Synthesis of prostaglandin I2 (prostacyclin) by cultured human - PNAS

Proc. Natl. Acad. Sci. USA Vol. 74, No. 9, pp. 3922-3926, September 1977

Cell Biology

Synthesis of prostaglandin I2 (prostacyclin) by cultured human and bovine endothelial cells (prostaglandin X/inhibition of platelet aggregation/thrombogenesis)

BABETTE B. WEKSLER*, AARON J. MARCUSt,

AND

ERIC A. JAFFE*

* Division of Hematology, Department of Medicine, Cornell University Medical College, New York, New York 10021; and t New York Veterans Administration Hospital, New York, New York 10010

Communicated by Alexander C. Bearn, July 7, 1977

Cultured endothelial cells derived from human ABSTRACT umbilical veins or bovine aorta produce a potent inhibitor of platelet aggregation. The inhibitor is synthesized from sodium arachidonate or prostaglandin endoperoxides by a microsomal enzyme system. Tranylcypromrine, a specific antagonist of prostacyclin synthetase, suppresses production of the inhibitor by endothelial cells. The inhibitor, which is ether extractable, has been identified using a two-step thin-layer radiochromatographic procedure and a synthetic prostaglandin I2 standard. With this procedure, we have shown that human and bovine endothelial cells convert sodium [3H]arachidonate to radiolabeled prostaglandin I2 and 6-keto-prostaglandin Fla, as well as prostaglandin E2. Thus, endothelial cells may be non-thrombogenic in vivo because they synthesize and release prostaglandin I2, a potent inhibitor of platelet aggregation. Fragments of blood vessel walls acting upon prostaglandin (PG) endoperoxides or arachidonic acid produce an unstable factor that prevents platelet aggregation and release (1, 2). This factor, recently identified as (5Z)-9-deoxy-6,9-a-epoxy A5-PGFIa, (prostacyclin or PGI2) (3), is much more potent as an inhibitor of platelet aggregation than PGEI or PGD2 (4) and also relaxes vascular smooth muscle (1, 5-7). PGI2 can be synthesized from prostaglandin endoperoxides or arachidonic acid by microsomes derived from blood vessel walls (1, 5, 8), but the specific cells possessing this synthetic activity have not as yet been identified. The cellular and acellular components of blood vessel walls are known to differ markedly in their interactions with platelets. Normal endothelium is non-adherent for platelets both in vio and in vitro in tissue culture (9). However, intimal and medial components such as collagen and microfibrils induce platelet aggregation (10). Endothelial cells have previously been shown to synthesize PGE-like prostaglandins (11). It seemed likely therefore, that endothelial cells should be capable of converting prostaglandin endoperoxides to PGI2 and/or of synthesizing PGI2 directly from arachidonic acid. The studies to be reported were undertaken to localize PGI2 synthesis at a cellular level. We have demonstrated that cultured human and bovine endothelial cells synthesize and release PGI2 after incubation with arachidonic acid and also convert exogenous prostaglandin endoperoxides to PGI2.

MATERIALS AND METHODS Culture and Preparation of Endothelial Cells. Human endothelial cells derived from umbilical cord veins were cultured using methods previously described (12). The culture medium consisted of medium 199 containing 20% heat-inactivated rabbit serum or 20% heat-inactivated pooled human serum, penicillin (100 units/ml), streptomycin (100 ,gg/ml), and 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. §1734 solely to indicate this fact.

L-glutamine (2 mM). Cells were grown in T-25 or T-75 flasks (Corning) at 370 under 5% C02/95% air until confluence. Both primary cultures and serial passages were utilized. Initial cultures of bovine endothelial cells were kindly provided by Alan Quarfoot and Francois Booyse (13) and were cultured in RPMI 1640 medium containing 10-20% heat-inactivated fetal calf serum and penicillin, streptomycin, and L-glutamine as above. Cells were harvested from culture flasks by treatment with 0.1% collagenase-0.01% EDTA for 10 min at 370, washed three times in buffer A [137 mM NaCl/4 mM KCl/11 mM glucose/10 mM N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (Hepes), pH 7.5] and finally resuspended in the same buffer at 6 X 106 cells per ml. Culture and Preparation of Other Cells. Human fibroblasts were grown in tissue culture, harvested with trypsin (0.25%) for 10 min, washed similarly, and resuspended in buffer A at the same protein concentration as the endothelial cells. Peripheral blood mixed leukocytes (70-80% neutrophils) were separated from heparin-treated whole blood by dextran sedimentation, washed, and resuspended in -buffer A at a similar protein concentration. Platelet Aggregation Experiments. Platelet-rich plasma (PRP) was prepared from venous blood drawn into 'ho volume of 3.2% trisodium citrate by methods previously described (14) and was kept tightly capped at 220 under 5% C02/95% air until use. Compounds being tested (0.1 ml) were added to 0.4 ml of PRP in the cuvette of a Payton aggregation module (Payton Associates, Buffalo, NY). Aggregating agents used included arachidonic acid (as the sodium salt in 0.1 M Na2COA), adenosine diphosphate (ADP) and collagen (Sigma Chemical Co., St. Louis, MO), and bovine thrombin (Parke Davis Co., Detroit, MI). Synthetic PGI2 (15) and 9,11-azoprostanoid III (16) were the generous gifts of E. J. Corey (Harvard University). Other PGs were a gift from J. Pike of Upjohn Co. Preparation of Endothelial Cell Subcellular Fractions. Endothelial cells in buffer A were frozen and thawed in dry ice/acetone and briefly sonicated using a microtip sonicator (Kontes, Inc., Vineland, NJ). The broken cell suspension was centrifuged at 8000 X g for 2 min (220) and the pellet was discarded. The supernatant was tested for its ability to inhibit platelet aggregation, and if active was then ultracentrifuged at 105,000 X g for 1 hr (40). The post-microsomal supernatant was removed and stored on ice until used. The microsomal pellet was resuspended by brief sonication in a volume of buffer A equal to the original sample volume and stored on ice until used. Assay of Whole Endothelial Cells and Subcellular Fractions for the Ability to Inhibit Platelet Aggregation. Intact endothelial cells (2 to 20 X 104) were incubated with PRP in a final volume of 0.5 ml in an aggregation module cuvette at 370 Abbreviations: PG, prostaglandin; PGI2, (5Z)-9-deoxy-6,9-a-epoxyA5-PGFI,; PRP, platelet-rich plasma. 3922

Proc. Natl. Acad. Sci. USA 74 (1977)

Cell Biology: Weksler et al. with stirring at 1000 rpm for up to 5 min. Sodium arachidonate other aggregating agents were then added in an amount that induced aggregation in PRP in the absence of endothelial cells. Control experiments utilizing other cell types in place of endothelial cells were also carried out. Supernatants from sonicated, frozen-thawed endothelial cells, resuspended microsomal pellets, and post-microsomal supernatants were similarly incubated with PRP, followed by the addition of an aggregating agent. In some experiments endothelial cell preparations were preincubated with theophylline (Sigma), indomethacin (Merck,

3923

CAA

or

c

0

E c

.C

Sharp and Dohme), or tranylcypromine (a gift from H. Green; Smith, Kline & French) prior to addition to PRP. As controls for the effects of drug carryover, equal amounts of drug mixed with buffer were added to PRP before challenge with an aggregating agent; no inhibition of aggregation occurred in such controls. Ether Extraction. Endothelial cells or microsomes were incubated with sodium arachidonate or buffer for 2-10 min (220) with gentle agitation. The incubation mixture was extracted with ten volumes of cold diethyl ether and the extraction was repeated once. The ether layers were combined and dried under nitrogen. The residue was immediately either redissolved in a small volume of buffer A and tested with PRP for its ability to inhibit platelet aggregation or dissolved in acetone and stored

-70°. Thin-Layer Radiochromatography. [5,6,8,9,11,12,14,153H(N)]Arachidonic acid, specific activity 64 Ci/mmol (New England Nuclear, Boston, MA) was converted to the sodium salt (17). Twenty microcuries of sodium [3H]arachidonate was mixed with unlabeled sodium arachidonate in 0.1 M Na2CO3 to yield a final concentration of 1 MM in the incubation mixture with the endothelial cell microsomal preparation. Four milliat

liters of the endothelial cell microsomal fraction in buffer A was then added and the mixture was gently agitated for 10 min at room temperature, after which unlabeled PGI2 standard (191

,g)

was

added. The incubation mixture was extracted three

times with cold ether (20, 10, and 10 ml). The combined extracts were dried under nitrogen and analyzed in System 1 (4). System I was run as follows. The dried ether extract was redissolved in chloroform/methanol (1:1, vol/vol), and applied under nitrogen to a silica gel G plate previously activated at 1100 for 30 min. The solvent system was chloroform/methanol/acetic acid (90:10:2, vol/vol) (4). Also included on parallel lanes on the plate were the following standards: PGI2, PGE2, PGD2, and arachidonic acid. Following chromatography, the plate was broken so that lanes containing the standards and one sample of the extracted incubation mixture could be stained. The plate was heated to 1200 (4 min) and stained with 10% phosphomolybdic acid in ethanol, followed by reheating, spraying with 50% (vol/vol) sulfuric acid, and heating for 30 sec (18). The area (in the unstained sample lanes) corresponding to the PGI2 standard was scraped off, eluted three times with chloroform/methanol (1:1), and processed in System II as described below. Recovery of PGI2 from the original incubation mixture was further improved by processing the remaining aqueous portion following the original ether extraction procedure. After the pH of the aqueous layer had been adjusted to 6.5 with 0.2 M citric acid, two extractions with cold ethyl acetate (8 ml each) were performed. The combined extracts were dried under nitrogen and processed by thin-layer radiochromatography in System I as above. System 11 (15) was run as follows. The chloroform/methanol eluates from System I were dried under nitrogen and dissolved in 50 ,l of methanol and 450,Ml of freshly prepared ethereal diazomethane (19, 20). Methylation was performed in the dark for

15

min at 15°. The dried methyl esters were dissolved in 40

1

min

FIG. 1. Inhibition of platelet aggregation by endothelial cells. PRP (0.4 ml) was stirred with 0.1 ml of endothelial cells (EC), fibroblasts, or leukocytes for 5 min and then sodium arachidonate (AA, final concentration of 0.5 mM) was added.

Al of cold ether and chromatographed along with methylated standards on an unactivated silica gel G plate that had been pre-treated with ammonia-saturated ether. The solvent system was ether/acetone (3:1, vol/vol) (15). Ether only partially solubilized the methyl esters; the ether-insoluble residue was dissolved in chloroform/methanol (1:1) and this material was chromatographed separately. The entire plate was processed and stained as above. Lanes were divided into zones based on positions of the internal standards, scraped into vials containing Aquasol (New England Nuclear), and assayed for radioactivity in a liquid scintillation counter.

RESULTS Inhibition of Platelet Aggregation by Intact Cultured Endothelial Cells. Intact cultured human endothelial cells, stirred with PRP for 5 min, completely inhibited platelet aggregation induced by sodium arachidonate (Fig. 1). However, fibroblasts or human blood leukocytes at similar cell protein concentrations did not inhibit platelet aggregation. None of the cell preparations caused platelet aggregation in PRP when sodium arachidonate was omitted. Endothelial cell pre- and post-culture media were not inhibitory. Endothelial cells also inhibited platelet aggregation induced by ADP, collagen, prostaglandin endoperoxide analogues, thrombin, or epinephrine (Table 1). The release of [14C]serotonin induced by these aggregating agents was also suppressed by endothelial cells (Table 1). Parallel experiments were carried out using cultured bovine endothelial cells with similar results. The bovine endoTable 1. Inhibition by endothelial cells of platelet aggregation and release of [14C]serotonin in platelet-rich plasma Platelet responses*

Aggregating agent ADP (2,iM) Sodium arachidonate (0.3 mM) Collagen (30 ,g) 9,11-Azoprostanoid III (0.13 Ag/ml) Thrombin (0.25 unit)

Aggregation 41 0 30 0 0 0

[14C]Serotonin release 0 1 17 2 3

Epinephrine (0.45 ,gM) Endothelial cells (1 X 104) were incubated for 5 min with PRP (0.5 ml), following which a threshold concentration of the particular aggregating agent was added. * As percent of control.

3924


Cell Biology: Weksler et al. EC

+

buffer

>AA

c

0

W

E

EC + tranylcypromine

c 4)

._T -J

1 min

FIG. 2. Suppression by tranylcypromine of the inhibitory effect of intact endothelial cells on platelet aggregation. Intact endothelial cells (EC) were preincubated for 2 min with tranylcypromine (500 ,g/ml) and then sodium arachidonate (AA, final concentration 0.5 mM) was added. Control experiments were performed in which either the cells or the tranylcypromine was replaced by buffer.

thelial cells possessed, however, more inhibitory activity (about 100-fold more) than did the human endothelial cells. To demonstrate that the inhibition of platelet aggregation by endothelial cells was caused by PGI2, experiments were performed using tranylcypromine, a specific inhibitor of PGI2 synthesis (4). As a control, endothelial cells were pre-incubated for 2 min with buffer and added to PRP. Aggregation was then induced by sodium arachidonate. As expected, platelet aggregation was inhibited (Fig. 2). However, if the endothelial cells were pre-incubated for 2 min with tranylcypromine (500 ,ug/ml), inhibition of platelet aggregation was completely suppressed, i.e., the platelets aggregated normally in response to sodium arachidonate. Tranylcypromine alone had no effect upon platelet aggregation. Intact endothelial cells pre-incubated for 30 min with indomethacin (0.1-10 ,g/ml), an inhibitor of cyclooxygenase, were able to inhibit platelet aggregation in PRP to the same extent as did untreated endothelial cells. Theophylline, a phosphodiesterase inhibitor, potentiated the inhibition of platelet aggregation by endothelial cells when used at a concentration (100 MM) that alone had no effect on platelet behavior. Inhibition of Platelet Aggregation by Subeellular Fractions of Endothelial Cells. Supernatants prepared from disrupted human endothelial cells inhibited platelet aggregation ordinarily induced by the agents shown in Table 1. Inhibition of platelet aggregation by the supernatant fraction of endothelial cell sonicate occurred within 15-60 sec after addition to PRP (Fig. 3), in contrast to the 3-5 min required for inhibition by intact endothelial cells (Fig. 2). Endothelial cell microsomes incubated with PRP also completely inhibited platelet aggregation induced by sodium arachidonate (Fig. 3). However, neither the post-microsomal supernatant nor the buffer control inhibited platelet aggregation. On the basis of protein content, microsomes generated 5-fold more inhibitory activity than did intact cells and 2-fold more than crude cell sonicate (Table 2). Sonicates of bovine endothelial cells were also highly inhibitory.

Preincubation of endothelial cell microsomes with 500,4g/ml of tranylcypromine resulted in the loss of their capacity to inhibit platelet aggregation. In control experiments the addition of tranylcypromine to synthetic PGI2 did not alter the inhibitory effect of PGI2 on platelet aggregation. However, microsomes prepared from endothelial cells cultured for 24 hr with indomethacin (0.1 ,ug/ml) inhibited platelet aggregation as well as those from untreated cells. Boiling the microsomes for 30 sec destroyed their ability to inhibit platelet aggregation. Endothelial Cells Release an Inhibitor of Platelet Aggregation. When untreated human endothelial cells were incu-

FIG. 3. Inhibition of platelet aggregation by subcellular fractions of endothelial cells. PRP (0.4 ml) was stirred with 0.1 ml of the fractions specified below, and then sodium arachidonate (AA, final concentration 0.5 mM) was added. (A) Endothelial cell sonicate, (B) endothelial microsomes, (C) endothelial cell post-microsomal supernatant, and (D) buffer A.

bated in buffer A and then centrifuged, the supernatant did not contain an inhibitor of platelet aggregation. However, endothelial cells incubated with sodium arachidonate (1-5 AuM) for 2 min (220) yielded supernatants that inhibited platelet aggregation induced by sodium arachidonate, 9,11-azoprostanoid III, collagen, and ADP. When suspensions of endothelial cells were incubated for 2 min (220) with sodium arachidonate and then extracted with ether, the ether extract inhibited platelet aggregation (Fig. 4). Ether extracts of endothelial cells incubated with buffer alone or extracts of sodium arachidonate alone did not inhibit platelet aggregation. The ether extracts, stored in dry acetone at -70°, retained inhibitory activity for at least 1 week. Inhibitory activity was abolished by heating to 1000 for 30 sec or by acidifying to pH 4. Endothelial cells preincubated for 15 min with indomethacin (1 ,g/ml) were subsequently incubated with sodium arachidonate (1 MM) for 2 min (22°) and then extracted with ether. The ether extract did not inhibit platelet aggregation. Parallel experiments using bovine endothelial cells gave similar results. Synthesis of PGI2 and 6-Keto-PGFI, by Endothelial Cells. Following incubation of both human and bovine endothelial cell sonicates with sodium[3H]arachidonate, synthesis of PGI2 and its stable derivative 6-keto-PGFa was demonstrated by thin-layer radiochromatography of lipid extracts derived from the incubation mixtures. A two-step thin-layer chromatographic system was necessary to resolve the radioactive products. When the lipid extracts were analyzed in System I, both the internal and external PGI2 standards migrated as single spots with an RF value (0.52) similar to that of the PGE2 standard. PGD2 migrated ahead of the latter standards (RF = 0.62). Unconverted arachidonic acid was identified near the solvent front. Table 2. Minimal inhibitory amount of endothelial cell fraction protein Minimal inhibitory amount, Igt Endothelial cell preparation* 75.0 + 4.4 Intact cells 35.4 : 5.5 Sonicated cell supernatant 16.5 ± 1.3 Microsomes * Intact cells suspended in buffer A at 6 X 106/ml. All fractions were prepared from this cell density and returned to initial volume. t Mean + SD of four to six separate preparations. This quantity inhibited platelet aggregation in PRP induced by 0.5 mM sodium

arachidonate.


Cell Biology: Weksler et al.

3925

PGE2

A

800

c

0

4n

E C

co

41

600-

J-

400 1 min

FIG. 4. Inhibition of platelet aggregation by ether extracts of sodium arachidonate-treated endothelial cells. Intact endothelial cells (EC) were incubated with sodium arachidonate (2 MM) and extracted with cold ether, and the extract was tested for its ability to inhibit platelet aggregation induced by sodium arachidonate (AA, final concentration 0.5 mM). A control experiment was performed by omitting the sodium arachidonate used to stimulate the endothelial cells.

200-

PGI2

6-keto-PG F 1 i

C

0 N

C

E

0

2

S

ca o

800-

4

6

8

{A1< 10

2

14

B

0

Thus, chromatography in System I provided a convenient method for the removal of unconverted [3H]arachidonic acid. To resolve PGI2, PGE2, and 6-keto-PGFIa, the zone from the System I thin-layer chromatographic plate containing the internal PGI2 standard was analyzed in System II, which clearly separated the methyl esters of these three prostaglandins. Partial conversion of the standard PGI2 to the 6-keto derivative was apparent on these chromatograms. As shown in Fig. 5, peak areas of radioactivity correspond to the location of PGI2, 6keto-PGFIa, and PGE2. The chromatographic patterns of the ether extracts of the incubation mixtures were similar to those of the ethyl acetate extracts. DISCUSSION Platelets do not adhere to normal vascular endothelium in vivo or to intact endothelial cells grown in tissue culture, but do adhere readily to injured endothelium or subendothelial components (9, 10). Saba and Mason noted (21) that endothelial cells released material that inhibited platelet aggregation, and Harker et al. (22) recently confirmed that endothelial cells added to PRP inhibit aggregation. Several groups (1, 5, 7) have recently found that microsomal fractions of porcine, human, and bovine arteries incubated with arachidonic acid or with prostaglandin endoperoxides produce a new prostaglandin (PGI2) that inhibits platelet aggregation. Because the acellular components of vascular walls promote platelet aggregation, these workers speculated that PGI2 was synthesized by endothelial cells. When we incubated intact human endothelial cells with PRP, platelet aggregation and serotonin release were markedly inhibited (Fig. 1 and Table 1). Complete inhibition of arachidonate or collagen-induced aggregation was produced by 2 to 20 X 104 human or 2 X 103 bovine endothelial cells per ml of PRP. Only a small, reversible primary wave was seen when aggregation was induced by threshold concentrations of ADP. Inhibitory activity was also generated by the supernatant fraction from frozen-thawed, sonicated endothelial cells (after removal of cell debris by centrifugation at 8000 X g and by endothelial cell microsomes (Fig. 3). Platelet aggregation was inhibited more quickly (15 sec) by subcellular fractions of endothelial cells than by intact cells (3-5 min). On the basis of protein content, the microsomal pellet was about 5 times more potent than intact cells in its capacity to generate inhibitory activity (Table 2). Only 15 gg of endothelial cell microsomal protein completely

6001-

PGI

400-

6-keto-PG F la

200-

2

4 6 12 8 10 Distance from origin, cm

1

FIG. 5. Thin-layer radiochromatography (System II) of methylated products derived from the zone in System I that chromatographed with standard PGI2 and PGE2 after extraction of endothelial cells incubated with sodium [3H]arachidonate. (A) Human and (B) bovine endothelial cells. Zones were demarcated according to positions of internal and external standards. Standards appeared as circular spots. The entire zone encompassing each spot was scraped and eluted. In this system, PGI2 was completely separated both from its stable end product 6-keto-PGFi,,, and from PGE2. The ordinate represents the total cpm in each zone scraped from the plate. The abscissa represents the distance along the plate referable to the zones. Radioactivity migrating ahead of PGI2 was not identified. Comparable zones from adjacent blank lanes, used as controls, contained negligible radioactivity (15 cpm). Material chromatographed in A was from an ethyl acetate extract and that in B from an ether extract. Differences in apparent RF reflect humidity conditions in separate runs.

inhibited platelet aggregation induced by sodium arachidonate, while 6-30 mg of vascular wall fragment was required for similar inhibition (4). Once generated, the inhibitory activity was extractable into ether and stable on storage in acetone at -70°. It was destroyed by boiling for 30 sec or by acidification. Inhibitory activity could not be extracted from endothelial cells incubated in buffer, but was extractable after the cells were briefly incubated with sodium arachidonate. Indomethacin-treated endothelial cells did not generate ether-extractable inhibitor from sodium arachidonate, but still inhibited platelet aggregation when added to platelet-rich plasma. These findings suggested that endoperoxides generated by platelets could be converted to the


3926 Cell Biology: Weksler et al.

inhibitor. However, endothelial cells pretreated with tranylcypromine, a specific inhibitor of prostacyclin synthetase (4), failed to inhibit platelet aggregation. The ability to use sodium arachidonate or endoperoxide as a substrate, the production of an ether-soluble inhibitor of platelet aggregation, and the heat and acid instability of the inhibitory product all suggested that endothelial cells were producing PGI2. When endothelial cells and synthetic PGI2 were compared for their inhibitory effects on platelet aggregation, 2 X 104 intact human endothelial cells (or 15 gg of microsomal protein) were as inhibitory as 0.1 ng of synthetic PGI2. PGI2, like PGE1, has been shown to increase platelet intracellular cyclic AMP levels (23, 24). Elevated intracellular cyclic AMP in turn probably inhibits the availability in platelets of arachidonate for oxidation by the prostaglandin synthetase system (25). Our findings that theophylline potentiated the inhibition of platelet aggregation by intact endothelial cells suggested that the inhibitor acted by increasing platelet intracellular cyclic AMP. When endothelial cell microsomes were incubated with sodium [3H]arachidonate and extracted with ether or ethyl acetate, the extract contained a radioactive lipid that cochromatographed with the PG12 standard in a two-step, thin-layer chromatographic procedure. In order to recover PGI2 in an unchanged form, the initial ether extraction was performed immediately after incubation without acidification. In addition, because prostaglandins are more readily extracted into organic solvents as the free acids, we attempted to improve the yield of PGI2, without producing total conversion to 6-keto-PGF1(,, by slight acidification. Thus, the aqueous layer remaining after the original ether extraction was acidified to pH 6.5 and extracted with ethyl acetate. These procedures enabled us to separate and identify radiolabeled PGI2, 6-keto-PGFia, and PGE2 in material derived from human and bovine endothelial cells incubated with sodium [3H]arachidonate. Thus, we have demonstrated synthesis and release of PGI2 and its derivative 6-keto-PGFia (as well as PGE2) in material that completely blocked platelet aggregation. It is of interest, however, that the biological assay, inhibition of platelet aggregation, was at least 1000-fold more sensitive in detecting PGI2 activity than the

radiochromatographic procedure. It is clear from the present studies that endothelial cells possess both the capacity to generate PGI2 directly from exogenous sodium arachidonate via cyclooxygenase and the capacity to convert externally formed prostaglandin endoperoxides to PGI2. Although we were unable to demonstrate it, it is possible that untreated endothelial cells continuously synthesize and release low baseline levels of PGI2 which in our incubation system were rapidly converted to a biologically inactive form, 6-keto-PGFia. The ability of endothelial cells to synthesize and release PGI2 and thus to inhibit platelet aggregation and the release reaction suggests that PGI2 functions to limit thrombus size. Because platelets do not adhere to intact endothelium in vivo or in vitro, PGI2 may also prevent platelet adhesion, though this remains to be proven. Blood vessel thromboresistance is probably a composite function involving endothelial membrane surface glycoproteins and mucopolysaccharides, membrane enzymes, and secreted products (of which PGI2 is one), as well as platelet reactivity. A balance may exist between the effects of endoperoxides and thromboxane A2 (produced by platelets in response to vascular damage) and the effects of PGI2 (produced by endothelium). The prostaglandin endoperoxides may serve, therefore, as pivotal substrates because they can be converted by platelets or by endothelial cells into products with opposing physiologic properties. Damage to or loss of endothelial cells in a local area may promote thrombosis both by exposing subendothelium and by ablating PGI2 production.

Subsequent adhesion of platelets to the subendothelium would lead to aggregation, degranulation, and release of factors that stimulate smooth muscle cell growth and enhance vascular permeability. Repetition of this process, over a long period of time, may result in atherosclerosis (10, 26). Prevention of these phenomena could depend upon the normal synthesis and release of PGI2 by endothelial cells. We are grateful to Dr. E. J. Corey of the Department of Chemistry, Harvard University, for his help and encouragement and gifts of 9,11-azoprostanoid III and PGI2. We acknowledge the excellent technical assistance of Ellen Hindy, Lenore Safier, Harris Ullman, Christine Baranowski, and Christopher Ley. This work was supported by the National Institutes of Health through a Specialized Center for Thrombosis Grant (HL 18828), the New York Heart Association, the Arnold R. Krakower Hematology Foundation, and a Veterans Administration Research Grant. E.A.J. is the recipient of a National Institutes of Health Research Career Development Award (KO4 HL 00237) and a Career Scientist Award from the Irma T. Hirschl Trust. 1. Moncada, S., Gryglewski, R., Bunting, S. & Vane, J. R. (1976)

Nature 263,663-665. 2. Moncada, S., Higgs, E. A. & Vane, J. R. (1977) Lancet i, 1821. 3. Johnson, R. A., Morton, D. R., Kinner, J. H., Gorman, R. R., McGuire, J. C. & Sun, F. F. (1976) Prostaglandins 12, 915929. 4. Gryglewski, R., Bunting, S., Moncada, S., Flower, R. J. & Vane, J. R. (1976) Prostaglandins 12,685-713. 5. Bunting, S., Gryglewski, R., Moncada, S. & Vane, J. R. (1976) Prostaglandins 12, 897-913. 6. Dusting, G. J., Moncada, S. & Vane, J. R. (1977) Prostaglandins 13, 3-15. 7. Raz, A., Isakson, P. C., Minkes, M. S. & Needleman, P. (1977) J.

Biol. Chem. 252,1123-1126. 8. Ho, P. P., Herrmann, R. D., Towner, R. D. & Walters, C. P. (1977) Biochem. Biophys. Res. Commun. 74,514-519. 9. Booyse, F. M., Bell, S., Sedlak, B. & Rafelson, M. E., Jr. (1975) Artery 1,518-539. 10. Stemerman, M. B. (1974) in Progress in Hemostasis & Thrombosis, ed. Spaet, T. H. (Grune & Stratton, New York), Vol. 2, pp. 1-48. 11. Gimbrone, M. A. & Alexander, R. W. (1975) Science 189, 219-220. 12. Jaffe, E. A., Nachman, R. L., Becker, C. G. & Minick, C. R. (1973) J. Clin. Invest. 52,2745-2756. 13. Booyse, F. M., Sedlak, B. J. & Rafelson, M. E., Jr. (1975) Thromb. 14.

Diath. Haemorrh. 34,825-839. Weksler, B. B., Gillick, M. & Pink, J. (1977) Blood 49, 185-

196. 15. Corey, E. J., Keck, G. E. &

Szekely, I. (1977) J. Am. Chem. Soc.

98,2006-2008. 16. Corey, E. J., Nicolaou, K. C., Machida, Y., Malmsten, C. L. & Samuelsson, B. (1975) Proc. Natl. Acad. Sci. USA 72, 3553358. 17. Smith, J. B., Ingerman, C., Kocsis, J. J. & Silver, M. J. (1974) J. Clin. Invest. 53, 1468-1472. 18. Green, K. & Samuelsson, B. (1964) J. Lipid Res. 5, 117-120. 19. Pace-Asciak, C. (1976) Experientia 32,291-292. 20. Fales, H. M., Jaouni, T. M. & Babashak, J. F. (1973) Anal. Chem. 45,2302-2304. 21. Saba, S. R. & Mason, R. G. (1974) Thromb. Res. 5,747-757. 22. Harker, L. A., Striker, G. E., Wall, R. T. & Quadracci, L. J. (1977) Clin. Res. 25, 518A. 23. Gorman, R. R., Bunting, S. & Miller, 0. V.

(1977) Prostaglandins Tateson, J. E., Moncada, S. & Vane, J. R. (1977) Prostaglandins 13,389-397. Minkes, M., Stanford, N., Chi, M.-Y., Roth, G. J., Raz, A., Nee13,377-388.

24.

25.

dleman, P. & Majerus, P. W. (1977) J. Clin. Invest. 59, 449454. 26. Ross, R. & Glomset, J. A. (1976) N. Engl. J. Med. 295,369-377,

420-425.