Signal Transduction via the Mitogen-activated Protein Kinase Pathway Induced by Binding of Coagulation Factor VIIa to Tissue Factor*

The putative role of tissue factor (TF) as a receptor involved in signal transduction is indicated by its sequence homology to cytokine receptors (Bazan, J. F. (1990) Proc. Natl. Acad. Sci. U. S. A. 87, 6934–6938). Signal transduction induced by binding of FVIIa to cells expressing TF was studied with baby hamster kidney (BHK) cells stably transfected with TF and with a reporter gene construct encoding a luciferase gene under transcriptional control of tandem cassettes of signal transducer and activator of transcription (STAT) elements and one serum response element (SRE). FVIIa induced a significant luciferase response in cells expressing TF, BHK(+TF), but not in cells without TF. The BHK(+TF) cells responded to the addition of FVIIa in a dose-dependent manner, whereas no response was observed with active site-inhibited FVIIa, which also worked as an antagonist to FVIIa-induced signaling. Activation of the p44/42 MAPK pathway upon binding of FVIIa to TF was demonstrated by suppression of signaling with the specific kinase inhibitor PD98059 and demonstration of a transient p44/42 MAPK phosphorylation. No stimulation of p44/42 MAPK phosphorylation was observed with catalytically inactive FVIIa derivatives suggesting that the catalytic activity of FVIIa was obligatory for activation of the MAPK pathway. Signal transduction caused by a putative generation of FXa activity was excluded by experiments showing that FVIIa/TF-induced signaling was not quenched by tick anticoagulant protein, just as addition of FXa could not induce phosphorylation of p44/42 MAPK in BHK(+TF) cells. These results suggest a specific mechanism by which binding of FVIIa to cell surface TF independent of coagulation can modulate cellular functions and possibly play a role in angiogenesis and tumor metastasis as indicated by several recent observations.

The putative role of tissue factor (TF) as a receptor involved in signal transduction is indicated by its sequence homology to cytokine receptors ( The extrinsic pathway of blood coagulation is initiated when FVIIa 1 circulating in plasma binds to the integral membrane protein, tissue factor (TF), exposed to the blood upon injury of the vessel wall. The biology of TF in blood coagulation has been studied extensively (see Refs. 1 and 2 for reviews). Initiation of coagulation as a result of FVII/TF activity is an extracellular event confined to the outer leaflet of the plasma membrane on TF-expressing cells. This was firmly established when it was shown by Paborsky et al. (3) that truncation of the cytoplasmic C-terminal of TF did not affect its cofactor function in the coagulation process. Still a number of other observations suggested an intracellular function of TF. It was found that TF showed sequence homology to the cytokine receptor superfamily (4). All members of this family are involved in signal transduction. Subclass II, which includes the receptor for interleukin 10 and receptors for interferon ␣/␤ and ␥, shows the highest homology to TF (5). The crystal structure of TF (6, 7) further substantiated its structural resemblance to the cytokine receptors. This homology might imply a possible functional role for TF as a receptor involved in signal transduction. Studies on a putative intracellular activity induced by FVII/FVIIa have been elusive. Two recent studies (8,9) reported that FVIIa could induce oscillations in intracellular free calcium in various TF-expressing cells. The authors proposed that this involved activation of phospholipase C; however, they failed to demonstrate directly phosphorylation caused by activation of intracellular kinases. Other studies have provided evidence that serine residues of the cytoplasmic domain of TF can be phosphorylated in TF-transfected cells (10) and that this domain works as a substrate for protein kinase activity when exposed to cell lysates (11). Signal transduction was also indicated in studies with cultured human monocytes (12) showing that addition of FVIIa could induce a transient tyrosine phosphorylation of several polypeptides. Finally very recent results suggested that FVIIa induced alteration in gene expression in human fibroblasts (13). With these diverse observations it was of interest to characterize a putative FVIIa-induced signal transduction pathway in further detail.
Ligand-induced oligomerization of receptor subunits, which juxtaposes to engage an intracellular signaling machinery, is a common element in signal initiation from cytokine receptors (14). This may lead to rapid phosphorylation of a subset of intracellular receptor-associated proteins and mitogen-activated (Ser/Thr) kinases (MAPK) (15,16). These kinases are arranged in several parallel signaling pathways, the induction of which may eventually activate the serum response element (SRE) and lead to transcription (17). Cytokine receptors, which like TF lack intrinsic kinase catalytic domains, may couple ligand binding to tyrosine phosphorylation by using noncovalently associated protein kinases, the Janus kinases, which phosphorylate signal transducers and activators of transcription (STATs) to induce binding to specific DNA elements and gene transcription. The JAK-STAT pathway may be engaged in * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

EXPERIMENTAL PROCEDURES
Cell Culture-The baby hamster kidney cell line BHK-21 tk Ϫ ts13 (ATCC CRL 1632) was cultured in Dulbecco's modified Eagle's medium containing 10% FCS, 100 IU/ml penicillin, 100 g/ml streptomycin. A human endothelial cell line ECV-304 (ATCC CRL-1998) was cultured in Medium 199 with Earle's salts containing 10% FCS, 100 IU/ml penicillin, 100 g/ml streptomycin. An epithelial-like canine kidney cell line MDCK-II was kindly provided by Professor Bo van Deurs, University of Copenhagen, Denmark and cultured in Dulbecco's modified medium supplemented with 5% FCS, 100 IU/ml penicillin, and 100 g/ml streptomycin. All cell lines were grown in T-75 or T-175 flasks and subcultured into 24-well tissue culture plates or 23.8-cm 2 single wells.
Proteins-Human recombinant FVII and FVIIa was expressed and purified as described (19). FVIIai was obtained by blocking of FVIIa in the active site with D-Phe-L-Phe-L-Arg chloromethyl ketone as described previously (20). FVIIa and FVIIai were iodinated by the lactoperoxidase/H 2 O 2 method (21) and purified by size exclusion chromatography on a Pharmacia Biotech Inc. NAP 5 Sephadex G-25 with 0.1 M NH 4 HCO 3 , 0.5% (w/v) human serum albumin as the mobile phase. FX and FXa were from Enzyme Research (Lafayette, IN). Recombinant tick anticoagulant protein (rTAP) was kindly provided by Dr. G. P. Vlasuk, Corvas (San Diego, CA). The inhibitor PD98059 and the phosphospecific antibody against p44/42 MAPK (Thr 202 /Tyr 204 ) and Western blot detection kit were from New England Biolabs (Beverly, MA).
Transfection of BHK Cells with TF and Luciferase Reporter Constructs-The complete human TF cDNA was cloned into the mammalian Zem219b expression vector (22). BHK cells were transfected with the TF expression plasmid using the calcium phosphate coprecipitation procedure, essentially as described (23). Cells with stably integrated constructs were selected with 1 M methotrexate. The KZ136 reporter construct contains an inducible firefly luciferase expression unit and the neomycin resistance gene for stable selection. Luciferase coding sequences are preceded by TATA and transcription start sequences of the human c-fos gene (nucleotides 649 -747, GenBank TM accession number K00650). The luciferase promoter consists of 4 STAT binding elements and a SRE, which were synthesized as oligonucleotides. The STAT elements used were those found in the c-fos (M67 version), p21 WAF1 , ␤-casein, and the Fcg RI genes. BHK cells with and without TF at 80% confluence were transfected with KZ136 (40 g of DNA in a 21-cm 2 culture dish), and 24 h later G418 was added to a final concentration of 1 g/ml. The selective pressure was changed to 0.5 g/ml after 2 weeks.
FVIIa Binding to Cell Surface TF and FX Activation Assay-This was performed as described previously (20).
Measurements of Gene Transcription by Stimulation of Luciferase Activity-Cells were grown in white view plates (Packard, Pangbourne, Berks, UK) for approximately 2 days to obtain 95-100% confluence. Cells were washed and grown for an additional 16 -24 h in medium with FCS reduced to 0.2%. The cells were again washed and exposed to FCS-free medium (100 l/well) containing the test compounds. Following incubation at 37°C for 6 h, 100 l of lysis and luciferase assay buffer (Packard) was added. After an additional 30 -180 min at room temperature, the luciferase activity was measured in a 1450 MicroBeta Trilux luminescence counter (Wallac, Finland) using 1 s of integration/well.
Cell Lysis and Western Blot Analysis-The total amount of p44/42 MAPK or phosphorylated p44/42 MAPK was detected using a Phos-phoPlus TM MAPK antibody kit (Biolabs) according to the manufacturer's protocol. BHK(ϩTF) cells were cultured in medium with 0.1% FCS for 24 h prior to the experiment. The experiment was performed by adding media with 0.1% FCS containing the various ligands for the indicated periods of time to the cells. As a positive control of cell signaling cells were treated with 15% FCS for 15 min in each experiment. Cells were lysed in 150 l of SDS sample buffer (9.2% (w/v) SDS, 25 mM Tris-HCl, 40% (v/v) glycerol, 80 mM EDTA, 1.2% (w/v) bromphenol blue, pH 6.8) supplemented with 3 mM sodium orthovanadate and 24.2 mM dithiothreitol. Lysates were loaded on a 12% SDS-polyacrylamide gel. A biotinylated protein marker was loaded on each gel, and fully phosphorylated p44/42 MAPK was loaded as a positive control in some experiments.

RESULTS AND DISCUSSION
Characterization of Transfected BHK Cells-Baby hamster kidney (BHK) cells without detectable amounts of TF were stably transfected with the TF expression vector. These BHK(ϩTF) cells contained approximately 3 ϫ 10 6 TF/cell as revealed by a binding assay with 125 I-labeled FVIIa and expressed functionally active TF working as a cofactor for FVIIamediated activation of FX with an EC 50 ϭ 1.0 nM. No significant activation was observed with untransfected BHK cells (results not shown). Subsequently the cells were stably transfected with the KZ136 reporter plasmid encoding a luciferase gene under transcriptional control of tandem cassettes of STAT1 and -3 and STAT4, -5, and -6 and one SRE. Serum contains a multitude of growth factors and activates several of these elements. Addition of 15% serum to starved cells transfected with the KZ136 reporter construct resulted in a 5-8-fold increase in luciferase activity independent of coexpression of TF (results not shown). Fig.  1A shows that TF-transfected BHK cells, BHK(ϩTF), with the reporter construct responded to the addition of FVIIa, whereas a significant response was not observed when FVIIa was added to BHK cells without TF. This suggested that FVIIa was involved in TF-dependent signal transduction resulting in gene transcription. The FVIIa-induced response was saturable with an apparent EC 50 of approximately 20 nM as indicated by the results shown in Fig. 1B.

Effect of FVIIa, FVII, and FVIIai on BHK(ϩTF) Cells Transfected with the KZ136 Signaling Reporter Gene Construct-
The mobilization of Ca 2ϩ stores in TF-expressing cells observed by Rottingen et al. (8), as well as transient phosphorylation of intracellular monocyte polypeptides observed by Masuda et al. (12), appeared to require the participation of catalytically active FVIIa. Hence, we have investigated whether proteolytically active FVIIa was also mandatory for TF signal transduction monitored by the luciferase reporter system. As shown in Fig. 2 the addition of zymogen, one-chain FVII, induced a luciferase response comparable with that of the

FIG. 1. FVIIa-induced stimulation of luciferase response in BHK cells transfected with TF and the KZ136 reporter construct.
Cells were lysed after a 6-h incubation with FVIIa at 37°C, and the luciferase activity was measured as described under "Experimental Procedures." A shows the effect of TF transfection by comparing the response obtained by addition of 400 nM FVIIa with KZ136transfected BHK cells with TF, BHK(ϩTF), and without TF, BHK(ϪTF). B shows a dose-response curve with exposure of BHK(ϩTF) cells to increasing concentrations of FVIIa as indicated. activated protease, FVIIa. However, since FVII bound to cell surface TF could be activated to FVIIa (24,25), it was impossible to exclude a response caused by generation of FVIIa. To circumvent this possibility we used an inactive FVIIa derivative, FVIIai, blocked in the active site by FFRck. In contrast to native FVII this variant did not induce a significant luciferase response.
Since FVIIa catalytic activity may thus be required for TFdependent signaling it was important to rule out that signaling occurred indirectly due to generation of FXa in trace amounts catalyzed by FVIIa bound to cell surface TF. If FXa was generated it might induce signaling through a FXa receptor (26 -28) or further downstream activation of the coagulation cascade ultimately resulting in activation of the thrombin receptor (29,30). In this context it is important to point out that the cells were exposed to a brief EDTA wash to remove possible traces of vitamin K-dependent coagulation factors before the addition of FVIIa. Further, Fig. 2 shows that FVIIa/TF-induced signal transduction was not prevented by TAP (tick anticoagulant protein) which specifically blocks the active site in FXa. Moreover, although the addition of FXa was capable of stimulating the luciferase activity, this response was not TF-dependent since a similar response was also seen in BHK cells without TF transfection (results not shown). Additionally, in contrast to the response induced by FVIIa, the response induced by FXa was fully suppressed by TAP (Fig. 2). These data strongly suggested that FVIIa/TF-induced signal transduction did not proceed via FXa or thrombin.
Since FVIIai did not induce a luciferase response it was of interest to see whether a FVIIa-induced response could be inhibited by binding of FVIIai to TF. Fig. 3 shows that addition of increasing concentrations of FVIIai quenched the signal induced by 20 nM FVIIa in a manner expected when both proteins compete for the same site(s) on TF.
Test of the Signaling Pathway by Means of the MAPK Inhibitor PD98059 -Activation of the luciferase reporter construct might occur via one of the known MAPK pathways. Fig. 4 demonstrates that the FVIIa-induced luciferase response in BHK(ϩTF) cells was completely inhibited by the specific inhibitor PD98059. This inhibitor binds to the inactive form of MAPKK (MEK) preventing its transformation into an active kinase (31). Phosphorylation of p44/42 MAPK and further downstream activation is thereby inhibited. The results shown in Fig. 4 thus suggested that stimulation of gene transcription by FVIIa/TF might occur via the MAPK pathway.
FVIIa-induced Phosphorylation of p44/42 MAPK in BHK(ϩTF) Cells-Using a specific antibody against the phosphorylated Thr 202 /Tyr 204 residues of p44/42 MAPK it was possible to confirm this conclusion. BHK(ϩTF) cells were grown to 90% confluence and then starved in 0.1% FCS for 24 h. Fig. 5  (panels A and B) shows a Western blot of cell lysates from BHK(ϩTF) cells exposed to 100 nM FVIIa for 0, 3, 5, 7, 10, and 40 min (lanes 2-7). Specific antibodies were used to visualize the amount of activated MAPK (panel A) and of total MAPK (panel B). Fig. 5A demonstrates that the MAP kinase was transiently phosphorylated peaking at approximately 10 min, whereas the amount of total MAPK remained essentially constant (panel B).
When a similar experiment was performed with nontransfected BHK cells the MAPK was not activated by FVIIa, but a phosphorylated MAPK response was obtained with 15% serum (results not shown).
The results shown in Fig. 5 (panels C and D) were obtained with BHK(ϩTF) cells exposed to 100 nM FVII, FVIIa, FVIIai, [Ala 344 ]FVII, or FXa for 15 min. A profound activation was seen with FVIIa, less so with FVII, and no significant activation was induced by FVIIai or an inactive FVII variant, [Ala 344 ]FVII, in which the active site serine was changed by site-directed mutagenesis (panel C, lanes 2-5). It was interesting to note that the addition of FXa (panel C, lane 6) also failed to induce Antibodies against phosphorylated SAPK or p38 MAPK were tested in a similar way in separate experiments; however, none of these pathways appeared to be stimulated by binding of FVIIa to TF (results not shown).
FVIIa-induced Phosphorylation of p44/42 MAPK in ECV 304 and MDCK Cells-FVIIa/TF-induced stimulation of gene transcription was also studied in the human immortalized endothelial cell line, ECV 304. After starvation for 24 h, ECV 304 cells expressed about 25,000 TF molecules per cell as estimated from the 125 I-FVIIa binding capacity. Cells exposed to 20 nM FVIIa in fresh equilibrated media were lysed and harvested after 0, 5, and 40 min. Western blots of these lysates are show in Fig. 6, A and B. A clear enhancement of the phosphorylated p44/42 band was observed after exposure to FVIIa for 5 min followed by a decreased response at 40 min.
Camerer et al. (1) showed that FVIIa induced relatively strong intracellular Ca 2ϩ oscillations in Madin-Darby canine kidney (MDCK) cells constitutively expressing TF capable of binding human FVIIa. They failed, however, to observe any phosphorylation and could not inhibit the oscillations with known tyrosine kinase inhibitors. It was therefore of interest to see whether it was possible to observe FVIIa/TF-induced phosphorylation of p44/42 MAPK in these cells. The results shown in Fig. 6, C and D, suggested that this was in fact the case. When approximately 80% confluent cells were exposed to 20 nM FVIIa for various time periods phosphorylation of p44/42 MAPK was clearly induced but appeared to be delayed compared with BHK(ϩTF) and ECV 304 cells. The results obtained with these cells confirmed that signaling capability was not confined to the somewhat artificial test system constituted by the transfected BHK cells. CONCLUSION Several recent studies have provided circumstantial evidence that FVIIa/TF may be involved in signal transduction and gene transcription. The present work demonstrates that this is the case and that it occurs in a FVIIa-and TF-dependent reaction via the p44/42 MAPK pathway. The results also suggest that the catalytic activity of FVIIa is mandatory for this process and that an indirect signaling pathway via FX activation can be excluded. The target substrate for the FVIIa activity has, however, not been identified, just as it remains to be shown whether TF as such works as a transmembrane signal transducer either alone or in combination with a putative ␤-subunit. The experiments with the effect of PD98059 on the luciferase response (Fig. 4) showed that this specific inhibitor of the MAPK pathway inhibited the stimulation induced by FVIIa as well as the background activity. This indicates that MAPK phosphorylation is a probable route to FVIIa-induced gene transcription although the existence of parallel pathways or cross-talk between pathways cannot be excluded. Recent studies have suggested a role for TF in angiogenesis (32,33), embryo vascularization (34), tumor metastasis (35)(36)(37), and smooth muscle cell migration (38). FVIIa/TF-induced signal transduction might provide a common molecular mechanism linking these cellular events together. The present work has uncovered important details about the signaling process and provided clues to further elucidate the mechanism.  1-4). In both experiments lane 5 shows biotinylated marker proteins and lane 6 the positive control when cells were exposed to 15% serum for 15 min.