Evidence That the Effects of Thrombin on Arachidonate Metabolism in Cultured Human Endothelial Cells Are Not Mediated by a High Affinity Receptor*

The effect of thrombin and its derivative, diisopro- pylphosphoryl-thrombin on [3H]arachidonic acid metabolism is studied in cultured umbilical vein endothe- lial cell monolayers. Thrombin causes a dose-depend-ent release of radioactivity from endothelial cells fed [3H]arachidonate. Thin layer radiochromatography of acidified supernatants reveals that most of the radio- activity is [3Hlarachidonate and its metabolites, 6-keto-prostaglandin F1, and prostaglandin Ez. Diisopropyl- phosphoryl-thrombin, which is enzymatically inactive, does not cause release of arachidonic acid or metabo- lites. A 50-fold excess of diisopropylphosphoryl-throm-bin, despite causing 98% inhibition of binding of “‘I- thrombin to its high affinity binding sites, does not inhibit thrombin-induced release. We conclude that the high affinity, active site-independent thrombin binding sites are not involved in thrombin-induced mobilization of esterified arachidonic acid. manner Binding rapid, Scatchard plot yielding


grade.
Endothelial Cell Cultures-Primary cultures of human umbilical vein endothelial cells were provided by the laboratory of Dr. John Hoak as described previously (1) as a modification of the method of Jaffe et al. (10). The experiments reported used batches of endothelial monolayers derived from pooling several umbilical cords.
Thrombin Preparation-Human thrombin and DIP-thrombin were prepared as described (11). The enzyme had 2700 United States units/mg based on an extinction coefficient (E;?,) of 19.5 (12).2 Clotting activity was determined by comparison with a standard curve using National Institutes of Health thrombin (13). and human fibrinogen was purified via the method of Straughn and Wagner (14). Also, in some experiments, thrombin was labeled with Iz5I by the lactoperoxidase method of Thorell and Johansson (15), as described previously (l), with the exception that the iodination was carried out in chloride-free sodium acetate buffers. This resulted in a product with a clotting activity which was stable at 4°C for at least 1 week.
Assay for Release of [3H/Arachidonate and Metabolites by Endothelial Cells-Three-or four-day-old endothelial cell cultures, consisting of confluent monolayers on 35-mm Petri dishes, were washed three times with Hanks' balanced salt solution buffered with 15 mM 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid (Hepes), pH 7.4, plus 0.1% albumin (Buffer A). The [3H]arachidonic acid, as soon as it had been received from the manufacturer, was diluted further into toluene and stored in 25-pCi aliquots at -80°C. Before each experiment, an aliquot was dried under NP, then eluted several times with Buffer A, and finally diluted with Buffer A to a final concentration of 1 pCi/ml. The monolayers were incubated with 1 ml of r3H]arachidonate in Buffer A. Preliminary experiments indicated that uptake of [3H]arachidonate was most rapid when the dishes were rocked at 37°C. After 1 h, 70 to 80% of the tritium was cell-associated, after which the level reached a plateau. Thus, in all experiments described, the r3H]arachidonate was added to rocked dishes for 1 h at 37°C except when aspirin or indomethacin was used (see below).
After incubation with [3H]arachidonate, the cells were washed three times in Buffer A, and then test substances or control solutions were applied in a total volume of 2 ml at room temperature. The dishes were swirled once after addition of test substances and were treated in every other way as binding experiments (see below). At intervals, samples (100 pl) were taken and pipetted directly into a toluene-based aqueous scintillation fluid containing Triton X-100.
In experiments where aspirin (1 m) (16) or indomethacin (20 p~) (17) were used, the cells, after incubation with [3H]arachidonate, were incubated an additional hour with drug (i.e. total incubation time of 2 h). Indomethacin was dissolved in absolute ethanol prior to addition of buffer to a final ethanol concentration of 0.1% (v/v). In control experiments, this concentration of ethanol had no effect on I The abbreviations used are: PG, prostaglandin; DFP, diisopropylfluorophosphate; DIP-thrombin, diisopropylphosphoryl-thrombin.
'The National Institutes of Health unit of thrombin has been redesignated the United States unit by the Bureau of Biologics, Food and Drug Administration. 803 I thrombin-induced release of tritiated products. Also, control experiments comparing cells incubated for 2 h in the absence of drug with cells incubated for 1 h only revealed no difference in the amount of radioactivity released.
Thin Layer Chromatography-After maximal release of tritium was obtained, the endothelial cell supernatants were removed (1 to 1.5 m l ) and acidified with 0.1 volume of glacial acetic acid. The supematants were extracted twice with ethyl acetate and dried under Nz. The samples were then redissolved in a small amount of ethyl acetate, spotted on Eastman Chromagram silica gel thin layer plates and developed in a solvent system containing chloroform/methanol/ acetic acid/H20, 90:8:1:0.8 (18). Also included in parallel lanes were arachidonic acid and prostaglandin standards PGAz (RF = OB), PGEz (RF = 0.6), PGFz, (RF = 0.4), and 6-keto-PGF1, (RF = 0.6). The developed plates containing standards were dried at 100°C on a hot plate, sprayed with 10% phosphomolybdic acid in ethanol, reheated at 100"C, and sprayed with concentrated sulfuric acid (19). Lanes containing radioactive material were not charred and were cut into 5mm slices and placed in a toluene-based scintillation fluid.
'25Z-Thrombin Binding to Endothelial Cells-Binding was carried out as described (1). Briefly, monolayers were washed three times with Buffer A, and I-ml aliquots containing 1251-thrombin and various concentrations of nonradioactive DIP-thrombin were added. The dishes were swirled once and then allowed to incubate unstirred for 2 min at room temperature. The cells were then rapidly washed twice with Buffer A and solubilized in 2% sodium dodecyl sulfate and 2% sodium carbonate, and radioactivity was determined. The washing procedure took 9 & 2 s. Preliminary experiments revealed the dissociation rate constant to be -0.9 min-I; thus, there was less than 15% dissociation during the washing procedure.

RESULTS
The addition of thrombin to [3H]arachidonate-fed monolayers resulted in a release of tritiated material into the supernatant (Fig. 1). This release was complete within 8 to 10 min. Both the rate and the amount of radioactivity released were dose-dependent and saturable, reaching a maximum at 1 United States unit of thrombin/ml (=lo-* M). The addition of buffer alone also caused a small release, presumably from mechanical stimulation of the cells by the addition of fluid (8). The saturability of the response was not related to substrate depletion since addition of trypsin (5 X lo-* M) (8) after a 10-min exposure to thrombin resulted in a release similar in magnitude to the original release by thrombin (data not shown). This concentration of trypsin does not affect endothelial cell morphology or viability (as tested by erythrosin B exclusion).
Both the amount of radioactivity released and the nature of radioactive products released were dependent on the presence or absence of albumin. In the presence of albumin, there was a 2to %fold increase in amount of radioactivity released. Thin layer radiochromatography revealed that greater than 90% of the radioactivity migrated in the region of arachidonate. In the absence of albumin, there was less release of tritium and more than 80% of the product migrated (RF = 0.6) with 6-keto-PGF1, and PGE2. Similar results have been reported in platelets (20, 21) and fibroblasts (22). As expected, aspirin (1 mM) and indomethacin (20 p~) inhibited the formation of prostaglandins by greater than 95% and had no effect on the release of arachidonate. All further experiments were done in the presence of 0.1% albumin to obtain maximum sensitivity of the release response.
To determine whether the high a f f i t y thrombin receptor on endothelium is involved in arachidonate metabolism, DIPthrombin was used. This derivative has a diisopropylphosphoryl group covalently bonded to its active site serine with no other alterations in the structure of thrombin. It neither clots fibrinogen nor cleaves synthetic substrates but has been shown to bind to endothelium in a manner indistinguishable from thrombin (1). At a concentration of 5 X lo-' M (equivalent to 50 units/ml), DIP-thrombin caused no release of tritium over control (Fig. 2). This concentration is 500 times the amount of active thrombin needed to cause measurable release. However, the presence of 50 units/ml of DIP-thrombin had no effect on the release of tritium by 1 unit/ml of thrombin (Fig. 2), even though it caused 98% inhibition of specific binding of 1 unit/ml of '251-thrombin to endothelium (Fig. 3). Likewise (data not shown), lower concentrations of DIP-thrombin had no effect.

DISCUSSION
In this paper, we provide evidence that the high affinity active site-independent binding sites for thrombin on endothelial cells are not involved in thrombin-induced release of arachidonate and arachidonate metabolites. If thrombin binding were involved, then an agent which inhibits thrombin binding (i.e. DIP-thrombin) should behave as a competitive antagonist.
However, blocking the binding sites with DIPthrombin has no effect on thrombin-induced changes in ar-a&donate metabolism, which makes it unlikely that these binding sites are involved. It could be argued that DIP-thrombin blocked less than 100% of the binding sites and occupancy of the remaining small number was responsible for the full effect. This "spare receptors" argument would not be consistent with the dose-response relationship shown in Fig. 1, where doses of less than 1 unit/ml of thrombin (which was used in the experiment shown in Fig. 3) result in progressively less release of tritium.
A similar phenomenon has been reported in platelets (23) where DIP-thrombin binds with equal affinity to the same receptor as active thrombin, but neither causes [%]serotonin release nor blocks thrombin-induced [%]serotonin release. Although the authors claim that the binding sites they measure are responsible for thrombin-induced aggregation and secretion (24), this argument has been attacked by others (25). In addition, it has been shown recently that ADP binds to platelets with all the characteristics of a high affinity receptor, but this binding is apparently not related to ADP-induced platelet aggregation since binding can be blocked by p-chloromercuribenzene sulfonate without affecting platelet aggregation (26). Similarly, the binding of [3H]norepinephrine to a variety of tissues was initially thought to be causally related to its biological effect. This was seriously questioned when a variety of compounds (e.g. D isomers of catecholamines, pyrocatechol, dihydroxymandelic acid) where found to inhibit binding without affecting biological response (27,28), and it now appears that a different class of binding sites (with higher affinity) is involved in the response (29).
A high affinity thrombin receptor has been proposed to be responsible for other effects of thrombin on endothelial cells, which include mitogenesis (2, 3), inactivation of plasminogen activator (5), and release of fibronectin (6), but cause-effect relationships have neither been demonstrated nor ruled out by experiments similar to those described in this paper.
Thus, the initial step on the pathway of thrombin stimulation of prostacyclin production is not known. Since stimulation is active site-dependent, proteolytic cleavage of a plasma membrane substrate may be involved. Presumably, then, this enzymatic process would involve the presence of an enzyme. substrate (Michaelis) complex on the plasma membrane. The cellular response conceivably would depend on the turnover rate of the enzyme. This is in contrast to models of biological response to classical bimolecular ligand-receptor interactions, where the amount of bound receptor in part governs the magnitude of response (30). The amount of enzyme-sub-strate complex on the membrane might be small relative to amounts of ligand. receptor complexes usually seen (lo3 to IO5 ligand molecules bound/cell at saturation). If this were true, then standard radioligand binding assays might not be sensitive enough to detect the presence of enzyme n substrate complexes.
Another possibility is that there is an active site-dependent receptor for thrombin which is distinct from the receptor that binds both DIP-thrombin and active thrombin. The model for an active site-dependent receptor would differ from that for the membrane-associated proteolytic substrate in that cellular response would not involve proteolysis of the receptor. If the affinity of thrombin for this receptor were low or the number of receptors smaller than the active site-independent binding sites, then binding of thrombin to the receptor would be difficult to distinguish from "nonspecific" binding seen in all receptor assay systems.
In conclusion, the interaction of thrombin with cell membranes is more complex than previously assumed. Caution should be exercised when ascribing cause and effect relationships to the interaction of thrombin with cell surfaces and biological response.