The ERK/MAPK Pathway Regulates the Activity of the Human Tissue Factor Pathway Inhibitor-2 Promoter*

Human tissue factor pathway inhibitor-2 (hTFPI-2) is a 32-kDa serine protease inhibitor that is associated with the extracellular matrix. hTFPI-2 inhibits several extracellular matrix-degrading serine proteases and may play a role in tumor invasion and metastasis. To study the signal transduction pathway that leads to the activation of the hTFPI-2, we cloned the potential promoter region of this gene adjacent to a heterologous luciferase reporter gene. Phorbol 12-myristate 13-acetate (PMA) induced the luciferase reporter gene in HEK293 cells and other epithelial cell lines, such as the human lung carcinoma A549 cells, the breast carcinoma MCF7 cells, and the cervical HeLa cells. This PMA induction was blocked with the MEK inhibitor UO126, suggesting that the PMA-induced activation of the hTFPI-2 promoter is mediated through MEK. Furthermore, epidermal growth factor induced the luciferase reporter gene in HeLa cells. Cotransfection of the luciferase construct with constitutively active components of the Ras/Raf/MEK/ERK pathway in EcR-293 cells lead to a 7- to 92-fold induction of the luciferase reporter gene, indicating that regulation of hTFPI-2 is mediated through this pathway. A series of luciferase reporter gene constructs with progressive deletions of the 5′-flanking region suggested that the minimal basal promoter activity is located between nucleotide positions −89 and −384, whereas the minimal inducible promoter activity is between −89 and −222. We have used the computer program TFSEARCH and mutagenesis to analyze potential transcription factor binding sites. We identified an AP-1 binding site at nucleotide position −156 (inducible activity) and a Sp1 site at position −134 (basal activity) as potential cis-acting elements in the promoter region of the hTFPI-2.

Human tissue factor pathway inhibitor-2 (hTFPI-2) is a 32-kDa serine protease inhibitor that is associated with the extracellular matrix. hTFPI-2 inhibits several extracellular matrix-degrading serine proteases and may play a role in tumor invasion and metastasis. To study the signal transduction pathway that leads to the activation of the hTFPI-2, we cloned the potential promoter region of this gene adjacent to a heterologous luciferase reporter gene. Phorbol 12-myristate 13-acetate (PMA) induced the luciferase reporter gene in HEK293 cells and other epithelial cell lines, such as the human lung carcinoma A549 cells, the breast carcinoma MCF7 cells, and the cervical HeLa cells. This PMA induction was blocked with the MEK inhibitor UO126, suggesting that the PMA-induced activation of the hTFPI-2 promoter is mediated through MEK. Furthermore, epidermal growth factor induced the luciferase reporter gene in HeLa cells. Cotransfection of the luciferase construct with constitutively active components of the Ras/Raf/MEK/ERK pathway in EcR-293 cells lead to a 7-to 92-fold induction of the luciferase reporter gene, indicating that regulation of hTFPI-2 is mediated through this pathway. A series of luciferase reporter gene constructs with progressive deletions of the 5flanking region suggested that the minimal basal promoter activity is located between nucleotide positions ؊89 and ؊384, whereas the minimal inducible promoter activity is between ؊89 and ؊222. We have used the computer program TFSEARCH and mutagenesis to analyze potential transcription factor binding sites. We identified an AP-1 binding site at nucleotide position ؊156 (inducible activity) and a Sp1 site at position ؊134 (basal activity) as potential cis-acting elements in the promoter region of the hTFPI-2.
Growth hormones and the tumor promoting agent PMA 1 initiate diverse intracellular signaling pathways that lead to the phosphorylation of transcription factors and ultimately to the regulation of target genes. Among the pathways often used to transduce signals are the mitogen-activated protein kinase (MAPK) cascades. These cascades consist of a three-kinase module that includes an MEK kinase (MEKK), which activates an MAPK/ERK kinase (MEK), which in turn activates a MAPK (1). Three well characterized MAPKs have been described in mammalian cells: the mitogen-responsive ERK, the stress-responsive JNK/SAPK, and the p38 MAPK. The Ras/Raf/MEK/ ERK signaling cascade regulates cell proliferation and differentiation (2). Components of this pathway are often activated in human tumors and oncogenic Ras, and constitutively activated ERKs have been found in a large variety of malignancies (3)(4)(5). We have used transcript profiling to identify genes that are differentially regulated by this pathway. Among the many activated genes, we have identified the human tissue factor pathway inhibitor-2 (hTFPI-2) as a gene that is highly upregulated by the ERK/MAPK pathway.
hTFPI-2 is a 32-kDa serine proteinase inhibitor with three tandem Kunitz-type domains (6,7) and has high homology to hTFPI-1, a regulator of the extrinsic blood coagulation pathway. The second Kunitz-type domain of hTFPI-1 binds to factor Xa, and this complex inhibits the activity of the factor VIIatissue factor complex through interaction of the first Kunitztype domain in hTFPI-1 and the active site of VIIa/TF (8). Despite the high homology of hTFPI-2 to hTFPI-1, hTFPI-2 is a weak inhibitor of the activation of factor X (9) and hTFPI-2 poorly inhibits tissue factor. However, hTFPI-2 inhibits the tissue factor-factor VIIa complex and a variety of serine proteases, including trypsin, plasmin, plasma kallikrein, chymotrypsin, and cathepsin G, but it does not inhibit thrombin, urokinase-type plasminogen activator, and tissue-type plasminogen activator (6,9). Most of the hTFPI-2 expressed in dermal fibroblasts and endothelial cells localizes within the extracellular matrix, probably bound to heparan sulfate (10 -12). hTFPI-2 can prevent the conversion of Pro-MMP-1 (matrix metalloprotease 1, interstitial collagenase) and Pro-MMP-3 (matrix metalloprotease 3, stromelysin-1/transin-1) into MMP-1 and MMP-3 by plasmin and trypsin (13) and therefore might indirectly regulate matrix proteolysis and connective tissue turnover.
The role of hTFPI-2 in cancer progression is not completely elucidated. On one hand, hTFPI-2 has an anti-invasive effect that might be mediated via inhibition of plasmin that activates proteases promoting degradation of the extracellular matrix and tumor invasion. Several tumor cell lines were less invasive when they were stably transfected with hTFPI-2 cDNA (14 -17). On the other hand, hTFPI-2 has been shown to have a pro-invasive effect in hepatocellular carcinoma cells (18).
In this study, we investigated the signaling pathway and transcriptional elements that regulate the expression of the hTFPI-2 gene in epithelial cells. Although it has been shown that PMA can stimulate hTFPI-2 gene expression in glioma cells and that the promoter region Ϫ312 to ϩ1 is critical for minimal and inducible promoter activity of hTFPI-2 (19), the signal transduction pathway by which PMA induces gene expression of hTFPI-2 and the promoter elements involved in hTFPI-2 regulation have not been studied in detail. Here we show that the hTFPI-2 gene expression is regulated by the ERK/MAPK signaling pathway and that the activity of this pathway is directed to an AP-1 site in the promoter of the hTFPI-2 gene.

EXPERIMENTAL PROCEDURES
Materials-The Phorbol 12-myristate 13-acetate was ordered from Sigma-Aldrich Chemicals Co. (St. Louis, MO) and recombinant human epidermal growth factor from Austral Biologicals (San Raman CA). The LipofectAMINE Plus reagent as well as the Ecdysone-Inducible Mammalian Expression system, including EcR-293 cells, zeocin, and pronasteron A, and the expression vector pIND were purchased from Invitrogen Corp. (Carlsbad, CA). The MEK inhibitor UO126 and the Luciferase assay system were obtained from Promega Corp. (Madison, WI). The Phospho-p44/42 MAPK (Thr 202 /Tyr 204 ) antibody was purchased from Cell Signaling Technology (Beverly, MA). The c-Myc (9E10) monoclonal antibody and the MEK-1 (C-18) and Raf-1 (C-12) polyclonal antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). The QuikChange TM site-directed mutagenesis kit was from Stratagene (La Jolla, CA). The FuGENE6 transfection reagent was obtained from Roche Diagnostics Corp.
Cell Culture-The HEK293, HeLa, A549, and MCF7 cell lines were obtained from American Type Culture Collection (ATCC, Rockville, MD). All the cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM) (Wisent Inc., St. Bruno, Quebec, Canada) with 10% fetal bovine serum (HyClone Laboratories, Mississauga, Ontario, Canada), except for the A549 cell line, which was cultured in DMEM supplemented with 5% fetal bovine serum.
Plasmids-The plasmid pRK5 containing the myc-RasV12 gene with a glycine to valine mutation at amino acid position 12 and a sequence (EQKLISEEDLGS) containing a myc epitope inserted between a methionine and a threonine (amino acid positions one and two) was a generous gift from Nathalie Lamarche-Vane, McGill University, Montreal, Canada. The plasmid was digested with the restriction enzyme ClaI and the 5Ј-overhang filled in with Klenow. The myc-RasV12 fragment was subsequently released by ApaI and cloned into the EcoRV/ApaI site of pIND.
pAN130-containing Raf-1 (20) served as a template to amplify the carboxyl-terminal, catalytic domain of Raf-1 (RafCT) using the oligonucleotides O-1 and O-2 listed in Table I. At the same time an EcoRI site (printed in boldface) was created to facilitate the cloning of the amplified 962-bp fragment into the EcoRI/XhoI sites of pcDNA3. A Kozak sequence (printed in italic) was inserted in-frame by cloning the phosphorylated, annealed oligonucleotides O-3 and O-4 (Table I) into the EcoRI site of RafCT, creating an AflII site (printed in boldface). The nucleotide sequence was verified by sequencing. The Kozak-RafCT fragment was released by AflII/XhoI and cloned in the corresponding sites of pIND.
MEK-1 was released from pAN104 (20) by BamHI/XhoI restriction digest and cloned into the corresponding sites of Bluescript KS. MEK-1SD was created by site-directed mutagenesis (QuikChange TM sitedirected mutagenesis kit, Stratagene) using two complementary oligo-nucleotides. The nucleotides that were changed to obtain a serine to aspartic acid mutation in MEK-1SD are underlined in O-5 in Table I. The nucleotide sequence was verified by sequencing, and the MEK-1SD was released from Bluescript with BamHI/XhoI and cloned into the corresponding sites of pIND.
The plasmid mycCMV5-ERK2-MEK1-LA was kindly provided by Melanie H. Cobb, University of Texas Southwestern, Dallas, TX. The plasmid was digested with HindIII, the 5Ј-overhand was filled in with Klenow fragment, and ERK2-MEK1-LA was subsequently released by KpnI and cloned into the KpnI/EcoRV sites of pIND.
Amplification of the hTFPI-2 Promoter Region-Genomic DNA was isolated from 293 cells. The cells were washed in PBS, lysed in a buffer containing 10 mM NaCl, 10 mM EDTA, 0.5% Sarkosyl, and 10 mM Tris, pH 8, and incubated with proteinase K (10 mg/ml) at 50°C overnight. After two phenol and two chloroform extractions, the genomic DNA was ethanol-precipitated and dissolved in TE, pH 8. A 1.5-kb fragment of the 5Ј-flanking region of the hTFPI-2 was amplified with an Expand High Fidelity PCR system (Roche Molecular Biochemicals) using the oligonucleotides O-7 and O-8 (Table I) to create a KpnI and a BglII restriction site (printed in boldface). This KpnI/BglII fragment was cloned into the plasmid pXP2, which contains the firefly luciferase reporter (21) and was a generous gift from Mark Featherstone, McGill University, Montreal, Canada. The resulting construct was named p-1511-luc. The sequence was found identical to the one published by Kamei (22) except for a C to A change at nucleotide position Ϫ47 relative to the translation start site.
Mutagenesis of Single Potential Transcription Binding Sites-The putative Sp1 transcription factor binding site GGGGCGG between nucleotide positions Ϫ190 and Ϫ184 was changed to GGGGCAA, the putative AP-1 site TGAATCA between nucleotide positions Ϫ162 and Ϫ156 was altered to GCTAGCA, and the overlapping Sp1/AP-2 and GC box GGCTCCGCCCCGGCGGGGG between nucleotide positions Ϫ144 and Ϫ126 was modified to GGCTTTGCCCCAACGGGGG. Doublestranded oligonucleotides containing the corresponding nucleotide changes were phosphorylated, annealed, and cloned as HindIII/EagI fragments into the corresponding sites of p-1055-luc.  (Table I). p-222MAP1/MSP1B-luc and p-222MAP1/MAP2-luc were created by PCR using the Expand High Fidelity PCR system using the oligonucleotides O-15 and O-16 or O-17, respectively. In both constructs the putative AP-1 site TGAATCA between nucleotide positions Ϫ162 and Ϫ156 was altered to GCTAGCA, and either the putative Sp1 site between nucleotide positions Ϫ140 and Ϫ134 CCGCCCC altered to TTGCCCC or the putative AP-2 site between nucleotide positions Ϫ136 and Ϫ126 CCCGGCGGGGG was altered to CCCGGCAAGGG. The sequence for the HindIII and EagI restriction sites are printed in boldface, and the nucleotide changes are underlined (Table I). The PCR products were verified by sequencing and subcloned as HindIII/EagI fragments hTFPI-2 Gene Is Regulated by the ERK/MAPK Pathway into the corresponding sites of p-1055-luc. Transient Transfection-For inhibitor studies, the HEK293 cells (150,000 cells/well) were seeded the day prior to transfection. The cells were cotransfected with 1 g of p-1511-luc and 70 ng of a GFP-spectrin control plasmid (23). 24 h after transfection, the cells were serumstarved for 24 h. Where indicated, the MEK1/2-specific inhibitor UO126 was added 15 min prior to PMA induction. The cells were harvested in phosphate-buffered saline (PBS) and divided into two tubes. Cells in one tube were lysed using cell culture lysis reagent (Promega) and centrifuged at 12,000 ϫ g for 5 min, and the cell extract was assayed for firefly luciferase activity using the luciferase reporter assay system (Promega). Light intensity was measured by using a microtiter plate luminometer (DYNEX Technologies, Inc., Chantilly, VA). The cells in the second tube were trypsinized and washed, and the percentage of cells that express GFP was determined by cytofluorimetry (EPICS XL-MCL) to control for transfection efficiency. Determination of protein by the Bradford assay (Bio-Rad) was carried out, controlling for harvesting efficiencies. Luciferase activity was expressed as firefly light units/g of protein and normalized for transfection efficiency.
HEK293, HeLa, A549, and MCF7 cells were seeded into six-well plates at 150,000, 250,000, 250,000, and 400,000 cells/well, respectively, 1 day prior to transfection. All cells were transfected with 1 g of the reporter gene constructs. HEK293 and HeLa cells were transfected with FuGENE6 at a 3 l of FuGENE/1 g of DNA ratio, and A549 and MCF7 cells were transfected with LipofectAMINE Plus according to the manufacturer's protocol. 24 h after transfection, the cells of each well were split into six wells of a 24-well plate. 7 h later, the cells were serum-starved overnight and subsequently treated with PMA (250 nM) or EGF (50 ng/ml) as indicated. Cell extracts were lysed and analyzed as described above. The -fold stimulation of luciferase was calculated as firefly light units/g of protein of PMA-or EGF-treated cells divided by the firefly light units/g of protein of nontreated cells.
Two days prior to transfection, EcR-293 cells cultivated in DMEM supplemented with 10% fetal bovine serum and Zeocin (400 g/ml) were seeded into six-well plates at 200,000 cells/well. EcR-293 cells stably express the modified ecdysone receptor. A gene of interest, cloned into a pIND-based inducible expression vector, can be induced with the ecdysone analog pronasteron A that binds to the ecdysone receptor. Cells were transfected with 6 l of FuGENE6 and 1.07 g of total DNA (0.9 g of RasV12, RafCT, MEK-1SD, ERK2-MEK1-LA, or pIND control vector, 0.1 g of reporter plasmid, and 70 ng of GFP control plasmid). The medium containing fetal bovine serum was replaced 24 h after transfection with DMEM without serum, pronasteron A was added 4 h later to the medium to induce protein expression at a concentration of 6 M unless indicated differently, and cells were harvested 20 h later. Luciferase activity was expressed as firefly light units/g of protein and normalized for transfection efficiency.

RESULTS
We have observed that the hTFPI-2 gene was highly upregulated on microarrays that were probed with cDNAs originating from HEK293 cells treated with 250 nM PMA for 4 h (data not shown). Recently, PMA induction of a hTFPI-2-luciferase reporter gene has also been shown in glioma cells (19), and mRNA of hTFPI-2 is up-regulated in BeWo and JEG-3 trophoblast cells treated with PMA (24).
To study the signaling pathway that leads to the induction of hTFPI-2 gene expression by PMA and to investigate important promoter elements involved in its transcriptional regulation, we PCR-amplified a 1511-bp fragment of the 5Ј-flanking region of the hTFPI-2 gene from genomic DNA isolated from HEK293 cells and cloned the 1511-bp fragment adjacent to the firefly luciferase reporter gene (p-1511-luc).
hTFPI-2 Promoter Activity Induction by Phorbol Esters and Inhibition of the PMA-dependent Induction by the MEK Inhibitor UO126 -To assess whether this potential promoter region allows transcription of luciferase, we transiently transfected HEK293 cells with the reporter plasmid p-1511-luc and monitored changes in luciferase activity of PMA-treated and untreated cells. Cells incubated with 250 nM PMA showed a 10-fold stimulation of luciferase compared with cells without PMA treatment (Fig. 1A). Induction of the luciferase reporter gene was decreased by 90% if cells were preincubated with the MEK1 inhibitor UO126 at a concentration of 10 M. UO126 completely inhibited PMA induction at a concentration of 100 M, suggesting that PMA activates the hTFPI-2 promoter through the MEK signaling pathway (Fig. 1A).
As shown in the Western blot in Fig. 1B, PMA activated p44/ p42 MAPK in HEK293 cells and the phosphorylation of p44/p42 could be inhibited by the MEK1-inhibitor UO126 in a dose-dependent manner. This result indicates that MEK activation is necessary for the induction hTFPI-2 promoter activity by PMA.

Phorbol Esters and the Growth Factor EGF Can Stimulate the Luciferase Reporter Gene in Epithelial Carcinoma Cell
Lines-HEK293 cells have very recently been reported as atypical epithelial cells and may originate from neuronal cells (25). To determine whether the PMA induction of the hTFPI-2 gene observed in HEK293 cells was unique to this cell line or whether PMA promotes up-regulation of the hTFPI-2 gene in other epithelial cells lines as well, we transiently transfected human lung carcinoma A549 cells, breast carcinoma MCF7 cells and cervical carcinoma HeLa cells with the luciferase reporter plasmid p-1511-luc. A 3.7-fold PMA-dependent induction of the hTFPI-2 promoter activity was observed in A549 cells, whereas PMA induced the hTFPI-2 promoter activity in MCF7 and HeLa cells 30-or 20-fold, respectively (Fig. 2). Because PMA has been reported to transactivate the epidermal growth factor receptor (26), we tested whether the hTFPI-2 gene could be induced by the growth factor EGF. Although EGF up-regulated the hTFPI-2 promoter activity in HeLa cells 10fold, no substantial stimulation of the luciferase reporter gene was obtained in A549 and MCF7 cells (Fig. 2).
Ras, Raf, MEK, and ERK Can Induce the hTFPI-2 Promoter Activity-To assess the importance of the components of the ERK/MAPK signaling pathway in the regulation the hTFPI-2 promoter activity, we cotransfected EcR-293 cells with the plasmid p-1511-luc and vectors containing constitutively activated signaling components, such as RasV12, RafCT, MEK-1SD, ERK2-MEK1-LA, or the empty control vector (pIND). The -fold stimulation of luciferase was calculated as normalized luciferase activity obtained in cells expressing active signaling components divided by the luciferase activity of samples originating from vector-transfected control cells (Fig. 3A). Protein expression of RasV12, RafCT, MEK-1SD, and ERK2-MEK1-LA in transiently transfected EcR-293 cells is shown on Western blots in Fig. 3B. All the constitutively activated signaling components were well expressed. Expression of RasV12 stimulated the luciferase reporter gene 37-fold compared with the control vector. Expression of constitutively active signaling MAPK components further downstream of Ras, such as RafCT, MEK-1SD, and ERK2-MEK1-LA, induced the luciferase reporter the gene 7-, 92-, or 39-fold, respectively (Fig. 3A), indicating that the hTFPI-2 gene expression can be regulated by the Ras/Raf/ MEK/ERK pathway in EcR-293 cells. The highest activation of the hTFPI-2 promoter was obtained by MEK-1SD containing aspartic acids at amino acid positions 218/222 (27), whereas the RasV12 and ERK2-MEK1-LA fusion protein (28) activated the hTFPI-2 promoter to a similar extent. In conclusion, these results demonstrate that the Ras/Raf/MEK/ERK signaling pathway mediates regulation of the hTFPI-2 gene.
The Minimal Inducible Promoter Activity Is Located within the Ϫ89/Ϫ222-bp Region of the hTFPI-2 Promoter-We transiently cotransfected EcR-293 cells with a series of luciferase reporter gene constructs containing progressive deletions of the 5Ј-flanking region with either a vector containing RasV12 or the empty control vector pIND. The -fold stimulations of the Ras versus control vector are shown in Fig. 4A. 35-to 51-fold stimulations of luciferase were obtained with constructs p-1511-luc through p-222-luc, whereas no inducible luciferase activity was obtained with construct p-89-luc, suggesting that the minimal inducible promoter activity is located between nucleotide positions Ϫ222 and Ϫ89 (Fig. 4A). Similar results were obtained in PMA-treated HEK293 cells expressing the deletion constructs. PMA stimulated p-1511-luc through p-222luc 5-to 8-fold, whereas p-89-luc was not induced by PMA (Fig.  4B). Therefore, we decided to investigate the promoter elements in the Ϫ222/Ϫ89 region that are responsible for the 51-fold Ras and 8-fold PMA stimulations as compared with control cells.
The minimal basal promoter activity includes the Ϫ89/ Ϫ384-bp region, because p-384-luc showed similar basal luciferase activity as longer constructs, whereas no basal activity was obtained with construct p-89-luc (Table II). As shown in Table II, the luciferase activity dropped by 74% in vector (pIND)-transfected cells expressing p-222-luc compared with cells expressing p-384-luc, indicating that the region, Ϫ384 to Ϫ222 bp, contains transcription factor binding sites important for basal activity. Similarly, luciferase light units were higher in samples originating from cells transfected with RasV12 and a reporter gene construct containing 384 bp of an upstream promoter segment compared with cells transfected with RasV12 and a reporter construct containing only a 222-bp promoter segment (Table II). This loss of activity is similar to the loss of basal activity in the Ϫ222/Ϫ384 region and therefore may be due to the absence of transcription factor binding sites important for basal activity. However, we can not rule out that the region, Ϫ384/Ϫ222 bp, may contain additional enhancer elements that can contribute to the inducible activity.
Identification of an AP-1 Site as a Specific Inducible DNA Response Element-To determine potential cis-acting elements responsible for the PMA and Ras inducibility of the Ϫ89/ Ϫ222-bp promoter region, we used the computer program TF-SEARCH version 1.3 that searches highly correlated sequence fragments versus the TFMATRIX transcription factor binding site profile data base by E. Wingender, R. Knueppel, P. Dietze, and H. Karas (GBF-Braunschweig). Several putative transcription factor binding sites were identified, including a potential Sp1 site (CGGGGGCGGCGGGG) between nucleotide positions Ϫ192 and Ϫ179, a potential AP-1 site (ATGAATCA) between positions Ϫ163 and Ϫ156, and an overlapping Sp1 (under-lined)/AP-2 and GC box (GGCTCCGCCCCGGCGGGGG) between positions Ϫ144 and Ϫ126 (Fig. 5).
To assess the importance of these sites, we mutated the presumptive Sp1 site (GG at position Ϫ185/Ϫ184 mutated to AA, printed in boldface above), mutated the presumptive AP-1 site (TGAAT at positions Ϫ162 to Ϫ158 mutated to GCTAG), or altered the presumptive Sp1 and AP-2 sites (CC at positions Ϫ140/Ϫ139 mutated to TT, and GG at positions Ϫ133/Ϫ132 mutated to AA; printed in boldface above). The resulting reporter gene plasmids p-198MSP1A-luc, p-198MAP1-luc, and p-198MSP1B/MAP2-luc lacked the consensus Ϫ184 Sp1A site, the consensus Ϫ156 AP-1 site, or the consensus Ϫ134/Ϫ126 Sp1B/AP-2 sites, respectively. As illustrated in Fig. 6, mutation of the putative Sp1A site in a 198-bp 5Ј-flanking promoter region resulted in a 40% reduction in the basal and inducible activity as compared with p-222-luc. The reporter construct p-198MSP1A was still 52-fold induced by RasV12 as compared with the control vector (Table II). The loss of some basal activity might be due to the mutation of the consensus Sp1A site or the absence of the Ϫ198 to Ϫ222 region. Mutation of the consensus AP-1 site caused a considerable decrease in inducible activity, whereas substantial basal activity was still retained, indicating that this putative AP-1 site is important for inducible activity. Furthermore, mutation of the overlapping consen-  1-4) or 12% (lanes 5-7) SDS-polyacrylamide gels and subjected to Western blotting with anti-myc (9E10) (lanes 1 and  2), anti-Raf-1 (C12) (lanes 3 and 4), or anti-MEK1 (C18) antibodies (lanes 5-7).
sus Sp1B/AP-2 site affected basal and inducible activity severely (Fig. 6). Deletion of the consensus Ϫ156 AP-1 and the consensus Ϫ126 AP-2 site caused only a minor decrease in inducible activity as compared with p-198MAP1-luc, a construct that had only the consensus Ϫ156 AP-1 site mutated, suggesting that the putative Ϫ126 AP-2 site may have a subtle effect on inducible promoter activity. Substantial basal activity, however, was still retained comparable to p-198MAP1-luc (Fig. 6). Deletion of the consensus Ϫ156 AP-1 and the consensus-134 Sp1 sites in the Ϫ222-bp 5Ј-flanking region of the hTFPI-2 promoter strengthened our previous observation. p-222MAP1/MSP1B-luc showed no basal activity above background and was also not stimulated by RasV12, suggesting that these sites are cis-acting elements critical for basal and inducible activity (Fig. 6). Similar results were obtained with another MAPK signaling pathway component MEK-1SD or with PMA (data not shown).
Taken together these results provide evidence that the AP-1 site at position Ϫ156 to Ϫ162 represents a cis-acting element that is critical for induction of the hTFPI-2 promoter activity by the MEK signaling pathway and the Sp1 site at position Ϫ134 to Ϫ140 is essential for basal promoter activity. DISCUSSION A schematic summary of the results presented in this study is shown in Fig. 7. We showed that the hTFPI-2 gene is upregulated in several epithelial cells following stimulation by PMA (Fig. 2). Additionally, EGF, a growth factor, was able to stimulate promoter activity of the hTFPI-2 gene as shown in HeLa cells (Fig. 2). In HEK293 cells, induction of the promoter activity by PMA could be blocked by the MEK-specific inhibitor UO126 (Fig. 1), suggesting that PMA induction of hTFPI-2 is mediated through a pathway that involves MEK. Indeed, activated Ras, Raf, MEK, and ERK were able to promote gene transcription (Fig. 3), indicating that the Ras/Raf/MEK/ERK signaling pathway is necessary for promoter activation. The ERK/MAPK pathway activates Fos proteins, which dimerize with Jun proteins and bind as the AP-1 complex to a consensus DNA sequence 5Ј-TGA(G/C)TCA-3Ј (29). We located an AP-1 consensus site at position Ϫ156 to Ϫ162 as a cis-acting element essential for inducible promoter activity. Additionally, a Sp1 consensus site at positions Ϫ134 to Ϫ140 of the hTFPI-2 promoter was essential for basal promoter activity (Fig. 6).
Our study in HEK293 cells indicates that the hTFPI-2 gene is regulated by the Ras/Raf/MEK/ERK signaling pathway. Each constitutively active signaling component of this pathway was able to induce high transcriptional activity of the hTFPI-2 gene. In agreement with these reporter gene studies, we observed up-regulation of the hTFPI-2 gene in preliminary microarray experiments comparing EcR-293 cells expressing  hTFPI-2 Gene Is Regulated by the ERK/MAPK Pathway RasV12, RafCT, MEK-1SD, or ERK2-MEK1-LA to nontransfected cells (data not shown). RasV12 is a strong activator of the Raf/MEK/ERK pathway, because it contains a glycine to valine mutation at amino acid position 12 and therefore remains in the active GTP-bound state (30). In contrast, RafCT, the carboxyl-terminal part of Raf that lacks the regulatory amino-terminal domain, is a moderate activator of the MEK/ ERK pathway (31). Consistent with this, we observed a 37-fold induction of hTFPI-2 promoter activity by Ras and a 7-fold induction by RafCT (Fig. 3). Similarly, the stronger activator Raf-CAAX, a full-length Raf containing a CAAX-box (32) induces higher (15-fold) induction of hTFPI-2 promoter activity than RafCT (data not shown).
The tumor-promoting agent PMA often serves as a model agent to study the mechanism by which growth factors regulate growth and differentiation of the cells. PMA mimics the action of diacylglycerol, the endogenous activator of protein kinase C (PKC) (33,34), and PKC activates Raf by direct phosphorylation (35). However, activation of Raf by PKC is Ras-dependent in several cell lines (36,37). Not all the phorbol ester responses can be explained by the PKC action, and some responses may be attributed to nonkinase phorbol ester receptors such as the ␣and ␤-chimaerins or to Ras-GRP, a phorbol ester receptor that plays a role in PMA activation of Ras (38). Furthermore, PMA has been shown to transactivate the epidermal growth factor receptor in mouse epidermal JB6 cells, and PMA-induced tumor promotion may be partially mediated through this receptor (26). PMA can activate distinct groups of MAPKs, such as the mitogen-responsive ERKs (extracellular signal-regulated kinases), the stress-responsive JNK/SAPKs (c-Jun amino-terminal kinase/stress-activated protein kinases), and p38 MAPKs. The pathways induced by PMA are cell line-dependent; for example PMA activates the JNK/MAPK pathway in normal oral keratinocytes but not in immortalized/transformed keratinocytes, HeLa cells, or HEK293 cells (39,40). In this report we provide evidence that hTFPI-2 gene expression induced by PMA is regulated through the Ras/Raf/MEK/ERK signaling pathway. The specific MEK inhibitor UO126 could block PMA induction, indicating that MEK/ERK activation is essential for induction of hTFPI-2 promoter activity by PMA. Because UO126 inhibited PMA promoter activation at the commonly used concentration of 10 M, we hypothesize that the ERK/MAPK pathway may be sufficient for PMA induction of hTFPI-2. Besides affecting the PMA-dependent hTFPI-2 promoter activation, UO126 had a minor negative effect on basal levels of the hTFPI-2 gene transcription, probably due to the low level of activation of hTFPI-2 by endogenous MAPK signaling components.
Activation of the ERK/MAPK causes induction of fos genes through phosphorylation of ternary complex factors (41). Indeed, we have observed up-regulation of FosB when PMAinduced HEK293 cells were probed on microarrays (data not shown). Fos heterodimerizes with Jun to form the AP-1 complex, which activates gene transcription by binding to the AP-1 element. Furthermore, Jun/Fos heterodimers can lead to increased c-jun transcription through binding to the AP-1 sites in the c-jun promoter (41). In agreement with this, we located a putative AP-1 site at position Ϫ156 to Ϫ162 as a cis-acting element essential for inducible promoter activity.
Our results suggest that basal promoter activity is located in the region, Ϫ384 to Ϫ89 bp, of the hTFPI-2 promoter consistent with a previous promoter deletion study in human transformed bone marrow microvascular endothelial cells (22), identifying an 85-bp fragment (corresponding to the Ϫ299 to Ϫ214 region in our study) that contained most of the basal activity. In addition to this, we have identified a consensus Sp1 site at Ϫ134 to Ϫ140 that is essential for the basal activity in the promoter region, Ϫ222 to Ϫ89 bp.
The region, Ϫ222 to Ϫ89 bp, is sufficient for a 52-fold Ras and an 8-fold PMA induction of the hTFPI-2 promoter activity FIG. 6. Mutagenesis of potential transcription factor binding sites. EcR-293 cells were transiently cotransfected with RasV12 or control vector pIND and luciferase reporter gene constructs containing modified potential transcription factor binding sites. Cotransfection with a vector containing GFP controlled for transfection efficiency. The cells were serum-starved 4 h prior to induction of protein expression with pronasteron A for 20 h. Luciferase activity was calculated as light units/g of protein and normalized for transfection efficiency. Data are shown as mean Ϯ S.D. from at least four independent transfection experiments. as compared with basal level (Fig. 4). We therefore investigated the candidate transcription factor binding sites in this promoter region and found that the consensus AP-1 site at position Ϫ156 to Ϫ162 is essential for induction of promoter activity by Ras (Fig. 6). Recently, Konduri (19) identified a 231-bp region between Ϫ312 and Ϫ81 in the hTFPI-2 promoter region that is responsive to PMA. Investigating hTFPI-2 promoter regulation in glioma cells, Konduri found a strong repressor in the region between Ϫ927 to Ϫ1181 and enhancer elements between Ϫ1511 and Ϫ1181. These repressor and enhancer elements found in glioma cells might be tissue-specific regulatory elements, because we did not find such elements in HEK293 cells. Similar luciferase activities were observed in constructs of various lengths between 1511 and 384 induced by Ras (Table II) or PMA (data not shown). Konduri suggested that three potential AP-1 sites (corresponding to Ϫ309 to Ϫ299, Ϫ213 to Ϫ203, and Ϫ162 to Ϫ156 in our study) may be involved in PMA-related gene induction (19). Although we identified the Ϫ156 to Ϫ162 consensus AP-1 site as essential for transcription of the hTFPI-2 gene, our results suggest that the Ϫ213 to Ϫ203 promoter segment may not be essential for Ras-related gene induction in EcR-293 cells, because the promoter segment of construct p-198MSP1A-luc, which lacks the Ϫ213/Ϫ203 segment, is sufficient for a 52-fold Ras induction similar to longer constructs (Table II).
The ERK/MAPK pathway stimulated by growth factors and phorbol esters is associated with cellular proliferation and differentiation. The biological consequences of hTFPI-2 induction by agents stimulating the ERK/MAPK pathway and the role of hTFPI-2 in tumor progression is not fully understood. On one hand, several cell lines stably overexpressing hTFPI-2, such as glioma SNB19 cells, choriocarcinoma JAR cells, amelanotic melanoma cells, and prostate cancer cells, were less invasive than their parental cells (14 -17) and addition of recombinant hTFPI-2 inhibited the invasiveness of SNB19 glioma cells as measured in a Matrigel assay (42). On the other hand, hTFPI-2 has been shown to have a pro-invasive effect in cancer. Recombinant hTFPI-2 has been shown to be a mitogen for vascular smooth muscle (43), and hTFPI-2 potentiates hepatocyte growth factor-induced invasion of human hepatocellular carcinoma cells and is capable of inducing invasion of human hepatocellular carcinoma on its own (18). Bimodal function has also been described for tissue inhibitors of metalloproteinases, another family of tissue inhibitors. Tissue inhibitors of metalloproteinases inhibit the activity of matrix metalloproteinases by binding to their zinc binding site, but they are also involved in the cell surface-targeted MMP activation cascade, stimulate growth, and are anti-angiogenic (44). Similarly, the TFPIs may also perform a very wide variety of biological activities, and transcriptional regulation of TFPIs may play an important role in the balance of protease activities resulting in the metastatic/ invasive phenotype.