Response of Human Coronary Arteries to Aggregating Platelets: Importance of Endothelium-Derived Relaxing Factor and Prostanoids

Transcription

1 306 Response of Human to Aggregating Platelets: Importance of Endothelium-Derived Relaxing Factor and Prostanoids Ulrich Forstermann, Andreas Miigge, Stefanie M. Bode, and Jiirgen C. Frolich Epicardial human coronary arteries (HCA) were obtained during heart transplantation. Strip preparations were mounted for isometric tension recording. Tension was induced with prostaglandin F^ (0.3-3 fim). Aggregating human platelets (107ml and K^/ml) caused marked relaxation if the endothelium of the HCA was intact, but produced modest additional constriction if the endothelium was removed. The endothelium-dependent relaxations to platelets were slightly enhanced in the presence of methysergide (1 fim) or after treatment of the platelets with the thromboxane synthase inhibitor dazmegrel (1 mm, 30 minutes). Relaxations to platelets were markedly inhibited or abolished in the presence of apyrase (1 unit/ml), after selective pretreatment of HCA with gossypol or methylene blue (both 30 /im, 30 minutes), or after addition of hemoglobin (20 fim) to the organ bath. Pretreatment of HCA (not the platelets) with aspirin (30 fim, 30 minutes) had no significant effect on platelet-induced relaxations. Adenosine 5'-diphosphate ( fim) induced marked relaxations in endothelium-intact and much smaller relaxations in endothelium-denuded HCA. A low concentration (10 nm) of serotonin (5-HT) produced modest endothelium-dependent relaxations, higher concentrations (0.1-1 fim) led to increases in tension (also in the presence of endothelium). The thromboxane A 2 mimetic U44069 (1-100 nm) was the most potent constrictor of HCA irrespective of the presence or absence of endothelium. After inhibition of thromboxane synthase, platelets produced large amounts of prostaglandins E^ and F^. Both prostaglandins constricted HCA, which can explain the limited effects of dazmegrel. These data suggest that adenine nucleotides released by aggregating platelets stimulate the endothelial cell layer of HCA to produce endothelium-derived relaxing factor. The factor is able to override the direct vasoconstrictor effects of other platelet products (5-HT, thromboxane A 2, and prostaglandin EJ and may be an important defense mechanism against coronary vasospasm. (Circulation Research 1988; 63: ) Spontaneous vasospasm of human coronary arteries (HCA), associated with angina, has been documented clinically. 1 Atherosclerotic HCA may be predisposed to vasospasm, 12 and inhibition of the endothelium-dependent vasodilator mechanism in atherosclerosis may be one causal factor favoring these events. 3-4 Under clinical conditions, the spasms can be complicated by thrombosis leading to vessel occlusion and myocar- From the Department of Clinical Pharmacology (U.F., S.M.B., and J.C.F.) and the Department of Cardiology (A.M.) at the Hannover Medical School, Hannover, FRG. Supported by Grant Fo 144/1-2 from the Deutsche Forschungsgemeinschaft. Address for reprints: Dr. Ulrich F6rstermann, Department of Clinical Pharmacology, Hannover Medical School, PO Box , D-3000 Hannover 61, FRG. Received August 10, 1987; accepted February 11, dial infarction. Therefore, interactions between platelets and the vessel wall of HCA may be of great importance. Recently, aggregating platelets have been shown to induce endothelium-mediated vasodilation in canine coronary arteries. Arteries denuded of endothelium contracted to platelets. 3-6 However, canine coronary arteries may differ from HCA with respect to their response to certain endotheliumdependent vasodilators. Endothelium-intact canine coronary arteries always relax to acetylcholine. 5 In intact HCA, the effect of acetylcholine is controversial. Both constricting effects 7-9 and vasodilator effects 310 have been reported. Therefore, this study was designed to investigate the effects of aggregating human platelets on HCA with the endothelium intact or removed. Endothelial cells can produce two potent vasodilators, namely prostacyclin" and endothelium-

2 Forstermann et al Coronary Artery Response to Platelets 307 derived relaxing factor (EDRF, whose active principle is likely to be nitric oxide). 12 -' 4 Both compounds relax HCA. 4 Also aggregating platelets form and/or release a number of vasoactive substances. Among these are not only vasodilators like adenine nucleotides but also vasoconstrictors like serotonin and thromboxane A 2."- 15 Also other prostanoids (like prostaglandins Ej and F2J can be formed by platelets under certain conditions. 16 Therefore, the second goal of this study was to characterize the substances involved in the vascular responses observed. Materials and Methods Preparation of Human Human epicardial coronary arteries (2-3 mm outer diameter) were obtained from recipient hearts during heart transplantation. Patients (aged years) received heart transplantation for congestive cardiomyopathy or, in one case, for coronary artery disease. The arteries were obtained 1-5 minutes after removal of the heart from the patient and immediately rinsed and immersed in ice-cold modified Krebs solutions. After removal of adventitial fat and connective tissue, transverse strips (cutopen rings of about 1 mm width) were prepared from the arteries; 95 strips from 13 patients were studied. Arteries with macroscopic atherosclerotic lesions (such as plaques and lipid and calcific debris in the intima) were not used in this study. Functional endothelial integrity was tested in each preparation by exposing it to the endotheliumdependent relaxant substance P (for details see below). Organ Bath Experiments With Human During the preparation of the arterial strips, any contact with the luminal surfaces was avoided to preserve endothelial integrity. Strips were suspended in 5-ml organ baths filled with oxygenated (95% O r 5% CO 2 ) modified Krebs solution (ph 7.4, 37 C) of the following composition (mm): Na , K , Ca , Mg , Cl" , HCO3-25.0, H 2 PO 4-1.2, SCV" 1.2, glucose 10.7, and disodium EDTA The vascular preparations were connected to force transducers for isometric tension recording; resting force was 1.5 g. The strips were contracted with prostaglandin (PG) F^ (0.3-3 fim) to induce a tension of 1-2 g. After the tension had stabilized (plateau usually reached after 7-10 minutes) the endothelium-dependent relaxing agent substance P (1 nm) was added to the organ bath as a functional test of endothelial integrity. Substance P (Serva, Heidelberg, FRG) was dissolved (1 mg/ml) in phosphate-buffered saline (8.1 mm Na 2 HPO 4, 1.9 mm NaH 2 PO 4, 150 mm NaCl, ph 7.4) containing 0.1% (wt/vol) gelatin. It was added to the organ bath in 50 fil. Only preparations showing more than 70% relaxation to this concen- +Endothellum D 108 SP {log M) Platelets (1/ml} FIGURE 1. Effect of substance P (1 nm), used as a functional test of endothelial integrity, and human platelets (10 7 /ml and 10'Iml) on strips of human coronary arteries with the endothelium intact (+Endothelium, 44 strips from seven patients) or removed ( Endothelium, 22 strips from four patients). Response is expressed as percent deviation from the contraction plateau induced by prostaglandin F 2a (PGF 2a, fim). Induced tension was 1.5 ±0.1 g in endothelium-intact and 1.3 ±0.1 g in endothelium-denuded preparations. *Significant differences between + Endothelium and Endothelium (p<0.05). tration of substance P were considered endotheliumintact (as previously determined 4 ) and used in the subsequent study (Figure 1). From other strips the endothelium was carefully removed with the blunt side of a scalpel (rubbing procedure). These rubbed preparations showed no relaxation to substance P (1 nm) (Figure 1). The presence of endothelium or the completeness of endothelial removal was confirmed in sample preparations by en face light microscopy after silver staining (method modified from Poole et al 17 ). In endothelium-intact preparations (as judged from the functional test) endothelium was preserved by at least 70%. Preparation of Platelets for the Organ Bath Experiments Fresh blood, anticoagulated with trisodium citrate (10.6 mm final concentration) was centrifuged at 200g for 15 minutes. The supernatant was recentrifuged at 400g for 25 minutes. This second supernatant (plasma) was discarded. The platelets (pellet) were suspended in phosphate-buffered saline containing 10.6 mm trisodium citrate and were recentrifuged (washing). The resulting pellet was resuspended in a small volume of the above buffer mixture, counted and diluted with further Ca 2+ -free buffer to yield a platelet count of 5 x 10 9 /ml. Then, 10 fi\ and 100 /xl of this platelet suspension were injected into the 5 ml organ bath resulting in final platelet counts of 10 7 /ml and lovml, respectively. Drugs Used in the Organ Bath Experiments The effect of aggregating platelets on PGF^preconstricted HCA was also tested in the presence

3 308 Circulation Research Vol 63, No 2, August 1988 HPL Endothelium PGF 2(/ -6 PGF 2a -6 2 min HPL 10 PGF 2f, -6 PGF 2 Moth -6 FIGURE 2. Representative traces of isometric tension recordings of strips of human coronary arteries with the endothelium intact (+Endothelium, upper panel,) or removed (, lower panel,). Human platelets (HPL, 10 7 lml and W'/ml) were added to the coronary tissues under tension induced by prostaglandin F 2a (PGF 2a, 1 fim). The same experiment was repeated with HPL treated for 30 minutes with the thromboxane synthase inhibitor dazmegrel (Daz, 1 mm), with normal HPL in the presence of the serotonin receptor antagonist methysergide (Meth, 1 fim), and with normal HPL in the presence of the ATP and ADP hydrolyzing enzyme apyrase (Apyr, 1 unit/ml ADPase). Concentrations of drugs are indicated as logarithms of molar concentrations; w, wash, change of bath medium. of the serotonin receptor antagonist methysergide (1 ^,M dimaleinate, Sandoz, Basel, Switzerland). Methysergide was dissolved (1 mg/ml) in distilled water. Further, the effect of platelets was tested in the presence of apyrase Grade VIII (Sigma, Munich, FRG; spec. act. 250 unit/mg ATPase, 54 unit/mg ADPase, <0.1 unit/mg AMPase). The enzyme was dissolved in phosphate-buffered saline; the final bath concentration was 1 unit/ml ADPase. Activity was calculated in terms of ADPase, as ADP is the predominant adenine nucleotide released by human platelets. 15 The compounds were added to the organ baths as shown in Figure 2. In other experiments, the platelets were pretreated with the thromboxane synthase inhibitor dazmegrel (1 mm for 30 minutes) prior to addition to the organ bath. Dazmegrel (Pfizer, Munich, FRG) was dissolved (3 mg/ml) in 10 mm NaOH and the ph of the solution was subsequently adjusted to 7.5 with 1 M HC1. The effects of aggregating platelets were also tested after selective treatment of HCA with the irreversible cyclo-oxygenase inhibitor aspirin (30 fim, 30 minutes), with the inhibitor of EDRF production and/or release gossypol (30 fim, 30 minutes), with the inhibitor of soluble guanylate cyclase, methylene blue (30 ^M, 30 minutes), or after addition of hemoglobin (20 fim) to the organ bath. Aspirin was used as the water-soluble lysinemonoacetylsalicylate (Bayer, Leverkusen, FRG). Gossypol (Sigma) was dissolved in DMSO and added to the bath solution in 10 /xl. The resulting DMSO concentration of 0.2% had no influence on the subsequent contraction-relaxation processes. Methylene blue (Sigma) was dissolved in distilled water. Human oxyhemoglobin was prepared by reducing the commercial hemoglobin/methemoglobin mixture (Sigma) with sodium dithionite as described by Martin et al. 18 Gossypol, methylene blue, and aspirin were removed from the organ bath and the HCA strips were washed several times with fresh Krebs' solution before the second exposure to platelets (Figure 4). Hemoglobin was added to the organ bath 3 minutes prior to the platelets and remained in the organ bath together with the platelets. Serotonin (HC1, Sigma) was dissolved (1 mg/ml) in 1 mm HC1 containing 1 mg/ml ascorbic acid; adenosine 5'-diphosphate (ADP, Sigma) was dissolved in phosphate-buffered saline. The stable thromboxane A 2 mimetic U44069, and the prostaglandins F^ and E2 (Upjohn, Kalamazoo, Michigan) were dissolved in 70% (vol/vol) ethanol. Glyceryl trinitrate (10% triturated mixture in lactose; Mack, Illertissen, FRG) was dissolved in distilled water.

4 Forstermann et al Coronary Artery Response to Platelets 309 Appropriate dilutions of these compounds in water were added to the organ bath in 5-50 /il. Statistics All data are given as mean±sem. Differences between responses of HCA with and without endothelium or before and after inhibitors were analysed for statistical significance using one way analysis of variance followed by the Fisher least significant difference test for the comparison of different means. 19 A p value <0.05 was considered significant. Results Effect of Human Platelets on Human When human platelets (prepared in Ca 2+ -free medium as described in "Materials and Methods") were injected into the organ bath containing a strip of human coronary artery, they aggregated spontaneously upon exposure to the Ca 2+ in the Krebs solution, to the collagen of the cut strip preparation, and to the glass of the organ bath. 6 The turbid suspension became clear within minutes. These aggregating human platelets produced concentrationdependent relaxations of strips of HCA under induced tension, if the endothelium was intact (Figures 1 and 2). HCA strips without endothelium showed either no response to aggregating platelets or small additional contractions were observed (Figures 1 and 2). The endothelium-dependent relaxation of HCA to platelets was slightly enhanced in the presence of the serotonin receptor antagonist methysergide (1 fim) (Figures 1 and 3). Methysergide (1 fim) sometimes produced modest increases in tension of HCA when added to the organ bath (Figure 2), which may be due to partial agonist activity at the 5-HT 2 receptor of smooth muscle cells. The endothelium-dependent relaxation of HCA to platelets was also slightly enhanced after pretreatment of the platelets with the thromboxane synthase inhibitor dazmegrel (1 mm, 30 minutes) (Figures 2 and 3, top panels). Dazmegrel (at the resulting approximate organ bath concentration of 10" 3 M) had no direct effect on tension. Dazmegrel treatment did not inhibit platelet aggregation as judged from the clearing of the turbid platelet suspension and from the increased effect of these platelets on endothelium-intact HCA (Figures 2 and 3). Apyrase is an enzyme that hydrolyzes ATP and ADP to AMP and adenosine. In the presence of apyrase, (1 unit/ml ADPase) the endotheliumdependent relaxations to platelets were markedly inhibited and sometimes even converted into contractions (Figures 2 and 3). Selective pretreatment of endothelium-intact HCA with the inhibitor of EDRF production and/or release, gossypol (30 /im, 30 minutes), abolished the endothelium-dependent relaxations to platelets or converted them into contractions (Figure 4). Also o o. c o 2 I I PUteleU {1/ml} Endothelium O Control +Apyrase +Dazmegrel O +Methyserglde <> Control * +Apyrase -fdazmegrel O +Methysergkte Platelets {1/ml} FIGURE 3. Effect of human platelets (HPL, 10 7 lml and 10'lml) on strips of human coronary arteries with the endothelium intact ( + Endothelium) or removed (). Response is expressed as percent deviation from the contraction plateau induced by prostaglandin F 2a (PGF 2a, (JLM). Experiments were performed with HPL alone (control), with untreated platelets in the presence of apyrase (1 unit/ml ADPase), with platelets treated for 30 minutes with dazmegrel (1 mm), and with untreated platelets in the presence of methysergide (1 fxm). Data are given as mean±sem of strips from four different patients. Drugs were added as shown in Figure 2. Some y-axis error bars have been omitted for clarity. pretreatment of HCA with the inhibitor of soluble guanylate cyclase, methylene blue (30 fim, 30 minutes), blocked platelet-induced relaxations (Figure 4). Relaxations were also inhibited in the presence of hemoglobin (20 jim), which is likely to bind and inactivate EDRF 18 (Figure 4). In contrast, pretreatment of the HCA with aspirin (30 ^M, 30 minutes) had no significant effect on platelet-induced endothelium-dependent relaxation (Figure 4). Effect of Platelet Products on Human In further experiments, preconstricted strips of HCA with the endothelium intact or removed were exposed to ADP (the major adenine nucleotide released by human platelets), to serotonin, and to

5 310 Circulation Research Vol 63, No 2, August 1988 a. c o +Endothelium s D Endothelium I Co Asp Goss MB Hb FIGURE 4. Effect of aggregating platelets (Kflml and 108/ml) on relaxations of endothelium intact strips of human coronary artery treated with different inhibitors (the platelets were not exposed to these inhibitors). Strips were preconstricted with prostaglandin F2a, (PGF2a, 0.33 i*m). The inhibitors were aspirin (Asp, 30 fjm); gossypol (Goss, 30 fim); methylene blue (MB, 30 fim); and hemoglobin (Hb, 20 ym). Aspirin, gossypol, and methylene blue were added to the organ bath for 30 minutes. They were removed from the organ bath by repeated washing with fresh Krebs solution before the second exposure of the coronary arteries to platelets (first exposure was before the inhibitors). Hemoglobin was added to the preconstricted strips 3 minutes prior to the platelets and remained in the bath together with the platelets. Columns represent mean±sem of 5-13 strip preparations from three to five different patients. Control columns (Co) are pooled from controls for all inhibitors. *Significant difference from respective control (p<0.05). the thromboxane A2 mimetic U ADP ( nm) produced concentration-dependent relaxations that were markedly reduced after removal of endothelial cells (Figure 5). A low concentration of serotonin (10 nm) produced modest endotheliumdependent relaxations in some HCA preparations; however, at higher concentrations (0.1-1 /xm), serotonin always increased the tension of HCA. Endothelium-denuded HCA showed additional constriction to all concentrations of serotonin (Figure 5). The thromboxane A2 mimetic U44069 (1-100 nm) increased the tension of HCA, irrespective of the presence or absence of endothelium (Figure 5). Glyceryl trinitrate (10 or 100 nm), a vasodilator that does not require an intact endothelium to elicit its relaxing effect in animal arteries,20 produced similar relaxations in endothelium-intact and denuded HCA (not shown). Effect of Prostanoids on Human When added to endothelium-intact HCA under resting tension, the thromboxane A2 mimetic U44069 proved to be the most potent vasoconstrictor prostanoid. It was almost 100 times more potent than (Figure 6). Surprisingly, PGF^ (which is the m Endothelium Endothellum -7 U (log M) FIGURE 5. Effect of different compounds (released during platelet aggregation) on strips of human coronary arteries with the endothelium intact (+Endothelium) or removed ( Endothelium). Strips were preconstricted with prostaglandin F2a (PGF2a, 1 fim) and tested with adenosine 5'-diphosphate (ADP), serotonin (5-HT), and the stable thromboxane A2 mimetic U All data are given as mean±sem of 7-14 strips from three to four different patients. ^Significant differences between +Endothelium and (p<0.05). major prostanoid formed by platelets after inhibition of thromboxane synthase16) proved to be equipotent to PGFi, in contracting HCA. However, its efficacy was lower than that of the other two prostanoids (Figure 6). Discussion Aggregating platelets, at concentrations lower than those normally present in blood, produced

6 Forstermann et al Coronary Artery Response to Platelets a u O PGF 2a PGE-, Prostanold {log M} FIGURE 6. Effect of three different prostanoids (formed during platelet aggregation) on endothelium-intact human coronary arteries under resting tension: the stable thromboxane A 2 mimetic U44069; prostaglandin F 2a (PGF^); prostaglandin E 2 (PGE 2 ). Response to I ym PGF 2a was always set 100% (=1.6±0.2 g). Data are given as mean±sem of 7-20 strips from two to four patients. relaxations of HCA that were strictly dependent on the presence of vascular endothelium. The inability of endothelium-denuded HCA to relax to aggregating platelets is unlikely to be due to damage of vascular smooth muscle, as relaxations to glyceryl trinitrate were unimpaired after the rubbing procedure. The endothelium-dependent relaxations to platelets were abolished if HCA was treated with the inhibitor of EDRF production and/or release, gossypol, or with the inhibitor of soluble guanylate cyclase, methylene blue 24 (Figure 4). It was also blocked by hemoglobin, which binds and inactivates EDRF. 18 These findings strongly suggest that the relaxations are mainly, if not exclusively, mediated by EDRF. Prostacyclin, which is also a potent vasodilator in HCA, 4 is unlikely to make a major contribution to this relaxation, because treatment of HCA (not the platelets) with aspirin did not have a significant effect on the platelet-induced endothelium-mediated relaxations (Figure 4). Treatment of the platelets with aspirin was not done in this model because it markedly delayed or prevented the spontaneous platelet aggregation (see "Materials and Methods"). The three major platelet products secreted or synthesized during aggregation of human platelets are adenine nucleotides (mainly ADP), serotonin, and thromboxane A ADP was a vasodilator in HCA and, similar to its action on animal arteries, exerted most of its relaxant effect through an action on endothelial cells (Figure 5). Apyrase (which hydrolyzes ATP and ADP to AMP and adenosine) produced partial inhibition of the relaxant effect of the low platelet concentration (10 7 /ml). The relaxant effects of the high platelet concentration (lovm!) was blocked completely or even converted into a contraction (Figures 2 and 3). This indicates that adenine nucleotides are likely to be the principal mediators stimulating the EDRF production in the endothelium of HCA. At low platelet concentrations (lovml), there may be a participation of serotonin in this relaxing effect, as low concentrations of serotonin were found to produce moderate endothelium-dependent relaxations (Figure 5). This could explain why apyrase did not block this relaxation completely. At higher platelet concentrations (lo^ml), the contribution of serotonin to the sum response of HCA seems to be vasoconstriction because the endothelium-dependent relaxation to the platelets was enhanced rather than inhibited in the presence of the serotoninreceptor antagonist methysergide (Figures 2 and 3). Methysergide has been reported to cause partially insurmountable, noncompetitive antagonism of 5- HT 2 receptor-mediated contractions in calf coronary arteries. 27 In endothelium-denuded HCA, the small endothelium-independent relaxing effect of ADP (Figure 5) is apparently overridden by the vasoconstrictor compounds serotonin and thromboxane A 2 (Figures 2 and 3). The thromboxane A 2 mimetic U44069 was found to be a potent vasoconstrictor irrespective of the presence or absence of endothelium (Figure 5). However, treatment of human platelets with the thromboxane synthase inhibitor dazmegrel produced only a moderate increase in the endothelium-dependent relaxation to platelets (Figure 3) and slightly attenuated the platelet-induced vasoconstriction in endothelium-denuded HCA (Figure 3). These modest effects are surprising in view of the potent constrictor action of compound U44069 (and thromboxane A 2, respectively). However, it has been reported previously that complete inhibition of thromboxane synthase in human platelets leads to increased production of other cyclo-oxygenase products (like cyclic endoperoxides and PGEz and PGF^.) 16 Platelet aggregation was not inhibited under these conditions (as confirmed in the present experiments), probably because other proaggregatory cyclo-oxygenase products can take over the functions of thromboxane A The same phenomenon may explain why treatment of human platelets with dazmegrel only slightly shifted the balance of their vasoactive effects towards vasodilation (Figure 3). Both PGF^ and PGF^ are vasoconstrictors in HCA (Figure 6), and the same probably applies to cyclic endoperoxides. Therefore, our findings suggest that inhibition of thromboxane synthase may be of limited benefit for the prevention of vasospasm in HCA (as it is for the prevention of platelet aggregation). In conclusion, the present results confirm findings previously reported in animal models 56 for human platelets and HCA. They demonstrate that endothelium-dependent relaxations of HCA to human platelets are likely to be mediated mainly by EDRF. The EDRF produced in response to aggregating platelets overrides the direct vasoconstrictor effect of the platelet products thromboxane A 2 and serotonin, the

7 312 Circulation Research Vol 63, No 2, August 1988 sum effect being vasodilation. Endothelial damage or dysfunction would markedly shift the balance of vasoactive compounds toward vasoconstriction. We therefore speculate that the ability of the vascular intima to produce EDRF could be an important mechanism in the prevention of coronary vasospasm. Note added in proof: During the revision process of this manuscript we have measured prostanoid production of aggregating platelets using specific radioimmunoassays. Aggregation was induced in the washed, decalcified platelets by addition of Ca 2+ (1.7 mm) and collagen (95 /xg/ml). Control platelets produced 12.7 ±2.9 ng/10 8 platelets/4 min thromboxane B 2, 0.47 ±0.12 ng PGF^ and 0.22 ±0.07 ng PGE^ (means ±SEM, n = 8). Platelets treated for 30 minutes with dazmegrel (1 mm) only produced 0.24 ± 0.04 ng thromboxane B 2, but 14.8±4.3 ng PGF^ and 17.2 ±4.5 ng PGE? (mean±sem, n = 8). This confirms the above suggestion that other constrictor prostanoids can take over the function of thromboxane A 2 after exposure of platelets to dazmegrel. Acknowledgments We are indebted to Professor H. G. Borst and the Cardiac Surgery Transplant Team of Hannover Medical School for their help in procurement of the human coronary arteries studied. We are also grateful to the staff of the Blood Transfusion Center of Hannover Medical School for the supply of fresh blood conserves for the platelet experiments. The technical assistance of Mrs. K. Burgwitz and Mrs. I. Reich wehr and the secretarial help of Mrs. B. Thiirnau are gratefully acknowledged. References 1. Schroeder JS, Bolen JL, Quint RA, Clark DA, Hayden WG, Higgins CB, Wexler L: Provocation of coronary spasm with ergonovine maleate. Am J Cardiol 1977;40: Waters DD, Szlachcic J, Boran R, Miller DD, Dauwe F, Theroux P: Comparative sensitivity of exercise, cold pressor and ergonovine testing in provoking attacks of variant angina in patients with active disease. Circulation 1983;67: Bossaller C, Habib GB, Yamamoto H, Williams C, Wells S, Henry PD: Impaired muscarinic endothelium-dependent relaxation and cyclic guanosine 5'-mono-phosphate formation in atherosclerotic human coronary artery and rabbit aorta. J Clin Invest 1987;79: Forstermann U, Mugge A, Alheid U, Haverich A, Frdlich JC: Selective attenuation of endothelium-mediated vasodilation in atherosclerotic human coronary arteries. Ore Res 1988;62: Houston DS, Shepherd JT, Vanhoutte PM: Adenine nucleotides, serotonin, and endothelium-dependent relaxation to platelets. Am J Physiol 1985;248:H389-H Houston DS, Shepherd JT, Vanhoutte PM: Aggregating human platelets cause direct contraction and endotheliumdependent relaxation of isolated canine coronary arteries. J Clin Invest!986;78: Fflrstermann U, MOgge A, Frolich JC: Endotheliumdependent relaxation of human epicardial coronary arteries: Frequent lack of effect of acetylcholine. Eur J Pharmacol 1986;128: Ku DD, Caulfield JB: Human coronary aortic bypass grafts, but not coronary arteries, responded to acetylcholine with endothelium-dependent vasorelaxation (abstract). Circulation 1987;76(suppl IV):IV Newman CM, Hackett DR, Fryer M, El-Tamini HM, Davies GJ, Maseri A: Dual effects of acetylcholine on angiographically normal human coronary arteries in vivo (abstract). Circulation 1987;76(suppl IV):IV Ludmer PL, Selwyn AP, Shook TL, Wayne RR, Mudge GH, Alexander RW, Ganz P: Paradoxical vasoconstriction induced by acetylcholine in atherosclerotic coronary arteries. N Engl J Med 1986^15: Moncada S, Vane JR: Pharmacology and endogenous roles of prostaglandin endoperoxides, thromboxane A 2 and prostacyclin. Pharmacol Rev 1979^0: Furchgott RF, Kahn MT, Jothianandan D: Comparison of endothelium-dependent relaxation and nitric oxide-induced relaxation in rabbit aorta (abstract). Fed Proc 1987;46:396A 13. Kahn MT, Furchgott RF: Similarities of behaviour of nitric oxide (NO) and endothelium-derived relaxing factor in a perfusion cascade bioassay system (abstract). Fed Proc 1987;46:397A 14. Palmer RMJ, Ferrige AG, Moncada S: Nitric oxide release accounts for the biological activity of endothelium-derived relaxing factor. Nature 1987;327: Meyers KM, Holmsen H, Seachord CL: Comparative study of platelet dense granule constituents. Am J Physiol 1982; 243:R454-R BerteK V, Falanga A, Tomasiak M, Dejana E, Cerletti C, de Gaetano G: Platelet thromboxane synthetase inhibitors with low doses of aspirin: Possible resolution of the "aspirin dilemma." Science 1983^220: Poole JCF, Sanders AG, Florey UW: The regeneration of aortic endothelium. J Pathol Bacteriol 1958;75: Martin W, Villani GM, Jothianandan D, Furchgott RF: Selective blockade of endothelium-dependent and glyceryl trinitrate-induced relaxation by hemoglobin and by methylene blue in the rabbit aorta. J Pharmacol Exp Ther 1985;232: Snedecor GW, Cochran WG: One way classificationsanalysis of variance, in Snedecor GW, Cochran WG (eds): Statistical Methods, ed 6. Ames, Iowa, Iowa State University Press, 1967, pp Furchgott RF: The role of endothelium in the response of vascular smooth muscle to drugs. Annu Rev Pharmacol Toxicol : Forstermann U, Goppelt-StrQbe M, FrOlich JC, Busse R: Inhibitors of acyl-coenzyme A: Lysolecithin acyltransferase activate the production of endothelium-derived vascular relaxing factor. J Pharmacol Exp Ther 1986;238: Busse R, Fflrstermann U: Inhibition of endothelial lysolecithin acyltransferase stimulates the formation of endotheliumderived relaxant factor, in Vanhoutte PM (ed): Mechanisms of Vasodilation IV. New York, Raven Press, Publishers, 1987 (in press) 23. F6rstermann U, Alheid U, FrOlich JC, Mulsch A: Mechanisms of action of lipoxygenase and cytochrome P-450-monooxygenase inhibitors in blocking endothelium-dependent vasodilatation. Br J Pharmacol 1988;93: Ignarro LJ, Kadowitz PJ: The pharmacological and physiological role of cyclic GMP in vascular smooth muscle relaxation. Annu Rev Pharmacol Toxicol 1985;25: Vanhoutte PM, Rubanyi GM, Miller VM, Houston DS: Modulation of vascular smooth muscle contraction by the endothelium. Annu Rev Physiol 1986;48: Rapoport RM, Draznin MB, Murad F: Mechanism of adenosine triphosphate-, thrombin-, and trypsin-induced relaxation in rat thoracic aorta. Circ Res 1984,55: Kaumann AJ, Frenken M: A paradox: The 5-HT r receptor antagonist ketanserin restores the 5-HT-induced contraction depressed by methysergide in large coronary arteries of calf. Naunyn Schmiedebergs Arch Pharmacol 1985^28: KEY WORDS human coronary arteries human platelets endothelium-derived relaxing factor thromboxane prostaglandin E 2 (dinoprostone) aspirin gossypol methylene blue hemoglobin