THE JOURNAL OF BIOLOGKXL. CHEMKWRV Vol. 254, No. 10, Issue of May 25. pp. 4092-4095. 1979 FGu.ed in U.S. A.

Binding of Human Thrombin to Cultured Human Endothelial Cells*

(Received for publication, July 27, 1978, and in revised form, January 2, 1979)

Brian J. Awbrey, John C. Hoak, and Whyte G. Owe@ From the Departments of Pathology, Internal Medicine and Biochemistry and the Cardiovascular Research Center, College of Medicine,SUniuersity if Zowa, Z&a City, Iowa 52242

Binding of thrombin to monolayer cultures of human umbilical vein endothelium is studied. Binding is meas- ured as inhibition by unlabeled ligand of the binding of ‘*%thrombin to the cells. Radioactivity bound to cul- tures at equilibrium is measured after draining but not washing the cells. To correct for unremoved superna- tant, 13%albumin is included as a second label in the medium. Equilibrium between bound and free throm- bin is attained within 1 min, and Scatchard analysis indicates a population of approximately 3 X lo3 sites/ cell with a dissociation constant of 10-l’ M, and a larger population with a dissociation constant greater than lo-@’ M. The two populations of sites are also indicated by a biphasic dissociation of bound label. Thrombin inactivated with diisopropyl fluorophosphate binds to the same receptor, with an affinity similar to that of active thrombin. Binding is unaffected by albumin (an acidic protein) and cytochrome c (a basic protein). Cul- tures of umbilical cord smooth muscle and fibroblasts bind thrombin at least 100 times more weakly than endothelium, and no binding to erythrocytes or a mono- layer culture of mouse neuroblastoma is detected.

Primary, spontaneous hemostasis results from a complex series of biological events which occur in immediate response to blood vessel injury. This response involves a triad of com- ponents: the damaged blood vessel, the blood platelets, and the plasma proteins involved in the blood coagulation process. It is generally accepted that the formation of thrombin rep- resents a key development during normal hemostasis, and in thrombosis. In addition to its roles as a zymogen activator and regulator in the coagulation cascade, thrombin binds to the surface of platelets (l-7), and has the capacity to cause platelets to aggregate and to release some of their internal components including serotonin and adenine nucleotides. Al- though binding of catalytically active thrombin to the platelet. membrane is required for activation of the platelet by throm- bin, binding per se does not require an active catalytic center and thus is physically discrete from subsequent events. Also of interest is the possibility that the vascular endothelium may interact with thrombin or other activated coagulation components. Such interactions might be components of a metabolic or protective mechanism, or conversely, might act to localize thrombin at the blood interface where it could exert a direct enzymatic effect. The present study addresses these questions, with standard approaches for the detection of re-

* This work was supported by Grant 14230-06 (Specialized Center of Research in Atherosclerosis). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ Career Development Awardee, National Institutes of Health. To whom correspondence should be addressed.

ceptor sites for human thrombin on monolayer cultures of human endothelial cells obtained from umbilical veins. The data presented demonstrate that human thrombin is bound reversibly and specifically by the plasma membrane of human endothelial cells.

MATERIALS AND METHODS

Sulfopropyl Sephadex, bovin serum albumin (essentially fatty acid- free), cytochrome c, and sodium dodecyl sulfate, were obtained from Sigma Chemical Co., St. Louis, MO. Diisopropyl fluorophosphate’ was purchased from Aldrich Chemical Co., Milwaukee, Wis. Carrier-free “‘1 (as KI) was obtained from Amersham-Searle and ‘“‘I-albumin (Albumotope-131) was the product of Squibb, Inc. Antiserum to human crz-macroglobulin was obtained from Miles Laboratories, Elk- hart, Ind., and an antiserum to human von Willebrand Factor (Factor VIII-related antigen) was prepared by the method of Zimmerman (8). Heparin (Panheparin) was obtained from Abbott Laboratories, North Chicago. Ill. Crude collagenase (sterile) was obtained from Worth- ington Biochemical Corp., Freehold, N. J.

Human thrombin was prepared as described previously (9). The product was homogeneous by gel electrophoresis in dodecyl sulfate. For some experiments, thrombin was inactivated at pH 7.5 with 1.0 mM DFP. Thrombin was labeled with “? by the lactoperoxidase method of Thorell and Johansson (10). The labeled protein was separated from excess “‘1 and lactoperoxidase reagents by chroma- tography on SP-Sephadex. For each experiment, 10 to 15 pg of thrombin were labeled. The reaction was quenched and the mixture was applied to a SP-Sephadex C-50 column (0.3 x 5 cm), equilibrated in 0.1 M NaCl, 0.02 M Tris-HCl, pH 7.5. The column was washed with equilibration buffer until less than 2 x 10” cpm/ml of “‘1 were detected in the effluent. With a single step of 0.6 M NaCl, the “‘I- thrombin w&$ eluted in a peak with a thrombin concentration suffi- cient to quantitate clotting activity. The product had a specific radioactivity of 9 X 10”’ cpm/mg of protein. The final yield of clotting activity averaged 20%, whether or not the protein was subjected to radioiodination, with virtually all of the loss occurring during chro- matography. When analyzed by gel electrophoresis in dodecyl sulfate, 95% of the ““I appeared in a peak wit.h a migration distance identical with that of unlabeled thrombin. A small (~5%) peak of /?-thrombin was detected, and no radioactivity was detected in the region of the tracking dye. The enzymatic activity of the product had a half-life in the frozen state of 12 to 24 h, a time during which unlabeled thrombin was stable. Such instability, presumably due to radiochemical dam- age, was reported also by Tollefson et al. (1). Therefore, thrombin was labeled routinely within 2 h of the beginning of an experiment.

Preparation of Cell Cultures-Primary cultures of human endo- thelial cells were prepared from human umbilical veins according to a slight modification of the method of Jaffe et al. (II). Umbilical cords were collected in Medium M-199 with 10% fetal calf serum. The umbilical vein was dilated and rinsed with 60 to 100 ml of phosphate- buffered saline. After clamping one end with a hemostat, the vein was filled with 0.2 to 0.3% collagenase. After 6 min, the cell suspension was drained into 5 to 10 ml of Medium M-199 with 20% fetal calf serum, and the vein was rinsed with additional 40 ml of Medium M- 199. The cell pellet resulting from centrifugation of the combined suspension was resuspended in fresh Medium M-199 with 20% fetal calf serum and seeded in Petri dishes (35 x 10 mm) at a density of 900,000 cells/dish. Cultures were incubated in a 5% CO2 atmosphere at 37°C. After 24 h, the monolayers were first washed with fresh ~.-___~ ___ ~ - -~~

’ The abbreviation used is: DFP, diisopropyl fluorophosphate.

4092

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

Thrombin Binding to Endothelium 4093

medium in order to remove excess red blood cells and were layered with a fresh 2-ml portion of Medium M-199 with 20% fetal calf serum. Confluent primary human endothelial cell monolayers of 7 to 8 x lo5 cells were used for the binding studies 3 to 5 days after the Petri dishes were seeded. Confluent cultures of human umbilical vein fibroblasts and smooth muscle cells were prepared as described pre- viously (12). Endothelium, fibroblasts, and smooth muscle cells were differentiated by phase-contrast microscopy and their ultrastructure was monitored by electron microscopy. Cell identification was con- sistent with published criteria for these types of cells (13-15).

Approximately 15 min before the beginning of an experiment, the cultures were washed three times with 1 ml/wash of Hanks’ Balanced Salts (16). modified by substituting 15 mM 4-(2-hydroxyethyl)-l-pi- perazineethanesulfonic acid, pH 7.4, for NaHC03. All subsequent steps were performed with the modified Hanks solution.

Binding of Thrombin to Endothelial Cells-Binding was evaluated by two methods.

For preliminary experiments, 1 ml of modified Hanks’ solution containina 0.005 NIH unit (1.6 nn) of ‘251-thrombin was &aced unon the washed endothelial cell cultures for 30 to 45 s at room tempera- ture. A 100~~1 sample of the incubation medium was then taken to determine free tbrombin concentration. The remaining medium was suctioned off and the cell layer was rinsed twice, as rapidly as possible, with 1 ml/wash of modified Hanks’ solution. The-cells were then solubilized with 1 ml of 1% sodium dodecvl sulfate. 0.01 M NaOH. 0.01 M EDTA. The cells were allowed to solubilize for 1 h, at which’time the solution was transferred to capped plastic test tubes and the radioactivity was determined.

To study binding under equilibrium conditions, bound thrombin was determined after removal of the medium, but without washing the cells. Such measurements are complicated by a small, but una- voidable trace of medium remaining on the cell monolayer, even with the most. thorough aspiration. The volume of unremoved medium, and accordingly the quantity of unbound thrombin which remained on the cells was determined by including ‘3’I-albumin as a second label in the incubation medium. Because of a 20% spill of 13’1 into the “‘1 channel of the scintillation spectrometer, ‘3’I-albumin concentra- tion was adjusted so that counts per min per ml in the 13’1 channel was approximately that in the “‘1 channel. One milliliter of the ?- thrombin-‘3’I-albumin incubation medium was placed on the washed endothelial cells for 30 to 45 s. A 100~~1 sample was then taken to determine free thrombin concentration. Remaining medium was suc- tioned off and without washing, the cells were solubilized.

Bound thrombin was calculated as:

(‘25I cpm, cells) - 9 cpm, cells

,3,1 cpm medium x ‘? cpm, medium 1 The ‘*‘I counts per min were corrected for 13’1 spill. Gel electrophoresis in dodecyl sulfate was performed by the method

of Laemmli (17). ‘?-Labeled products were determined by dividing the gels after electrophoresis into 2-mm slices.

RESULTS

Maximum binding of thrombin by the cells occurred within 30 s, in the range of thrombin concentrations used for these experiments (Fig. 1). With up to 10 min of additional incuba-

L . . * . . . . 123456789

MINUTES

FIG. 1. Time course of binding of thrombin to endothelium. Cells were incubated in 1 ml of buffer (see “Materials and Methods”) containing 0.005 unit of labeled thrombin. At the indicated times, medium was removed, and the cells were washed twice with 1 ml of buffer, dissolved, and counted. Each point is an average of three determinations. All subsequent binding experiments employed the double label method.

tion, there was no increase or decrease in bound radioactivity. Label bound in the presence of 1 unit/ml of unlabeled throm- bin averaged 30% of the total tracer bound, at either 30 s or at 10 min. Since maximum uptake was reached by 30 s, with no significant change in total uptake for 10 min, all binding experiments were terminated after 30 to 45 s of incubation. When endothelial cells were incubated for 30 s to 5 min with the labeled thrombin, washed, dissolved, and then analyzed by gel electrophoresis in dodecyl sulfate, a single peak of radioactivity with a migration rate indistinguishable from that of unlabeled thrombin, was detected.

Use of the double label to measure thrombin binding at equilibrium requires that the binding of albumin to the cells is insignificant compared to the binding of thrombin, and that albumin does not interfere with thrombin binding. If the medium is simply removed after exposure of the cells to both labels and cells are counted without washing, the proportion of thrombin counts remaining is about twice the proportion of albumin counts (Fig. 2); a single wash removes virtually all of the labeled albumin. The decrease in the difference with each wash reflects removal of bound thrombin from the membrane, as it is only with unwashed cells that bound and free thrombin are in equilibrium. Subsequent experiments determined that albumin does not affect thrombin binding. These data also demonstrated that less than 30% of bound thrombin is eluted with two washes, where virtually all of the albumin is removed.

To insure that possible differences in binding between un- labeled and labeled thrombin were not being measured, bind- ing was measured at a constant, tracer concentration of labeled

WASHES

FIG. 2. Elution of ‘%I-thrombin and ‘3’I-albumin from endothe- lium. Cells were incubated for 30 s with a mixture of 0.005 unit/ml of ‘*‘I-thrombin and 1 gg/ml of ‘3’I-albumin. Medium was removed by suction, with the dishes tilted to insure good drainage. The cells were either solubilized or rinsed the indicated number of times and solu- bilized. The cells were exposed to 1 ml of fresh buffer for approxi- mately 5 s per rinse. Data are expressed as per cent of incubation medium radioactivity remaining with the culture. Open bars are “‘1 cpm; shaded bars are 13’1 cpm.

I I. I 5 IO 20 40

BOUND THROMBIN ~lO’molecules/celll

FIG. 3. Scatchard analysis of the binding of thrombin to endothe- lium monolayers. Incubation medium contained 0.005 unit/ml of ‘?- thrombin, 1 gg/rnl of ‘3’I-albumin, and variable concentrations of unlabeled thrombin. Other details are described under “Materials and Methods.” Open circles are uncorrected data and closed circles represent data corrected by subtraction of lower affinity binding.

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

4094 Thrombin Binding to Endothelium

thrombin, and only the unlabeled thrombin concentration was varied. Use of this method also facilitated comparisons with other potential ligands, and minimized sampling and arith- metic error in the second label correction. An adsorption isotherm, transformed as described by Scatchard (18) is shown in Fig. 3. The data are averages taken from seven lots of cultured cells, with each point taken in duplicate or triplicate. Lower affinity binding (Ko > 2 X lo-’ M), indicated by upward curvature of the line (open circles) at high ligand concentra- tion, was subtracted (19) to demonstrate the high affinity binding shown by the closed circles. Least squares analysis (20) (N = 123, r - 0.89) was used to extrapolate to the axes to give 5 = 3300 sites/cell with Kd s lo-*’ M.

Competition with labeled tracer thrombin was used to evaluate the specificity of thrombin binding to the cells. In these experiments, unlabeled proteins were added with the labeled thrombin, and binding of label was determined. Fig. 4 shows the results obtained with unlabeled thrombin, cyto- chrome c, albumin, and thrombin which had been inactivated with DFP. Inactive thrombin competed equally with active thrombin for binding (Fig. 4, open circles). Conversely, no competition by either cytochrome c (Fig. 4, squares) or albu- min (Fig. 4, triangles) was observed. Since thrombin binds to hydrophilic cation exchangers (9), cytochrome c, which also binds to cation exchangers, was evaluated to determine that thrombin does not bind by nonspecific but high affinity, saturable ion exchange. Albumin was evaluated both as a representative plasma protein and to validate its use as a second label. The finding that albumin does not compete for binding with thrombin is consistent with the finding (Fig. 2)

0.2 0.4 0.6 0.6 I UNLABELED PROTEIN

FIG. 4. Effect of unlabeled thrombin, cytochrome c, albumin and DFP-thrombin on the binding of labeled thrombin to endothelium. Cells were co-incubated for 30 s with 0.005 unit/ml of ‘251-thrombin and unlabeled thrombin (O-----O, units/ml), cytochrome c (A.. . .A, pg/ml), albumin (Cl.. . . Cl, mg/ml), and DFP-thrombin (w, unit equivalents/ml). Binding was evaluated with the double labeled method. The data for unlabeled thrombin are averages of the data shown in Fig. 3.

II I 5 10 15 20 25 30

MINUTES

FIG. 5. Dissociation of labeled thrombin from endothelium. Cells were incubated for 30 s with 0.005 unit/ml of ‘251-tbrombin. Medium was removed with suction and then replaced with buffer. After the indicated intervals, buffer was removed and the cells were solubilixed.

A B C D E CELL TYPE

FIG. 6. Binding of thrombin to cell types other than endothelium. Monolayer cultures were incubated with 0.005 unit/ml of ‘251-throm- bin and 1 &ml of ‘3’I-albumin. Cultures indicated by the shaded bars contained 1 unit/ml of unlabeled thrombin. For studies with red blood cells, the cells were collected and washed on a 0.45 c Millipore filter. Cell types are: A, endothelium; B, umbilical cord smooth muscle; C, umbilical cord fibroblasts; D, mouse neuroblastoma; E, 2.5 x 10’ human red blood cells. Each cell type had grown to confluence on a 35-nun Petri dish, as determined by phase-contrast microscopy. The red cells had a calculated surface area approximately twice that of the confluent cell cultures.

that binding of albumin to the cells is insignificant compared to that of thrombin.

Dissociation of bound thrombin was biphasic (Fig. 5). The half-times for dissociation were approximately 1 mm and 60 mm for the fast and slow rates, respectively. The rates of dissociation were unaffected by addition of excess (3 pg/ml) unlabeled thrombin to the medium.

Of plasma proteins known to interact directly or indirectly with thrombin, two have been localized in endothelium: (~2‘ macroglobulin (21) and von Willebrand Factor (22, 23). To test the possibility that either protein could function as a thrombin receptor, cells were incubated for 5 min with anti- sera to cy2-macroglobulin or von Willebrand Factor (Factor VIII-related antigen), with antibody concentrations sufficient to detect antigen in fixed cells by indirect immunofluores- cence. Neither antiserum had an effect on binding of thrombin to the cells. Likewise, pretreatment of the cells with heparin and then washing the monolayer had no effect on thrombin binding.

When compared with endothelial cells (Fig. 6A), monolayer cultures of umbilical cord smooth muscle (Fig. 6B) and fibro- blasts (Fig. SC) bound little thrombin per unit cell surface area, and the binding was not saturable, that is, was not inhibited by excess unlabeled thrombin in the range of con- centrations studied. Thrombin did not bind to a mouse neu- roblastoma monolayer culture (Fig. 6D) or to human eryth- rocytes (Fig. 6E).

DISCUSSION

The specificity of thrombin binding to endothelium suggests the presence on the cell surface of a thrombin receptor, and thus communication between the humoral hemostasis system and the blood vessel wall. Analysis of the binding by the method of Scatchard (18) demonstrated a population of bind- ing sites of approximately 3300 sites per endothelial cell with an apparent dissociation constant of 1 X lo-” M, and a second, larger population of binding sites with an apparent dissocia- tion constant of > 2 x lo-’ M. The latter were subtracted for analysis of high affinity sites, but there is no evidence that they are nonspecific. Although this analysis is performed with the assumptions of equilibrium and of absence of site-site interactions, the important conclusions, that binding is high affinity, saturable, and specific, stand independent of those assumptions.

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

Thrombin Binding to Endothelium 4095

Dissociation of more than 65% of the ‘251-thrombin within 30 min. when the cells are incubated in thrombin-free buffer, suggests that binding observed within short incubation periods is unlikely to represent thrombin taken into the cells. In addition, the localization of radioactivity only in the thrombin band on sodium dodecyl sulfate gel electrophoresis after sol- ubilization and lyophilisation of labeled cells showed that the labeled thrombin remained intact during the period of study. However, internalization of a fraction of the bound thrombin cannot be excluded. These findings, together with the failure of an antiserum to an-macroglobulin to affect thrombin bind- ing, exclude cuz-macroglobulin as a candidate for the receptor.

Labeled thrombin was competed for equally by DFP-throm- bin and unlabeled tbrombin. This indicates that the active site of the thrombin molecule is not involved in binding to endothelial cells. Human blood vessel smooth muscle cells and blood vessel fibroblasts, in contrast to endothelial cells, lacked high affinity thrombin receptors, and human red cells and a mouse neuroblastoma cell line monolayer bound essen- tially no thrombin. Lower affinity binding to these cells was not evaluated. These findings provide further evidence that the uptake of thrombin by endothelial cells is specific.

The recent reports (12, 24, 25) that thrombin can cause release of PC& (prostacyclin) from endothelium are of partic- ular relevance. PGIt is the most potent known inhibitor of platelet aggregation and adherence, and its release by a po- tentially thrombogenic substance, thrombin, suggests that intact endothelium may be involved in the control of hemo- stasis and thrombosis. Studies are in progress to delineate the importance of thrombin receptors in this phenomenon.

Acknowledgments-The endothelial cultures were prepared and maintained by Ms. Glenna Fry. We thank Dr. Lawrence DeBault for generously providing the mouse neuroblastoma cultures.

REFERENCES

1. Tohefaen, D. M., Feagler, J. R., and Majerus, P. W. (1974) J. Biol. Chem. 249,2646-2651

2. Ganguly, P. (1974) Nature 247,306-307 3. Ganaulv. P.. and Sonnichsen. W. J. (1976) Br. J. Haematol. 34.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14. 15. 16.

17. 18. 19. 20.

21. 22.

23.

24.

25.

. ., , 291-301

Tohefsen, D. M., Jackson, C. M., and Majerus, P. W. (1975) J. Clin. Invest. 56, 241-245

Shuman, M. A., ToIIefsen, D. M., and Majerus, P. W. (1976) Blood 47,43-54

Shuman, M. A., and Majerus, P. W. (1975) J. Clin. Znuest. 56, 945-950

Mohammed, S. F., Whitworth, C., Chuang, H. Y. K., Lundhlad, R. L., and Mason, R. G. (1976) Proc. Natl. Acad. Sci. U. S. A. 73, 1660-1663

Zimmerman, T. S., Hoyer, L. W., Dickson, L., and Edgington, T. S. (1975) J. Lab. Clin. Med. 86, 152-159

Owen, W. G., Esmon. C. T.. and Jackson, C. M. (1974) J. Biol. Chem. 249; 594-605

ThoreII. J. I.. and Johansson. B. G. (1971) Biochim. Bionhvs. Acta 261,363-369

. ”

Jaffe, E. A., Nachman, R. L., Becker, C. G., and Minick, C. R. (1973) J. Clin. Invest. 52, 2745-2756

Czervionke, R. L., Hoak, J. C., and Fry, G. L. (1978) J. CZin. Invest. 62, 347-856

Jaffe, E. A., Nachman, R. L., Becker, C. G., and Minick, C. R. (1973) J. Clin. Invest. 62.2745-2756

Ross, R. (1971) J. Cell Bioi 50, 172-186 Gimbrone, M. A., and Cotran, R. S. (1975) Lab. Invest. 33, 16-27 Hanks, J. H., and Wallace, R. E. (1949) Proc. Sot. Exp. Biol.

Med. 71,196-200 Laemmh, U. K. (1970) Nature 227,680~685 Scatchard, G. (1949) Ann. N. Y. Acad. Sci. 51,660-672 Chamness, G. C., and McGuire, W. L. (1975) Steroids 26,538-542 Huntsberger, D. V. (1967) Elements ofStatistical Inference, AIIyn

& Bacon, Inc., Boston Becker, C. G., and Harpel, P. C. (1976) J. Exp. Med. 144, l-9 Hoyer, L. W., De Los Santos, R. P., and Hoyer, J. R. (1973) J.

Clin. Invest. 52,2737-2744 Bloom, A. L., Giddings, J. L., and WiIks, C. K. (1973) Nature New

Bial. 241,217-219 Weksler, B. B., Ley, C. W., and Jaffe, E. A. (1978) J. Clin. Invest.

62,923-930 Czervionke, R. L., Smith, J. B., Hoak, J. C., Fry, G. L., and

Haycraft, D. (1979) Thromb. Res. 14, 781-786

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

B J Awbrey, J C Hoak and W G OwenBinding of human thrombin to cultured human endothelial cells.

1979, 254:4092-4095.J. Biol. Chem. 

  http://www.jbc.org/content/254/10/4092Access the most updated version of this article at

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/254/10/4092.full.html#ref-list-1

This article cites 0 references, 0 of which can be accessed free at

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from