Coagulation factors VII and X induce Ca2+ oscillations in Madin-Darby canine kidney cells only when proteolytically active.

We have recently reported that the activated serine protease and blood coagulation Factor VII (FVIIa) can induce Ca2+ oscillations in Madin-Darby canine kidney cells. We now demonstrate a similar response by Madin-Darby canine kidney cells to the active coagulation Factor X (FXa), which is also a serine protease and a substrate of the tissue factor (TF)·FVIIa complex in the initiation of the coagulation cascade. The phosphatidyl inositol-specific phospholipase C inhibitor U73122 inhibited the signals elicited by both FVIIa and FXa. Lack of sensibility to the tyrosine kinase inhibitors herbimycin A, genistein, and the tyrphostin AG18 and discordance between TF expression and FVIIa responsiveness argued against TF acting as a cytokine-like receptor, with tyrosine kinase-mediated activation by FVIIa. As demonstrated using the protease inhibitor benzamidine and by specific active site inhibition with 1,5-dansyl-Glu-Gly-Arg chloromethyl ketone, both FVIIa and FXa lost their ability to elicit a calcium response when devoid of their proteolytic activity. Consistent with this, the native (zymogen) form of Factor X did not induce Ca2+ transients. Homologous but not heterologous inhibition of FVIIa- and FXa-evoked Ca2+ signals by 1,5-dansyl-Glu-Gly-Arg chloromethyl ketone-inactivated FVIIa and FXa suggested that each factor had its own specific cell surface anchoring receptor. The two coagulation factors did not show homologous desensitization as seen for thrombin stimulation. Studies with hirudin excluded involvement of the established activation pathway through thrombin itself. Lack of desensitization of the response to FVIIa or FXa by thrombin ruled out any involvement of proteinase activated receptor-1 (PAR-1), the thrombin receptor. We speculate that FXa and FVIIa may work via a receptor (possibly common) analogous to PAR-1 or its functional homologue PAR-2. Although TF is essential for the FVIIa-induced signaling event, its role in the phosphatidyl inositol-specific phospholipase C-mediated Ca2+ signal may be in anchoring FVIIa to the cell surface rather than in transmembrane signal mediation.

Tissue factor (TF) 1 is an integral membrane protein that acts as an essential cofactor for the coagulation protease factor VIIa (FVIIa). The cloning of TF cDNA (1)(2)(3)(4) revealed that it was a member of the cytokine receptor superfamily (5). Binding of FVIIa to TF activates the protease and also induces Ca 2ϩ oscillations in several different cell types (6). These oscillations were strictly dependent on TF (6). TF also plays a critical role in development as demonstrated by gene targeting experiments in mice (7).
It has been reported that coagulation Factor Xa (FXa) can also induce a Ca 2ϩ response in endothelial cells (8), and we report here that FXa induces Ca 2ϩ oscillations in MDCK cells in a similar manner to FVIIa. FX and FXa have a number of cell surface receptors including Factor Va, Mac-1 (CD11/CD18), and effector cell protease receptor 1 (9), and this activity may thus be independent of TF. Cross-linking of effector cell protease receptor-1 (but not FXa binding) has been shown to generate a Ca 2ϩ response (10).
The modulation of cellular function by thrombin is probably mediated by two receptors (11). One of them (the thrombin receptor, proteinase activated receptor-1 (PAR-1)) is G-protein linked and activates cells through a phospholipase C ␤ , inositol (1,4,5)-trisphosphate pathway to give a strong Ca 2ϩ response (12).
In this report we use a digital imaging system to examine the response of fura-2-loaded MDCK cells to stimulation with these three coagulation proteases with the aim of further characterizing the pathway(s) leading to these changes in intracellular Ca 2ϩ levels. We have looked specifically at the convergence of activation pathways utilized by factors VIIa and Xa and compared these to Ca 2ϩ signals in response to thrombin stimulation.
Cell Culture-The constitutively TF-expressing MDCK type I distal tubule-derived epithelial cell line was obtained from Professor K. Prydz (Oslo, Norway) and cultured in Dulbecco's modified Eagle's medium supplemented with 5% fetal calf serum and L-glutamine (380 g/ml) in 30-mm tissue culture dishes. For experiments approximately 2 ϫ 10 5 cells/ml were seeded on glass coverslips 1 day prior to observation if not otherwise stated. Human fibroblasts isolated from a healthy individual were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and L-glutamine (380 g/ml).
Measurement of Cytosolic Ca 2ϩ in Single Cells-The measurement of cytosolic free Ca 2ϩ in single cells was done as described previously (6). The cells were incubated for 60 min at 37°C with a solution of 5 M fura-2, 0.25% dimethyl sulfoxide, and 0.025% Pluronic F-127 in HSS. The cells were then washed once and incubated with 400 l of HSS. Additions to the cell cultures were done by injection of 100 l into the well. The Ca 2ϩ imaging and registration software has been developed in our laboratory (14  presented as means and their standard error (S.E.). A Ca 2ϩ signal was defined as a response if the difference between the maximal Ca 2ϩ level after the application of agonist, and the mean before application was greater than 200 nM. The maximal increase in Ca 2ϩ was calculated by computing the maximal Ca 2ϩ after the application of agonist and then subtracting from this value the average Ca 2ϩ level before application. The integral of the part of the response after application of agonist, which was above the average Ca 2ϩ level before application, was taken as the increase in cytosolic free Ca 2ϩ . This integral depended directly on the observation time, which was kept constant within but not between experiments. Spontaneous responders and cells with compromised permeability were excluded by not considering cells that during the first 35 s of observation had either a higher absolute Ca 2ϩ level than 700 nM or a difference of more than 100 nM between the maximal and minimal Ca 2ϩ level.
In the experiments on duration of the Ca 2ϩ signal (see Fig. 4) and on desensitization (see Figs. 9 and 10), the cells were in a stimulated state at the start of the observation period, and the average Ca 2ϩ level before this period could not be calculated. In these experiments the Ca 2ϩ signal was defined as a response if the difference between the maximal and the minimal Ca 2ϩ level in the observation period was greater than 300 nM. The integral of the Ca 2ϩ response was calculated by computing the average Ca 2ϩ level within the given observation period and then subtracting from this value the minimal Ca 2ϩ level in the same time span. Both integrals and response percentages are therefore comparable only within these experiments.
Noncytosolic fura-2 was present in MDCK cells up to about 20% of total intracellular content, as demonstrated in digitonin permeabiliza-tion experiments (16). This will lead to an overestimation of the resting Ca 2ϩ concentration by 200 -300 nM. Because we are concerned about the Ca 2ϩ signal, i.e. changes in the cytosolic Ca 2ϩ concentration, this overestimation will not conceivably invalidate our findings.
Antibody Inhibition of FXa and FVIIa-FVIIa (500 nM) was incubated for 2 h at 37°C with a rabbit polyclonal anti-FVIIa antibody produced and purified in our laboratory (final concentration, 2.5 mg/ ml). As determined using a two-stage assay (16) with human brain extract as source of TF (0.25 unit/ml), limiting amounts of antibodyinhibited or normal FVIIa (10 nM), FX purified from human plasma (0.5 unit/ml, Sigma), and a chromogenic substrate for FXa (FXa-1, Nycomed), the antibody reduced the FVIIa coagulant activity by Ͼ90%. Similar results were obtained with a mouse monoclonal anti-FVIIa antibody (10G2). The antibody-inhibited FVIIa was used at a final concentration of 100 nM, with identically treated FVIIa or antibody alone as controls.
FXa (2.5 units/ml) was incubated for 1-2 h at 4°C with a mouse monoclonal anti-FXa antibody (93 g/ml, American Diagnostica 5224). The antibody reduced the activity of FXa by Ͼ70%, as determined using a one-stage chromogenic substrate assay for FXa (FXa-1, Nycomed). Cells were stimulated with FXa (0.5 unit/ml) and FXa (0.5 unit/ml) subjected to inhibition by antibody or antibody alone.
Duration of Response-Cells loaded for 1 h with fura-2 were rinsed briefly with HSS, fresh HSS with the appropriate agonist was added, and the cells were left at 37°C for the time indicated (3, 30, or 60 min) before observation. No single well was observed more than once.
Internalization of TF in Transfected MDCK Cells-Iodinated FVIIa or an iodinated anti-TF monoclonal antibody (4504) was allowed to bind onto the surface of either untransfected MDCK cells or MDCK cells transfected with a construct coding for full-length human TF (MDCK hTF 1-263 ) (17). After saturating the binding at 4°C (1.5 h, 1 g/ml FVIIa or 2.5 g/ml anti-TF monoclonal antibody) and removal of unbound FVIIa or antibody, the cells were brought up to 37°C and incubated further for up to 1 h. Cell surface-bound FVIIa or antibody was then removed with a pH 4.0 washing buffer, and what remained in the cell fraction was taken to be internalized. We corrected for the amount of radioactivity that had dissociated from the cells during the incubation period.
Inhibitors-Inhibitors (U73122, herbimycin, genistein, AG18, hirudin, and benzamidine) and control (U73343 and HSS) at the given concentrations were added to the cells in HSS after the 1-h fura-2loading period and 3-5 min before the start of the fluorescence observation, or, if indicated, added in a volume of 100 l during observation. All the inhibitors were adjusted to the right osmolarity prior to use.
RVV Activation of FX-RVV in cephalin was used to activate human and bovine FX. RVV (containing approximately 6 g of FX-activating enzyme) was reconstituted in 3 ml of 0.15 M NaCl at 37°C. One volume was mixed with nine volumes of a solution of 8.3 units/ml FX and incubated for 1 h at 37°C. The one-stage chromogenic FXa substrate test showed essentially complete activation of FX using these conditions. At 30 s RVV-activated FX was added to a final concentration of 1.5 units/ml. Unactivated but otherwise identically treated FX or RVV were used as controls.
DEGRck Inactivation of FXa-FXa (5 units/ml) was incubated with DEGRck (5 mM) for 1 h at 37°C in HSS. As control we used the same concentration of FXa without DEGRck. After the inactivation period the samples were passed through a filter with a 10-kDa cut-off and washed three times with 4 ml of HSS to remove excess DEGRck. The one-stage chromogenic FXa substrate test showed complete inactivation of FXa, and control filtration of FXa alone excluded significant loss of DEGR-FXa during filtration. Cells were stimulated at 30 s with a final concentration of DEGR-FXa corresponding to 1 unit/ml of native FXa and  Desensitization Experiments-If otherwise not stated, incubation for desensitization lasted 1 h prior to the start of observation. The agonist (in 100 l of HSS) was added to the cells together with the 400 l of HSS with fura-2 for loading. The cells were then rinsed three or four times with HSS, pH 7.4 (or HSS, pH 4.0, where stated), and observed either with HSS alone or with a repeated addition of agonist. The cells were stimulated at either 30 or 180 s after the start of observation.

Effects of FVIIa and FXa on the [Ca 2ϩ ] c level of MDCK
Cells-We have recently reported (6) that human FVIIa induces dose-dependent synchronous calcium oscillations in MDCK cells. In further characterization of this signaling event, we included coagulation Factor Xa, which has been suggested to give a calcium response in endothelial cells (8). When MDCK cells were treated with increasing concentrations of FXa, dosedependent Ca 2ϩ responses were observed ( Fig. 1 and Table I). The number of responding cells, the integral of the Ca 2ϩ signal, the maximal amplitude of the response (Table I), and the frequency of the Ca 2ϩ oscillations (not shown) all increased with increasing concentrations of FXa. As observed earlier for FVIIa (6), FXa also induced synchronous Ca 2ϩ oscillations (not shown). The characteristics of the Ca 2ϩ response to FXa (Fig. 1) were similar to those observed for FVIIa (6). Ca 2ϩ spikes were observed even at 0.1 unit/ml of FXa, which is well below the normal plasma concentration of FX (1 unit/ml). At a concentration between 0.2 and 0.5 unit/ml FXa Ͼ90% of the cells showed Ca 2ϩ transients. Above 0.5 unit/ml FXa, the cells would frequently not reach down to base-line levels before entering a new spike. Neither the FXa responses observed here nor the FVIIa evoked responses reported previously were due to contaminants in our preparations, as demonstrated by the substantial inhibition of the signals by antibodies neutralizing the respective factor activities (Fig. 2).
The FVIIa concentrations used (100 -200 nM) are higher than the plasma concentration of Factor VII (10 nM). The Ca 2ϩ effect is observed even at 2 nM but then only in a fraction (ϳ40%) of the cells (6). To obtain the effect in all cells, a substantially higher concentration must be used. Binding of Factor VII to TF⅐phospholipid membranes may increase the local concentration in vivo. To mimic the effect of flowing blood providing new Factor VII to the local milieu, a higher concentration of FVIIa was necessary to obtain the response in all cells under the stationary conditions of our experiments.
Considering that both FVIIa and FXa can bind to TF and because both induce Ca 2ϩ oscillations in MDCK cells, we looked to see if the two clotting factors had additive effects on the Ca 2ϩ signal. When the cells were first treated with FVIIa and then after 5 min with FXa or vice versa, the frequency of the Ca 2ϩ oscillations as well as the mean level of cytosolic free Ca 2ϩ increased (not shown). Some of the base-line Ca 2ϩ oscillations were also converted to a sustained Ca 2ϩ response with sinusoidal oscillations when the second clotting factor was given.
Cell Density-Recent studies in our laboratory (17) have demonstrated that the major fraction of surface expressed TF (Ͼ90%) is localized on the basolateral surface of MDCK cells when these are grown under conditions allowing the formation of intercellular tight junctions. Many receptor-mediated signals are also down-regulated when cells are grown to confluence. We therefore examined whether cell density influenced the Ca 2ϩ signals induced by FVIIa and FXa in MDCK cells. MDCK cells were seeded at low density and used for experiments 48, 72, 96, and 120 h after seeding. The cells were more than 80% confluent after 72 h, and the cell layers were completely confluent after 96 h as seen in the light microscope. After 120 h, when the cells were at maximal density, both FVIIa and FXa were unable to mount a Ca 2ϩ response (Table II  and Fig. 3). This was not because the cells had lost their potential to respond with a Ca 2ϩ increase, because all cells responded to the addition of 10 M ATP (Fig. 3). Consequently, the Ca 2ϩ oscillations generated by FVIIa and FXa did not appear when the MDCK cells were at high density. At shorter times the Ca 2ϩ signals were reduced (Table II).
Duration of the Ca 2ϩ Response-Due to limitations in the capacity of our experimental setup and the problem of photobleaching and in order to avoid overexposure of the cells to incoming light, single cells were regularly not observed for more than 5-7.5 min. In order to investigate the duration of the Ca 2ϩ response, we therefore observed the cells for 5 min at a series of time points after the addition of agonist. Each well was never observed more than once. We looked at the duration of Ca 2ϩ signaling in cells stimulated with thrombin (1 unit/ml), FVIIa (200 nM), or FXa (1 unit/ml). In another cell line thrombin-induced Ca 2ϩ transients lasted for less than 1 h (18). The observations were taken at 3, 30, and 60 min (Fig. 4). The synchronous Ca 2ϩ oscillations observed in the presence of thrombin leveled off rather rapidly also in our system, whereas the signals generated by FVIIa and FXa were of a longer lasting nature (Fig. 4). The FVIIa-stimulated cells consistently showed strong, highly synchronous Ca 2ϩ oscillations even 1 h after initiation. The amplitudes of the FVIIa-induced oscillations were reduced somewhat after 1 h. This may be a nonspecific effect related to fura-2 efflux and sequestration or desensitization. Also the FXa-stimulated cells appeared to oscillate with a certain synchronicity after 60 min, although not with the same consistency and amplitude as the FVIIa-induced cells.
Internalization of Surface-bound FVIIa and TF in MDCK Cells-The long duration of the FVIIa response suggested the lack of an efficient ligand-induced internalization mechanism for the FVIIa receptor, TF. To look at the turnover of FVIIa and TF on the cell surface within the time period studied, iodinated FVIIa or an iodinated anti-TF monoclonal antibody was al- lowed to bind at 4°C onto the surface of either untransfected MDCK cells or MDCK cells transfected with a construct coding for full-length human TF (MDCK hTF 1-263 ) (17). A 37°C internalization period followed. Within 1 h, only 20% of FVIIa and antibody were internalized (not shown). This is consistent with the lack of down-regulation of the response.
Effects of FVIIa and FXa on Phosphatidyl Inositol-specific Phospholipase C-To investigate whether the Ca 2ϩ signals induced by FVIIa and FXa involved the activation of phosphatidyl inositol specific phospholipase C (PI-PLC) and thereby an increase in inositol (1,4,5)-trisphosphate, we took advantage of the PI-PLC inhibitor U73122 (19). A close analogue (U73343) (19) with no observed effects on PI-PLC was used as control. Pretreatment (3-5 min) of the cells with 5 M of the active inhibitor U73122 completely inhibited the appearance of Ca 2ϩ transients after subsequent exposure to FVIIa or FXa, whereas the same concentration of the inactive analogue U73343 was without effect (Fig. 5). Application of the inhibitor to cells already stimulated with FVIIa or FXa and therefore oscillating, also blocked the signal in most cells (not shown).
TF in FVIIa-induced Cellular Activation-We have earlier presented clear evidence of a TF dependence for the FVIIainduced Ca 2ϩ signal (6). We then suggested, based on structural homology of the single transmembrane glycoprotein TF with the cytokine receptor family (5), that TF may interact with a cytosolic tyrosine kinase pathway. Two observations have made us question such a role for TF.
Firstly, inhibitors of tyrosine kinases did not inhibit the FVIIa-induced Ca 2ϩ signal. 1-h incubation of MDCK cells with 10 or 100 M genistein, 0.1 or 1.0 g/ml herbimycin A, or 1 M of the tyrphostin AG18 reduced neither the number of responding cells nor the increase in Ca 2ϩ elicited by 100 or 200 nM FVIIa (not shown). This was in agreement with earlier findings where we could not detect any difference in the phosphorylation pattern between FVIIa-treated and untreated MDCK cells when examined by immunoblotting with anti-phosphotyrosine antibodies (6). How PI-PLC is activated remains an open question, although G-protein activation seems most likely.
In our earlier studies (6) there was a discordance between the number of cells expressing TF and those responding to FVIIa, especially evident in the case of J82 cells, where about 80% of the cells carry stainable surface (20) TF, but only about 30% responded to FVIIa binding with Ca 2ϩ signals. This suggested that TF and FVIIa alone may not be enough to induce the increase in [Ca 2ϩ ] c . Further evidence was obtained from studies of human fibroblasts, where no Ca 2ϩ response to FVIIa was observed (2% responding cells, n ϭ 42), although they clearly expressed TF on the surface (21) of almost all cells.
Is the Proteolytic Activity of FVIIa and FXa Necessary for Inducing the Ca 2ϩ Response?-Both FVIIa and FXa are serine proteases, and their zymogens are prone to activation. Using a purified human FX preparation, we obtained variable and un-  reproducible Ca 2ϩ responses. Substantial augmentation of the response was obtained by activation of the same FX preparation as well as a bovine FX preparation with RVV with essentially no Ca 2ϩ response induced by the venom itself (Fig. 6). We then tested whether inhibition of their serine protease activity would block the Ca 2ϩ signal evoked by FVIIa and FXa. A general serine protease inhibitor, benzamidine, inhibited FVIIa-induced Ca 2ϩ oscillations (Table III). Total inhibition was obtained at 10 mM, a concentration that also blocked the FXa-evoked Ca 2ϩ response (Table III). DEGRck binds to and blocks the active sites of both FVIIa and FXa. Neither DEGR-FVIIa nor DEGR-FXa, both of which were inactive in a chromogenic substrate test, evoked Ca 2ϩ signals in MDCK cells (Fig. 7). Both of these inactivated factors inhibited the effects of a subsequent addition of the homologous active factor (Fig. 8), suggesting the involvement of one or two saturable cell surface binding sites. Epinephrine and ATP still induced a Ca 2ϩ response in these cells, indicating that there was no general inhibition of Ca 2ϩ signaling (not shown). Little heterologous inhibition was observed for FXa addition after DEGR-FVIIa. DEGR FXa did not inhibit the action of FVIIa.
To see whether the long duration and the synchrony of the oscillations in cytosolic free Ca 2ϩ generated by FVIIa was due to continuous signal generation by FVIIa or to the triggering of an intracellular cascade of calcium release and/or influx independent of the initial agonist, we added benzamidine to cells that had already been oscillating for 1 h due to the continuous presence of FVIIa. This led to an abrupt termination of the oscillations (Fig. 8). The same was seen for FXa-induced oscillations (not shown). This was not the case for cells stimulated with bradykinin (Fig. 8), another good inducer of synchronous Ca 2ϩ oscillations in MDCK cells. A general effect of benzamidine on the ability of the cells to sustain Ca 2ϩ oscillations was therefore unlikely.  bin or via Thrombin Receptors?-Thrombin is a powerful inducer of intracellular Ca 2ϩ signals, and trace concentrations of prothrombin being converted to thrombin by FXa in our experiments should be excluded. Three lines of evidence indicate that this is indeed unlikely. First, the thrombin inhibitor hirudin (5 units/ml) inhibited almost completely the Ca 2ϩ signals induced by thrombin (0.2 unit/ml), whereas only a very moderate reduction was observed in the response to FVIIa and FXa (Table IV). The number of responding cells decreased with 10% or less, the increase in Ca 2ϩ levels decreased with 13-25%, and the total integral of the Ca 2ϩ response decreased with 10 -31%. This limited reduction may be accounted for by a small direct effect of hirudin on the response potential of the cells or on the activity/binding of FVIIa and FXa. Second, no thrombin-like activity was detectable in the cell culture medium even in prolonged incubations (not shown). Third, desensitization of the thrombin response (see Fig. 10) using 1 unit/ml thrombin for 1 h did not essentially impair the response to FVIIa or FXa. This also excluded direct activation of PAR-1 by FVIIa⅐TF or FXa and made any contribution from a second thrombin receptor less likely (11).
Desensitization of FVIIa and FXa Responses-Homologous and heterologous desensitization by FVIIa and FXa was investigated after prior incubation with the respective agonist for 1 h. The cells were then washed three times with HSS, pH 7.4 or 4.0, as indicated, and fresh HSS without additions was added followed by observation of the cells for 180 s. The thrombin-induced (Fig. 9) and the FXa-induced (Fig. 10) cells remained silent with respect to Ca 2ϩ oscillations during this period, whereas the FVIIa-induced cells showed synchronous oscillations even after the pH 7.4 washes (Fig. 10). The latter observation is consistent with earlier evidence from our laboratory that FVIIa is not effectively removed from the cell surface unless an acid wash (e.g. at pH 4.0) is performed. When washed at pH 4.0, the cells did not oscillate during the 180-s observation period (not shown). FVIIa-induced oscillating cells washed at pH 7.4 did not change their synchronous pattern upon restimulation with FVIIa, indicating that all available relevant binding sites were occupied. FXa was also unable to change the synchronous pattern of FVIIa-stimulated cells. Cells restimulated with FVIIa after a pH 4.0 wash responded erratically, most likely due to the low pH exposure. In many cells highly synchronous Ca 2ϩ oscillations were re-established, showing that no homologous desensitization had occurred.
Upon restimulation with FXa, FXa-induced and then silenced cells showed no homologous desensitization. The cells moved directly into highly synchronous oscillations. They were also stimulated by FVIIa, i.e. no cross-desensitization was observed. The slight reduction of the thrombin response after FVIIa and FXa incubation for 1 h was most likely due to the fact that the very strong normal response to thrombin requires cells that are unperturbed. DISCUSSION We demonstrated recently that binding of the coagulation factor VIIa to the surface of cells carrying TF elicited intracellular Ca 2ϩ spikes (6). We demonstrate here that FXa has a similar effect on MDCK cells. For both factors their proteolytic activity was necessary for the effect to occur. The Ca 2ϩ response depended on the concentration of agonist. Both responses were inhibited by neutralizing antibodies to the respective agonists. Possible involvement of thrombin in eliciting these responses has been excluded.
The induced oscillations gradually became synchronous for large numbers of cells. Similar synchrony induced by bradykinin was shown to depend on intercellular gap junctions. 2 These synchronous signals go on for more than 1 h and can be stopped by washing off the agonist (for FXa) or by adding an inhibitor of serine proteases (benzamidine). In contrast, thrombin-induced signals decay with time and disappear completely after 30 -45 min. A correspondingly low internalization of FVIIa and the absence of desensitization for FXa and FVIIa are consistent observations. The turnover may be slow because of lack of an essential component because it is likely to depend on the whole TF⅐FVIIa⅐FXa⅐tissue factor pathway inhibitor complex (22).
Increasing cell density led to decreasing signals, probably caused by formation of dense cell layers and tight junctions leading to down-regulation or basolateral sequestration of a factor essential for the signaling event. Studies of the TF distribution of the MDCK cell surface have revealed a predominant basolateral localization (17). Reduced Ca 2ϩ signals would then result from reduced or abolished access of the agonist to the receptors on the basolateral surface. This is consistent with a role for TF in the generation of these signals. Several observations suggest that TF is not the sole cellular component involved in initiating the Ca 2ϩ signal. Although constitutive TF producers, only a fraction of J82 cells responds to FVIIa binding with Ca 2ϩ oscillations (6). Another example is human fibroblasts, which carry surface TF but do not respond to ligand binding with Ca 2ϩ signals at all.
The obligatory proteolytic activity and the absence of tyrosine kinase activation suggest that a protease-activated receptor may be involved in generation of the Ca 2ϩ signal. This is clearly not PAR-1. The role of the newly cloned orphan proteinase-activated receptor, PAR-2 (23), will be examined. Its distribution among different cell types (24 -27) appears to be consistent with that of responsiveness to FVIIa. It is difficult, however, to reconcile these long lasting signals and their abrupt termination upon benzamidine addition with the hypothesis that a proteolytically activated receptor with a functional tethered ligand should be involved. There is no evidence of consumption or down-regulation of the putative receptor over a period of more than 1 h.
The intracellular pathway leading to the release of Ca 2ϩ is not known. Our data suggest that PI-PLC is involved, although measurements of inositol (1,4,5)-trisphosphate levels have been inconclusive (not shown). There is no indication of an involvement of tyrosine kinases, neither from inhibitor studies nor from direct immunoblots using anti-phosphotyrosine antibodies. G-protein-mediated activation remains an alternative.
Similarities in characteristics of and requirements for the FVIIa-and FXa-induced Ca 2ϩ signals suggest that the pathway leading to these signals may converge early in the signaling cascade. The U73122 inhibitor studies show that this occurs before or on the activation of PI-PLC. If the activation of PI-PLC is G-protein-mediated, convergence may occur at the G-protein, the signaling receptor, or possibly a cell anchoring receptor.
Binding of DEGR-FVIIa to MDCK cells prevented signaling when subsequent binding of FVIIa was attempted. Similar observations were made for FXa. Binding of DEGR-FVIIa did not, however, prevent Ca 2ϩ signaling induced by a subsequent addition of FXa. The simplest interpretation of these data is that FVIIa and FXa have separate and saturable binding sites, although not necessarily different receptors.
In conclusion, we suggest as a useful working hypothesis that the mode of action of FVIIa and FXa in eliciting the intracellular Ca 2ϩ signal involves a receptor system, one anchoring receptor with a task of concentrating, activating (in the case of FVII), and presenting the protease to a second receptor, which after a proteolytic event triggers the pathway leading to PI-PLC activation and Ca 2ϩ release.