The mitogen-activated protein kinase and JAK-STAT signaling pathways are required for an oncostatin M-responsive element-mediated activation of matrix metalloproteinase 1 gene expression.

Both astrocytes in the central nervous system and fibroblasts in somatic tissues are not only the major sources of extracellular matrix components but also of matrix metalloproteinases (MMPs), a family of enzymes directly involved in extracellular matrix breakdown. We have analyzed the regulation of the expression of MMPs and TIMPs (tissue inhibitors of metalloproteinases) in human primary astrocytes stimulated with oncostatin M (OSM) and other extracellular mediators in comparison with normal human dermal fibroblasts. It was found that OSM induced/enhanced transcription of MMP-1 (interstitial collagenase) and MMP-3 (stromelysin 1) in astrocytes, and MMP-1, MMP-9 (gelatinase B), and TIMP-1 in fibroblasts. Analysis of the signal transduction leading to activation of the MMP-1 gene revealed the presence of an OSM-responsive element (OMRE) encompassing the AP-1 binding site and the signal transducer and activator of transcription (STAT) binding element, which mediate activation by OSM. OMRE is also present in the TIMP-1 gene promoter and, although there are some differences in these two motifs, both appear to be targets for the simultaneous action of OSM-induced nuclear effectors. The induced enhancement of transcription by synergistically acting AP-1 and STAT binding elements in response to OSM is Raf-dependent. Cross-talk between the mitogen-activated protein kinase and JAK-STAT pathways is required to achieve maximal induction of the OMRE-driven transcription by OSM.

Astrocytes and fibroblasts each produce a complex mixture of extracellular matrix components (1), turnover of which is an integral part of many physiological processes such as development, cell migration, angiogenesis, tumor invasion (2), and neurodegenerative diseases (3,4). The transcriptional regulation of matrix metalloproteinases (MMPs) 1 has been observed in these processes (5,6), and after extracellular activation the activity of these enzymes can be controlled by specific TIMPs (5,7,8).
The expression of MMPs is greatly modulated by cytokines and growth factors (reviewed in Refs. 6 -8), which induce cellular responses by activating intracellular signaling cascades including MAP kinase (9,10) or JAK-STAT (11) pathways via specific cell surface receptors. Activated MAP kinases enter a nucleus and activate (or inactivate) transcription factors by phosphorylation of serine or threonine residues. Genetic and biochemical studies have revealed a series of MAP kinase nuclear substrates including those directly involved in MMPs transcriptional regulation such as the gene products of fos and jun oncogenes which compose the AP-1 transcription factor and ets family members (10,(12)(13)(14)(15). The coordinated up-regulation of MMP-1 and TIMP-1 genes by growth factors, as well as by several nuclear and non-nuclear oncogenes (6,16), can be explained by the combined actions of transcription factors operating through composite Ets/AP-1 motifs of the type first identified as ras-responsive elements in some mammalian promoters, including MMP-1 and TIMP-1 (16,17).
Oncostatin M (OSM) is a multifunctional cytokine that affects the growth and differentiation of a variety of cell lines and is produced by activated monocytes and T lymphocytes (18 -20). Recently, OSM was demonstrated to be an immediate early gene induced by multiple cytokines through the JAK-Stat5 pathway (21). This factor exerts a negative effect on the growth of some tumor cell lines and also acts as a positive growth regulator of normal fibroblasts, AIDS-related Kaposi's sarcoma and a TF-1 cell line (22)(23)(24). In addition, OSM induces differentiation of the M1 murine myeloid leukemia cell line (25), modulates the function of endothelial cells (26), up-regulates the expression of the TIMP-1 in human fibroblasts (27), and stimulates acute phase protein expression in hepatocytes (28).
Together with leukemia inhibitory factor, IL-6, and IL-11, OSM is a member of a family of pleiotropic cytokines that share (20,29,30) a common signal transducer, gp130, which is the ␤ component of a multiple subunit cell surface receptor (31). More recently, the JAK-STAT signaling pathway has been shown to play an integral role in intracellular signaling by the cytokine receptors (32). Signal transducers and activators of transcription (STATs) are a unique family of transcription fac-tors activated by many cytokine receptors (11). The ligandreceptor interaction brings the receptor-associated JAK kinases into apposition, enabling them to recruit a latent cytoplasmic STAT member family. The SH2 domains of STAT proteins interact with sites of receptor tyrosine phosphorylation and become accessible for the activated JAKs. Following activation by tyrosine phosphorylation, the STAT proteins form homo-or heterodimers that are nucleus translocated where they bind to specific regulatory elements in target genes (reviewed in Refs. 11 and 33).
Although astrocytes appear to perform a function in the central nervous system similar to that of fibroblasts in somatic tissue, only limited information is available about the regulation of MMP expression in these cells (54,74). Analysis of the expression of MMPs and TIMPs in human primary astrocytes stimulated with OSM and other extracellular mediators in comparison to human fibroblasts were, therefore, performed in order to determine whether this cytokine modulates MMP gene expression in these cell lines. The mechanisms by which OSM activates transcription of MMP-1 (interstitial collagenase) were analyzed, and an OSM-responsive element (OMRE) in which the AP-1 binding site and the SBE element cooperate together to achieve maximal inducibility by OSM has been characterized.
Recombinant Plasmid Constructions-In order to create a pGLMP1-Prom reporter vector, the human MMP-1 promoter (12) (Ϫ525/ϩ15 fragment) was generated by polymerase chain reaction with a human genomic DNA template and inserted into pGL2-Basic vector (Promega) in BglII site. In pGLMP1-Prom(muSBE), a substitutive mutation was introduced by employing the Deng and Nickoloff method (35), using as a mutagenic primer 5Ј-CACCTCTGGCTTTCTATCAGGGCAAGG-ACTC-3Ј.
The oligonucleotides containing the wild-type OMRE of the human MMP-1 (12) and TIMP-1 (36) promoter fragments, with or without mutation in either the SBE site or AP-1 site (see Table I), were inserted as five repeats in a head-to-tail orientation in an XhoI site of pGL2-Promoter reporter vector (Promega). Orientation, copy number, and sequences were verified by DNA sequencing.
Cell Culture and Treatment-Subconfluent monolayer cultures of normal human dermal fibroblasts (NHDF; Clonetics Co., San Diego, CA) were maintained in fibroblast growth medium. Experiments with NHDF cells were performed between 3-5 passages in fibroblast growth medium supplemented with 10% fetal bovine serum (FBS), unless otherwise noted. Human normal astrocytes were cultured in minimal essential medium containing 10% FBS, 1% glucose, 1 mM sodium pyruvate, for up to 4 -5 passages before they were used for experiments.
Transient Transfection-Plasmid DNA (3 g of total DNA) was transferred to exponentially growing cells in Dulbecco's modified Eagle's medium supplemented with 10% FBS on six-well tissue culture plates using the calcium phosphate-mediated transfection protocol (75). The cells were harvested and assayed for luciferase activity according to the Promega protocol. Transfection efficiency was assessed by cotransfection of a ␤-galactosidase expression plasmid pSV-␤-Gal (Promega Corp., Madison, WI); luciferase activity was normalized to ␤-galactosidase activity. Each transfection was performed in duplicate and repeated three times using multiple DNA preparations. Yield of transfection was determined by in situ ␤-galactosidase assays (37) and ranged from 2 to 4% or from 25 to 35%, in NHDF or human astrocytes, respectively.
Analysis of Primary Response Gene Expression-Total cellular RNA was extracted from cells by the guanidinium thiocyanate method of Chomczynski et al. (38). Northern blots were made as described previously (39) and analyzed in a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).

OSM Modulates Matrix Metalloproteinase Gene
Expression-We compared matrix metalloproteinase gene expression in two "stromal" human primary cell lines, which share some common functional characteristics despite their different origin. Analysis of non-activated cells had shown the presence of MMP-1, MMP-2, MMP-3, and MMP-9 in NHDF cells, but only MMP-2 could be observed in primary astrocytes at a detectable amount. In both cell lines, the level of the MMP-2 expression was unchanged after treatment with different activators (Fig.  1). IL-1 and TNF␣ induced MMP-1 and MMP-3 in NHDF cells after 7 h (data not shown), but not MMP-9, and this effect was enhanced after 24 h (Fig. 1). MMP-1 and MMP-9 expression was up-regulated by IL-1 in human astrocytes, whereas MMP-3 was not ( Fig. 1).
When NHDF cells were exposed to OSM, an increased level of MMP-1 mRNA (2.5 times) was observed (Fig. 1). The treatment of NHDF cells with OSM for 24 h in the presence of IL-1␤, TNF␣, or fibroblast growth factor resulted in enhancement of MMP-1 mRNA up to 12, 24, or 9.5 times, respectively, while mediators without OSM induced synthesis only 5.8, 8.8, and 4 times (Fig. 1). OSM alone or together with IL-1␤, TNF␣, and fibroblast growth factor transiently enhanced MMP-9 after 7 h in NHDF cells (data not shown). In human astrocytes, MMP-3 was induced only when cells were challenged with OSM and IL-1 simultaneously (Fig. 1). The cooperation between signals induced by OSM in presence of either IL-1 or PMA resulted in 100-fold or 365-fold, respectively, stimulation of MMP-1 in human astrocytes (Fig. 1). Northern blot analysis did not detect an increase of endogenous MMP-1 mRNA in response to OSM alone in these cells, perhaps, because basal MMP-1 expression was below the level of detection (Fig. 1). However, the MMP-1 promoter was responsive to OSM in NHDF cells as well as in human astrocytes (Fig. 2).
The role of OSM in the regulation of the human endogenous matrix metalloproteinase inhibitors, TIMP-1, TIMP-2 ( Fig. 1), and TIMP-3 (data not shown), was investigated. In human astrocytes, TIMP-1 was strongly activated by IL-1␤ and PMA, were transfected with pGLMP1-Prom or pGLMP1-Prom(muSBE) or pGLMP1-AP1-SBE(5x) as described in "Experimental Procedures." Cells were rescued for 12 h and then were untreated or treated with OSM, PMA, or both for various times. Shown is the 24-h activation time, which gave maximal stimulation of luciferase activity. A pGL2Promoter (Promega) was used as a control of transfection, and its transcriptional activity in unstimulated cells was defined as 1 unit of relative luciferase activity. and only weakly by OSM (Fig. 1). We did not detect any upregulation of the other two inhibitors in these cells, apart from TIMP-3 activation by PMA. In contrast, TIMP-1 was not induced by IL-1 or by TNF␣ in NHDF cells, similar to results obtained with both synovial fibroblasts and endothelium (42), but was increased significantly when the cells were exposed to either OSM or PMA.
OSM Induces Sequence-specific MMP-1 and TIMP-1 Promoter-binding Proteins-In order to determine the mechanism by which transcription of MMP-1 can be regulated by OSM, we addressed the question of which regulatory element(s) in the MMP-1 gene confers responsiveness to OSM. It was demonstrated that a 4-fold and 5-fold activation of the MMP-1 promoter (Ϫ525/ϩ15) by OSM in transfected NHDF cells and human astrocytes, respectively, could be attained (Fig. 2). Recently, the JAK-STAT signaling pathway has been clearly linked to OSM-mediated cell activation (32). Our sequence analysis has revealed the presence of an element homologous to the STAT binding site (SBE) (33,(43)(44)(45) in the 5Ј-flanking region of the human MMP-1 gene between Ϫ53 and Ϫ45. It should be pointed out that the SBE is only partially responsible for OSM-mediated activation of the MMP-1 promoter, since mutation of this site did not completely remove OSM inducibility (Fig. 2). Thus, SBE appears to function in conjunction with other elements of the MMP-1 promoter, which, together with SBE, could be required in order to achieve a full responsiveness to OSM. It has been demonstrated previously that the positive and negative effects of tumor promoters, oncogene products, and growth factors on MMP-1 transcription is at least in part mediated by the proximal AP-1 site (12)(13)(14)46). We demonstrated that the multimerized wild-type AP-1-SBE element conferred a very strong responsiveness to OSM, up to 7-fold or 16-fold in NHDF cells and human astrocytes, respectively (Fig. 2).
Using the probe containing both the SBE and AP-1 sites of the MMP-1 promoter and different oligonucleotides for competition, we mapped the OSM-induced pattern of DNA-binding proteins. The most retarded bands contained SBE elementbinding proteins, as these components were easily out-competed when oligonucleotides with an SBE site or SIE were present in excess during DNA-protein complex formation (Fig.  3). Significantly, the OSM-induced protein binding to the SBE element was activated very early, indeed within minutes. On the other hand, faster migrating complexes were composed of the AP-1-binding proteins, and addition of any oligonucleotide with a typical AP-1 site resulted in their complete disappearance. Increase in binding to the AP-1 site was also noted, although not as rapid as to the SBE element, and the maximal level of induction was reached in approximately 1 h. The binding of transcription factors to SBE site in response to OSM was not dependent on binding to the AP-1 site (and vice versa), as was demonstrated in EMSA using as probes AP-1/SBE motifs with either AP-1 or SBE sites mutated (data not shown). Very similar results were demonstrated in parallel experiments, where a DNA fragment corresponding to Ϫ97/Ϫ75 DNA sequence in the human TIMP-1 promoter was used as a probe (Fig. 3B), although some differences in binding to SBE motifs were detected.
Activation of JAK-STAT Signaling Pathway by OSM-In order to examine intracellular signaling in NHDF cells induced by OSM, we investigated the activation of an OSM receptorassociated tyrosine kinase. Tyrosine phosphorylation of Jak1 and Jak2, but not Tyk1, was observed as a result of cytokine activation (Fig. 4A).
In DNA gel shift assays (Fig. 3), the activation of the SBEbinding proteins by OSM was rapid and had a rather transient characteristic. In fact, the elevated DNA binding activity to the SBE element was observed only 2 min after OSM stimulation, being sustained for the next hour. However, after 1 h of incubation of NHDF cells with OSM, we could not detect any more activated STAT proteins. We also observed both Stat1 and Stat3 are nucleus translocated in response to OSM in human fibroblasts (data not shown), which correlated very well with the EMSA results.
Wagner et al. (43) and Zhong et al. (47) have determined the pattern and composition of proteins binding to the sis-inducible element of the fos promoter (SIE) in DNA mobility gel shift assays using cell extracts from interferon ␥or IL-6-induced cells. The composition analysis of these bands revealed that SIFA consists of a Stat3 homodimer, while SIFC contains a Stat1 homodimer and SIFB is a Stat1/Stat3 heterodimer (47). In Fig. 4 we show that OSM is able to induce SIFA, SIFB, and SIFC complexes in gel mobility shift assays where the SIE element (5Ј-TTTCCCGTAAA-3) (43) was used as a probe. We then examined OSM-induced DNA-binding proteins, testing as binding site and SBE were used as labeled with 32 P and used as probes in EMSA assays. NHDF cells were treated with OSM for given periods of time, and total protein extracts were prepared as described under "Experimental Procedures." In competition studies a 200-fold excess of unlabeled competitor (see Table I) was used. The bands corresponding to SBE or AP-1 binding complexes are marked on the right side. a probe the SBE element of the MMP-1 promoter 5Ј-ATTTCT-GGAAG-3Ј. It appeared that the Stat1 homodimer had the highest affinity to this element although we also observed some formation of a Stat3 homodimer and, presumably, a Stat1/3 heterodimer DNA complex. In addition, we noticed a new OSMinduced DNA binding complex, which was completely supershifted with antiStat1, while anti-Stat3 antibodies failed to interact. This complex could contain a multimer of Stat1 with another member of the STAT family or could represent an example of cross-interaction between a STAT protein and different transcriptional factors. In contrast, the SBE element of the TIMP-1 promoter 5Ј-CATCCAGGAAG-3Ј preferably bound the Stat3 homodimer and, to a lesser extent, the Stat1 homodimer, but bound very little if any of the Stat1/3 heterodimer. The transcriptional discrepancy between MMP-1 and TIMP-1 responsiveness to OSM is at least to some extent the reflection of differences in binding competence of these two SBE sites, resulting from changes in DNA sequences, but the mechanism underlying such an effect remains obscure.
OMRE Is the Target for Cooperative Action of STAT with AP-1 during Response to OSM-We have already demonstrated that both AP-1 and SBE sites in MMP-1 and TIMP-1 promoters are receiving an OSM-induced signal (Figs. 2 and 3). In order to demonstrate that OMRE is a mediator of transcriptional activation in response to OSM, we constructed a luciferase reporter gene with five repeats of OMRE (from MMP-1 and TIMP-1 promoters) in head-to-tail orientation inserted pGL2-Promoter: pMP1-AP1-SBE(5x)-luc and pT1-AP1-SBE(5x)-luc. Human astrocytes were transfected with these constructs and challenged with either OSM or PMA or both. PMA was chosen, since protein kinase C was shown to be an inducer of the AP-1 site in the MMP-1 promoter (12). We found that pMP1-AP1-SBE(5x)luc and pT1-AP1-SBE(5x)-luc were maximally induced after 24 h of treatment with OSM, resulting in 16-fold and 8-fold stimulation of transcription, respectively (Fig. 5). Remarkably, PMA, a protein kinase C activator, failed to cooperate with OSM in the transcriptional activation of the OMRE-driven luciferase expression vector. We do not understand the basis of this phenomenon, but it is possible that an elevated PMAinduced signaling pathway is competing out a common coacti-vator of transcription, as was demonstrated in case of the AP-1 and a nuclear receptor (48). Next, we addressed the question as to whether cis-acting regulatory sequences within OMRE of MMP-1 and TIMP-1 promoters cooperate with each other to achieve maximal induction of transcription. Mutant constructs were prepared, which had five repeats of OMRE in a head-totail orientation inserted into the pGL2-Promoter plasmid but were defective either in the AP-1 site or in the SBE element in a manner so that they failed to bind transcription factors (Table I, Figs. 2 (A and B) and 5). We showed that both cis-acting motifs in OMRE are fundamental to achieving an OSM-induced maximal rate of transcription of OMRE-driven genes (Fig. 5). When the SBE element was mutated, the responsiveness to OSM was reduced drastically, although not completely. The basal level of the gene expression and PMA-inducibility remained unchanged, indicating that the SBE site is not involved in PMA-mediated activation. On the other hand, the AP-1 binding site is essential for both OSM-and PMA-activation and seems to be indispensable for the basal level of gene expression in this model. These data leave no doubt that both the AP-1 and SBE sites are necessary to confer OSM responsiveness in OMRE-driven genes, and their synergistical cooperation is crucial to accomplishing maximal rates of transcription.
OSM-induced Activation of OMRE Is Raf-dependent-Since the activity of both AP-1 and STAT can be modulated by the MAP kinase pathway, we investigated whether there was a requirement for one of the members of the growth promoting signal cascade, Raf-1 (49), during OMRE activation by OSM. Raf-1 is believed to be a central cytoplasmic signal transducer in the signaling pathway induced by many extracellular activators, but its direct effect on OSM-induced transcriptional activation has not been previously studied. In order to check this possibility, human astrocytes were co-transfected with pMP1-OMRE(5x), with a dominant negative mutant of Raf-1, pRSV-RafC4, or pRSV-RafC4PM17 serving as a control (34), and cells were challenged with OSM. It was demonstrated that the Raf-1 dominant negative expression vector suppressed OSM induction of the OMRE-driven luciferase expression vector, whereas pRSV-RafC4PM17 was not effective (Fig. 7), FIG. 4. OSM induces JAK-STAT signaling pathway. A, Jak1 and Jak2 are tyrosine-phoshorylated after OSM treatment. Western blot analysis of activation of JAK/Tyk family kinases in NHDF cells for 12 min with OSM, IL-6, or interferon ␥, and then immunoprecipitated either with anti-Jak1 or anti-Jak2 or anti-Tyk2. Immunoprecipitates were subjected to SDS-PAGE and probed with anti-phosphotyrosine antibodies. The filters were stripped and reprobed with different anti-JAK antibodies. B, demonstration of Stat1 and Stat3 proteins in DNA binding complexes induced by OSM. 32 P-Labeled MP1-(muAP1)-SBE or T1-(muAP1)-SBE or hSIE DNA fragments were used in EMSA performed with total cellular extracts from NHDF cells exposed for 12 min to OSM as described under "Experimental Procedures." which is indicative of a Raf-1 involvement in OSM-induced signal transduction.
The recent demonstration of the activation of the STAT signaling pathway by the oncogene for Src raises the possibility that STAT proteins can contribute to oncogenesis by Src (50). Moreover, v-src and other oncogenes were found to activate the MMP-1 promoter (16). Src is a non-receptor tyrosine kinase, which was demonstrated to activate Raf-1 (51) and mediate regulation of a variety of transcription factors. Overexpression of the v-src oncogene in human astrocytes induced a very high transcriptional activity of the OMRE-driven luciferase expression vector and completely diminished further OSM inducibility. Cotransfection of human astrocytes with the v-src expression vector, the Raf dominant negative mutant, and the OMREdriven luciferase expression vector revealed that Raf is only in part involved in v-src OMRE element activation (Fig. 6). At this point the biological relevance of this observation is unclear. DISCUSSION Astrocytes constitute approximately 50% of the total cell number within the central nervous system (52). Aside from their pivotal function of elaborating extracellular matrix components in the brain, these cells play essential roles in the regulation of the extracellular environment by responding and releasing growth factors and cytokines, which are involved in maintaining homeostasis in the brain extracellular fluid (1,53). In our experiments it was found that human primary astrocytes can express several matrix metalloproteinases: MMP-1, MMP-3, MMP-2, and MMP-9, as well as their endogenous inhibitors TIMP-1, TIMP-2, and TIMP-3, although each was apparently under cell type-specific differential regulation (Fig. 1). In unstimulated cells, the only detected mRNA of all of the examined metalloproteinases was for MMP-2 (Fig. 1). Another gelatinase, MMP-9, could, however, be induced after stimulation with IL-1␤ or PMA, as was recently demonstrated in rat astrocytes (54). IL-1␤ and OSM synergize to stimulate MMP-1, particularly in human primary astrocytes, where they show a nearly 90-fold induction over that found with either cytokine alone (Fig. 1). Conversely, OSM, which is not an efficient inducer of TIMP expression in human primary astrocytes, has a strong positive effect on TIMP-1 in human fibroblasts. IL-1, synergistically with OSM, does induce MMP-3 expression in human astrocytes, whereas we have not observed this effect in NHDF cells. Since TIMP-1, -2, and -3 are not strongly up-regulated by OSM in human primary astrocytes, a net positive proteolytic activity will be created upon exposure of these cells to IL-1␤ and OSM. However, in NHDF cells, IL-1␤ and OSM, together, not only induce MMP-1 but also TIMP-1. Thus MMP regulation may be more tightly controlled in fibroblasts than in astrocytes, particularly when OSM and IL-1␤ work in concert with each other.
There is strong evidence linking neurodegenerative diseases with an imbalanced expression of proteinases and proteinase inhibitors (3,4,55), and an elevated metalloendopeptidase activity has been detected in Alzheimer-affected hippocampal tissue in comparison to age-matched controls (56 -58). Since MMPs have a broad specificity, it is not without reason to assume that up-regulation of their synthesis relative to that of controlling inhibitors by elevated level of inflammatory cytokines including OSM could allow then to play a major role in neurodegenerative diseases. This is consistent with the recent finding that overexpression of OSM in the central nervous system of transgenic mice was detrimental and lethal (59).
The data presented indicate that both of the cis-acting sequences, AP-1 and SBE, present in an OMRE not only are the targets for OSM-induced nuclear effectors but also cooperate together in transcriptional activation. We have also found this characteristic composite AP-1/SBE motif in promoters of two distally related human genes encoding human MMP-1 and its endogenous inhibitor TIMP-1. The requirement of AP-1 and SBE sites to confer OSM responsiveness onto the rat TIMP-1 promoter was also reported (60), but neither synergistic cooperation between AP-1 and SBE in response to the cytokine nor AP-1 activation in rat hepatocytes was noted previously. Composition analysis of SBE complexes revealed involvement of Stat1 and Stat3 in OSM-elevated transcriptional activation of MMP-1 and TIMP-1 genes (Fig. 4). Because of different DNA sequences, the SBE of the MMP-1 promoter had the highest affinity to the Stat1 homodimer, while the Stat3 homodimer preferably bound to the TIMP-1 promoter, although both SBE elements could bind to a lesser extent other homo-and heterodimers of STAT family members activated by OSM (Fig. 4). Whether this could explain how the JAK-STAT signal transduction pathway can accomplish specificity remains to be established. Moreover, OSM induces a transcription factor that binds to the AP-1 site (Fig. 2). The activity of AP-1 is the function of a large group of bZip transcriptional factors including those of the Fos and Jun families, which function as Jun-Jun or Jun-Fos dimers, forming a bimolecular DNA binding domains. Cross-family dimerization with other members of the FIG. 5. AP-1 cooperates with STAT to achieve maximal transcription activity of the OMRE-driven luciferase vector. Human primary astrocytes were transfected with pGLMP1-AP1-SBE(5x), pGLMP1-AP1-(muSBE)(5x), pGLMP1-(muAP1)-SBE(5x), pGLT1-AP1-SBE(5x), pGLT1-AP1-(muSBE)(5x), or pGLMP1-(muAP1)-SBE(5x) as described under "Experimental Procedures." Cells were rescued for 12 h and then challenged with OSM, PMA, or both for next 24 h. Cellular extracts were prepared and the luciferase activity measured as described under "Experimental Procedures." bZip family alters DNA binding specificity (61)(62)(63). It will be interesting to further investigate the composition of OSMinduced AP-1 binding complexes and their cooperation with SBE element-binding proteins in OMRE-driven transcription machinery.
A clue to the mechanism by which OSM might activate transcription of OMRE-driven genes is beginning to emerge (Fig. 7). Clearly, OSM elicits cellular responses that cannot be limited to any single signaling cascade (JAK-STAT). We postulate that cells must exert active MAP kinase and JAK-STAT pathways in order to maintain full OMRE inducibility by OSM. Activation and cooperation between these two signaling cascades in response to OSM is required to achieve maximal transcriptional activity of the OMRE-driven luciferase vector. Raf-1 is an activator of multiple nuclear effectors, including Fos, Jun, and activating transcription factors (49,64). Moreover, phosphorylation on serine 727 in Stat1 and Stat3 is required for their maximal transcriptional activity (65). ERK2, which is a serine/threonine kinase downstream from Raf in the MAP kinase pathway, was reported to directly convert Stat1 to its fully active form by phosphorylating serine 727 (66). Recently, it was shown that Jak2 was required for growth hormone-stimulated activation of ERK2/MAPK (67). This could explain how OSM induced signaling bifurcates to the ERK2/ MAPK and STAT signaling pathways and converges with Stat1 and Stat3 phosphorylation by ERK/MAPK, and by activation of OMRE. This is consistent with our observation that suppression of endogenous Raf-1 impaired OSM inducibility of the OMRE-driven luciferase vector (Fig. 6).
The synergistic enhancement of transcription by AP-1 and STAT proteins could be accomplished by utilizing a coactivator of transcription referred to as cointegrator (48) such as a nu-FIG. 6. OSM-induced AP1/SBE cooperation in transcription activation of the OMRE-driven luciferase construct is Raf-dependent. Human primary astrocytes were cotransfected with pGLMP1-AP1-SBE(5x) and pRSVRaf-C4 or pRSVRaf-C4PM17 or with an empty expression vector as described under "Experimental Procedures." Cells were rescued for 8 h, followed by 24 h of serum deprivation, and then exposed to OSM for the next 24 h. Cells cotransfected with the v-src expression vector were also examined. The cellular extracts were prepared and the luciferase activity was measured as described under "Experimental Procedures."  (68), the protein linked to transactivation by CREB, Jun/Fos, and nuclear receptors (48,69,70). Moreover, the binding of CREB-binding protein to CREB or AP-1 proteins is serine phosphorylation-dependent (69,70). The strong activation of OMRE by v-src and suppression of OMRE inducibility by dominant negative Raf-1 mutant (Fig. 6) support this hypothesis, since both Src and Raf-1 kinases trigger serine kinase activity (49,51,64).
Alternatively, OSM has been shown to stimulate a rapid but transient elevation of primary response genes including c-Jun (71), and products of this gene can enhance OMRE, already activated by STAT transcriptional factors, as a positive feedback regulatory loop. The mechanism by which OSM activates transcription of immediate-early genes is not always clear, but most likely this induction is mediated by the JAK-STAT pathway, and the newly biosynthesized transcriptional factors could further enhance transcription of later genes such as MMPs and TIMPs.
OSM is expressed by activated monocytes and T-lymphocytes (19,20), and it may be involved in modulation of stromal cell function at inflammatory sites since fibroblasts, astrocytes, and also endothelium (26) respond very strongly to this cytokine. The ability to bind gp130 directly as well as two other receptor subunits, leukemia inhibitory factor receptor and oncostatin M receptor (72,73), provides OSM with a variety of activities that any other member of this cytokines family. The recent finding that a tissue-specific targeted bovine OSM in transgenic mice had a profound and often lethal effects (59) raises the possibility that OSM plays a role in the regulation of development and homeostasis. The physiological function of OSM is still unknown, and the targeted gene disruption could be the experiment of choice in order to establish the primary function of this cytokine.