be detected using the hSIE (33). Using the hSIE and the same HepG2 extracts as above, we noted that treatment with OSM resulted in rapid nuclear factor binding with gel shift mobilities consistent with those observed for homodimers of STAT3 or STAT1 and the heterodimer of STAT3 with STAT1 (Fig. 3D). Each of these complexes were verified using antibodies to STAT3 or STAT1 in EMSA supershift assays (data not shown) as established previously (33). STAT1 was detected at early time points, whereas STAT3 persisted from 15 min to 12 h of OSM treatment. However, neither the AP-1-Ets or Ets sequences (probes 3 and 4) detected complexes with the same mobility and kinetics as compared with the hSIE in response to OSM.
The TIMP-1 AP-1/Ets (probe 3) complexes and STAT mobil-ity shifts were specifically competed by the 50-fold addition of unlabeled probe 3 or hSIE probe, respectively (Fig. 4). Neither AP-1 complexes 1 or 2 could be eliminated by competition with unlabeled hSIE probe (Fig. 4) or supershifted with anti-STAT1 or anti-STAT3 (data not shown). EMSA analysis with probe 3A, containing mutation of nucleotides in the AP-1 site, dramatically reduced complexes 1 and 2, and other specific complexes were not detectable (Fig. 5) over the time course of OSM stimulation. Thus, we could not detect STAT nuclear factors associated with the AP-1 complexes 1 and 2, nor binding to sequences immediately flanking the TIMP-1 AP-1-like element that show partial homology to a STAT DNA binding site. Further downstream sequences showed weak homology to GAS (differing from the consensus GAS site by a two nucleotide insertion); however, probe 5 (Ϫ38/Ϫ15) elicited no early or late induced gels shifts similar to STAT complexes (data not shown). Thus, we could not detect STAT-1/3 nuclear protein binding to TIMP-1 promoter sequences in these cells. Supershift analysis of TIMP-1 AP-1 bands were carried out to identify Fos and Jun components of these complexes. Nuclear extracts were prepared from HepG2 cells treated with OSM for 0, 1, and 12 h and incubated with polyclonal antibodies specific for AP-1 proteins prior to EMSA analysis. Using an antibody reactive against all Fos-related antigens (anti-pan-Fos), complex 1 was obliterated at 0, 1, and 12 h, and a Fos supershift was clearly visible at 1-and 12-h time points (Fig.  6A). The anti-pan-Fos antibody also supershifted complex 2 from HepG2 cells treated with OSM for 1 h (Fig. 6A). However, when an anti-c-Fos specific antibody was used, only complex 2 was supershifted and complex 1 remained unaffected, suggesting that c-Fos is a prerequisite for the OSM-stimulated assembly of complex 2. Of the anti-Jun antibodies used in EMSA binding reactions, anti-JunD and anti-JunB supershifted TIMP-1 AP-1 probe complexes (Fig. 6B). JunD was detected at time 0, 1, and 12 h of OSM treatment. JunB could also be detected in these complexes, whereas c-Jun could not be de-  (Table I) spanning the TIMP-1 promoter were tested by EMSA. The probes were as follows: A, AP-1 (probe 2); B, AP-1-Ets (probe 3); C, Ets (probe 4); and hSIE (D). HepG2 cells were treated with OSM from 0 to 12 h, and nuclear extracts were prepared and stored at Ϫ70°C until analysis. EMSA gels were dried and subjected to autoradiography. Longer gel profiles (as seen in Fig. 4) showed no other specific bands; thus, only the tops of the gels are shown here. tected at any time. Together, these data suggest that, whereas complex 1 consists of Fos and Jun, complex 2 specifically contains c-Fos, which likely complexes with JunD or JunB.
Because sequences within Ϫ95 to ϩ1 contained two putative SP1 binding sites, both of these were examined for the binding of HepG2 nuclear factors. Oligonucleotides spanning Ϫ95 to Ϫ66 (probe 1) or Ϫ19 to ϩ2 (probe 6) constitutively bound two specific complexes, which remained unchanged in response to OSM treatment (Fig. 7, A and B). Both complexes were competed by cold SP1 probe (Fig. 7A) and are consistent with the mobility of SP1 nuclear factor binding to a consensus SP1 DNA element (data not shown). Anti-SP1 antibody supershifted the slower migrating complex but did not appear to affect the faster migrating specific complex (Fig. 7C). This may indicate the binding of other SP1 family members (34). Binding of SP1 to the Ϫ95/Ϫ66 oligonucleotide was only detectable following a one week exposure of EMSA gels (data not shown), although weak binding to this probe may be due to insufficient sequence flanking the 5Ј end of the SP1 site. Thus, SP1 nuclear factors constitutively occupy binding sites within the TIMP-1 promoter and OSM does not appear to affect the SP1 DNA binding.
OSM, but Not Other Cytokines, Induces Complex 2-Given the overlapping biological functions of OSM and IL-6, these and other cytokines were examined together for their ability to induce DNA binding of complexes to the TIMP-1 AP-1 probes in EMSAs. Interestingly, OSM was the only cytokine examined that strongly induced complex 2 (Fig. 8A), whereas both OSM and IL-6 markedly induced activation of STAT binding to the hSIE probe (Fig. 8B). Thus, within the same assay, OSM and IL-6 utilize shared (STAT) and distinct (AP-1-complex 2) nuclear signaling pathways leading to protein-DNA interactions.
PMA has been shown to activate AP-1 (34) and stimulate TIMP-1 expression (24). When tested in HepG2 cells, PMA potently stimulated complex 2 formation (AP-1 probe 2) and to a greater extent than OSM (Fig. 8A). To examine the activity of PMA on transcription, we compared the effectiveness of PMA to OSM or IL-6 in stimulating CAT activity of the TIMP-1 Ϫ62/ϩ47 reporter gene construct B. Table II shows that PMA alone (or 20% FBS) was unable to up-regulate CAT activity through the TIMP-1 Ϫ62/ϩ47 promoter element whereas OSM induced 10-fold increases and IL-6 induced 4-fold changes. When used in combination with IL-6, PMA also failed to stimulate CAT activity beyond that of IL-6 alone. Thus, despite equivalent STAT activation and c-Fos induction as in OSMtreated cells, the combination of IL-6 and PMA was not sufficient to induce similar levels of transcriptional activation. OSM may induce qualitative differences in AP-1 factors distinct from that by PMA and requires additional regulatory elements for the pronounced induction of the TIMP-1 Ϫ62/ϩ47 reporter gene construct beyond that induced by IL-6.
Oncostatin M Stimulates c-Fos Protein Nuclear Accumulation-Because the data implicate c-Fos in gel-shifted complex 2, and OSM markedly stimulated early transient expression of c-Fos mRNA (Fig. 1), we examined c-Fos protein levels in nuclear extracts from HepG2 cells by Western blots (Fig. 8C). In the same extracts used for EMSA, OSM and PMA markedly up-regulated c-Fos protein levels in HepG2 cell nuclei. Expression of c-Fos protein was abrogated in the presence of protein synthesis inhibitors (puromycin and emitine) (Fig. 9B), and in addition, these agents blocked complex 2 formation in response to OSM (Fig. 9A). Complex 1 formation was not affected at 0.25 or 1 h; however, the inhibitors did reduce complex 1 at 12 h, which suggested newly synthesized proteins are involved in complex 1 at this later time point. OSM stimulated nuclear accumulation of c-Fos by 1 h, which persisted to 12 h of OSM treatment (Fig. 9B). Interestingly, the electrophoretic mobility of c-Fos was reduced at 12 h when compared with 1 h of OSM stimulation. This difference in mobility is first observed after 2 h of OSM treatment (data not shown), persists for up to 12 h, and may represent a change in the phosphorylation status of c-Fos.
In contrast to c-Fos, and consistent with their presence in AP-1 gel-shifted complexes, both JunB and JunD were constitutively present and JunD was moderately up-regulated at the 1 h time point of OSM treatment (Fig. 9, C and D). Puromycin and emitine had no effect on JunB at early time points of OSM stimulation and only affected JunD levels after 12 h of treatment. Detection of c-Fos, JunB, and JunD in Western blots was confirmed by specific competition of relevant peptides for each of the antibodies used, and c-Jun was undetectable from the same nuclear protein extracts (data not shown). Thus, OSM stimulates the nuclear accumulation of c-Fos protein, whereas JunB and JunD nuclear factors are constitutively resident within the nucleus and appear largely unaffected by OSM at the level of new protein synthesis. DISCUSSION The transcription factor c-Fos is one of several participants in various Fos/Jun complexes that bind AP-1 sites of gene promoters to regulate transcription (31). We have shown here that, upon stimulation by OSM in HepG2 cells, c-Fos is highly induced at the mRNA and protein level and takes part in an Stimulation of DNA binding activity was tested using probe 3, which contains the TIMP-1 AP-1 sequence (A), or the hSIE probe (B). Protein levels of c-Fos in HepG2 nuclear extracts (15 g) of cells treated with the above cytokines or PMA were examined by Western blotting using an anti-c-Fos specific primary antibody and a goat anti-rabbit horseradish peroxidase secondary antibody (C). c-Fos protein was detected by enhanced chemiluminescence followed by autoradiography.

TABLE II
Regulation of Ϫ62/ϩ47CAT promoter activity by OSM and IL-6 but not PMA HepG2 cells were transfected with 10 g of Ϫ62/ϩ47CAT, allowed to recover overnight and replated in six-well Costar plates. The cultures were then serum-starved for 6 H and stimulated with OSM, IL-6, PMA, IL-6 and PMA, or 20% FBS for 18 H in serum-free conditions. CAT activity was assayed as described in "Materials and Methods." -Fold change was calculated from phosphorimagery results and averaged from at least three separate experiments Ϯ S.D. AP-1 complex (complex 2) that is separate from that constitutively present (complex 1) using the AP-1-containing sequence (Ϫ59/Ϫ53) of TIMP-1 as a probe. OSM induced rapid and transient expression of c-Fos (peak at 30 min) and nuclear accumulation, which bound the AP-1 site in complex with Jun proteins (peaking at 1-2 h). Nuclear c-Fos protein levels coincided with the induction of complex 2 by OSM. Protein synthesis inhibitors (puromycin and emitine) inhibited c-Fos protein expression and formation of complex 2. Thus, c-Fos is prerequisite to complex 2 formation, and its nuclear accumulation in response to OSM is dependent upon new protein synthesis. Such AP-1 activation may contribute to effects of OSM on growth regulation or expression of other genes. Previous work has shown OSM-mediated induction of other immediate early genes such as egr-1, c-jun, and c-myc in fibroblasts (35), which also suggests a broad range of gene products could in turn be regulated by OSM stimulation.

Stimulus
Although functional redundancy within the IL-6-type family of cytokines is a common observation, OSM manifests both shared and distinct biological activities with other family members (IL-6, LIF, IL-11, ciliary neurotropic factor, and cardiotrophin-1), which could be attributed to differential expression of receptor complexes on different cells. Alternatively, the subunits unique to individual receptor complexes may contribute to the biological activities of these cytokines. OSM binds and activates cells through the LIF receptor (type I OSM receptor) and a newly identified complex of gp130 and OSM receptor ␤ chain (type II OSM receptor) (36) in human cells. Like other related family members, OSM stimulates components of the Ras-MAP kinase pathway, such as Ras, Raf-1, Grb2, Shc, and the p42 MAP kinase (4,37), and the JAK-STAT pathway (6,7). Other signaling pathways have also been implicated in mediating signals transduced by OSM, including those affecting phosphatidylinositol 3Ј-kinase and the canonical Src kinase (38). We have here confirmed that OSM potently induces mRNA expression of the TIMP-1 gene in HepG2 cells, and shown that OSM strongly enhances transcription of a CAT reporter gene flanked by Ϫ62 to ϩ47 of TIMP-1 promoter construct. Up-regulation of CAT activity was most strongly induced by OSM treatment, less so by IL-6, and not at all by 20% FBS, PMA, or the combination of IL-6 and PMA. Taken together, this implicates OSM in activating a signaling event(s) distinct from IL-6 or serum factors that likely act on target DNA elements within the TIMP-1 promoter for up-regulating its transcription and expression. Interleukin-6 also regulates the TIMP-1 promoter (Fig. 2) but appears not to activate c-Fos or form complex 2; nor did IL-6 regulate AP-1-containing promoter/CAT (Fig. 2) constructs to as great a degree as OSM. This is consistent with previous work, which showed a similar effect on the regulation of the rat TIMP-1 promoter, as well as an AP-1 site requirement for maximal responses (39), although AP-1 activation was not identified. A role for AP-1 in TIMP-1 promoter regulation has also been implicated in F9 cells overexpressing AP-1 genes (40). Because OSM is more effective at stimulating acute phase protein production by HepG2 cells than other gp130 cytokines (26), and regulates low density lipoprotein receptors on these cells (16), we suggest that c-Fos activation is an additional component of OSM signaling which contributes to differential effects. Alternatively, OSM or IL-6 may confer differences in Fos/Jun complexes through posttranslational modifications such as phosphorylation or alter the composition of associated dimers of AP-1.
In addition to the AP-1 site (27,40,41), the proximal TIMP-1 promoter also contains putative regulatory motifs including an Ets binding sequence (27,40) with homology to STAT elements, and SP1 elements (27). Both the Ets site (putative STAT site) and SP1 site appeared to contribute somewhat to basal transcription levels based on deletion analysis (Fig. 2); however, binding of factors to these sites was not altered upon OSM stimulation. Previous studies have shown that Ets can cooperate with AP-1 in regulating transcription in other cells (42,43). Others have shown that STATs can bind AP-1/Ets sequences of the rat TIMP-1 in HepG2 cells (39), and human TIMP-1 promoter in astrocytes (44) and that this site also contributes to transcription by OSM. Our results in HepG2 cells did not detect binding of STATs to this sequence despite the presence of activated STAT-1 and STAT-3 (hSIE binding) in the nuclear extracts and long exposures of gels. In addition, mutation of the AP-1 site in the AP-1-Ets probe completely eliminated detection of any OSM-inducible nuclear factors capable of binding this probe (Fig. 5). Differences between the levels of STAT proteins expressed in HepG2 cells and astrocytes could account for this result. The prominence of AP-1 nuclear factor binding to this sequence that we observe is consistent with previous studies (44), wherein AP-1 binding appeared dramatically greater than STAT binding in EMSA assays with an equivalent of an AP-1/Ets probe. We suggest that this abundance of AP-1 binding reflects a physiological importance among other factors participating in the regulation of this proximal TIMP-1 promoter.
Although deletion of the AP-1 motif (Ϫ59/Ϫ52) markedly affected OSM activation, this AP-1 element was not sufficient for full OSM-responsiveness because deletion of sequences between ϩ1 to ϩ47 also reduced transcription (Fig. 2). The TIMP-1 gene may utilize SP1 binding sites and a putative initiator element downstream of the transcription initiation site as described for other TATA-less promoters. We have observed a putative pyrimidine-rich initiator element immediately 3Ј to ϩ1 of the TIMP-1 gene that may be necessary for the assembly of a competent transcriptional machinery apparatus for transcription initiation (45). Alternatively, additional responsive elements 3Ј to the initiation start site may cooperate with the TIMP-1 AP-1 element at Ϫ59/Ϫ53 in response to OSM. Interestingly, Logan et al. (40) have characterized a weak AP-1 binding site within ϩ1/ϩ47, which may also be a target for OSM signals. We are currently examining this aspect.
Although c-Fos and its participation in complex 2 could also be stimulated by PMA (as assessed by supershifts; data not shown), PMA alone was unable to up-regulate TIMP-1 promoter activity as assayed by CAT reporter gene expression (Table II). In addition, we found that the costimulation of HepG2 cells with IL-6 (STAT induction) and PMA (c-Fos induction) could not appreciably up-regulate the CAT activity driven by the Ϫ62 to ϩ47 TIMP promoter. This also suggests that OSM induces an additional signal that regulates TIMP-1 in this system. STAT proteins may interact with SP1 (46) and Jun (47) proteins to cooperate in elevating transcription in other systems; thus, a similar mechanism could occur in HepG2 cells. Alternatively, OSM and PMA may differ qualitatively in affecting posttranslational modification of AP-1 nuclear factors, such as phosphorylation, that influence the transactivation potential of these transcription factors.
In summary, maximal expression of TIMP-1 by OSM (but not IL-6) may, at least in part, be attributed to the induction of c-Fos and its complex with AP-1 sites in the proximal promoter. This may also involve posttranslational modifications such as phosphorylation to these factors. Cooperation of nuclear proteins binding to SP1 and Ets DNA elements is also needed for maximal expression, and OSM signaling may recruit additional factors that interact with sequences downstream from ϩ1 to ϩ47. Our results also support the existence of shared and distinct signaling pathways by OSM and IL-6 in HepG2 cells.