Michael E. Bromberg, Temple University School of Medicine, Sol Sherry Thrombosis Research Center, 3400 North Broad Street, Philadelphia, PA 19140, USA. Tel.: +1 215 707 7235; fax: +1 215 707 2783; e-mail: firstname.lastname@example.org
Summary. Tissue factor (TF) is a transmembrane glycoprotein that initiates blood coagulation when complexed with factor (F)VIIa. Recently, TF has been shown to promote cellular signaling, tumor growth, angiogenesis, and metastasis. In the present study, we examined the pathway by which TF–FVIIa complex induces cellular signaling in human breast cancer cells using the Adr-MCF-7 cell line. This cell line has high endogenous TF expression as measured by flow cytometry and expression of protease-activated receptors 1 and 2 (PAR1 and PAR2) as determined by reverse transcriptase-polymerase chain reaction analysis. Both PAR1 and PAR2 are functionally active as determined by induction of p44/42 mitogen-activated protein kinase (MAPK) phosphorylation using specific agonist peptides. We found that MAPK phosphorylation in this cell line was strongly induced by the combination of FVIIa and factor (F)X, but not by FVIIa alone at a concentration of FVIIa that approaches physiological levels. Induction of MAPK phosphorylation involved the formation of TF–FVIIa–FXa complex and occurred by a pathway that did not require thrombin formation, indicating a critical role for FXa generation. In addition, induction of MAPK phosphorylation was found to be independent of PAR1 activation. We then examined whether TF–FVIIa complex formation could promote tumor cell migration using a modified Boyden chamber chemotaxis assay. The combination of FVIIa and FX, but not FVIIa alone, strongly induced migration of tumor cells by a pathway that probably involves PAR2, but not PAR1 activation. MAPK phosphorylation was found to be required for the induction of cell migration by the combination of FVIIa and FX. These data suggest that TF–FVIIa-mediated signaling in human breast cancer cells occurs most efficiently by formation of the TF–FVIIa–FXa complex. One of the physiological consequences of this signaling pathway is enhanced cell migration that is probably mediated by PAR2, but not PAR1 activation.
Tissue factor (TF) is a 47-kDa transmembrane glycoprotein that complexes with factor (F)VIIa to initiate blood coagulation . Although best characterized for its role in blood coagulation, TF has recently been shown to participate in a variety of physiological processes distinct from hemostasis, including embryogenesis, inflammation, cellular signaling, cell migration, tumor growth, metastasis, and angiogenesis [2,3]. We previously showed that overexpression of human TF by a human melanoma cell line, SIT1, resulted in a major increase in the metastatic potential of the cells in a murine model of metastasis by a mechanism that requires both the cytoplasmic and extracellular domains of the molecule [4,5]. This result has been confirmed by other investigators who have also shown that a proteolytically active TF–FVIIa complex is necessary for this process .
Formation of TF–FVIIa complex has been shown to induce cellular signaling events that include: calcium fluxes , p44/42 mitogen-activated protein kinase (MAPK) phosphorylation , p38 MAPK phosphorylation , protein kinase B phosphorylation  and upregulation of multiple genes such as early growth response gene-1 , Cyr61 and connective tissue growth factor gene . Proteolytically active FVIIa is required for signal transduction, and a unique family of G protein-coupled receptors known as protease-activated receptors (PARS) has been implicated in this pathway . Four PARs (PAR1–4) have been described to date. PAR1, PAR3, and PAR4 are activated by thrombin, whereas PAR2 is activated by trypsin and tryptase . Recently, formation of TF–FVIIa complex has been found to activate PAR2 directly . However, the significance of this pathway involving direct activation of PARs by TF–FVIIa complex is unclear, since high non-physiological levels of FVIIa were used . TF–FVIIa complex signaling can also occur through the generation of downstream blood coagulation intermediates, such as factor (F)Xa . Whether this signaling pathway involves PAR2 or other PARs also remains to be determined [12,14].
TF and PARs are expressed by a variety of tumor cells [15–17]. The physiological consequences of the TF–FVIIa signaling pathway involving PAR activation in tumor cell function are not fully known. High expression of PAR1 has been associated with increased tumor invasiveness and metastasis in some tumors [18,19,20], while the role of PAR2 in metastasis is not known. In the present study, the pathway by which TF–FVIIa complex transduces cellular signaling and the role of TF–FVIIa complex in the induction of cell migration, a key step in the metastatic cascade, were investigated in human breast cancer cells. We show that signal transduction occurs most efficiently by formation of the TF–FVIIa–FXa complex rather than by TF–FVIIa complex alone, and that p44/42 MAPK phosphorylation occurs through a pathway that is independent of PAR1. Moreover, formation of TF–FVIIa–FXa complex strongly induces migration of tumor cells by a pathway that probably involves PAR2, but not PAR1 activation.
Materials and methods
Recombinant human FVIIa was a gift from Dr L. Petersen (Novo Nordisk, Måløv, Denmark). Human FX and FXa were from Hematologic Technologies (Essex Junction, VT, USA). WEDE15 and ATAP2 anti-PAR1 antibodies were obtained from Coulter Corp. (Hialeah, FL, USA) and Santa Cruz Biotech (Santa Cruz, CA, USA), respectively. Anti-TF antibody (TF85G9) was a gift from Dr J. Morrissey (University of Illinois, Urbana-Champaign, IL, USA). Recombinant hirudin was from Calbiochem (San Diego, CA, USA). Recombinant Ancylostoma caninum Anticoagulant Peptide (AcAP) was prepared as previously described . Tick Anticoagulant Peptide (TAP) was kindly provided by Dr K. McLean (Berlex Biosciences, Richmond, CA, USA). PAR1 agonist peptide (TFLLRN), PAR2 agonist peptide (SLIGKV), and control, scrambled peptide (LSIGKV) were synthesized at the W.M. Keck Biotechnology Resource Center at Yale University School of Medicine. All oligonucleotides were also synthesized at Yale University.
The Adr-MCF-7 and MCF-7 parent cell lines were obtained from Dr M. Reiss (Cancer Institute of New Jersey, New Brunswick, NJ, USA) and grown in RPMI 1640 media supplemented with 10% fetal calf serum (FCS), 50 U mL−1 penicillin, and 50 µg mL−1 streptomycin. The MDA-MB-231 human breast carcinoma cell line was obtained from ATCC (Manassas, VA, USA) and maintained in Dulbecco's modified eagle media (DMEM) supplemented with 10% FCS, 50 U mL−1 penicillin, and 50 µg mL−1 streptomycin. DAMI, a human megakaryocytic cell line, was obtained form Dr D. Gilligan (Yale University School of Medicine) and maintained in RPMI 1640 media supplemented with 10% FCS, 50 U mL−1 penicillin, and 50 µg mL−1 streptomycin. Human umbilical vein endothelial cells (HUVEC) were obtained from Dr J. Pober (Yale University School of Medicine) grown in 199 media supplemented with 20% FCS. All cell lines were maintained in a humidified atmosphere containing 5% CO2.
Cells grown in culture were briefly trypsinized, washed in Ca2+- and Mg2+-free Hank's balanced salt solution, and labeled with a fluorescein-conjugated murine monoclonal antihuman TF antibody (no. 4508CJ; American Diagnostica, Stamford, CT, USA). Cells were also labeled with an isotype-matched fluorescein-conjugated mouse IgG, as a control. After washing to remove unbound antibody, the cells were analyzed using a Becton Dickinson FACS Star Analyzer (BD Biosciences, San Jose, CA, USA).
Northern blot analysis
Total RNA was isolated from cell lines using the RNeasy kit. Total RNA (10 µg) for each sample was fractionated by gel electrophoresis on 1% agarose/6% formaldehyde gel and transferred onto a nylon membrane (BrightStar; Ambion, Austin, TX, USA) as previously described . Hybridization was carried out with a 32P-labeled cDNA probe that corresponded to bp 647–929 of human TF cDNA (where A of the start ATG represents the first nucleotide). Membranes were also hybridized with a GAPDH cDNA probe (Ambion) that was 32P-labeled to verify equal RNA loading.
First strand cDNA synthesis was carried out using Advantage RT-for-PCR kit (Clontech, Palo Alto, CA, USA). DNase-treated total RNA (1 µg) was used to prepare cDNA with oligo (dT)18 primer. An aliquot of the reaction mixture was then used for polymerase chain reaction (PCR) amplification using PCR Master Mix (Promega, Madison, WI, USA), using the following primer sequences:
Thermocycling parameters were as follows: 95 °C for 2 min, then 29 cycles of 95 °C for 30 s, 56 °C for 1 min, 72 °C for 2 min. A final elongation step was carried out at 72 °C for 5 min. Equal amounts of cDNA synthesis were verified using GAPDH primers (Clontech) using the thermocycling conditions above except that 24 cycles of amplification were used.
PCR products were resolved on a 1.5% agarose gel for visualization.
P44/42 MAPK phosphorylation
Adr-MCF-7 cells were plated in 6-cm dishes and grown to 70% confluence in RPMI 1640 media supplemented with 10% FCS. The cells were then made quiescent by starvation in serum-free RPMI 1640 for 24 h prior to agonist stimulation. Cells were stimulated with agonist in serum-free RPMI 1640 media and analyzed for MAPK phosphorylation by Western blotting using a kit from Cell Signaling Technologies (Beverly, MA, USA). Phosphorylation levels were determined by scanning densitometry using NIH Image software with levels normalized to the level of total MAPK for each sample.
Cell migration assay
Boyden chamber chemotaxis assays were carried out using a Neuro Probe (Gaithersburg, MD, USA) 48-well micro chemotaxis chamber with 8 mm2 filter well area. Polycarbonate filters with 12 µm pore size were coated with mouse collagen type IV (BD Biosciences, Bedford, MA, USA) as described by Thompson et al. . Cells were harvested using non-enzymatic Cell Dissociation Media (Sigma, St Louis, MO, USA) and washed twice in serum-free RPMI containing 0.1% bovine serum albumin (BSA) and added to the upper chamber at 105 cells per well. The lower chambers contained either RPMI 1640 media containing 0.1% BSA (RPMI–BSA) or RPMI–BSA containing chemoattractants (coagulation proteins or PAR-specific peptides). Recombinant hirudin (100 nm) was added to both the upper and lower chambers to eliminate any effects of thrombin. The chambers were incubated in a humidified incubator at 37 °C containing 5% CO2 for 6 h. Non-migratory cells on the upper surface of the membrane were removed with a plastic scraper. The cells that migrated across the filter were fixed and stained using Diff-Quik (Dade Behring, Deerfield, IL, USA). The number of migratory cells per membrane was measured by light microscopy at 660 × total magnification. Each data point is the average of cells in three random fields. Each determination represents the average of at least four individual wells.
Statistical analyses were performed using an unpaired Student's t-test. Statistical significance was assumed to occur with P ≤ 0.05.
To investigate TF–FVIIa complex-induced cellular signaling in tumor cells, we chose to study the Adr-MCF-7 cell line, a multidrug-resistant subline of the human breast cancer cell line MCF-7, which has high endogenous levels of TF . Northern blot analysis revealed high TF mRNA levels in the Adr-MCF-7 cells compared with the parental MCF-7 cell line (Fig. 1). Surface expression of TF in the Adr-MCF-7 line was nearly 10-fold higher compared with MCF-7 line as measured by flow cytometry (Fig. 1). This increase corresponds to a nearly 100-fold increase in TF functional activity as measured by a two-stage clotting assay . Reverse transcriptase (RT)-PCR analysis for PAR expression in this cell line showed that only PAR1 and PAR2 were expressed (Fig. 2).
Activation of PAR1 or PAR2 has been shown to induce p44/42 MAPK phosphorylation . To test whether the Adr-MCF-7 has functional expression of PAR1 and PAR2, the cell line was grown to subconfluence, serum starved for 24 h, and stimulated with either the PAR1-specific agonist peptide, TFLLRN, or the PAR2-specific agonist peptide, SLIGKV. Treatment of the cells with the PAR1-specific agonist peptide induced maximal phosphorylation at 15 min, with a nearly 4-fold increase in MAPK phosphorylation compared with the time 0, unstimulated control that was sustained for the duration of the time course (Fig. 3). As shown in Fig. 4, phosphorylation was also induced using the PAR2 agonist peptide or serum (Fig. 4A), but not with the scrambled peptide, LSIGKV, as a control (Fig. 4B). Phosphorylation was maximal at 10 min with the PAR2 agonist peptide stimulation and was nearly 5-fold higher compared with the time 0, unstimulated control. Maximal phosphorylation levels were sustained for the duration of the 30-min time course. By contrast, treatment of the cells with the PAR4-specific peptide GYPGKF (500 µm) did not induce MAPK phosphorylation (data not shown). These results indicate that activation of PAR1 or PAR2 leads to p44/42 MAPK phosphorylation in the Adr-MCF-7 cells.
Other investigators have shown that high concentrations (50–100 nm) of FVIIa induce p44/p42 MAPK phosphorylation in some cell lines [9,25]. We first tested whether such a pathway occurs in this breast cancer cell line. Treatment of the Adr-MCF-7 cells with recombinant FVIIa induced MAPK phosphorylation in a dose-dependent manner with a concentration that induced a half maximal response (EC50) of ∼ 38 nm (Fig. 5). Essentially no change in the basal level of MAPK phosphorylation with cells treated with 10 nm FVIIa was found (Figs 5 and 6). In contrast, nearly a 3-fold increase in phosphorylation was observed by treatment of the cells with the combination of 10 nm FVIIa and 150 nm FX, which could be markedly inhibited by a specific active site inhibitor of FXa, AcAP , but not by the presence of hirudin (100 nm), a specific thrombin inhibitor. Similar results were also found using another specific FXa inhibitor, TAP  (data not shown). In addition, FXa was found to induce phosphorylation of MAPK strongly (nearly 4-fold) which could be blocked in the presence of AcAP (Fig. 6). These results indicate that at levels of FVIIa that approach physiological concentrations, generation of FXa is required for cellular signaling. Moreover, thrombin formation is not required in this pathway.
To test whether activation of PAR1 is required for TF–FVIIa signaling, the cells were treated with the combination of FVIIa and FX in the absence and presence of PAR1 blocking antibodies, WEDE15 and ATAP2 . As shown in Fig. 7, phosphorylation was increased 2-fold in the presence of the combination of FVIIa and FX that was only slightly (15%) inhibited in the presence of the blocking antibodies. By contrast, nearly 70% of the induction of MAPK phosphorylation by 2.5 nm thrombin was inhibited by antibody treatment. These results suggest that PAR1 activation is not required for TF–FVIIa-mediated signaling, and that PAR2 is likely to be involved. However, because of the lack of a suitable antibody that blocks cleavage activation of PAR2, we could not determine whether PAR2 activation solely mediates TF–VIIa signaling in this cell line.
To test whether TF–FVIIa-mediated signaling might alter tumor cell migration, a key step in the metastatic cascade, a modified Boyden chamber chemotaxis assay was carried out. The assays were performed in the presence of hirudin to avoid any effects of thrombin generation. The combination of FVIIa and FX induced a 10-fold increase in migration across filters coated with type IV collagen, whereas FVIIa alone did not significantly alter the number of migrating cell compared with BSA (Fig. 8). In contrast, FXa (150 nm) only slightly promoted cell migration (data not shown), suggesting that FXa generation on the surface of the cells is important for chemotaxis. These results indicate that formation of TF–FVIIa–FXa complex efficiently induces cell migration as it did MAPK phosphorylation, although the PAR(s) mediating these effects might not be the same.
To confirm that the effect of the combination FVIIa and FX is dependent upon TF complex formation, the chemotaxis assay was carried out in the absence and presence of a neutralizing anti-TF antibody. As shown in Fig. 9, the presence of the anti-TF antibody completely inhibited the migration induced by the combination of FVIIa and FX. Cell migration was also carried out in the presence of either the PAR1 or PAR2 activating peptides to determine which PAR receptor might be mediating the effect of FVIIa and FX (Fig. 9). Treatment of the cells with the PAR2 peptide induced a 3-fold increase in cell migration. By contrast, PAR1 activation of the cells led to an inhibition of migration by over 50%. These results suggest that the effect of the combination of FVIIa and FX might be mediated, in part, by PAR2 but not PAR1 activation.
Finally, to determine whether induction of MAPK phosphorylation by TF–FVIIa–FX complex is necessary for promotion of cell migration, cells were treated with U0126, an inhibitor of MEK1/2 that blocks MAPK phosphorylation . Dose-dependent inhibition of MAPK phosphorylation was observed with an IC50 of ∼ 0.2 µm (data not shown). Treatment of the cells with U0126 completely blocked cell migration, indicating that phosphorylation of p44/42 MAPK is required in this pathway (Fig. 10).
In this study, we first investigated how TF–FVIIa induces cellular signaling in human breast cancer cells using the Adr-MCF-7 cell line which has high endogenous expression of TF and expression of only PAR1 and PAR2. We found that low concentrations of FVIIa, which approach physiological levels, are insufficient to initiate cell signaling, and that FXa generation is critical for induction of MAPK phosphorylation by TF–FVIIa complex in AR-MCF-7 cells. Furthermore, thrombin formation is not required in this pathway. High concentrations of FVIIa alone were also found to induce MAPK phosphorylation in this cell line. Interestingly, this dose-dependent effect of FVIIa also appeared to be dependent upon FXa generation, since the effect could be abrogated in the presence of TAP (unpublished observations). These results are consistent with the recent studies of other investigators, who showed that efficient signaling with TF–FVIIa involves formation of the TF–FVIIa–FXa ternary complex [12,30]. In contrast, downstream components of the blood coagulation pathway might not be required for cellular signaling induced by high levels of FVIIa [14,25] in some cell lines. The significance of physiological effects induced by high concentrations of FVIIa, however, remains unclear.
A major function of TF in cellular signaling appears to be a docking site for FVIIa leading to the activation of PARs. The cytoplasmic domain of TF is not required for PAR activation alone , suggesting that signal transduction directly through TF does not occur in this pathway. Recently, PAR2 has been shown to be activated by TF–FVIIa complex ; however, other investigators have suggested that an as yet to be described PAR is involved . Results from our study show that PAR1 is not required for TF–FVIIa signaling and that PAR2 and/or an as yet to be described PAR is probably involved.
We then investigated whether FVIIa or the combination of FVIIa and FX might promote cell migration as a possible consequence of TF–FVIIa signaling. Others investigators have shown that FVIIa is capable of enhancing platelet-derived growth factor (PDGF)–BB stimulated chemotaxis of fibroblasts . Our study shows that migration of tumor cells is induced most efficiently by formation of TF–FVIIa–FXa complex, as in the induction of MAPK phosphorylation, and that other chemotactic stimuli are not required. Moreover, thrombin formation is unnecessary, since tumor cell migration experiments were carried out in the presence of hirudin. FXa, which induced robust MAPK phosphorylation, was found to promote cell migration only weakly. These results suggest that formation of TF–FVIIa–FXa complex might favorably position nascent FXa near PAR(s) that are responsible for induction of migration.
The role of PAR receptors in tumor function is not fully known. In our cell migration experiments, migration was enhanced by PAR2 activation and inhibited by PAR1 activation, suggesting that PAR2 and/or an as yet to be described PAR is responsible for inducing migration. Increased PAR1 expression has been shown to be associated with increased invasiveness and metastasis in some tumor cell lines [18–20]. However, our study is in agreement with the recent study of Kamath et al.  that showed PAR1 activation inhibits cell migration. PAR1 expression has also been found to be downregulated in some metastatic cells . Thus, the overall impact of PAR1 activation on the metastatic properties of different tumor cells remains to be determined and might be cell line specific.
Treatment of the Adr-MCF7 cells with either PAR1-activating peptide or PAR2-activating peptide led to phosphorylation of MAPK. However, PAR1 activation was found to inhibit migration whereas, PAR2 activation promoted migration. Because the combination of FVIIa and FX strongly promoted cell migration, our studies suggest that TF–FVIIa-mediated signaling, which probably involves PAR2, is distinct from the signaling pathway involving PAR1 activation. Moreover, we find that activation of MAPK phosphorylation is necessary, but not sufficient to induce cell migration. Other investigators have shown that MAPK phosphorylation is linked to myosin light chain phosphorylation, which is required for cell motility . Formation of TF–FVIIa–FXa complex and activation of PAR2 or another yet be described PAR might also lead to the downstream activation of the small GTPases, Rac and Cdc42, that are required for cell motility .
Upregulated TF expression occurs in highly invasive human breast cancer cell lines compared with less invasive breast cancer cell lines . Our study is the first to show that TF–VIIa signaling occurs in human breast cancer cells. Efficient signal transduction in this pathway involves formation of TF–FVIIa–FXa complex and probably PAR2 activation. While formation of TF–FVIIa–FXa complex was found to induce strongly migration of tumor cells, other physiological consequences of the TF–VIIa signaling pathway remain to be determined. The role of the cytoplasmic domain of TF, which is required for the full metastatic effect TF [4,5], also remains to be characterized in this pathway. Targeting TF–VIIa-mediated signaling in tumor cells could represent a novel strategy to prevent and/or control tumor metastasis.
This research received support from the WW Smith Charitable Trust (M.E.B.) and MO1 RR00349 from NCRR:NIH (M.E.B.).