Potassium-Induced Release of Endothelium- Derived Relaxing Factor From Canine Femoral Arteries

1098 Potassium-Induced Release of Endothelium- Derived Relaxing Factor From Canine Femoral Arteries Gabor M. Rubanyi and Paul M. Vanhoutte Downloaded from http://ahajournals.org by on January 13, 2019 Experiments were designed in a bioassay system to analyze the effect of elevated (from 5.9 mm to 7.5-45.9 mm) extracellular K + concentration on the release of endothelium-derived relaxing factor. Segments of canine femoral artery with endothelium (donor segment) were mounted in an organ bath and perfused with modified Krebs-Ringer bicarbonate solution; the effluent from the donor segment was used to superfuse a canine coronary artery ring without endothelium (bioassay tissue). Elevation of perfusate K + concentration by 1.6-15 mm by intraluminal infusion of potassium chloride upstream of the donor segment evoked further contractions of bioassay rings contracted with prostaglandin F^. In contrast, the bioassay rings progressively relaxed when increasing concentrations of potassium chloride (10-40 mm) were added extraluminally to the organ bath where the perfused segment was mounted. Extraluminal application of phenylephrine or prostaglandin F&, did not evoke relaxations in the bioassay ring. Removal of the endothelium from the donor segment or selective exposure of the segment (but not the bioassay ring) to Ca 2+ -deficient solution prevented the K + - induced relaxations. Treatment of the donor segment and the bioassay ring with inhibitors of known endogenous vasoactive substances (acetylcholine, norepinephrine, adenine nucleotides, and prostanoids) had no significant effect on the relaxation of the bioassay ring evoked by extraluminal application of potassium chloride. Simultaneous measurements of changes in isometric force in the donor segment and bioassay ring revealed that extraluminal elevation of K concentration relaxed the segments as well and that the relaxations could not be prevented by simultaneous intraluminal infusion of potassium chloride. Prolongation of the transit time between the donor segment and bioassay ring from 1 to 6 seconds depressed the relaxations of the bioassay tissue evoked by extraluminal elevation of K concentrations that could be prevented by superoxide dismutase (infused downstream of the donor segment). These findings indicate that K can stimulate the release of endothelium-derived relaxing factor from perfused canine femoral arteries but only when applied extraluminally. It is postulated that K triggers the release of a still unidentified mediator in the blood vessel wall that stimulates the release of the relaxing factor from the endothelial cells. (Circulation Research 1988;62:1098-1103) The endothelium plays an important role in modulating the tone of underlying vascular smooth muscle cells by releasing a relaxing substance (endothelium-derived relaxing factor [EDRF]) in response to stimulation by a variety of vasoactive agents. 12 Although extracellular Ca 2+ is required for the production and release of the relaxing factor, 3 " 7 the characteristics of the endothelial Ca 2+ channels involved in the response are still uncertain. Voltage-operated Ca 2+ channels can play a role in the transloeation of Ca 2+ in the endothelial cells, since the dihydropyridine calcium channel agonists Bay K 8644 and ( + )-202,791 8 " trigger the release of EDRF from canine femoral arteries. 10 Hence, activation of Ca 2+ transloeation through voltage-dependent Ca 2+ channels by elevation of extracellular K concentration (which evokes membrane depolarization in endothelial cells)" should also trigger the release of the relaxing factor from the endothelial cells. The present study From the Department of Physiology and Biophysics, Mayo Clinic, Rochester, Minnesota. Supported in part by National Heart, Lung, and Blood Institute grants HL-31183, HL-34634, and HL-31547. Address for correspondence: Gabor M. Rubanyi, MD, PhD, Beriex Laboratories Inc., 110 East Hanover Avenue, Cedar Knolls, NJ 07927. Received September 16, 1987; accepted January 6, 1988. was designed to test this possibility. It determines the effects of elevated extracellular concentration of K + on the release of EDRF from canine femoral arteries. Materials and Methods Experiments were carried out on femoral and left circumflex coronary arteries isolated from mongrel dogs (15-25 kg) anesthetized with sodium pentobarbital (30 mg/kg). The blood vessels were cleaned of loose adipose and connective tissue and studied in modified Krebs-Ringer bicarbonate solution (used as both the perfusate and bathing solution) with the following composition (mm): NaCl 118, KC1 4.7, CaCl 2 2.5, KH 2 PO 4 1.2, NaHCO 3 25.0, edetate calcium disodium 0.026, and glucose 11.1 (control solution). In some experiments, calcium chloride was omitted from the control buffer (Ca 2+ -deficient solution). Because pilot studies (n = 4) showed that inhibition of cyclooxygenase by indomethacin had no significant effect on K + -induced relaxation of the bioassay rings, the buffer solution in the present experiments always contained indomethacin (10~ 3 M) to prevent the synthesis of prostanoids. Bioassay Studies A bioassay technique described in detail elsewhere 12 was used. Side branches of a segment (3.0-3.5 cm

long) of the femoral artery were tied. In some segments, the endothelium was intentionally removed by gentle mechanical rubbing of the intimal surface. The segments were fixed to stainless steel cannulas (1.5 mm i.d.), placed in an organ chamber that was maintained at 37 C, and perfused at constant flow (2 ml/min) with control solution gassed (95% O 2-5% CO2) and heated (37 C) in a separate reservoir. A stainless steel tube was also placed in the organ chamber through which control solution was pumped at a rate of 2 ml/min from the same reservoir. A ring of coronary artery, in which the endothelium had been removed (bioassay ring), was suspended directly below the organ chamber by means of two stainless steel stirrups passed through its lumen. One stirrup was fixed, and the other was connected to a force transducer (model FT03D, Grass Instruments, Quincy, Massachusetts) for recordings of isometric force. The assembly of bioassay ring, stirrups, and force transducer could be moved freely below the organ chamber, allowing the preparation to be superfused with the perfusate from either the femoral artery with endothelium (endothelial superfusion) or without endothelium (vascular superfusion) or with the stainless steel tube (direct superfusion). The transit time between the perfused femoral artery and the superfused coronary artery ring was 1 second. In some experiments, the transit time was prolonged to 6 seconds. In separate experiments, a modified bioassay system was used for the simultaneous measurement of changes in isometric force generated by the perfused segment and the superfused bioassay ring (Figure 1). Two stainless steel stirrups were inserted into the wall of the femoral artery segment before mounting it horizontally in an organ chamber (50 ml, 37 C). One stirrup was anchored to the bottom of the chamber; the other was connected to a force transducer. Drugs and potassium chloride were either infused into the perfusate by means of Harvard infusion pumps (model 901, South Natick, Massachusetts) upstream Rubanyi and Vanhoutte K + -Induced Release of EDRF 1099 of the femoral artery (intraluminal application at site 1, Figure 1) or added to the organ chamber (extraluminal application). During intraluminal application, both the bioassay ring and the luminal surface of the donor segment were exposed to drugs and potassium chloride, but when they were applied extraluminally into the organ bath, only the adventitial surface of the donor segment was exposed to them. In some experiments, calcium chloride or superoxide dismutase was infused into the perfusate downstream of the donor segment (site 2, Figure 1) to avoid contact with the endothelium. The bioassay ring was superfused through the stainless steel tube with control solution for 60 minutes. During this interval, the ring and the perfused segment were stretched in a stepwise manner until the basal tension reached approximately 10 g, the optimal tension for active contraction in these vascular preparations. 1314 Drugs The following drugs were used: acetylcholine, apyrase, atropine, indomethacin, phenylephrine, dl-propranolol, prostaglandin F^, and superoxide dismutase (all from Sigma Chemical, St. Louis, Missouri). The drugs were dissolved in distilled water except for indomethacin, which was dissolved in a solution of 5 x 10~ 3 M Na 2 CO 3 by sonication. The concentration of the drugs is expressed as final molar concentration in the perfusate or in the solution in the organ chamber. The concentration of potassium chloride added to the perfusate (intraluminal application) or to the organ bath (extraluminal application) is expressed as the change (increase) of the millimolar K + concentration from 5.9 mm (already present in the control solution). Calculations and Statistical Analysis Changes in isometric force of the bioassay ring or donor segment induced by potassium chloride or drugs are expressed as a percentage of the initial contractions Roller pump FIGURE 1. Bioassay apparatus for simultaneous measurement ofabluminal (towards the media of the perfused segment) and intraluminal (detected by the bioassay ring) release of vasoactive substances from endothelial cells. Drugs and chemicals were infused into perfusate either upstream (site 1, intraluminal application) or downstream (site 2) of the perfused segment or were administered into the organ chamber (extraluminal application) where the perfused segment was mounted. Wotrt

1100 Circulation Research Vol 62, No 6, June 1988 evoked by prostaglandin F^ (PGF2J. Data are shown as mean ± SEM. In each series, n is the number of dogs from which arteries were taken. Statistical comparisons were performed by Student's t test for paired or unpaired observations. When p was smaller than 0.05, differences were considered to be statistically significant. Results Intraluminal Application of Potassium Chloride PGF^ (4 X 10~ 6 M; added to the perfusate and organ bath solution) evoked sustained contractions of the bioassay coronary artery rings during endothelial (7.0±0.7 g, n = 11) and vascular (9.2±0.8 g, n = 3) superfusion. An increase of K + concentration of the perfusate to 20.9 mm (by adding 15 mm KCl to the perfusate upstream of the perfused femoral artery) caused comparable further contractions of the bioassay ring during both endothelial (Figure 2A) and vascular (Figure 2C) superfusion. Intraluminal infusion of acetylcholine (10~ 6 M) caused relaxation (84.7 ±5.5% of PGF^ contraction, n = 9) of the bioassay rings during endothelial superfusion (Figure 2A) but only moderate further contractions during vascular superfusion (Figure 2C). Stepwise elevation of the perfusate K concentration by 1.6-15 mm caused comparable concentration-dependent contractions of the bioassay rings during direct and endothelial superfusion (Figure 3). Extraluminal Application of Potassium Chloride Elevation of the extraluminal K + concentration by 15 mm in the organ chamber induced statistically significant and reversible relaxation (34.5 ±11.1%, p<0.05, n = 4) of the bioassay rings during endothelial superfusion (Figure 2B). The relaxation started after a delay of 4.8 ±0.6 minutes (/i = 4). Removal of the PGF,«INTRALUMINAL ENDOTHELIAL SUPERFUSION ACh VASCULAR SUPERFUSION ACh EXTRALUMINAL FIGURE 2. Recording of isometric force of a bioassay coronary artery ring contracted by prostaglandin F 2a (PGF 2a, 4 X 10~ 6 M) during endothelial (Panels A and B) and vascular (Panels C and D) superfusion, demonstrating the effects of intraluminalty (site 1, Panels A and C) or extraluminalty (Panels B and D) administered potassium chloride (K+, 15 mm; total K + concentration was 20.9 mm in the buffer) or acetylcholine (10~ 6 M, ACh). Experiments were carried out in the presence of indomethacin (10~ 5 M). -70 5 10 20 30 40 KCl, mm FIGURE 3. Effect of cumulative increase ofpotassium chloride concentration in perfusate (intraluminal) or in the organ chamber (extraluminal) on the isometric force developed by bioassay coronary rings during direct (A), endothelial (intraluminal, ; extraluminal, o), and vascular (extraluminal, *) superfusion. Experiments were carried out in the presence of indomethacin (10~ s M). Data are shown as mean ± SEM of 3-11 experiments and expressed as percent change of the initial contraction evoked by prostaglandin F 2a [4 xlo~ 6 M PGF 2a, 10O% = 7.8±1.2 g (direct, n = 3); 8.1±0.8 g (vascular, D = 3); 8.2 ±0.6 g (endotheliallintraluminal, n = 9); and 7.0±0.7 g (endotheliallextraluminal, n = endothelium from the perfused femoral artery prevented the relaxation of the bioassay ring in response to extraluminal elevation of the K + concentration (Figure 2D). Extraluminal application of acetylcholine (10~ 6 M) had no effect on the bioassay rings during either endothelial or vascular superfusion (Figures 2B and 2D). Stepwise elevation of the potassium chloride concentration (by 10 40 mm) in the extraluminal organ bath solution produced progressive, concentrationdependent relaxations of the bioassay ring during endothelial superfusion (Figure 3). The relaxations were prevented by removal of the endothelium from the femoral artery (Figure 3). Prolongation of Transit Time and Effect of Superoxide Dismutase Prolongation of the transit time between the femoral artery segment and the bioassay ring from 1 to 6 seconds significantly reduced the relaxation of bioassay rings in response to extraluminal elevation of K + concentration by 20 mm (6.0±3.4%, n = 4). Infusion of superoxide dismutase (150 U/ml) into the perfusate downstream of the femoral artery (site 2) augmented the relaxation of the bioassay ring (to 27.9±4.8%, p<0.05, n = 4) in areversiblefashion; 15 minutes after cessation of infusion of superoxide dismutase the re-

laxations evoked by extraluminal elevation of K + concentration were reduced again (to 13.4 ±4.1%, n = 4). Extracellular Ca 2+ To selectively expose the donor segment (but not the bioassay ring) to Ca 2+ -deficient solution, the solution was used as both the perfusate and the organ bath solution, and appropriate amounts of calcium chloride were infused downstream of the femoral artery (site 2) so that the Ca 2+ concentration was restored to 2.5 mm before the perfusate reached the bioassay ring. Relaxations of the bioassay ring induced by extraluminally administered potassium chloride (40 mm, 52± 5.6%, n = 3) were prevented after 20 minutes of exposure of the femoral artery to Ca 2+ -deficient solution ( + 2.5 + 1.5%, w = 3). Changes in Force in the Perfused Segment To determine whether extraluminal elevation of K + concentration can evoke relaxation in the perfused segment itself, the isometric force generated in the wall of the femoral segments with endothelium was measured (Figure 1). In these experiments, the effects of intraluminal and extraluminal application of potassium chloride were compared in the same preparations (n = 3). Elevation of perfusate (intraluminal) K + concentration by 12 mm produced sustained contractions of the superfused bioassay rings and perfused donor segments (4.1 ±0.8 and 4.8 ±0.7 g, respectively; n = 3) (Figure 4). Extraluminal elevation of the K + concentration (by 15 mm; while maintaining the intraluminal infusion of 12 mm KC1) evoked further contraction in the perfused Coronary artery bioassay ring ( Intraluminal ) Perfused femoral artery segment ( ablumlnal ) Intraluminal KCI, 12 mm Extraluminal KCI, 15 mm 4 rrrin FIGURE 4. Simultaneous measurement of changes in isometric force of a superfused coronary artery ring without endothelium (upper tracing; detecting intraluminalty released vasoactive substances) and a perfused femoral artery segment with endothelium (lower tracing; detecting abluminalty released endothelium-derived vasoactive factors) during intraluminal (12 mm) and extraluminal (15 mm) application of potassium chloride. Experiment was carried out in the presence of indomethacin (10~ 5 M).ACh, intraluminal infusion of acetylcholine (10~ 6 M, site 1). Rubanyi and Vanhoutte K + -Induced Release of EDRF 1101 segment (7.5 ± 1.2 g, n = 3) followed by a transient relaxation (17.7±6.9% of the K + -induced contraction, /J<0.05). Extraluminal application of 15 mm KCI caused only relaxation in the coronary rings (59.6 ± 10.2%, n = 3). The relaxation of the segment and the bioassay ring started after a delay of 5.4± 0.8 minutes. Intraluminal infusion of acetylcholine (10~ 6 M) induced completerelaxationof both the donor segment and the bioassay ring (Figure 4). Elevation of extraluminal K + concentration by 10 mm (without simultaneous intraluminal elevation of K + concentration) evoked relaxation (29.8 ± 12.5%) in two of three femoral artery segments. Phenylephrine and Prostaglandin F 2a Extraluminal application of phenylephrine (10~ 6 M, n = 3) or PGF^ (4 x 10~ 6 M, n = 4) triggered contraction of the perfused segment (6.2±0.8 and 3.6±0.7 g, respectively) similarly to that evoked by extraluminal addition of 20 mm KCI (3.0±0.8 g). These agonists did not evokerelaxationof the bioassay ring. Pharmacological Antagonists of Endogenous Vasoactive Substances The perfused femoral artery segment and the bioassay ring were treated with antagonists (added simultaneously to the perfusate and the organ chamber solution) for 20 minutes before extraluminal administration of 20 mm KCI. Atropine (10~ 6 M, n = 3) did not attenuate the potassium chloride-induced relaxation of the bioassay ring (44.8 ±9.1%) but prevented the endothelium-dependent relaxations to 10 ~6 M acetylcholine. 15 The /3-adrenergic antagonist propranolol (3 X 10" 6 M, n = 3) or apyrase (0.4 U/ml, n = 3; which breaks down adenine nucleotides 16 ) had no significant effect on relaxations of the bioassay ring evoked by extraluminal application of 20 mm KCI (47.1 ±7.6% and 41.4±9.9%, respectively). Discussion The present study demonstrates that increases in extracellular K + concentration can trigger the release of a vasodilator mediator (or mediators) from isolated perfused canine femoral arteries. That therelaxingmediator is released from endothelial cells and not from other constituents of the vessel wall is demonstrated by the prevention of relaxations by removal of the endothelium from the femoral arteries. Because the relaxations were obtained in the presence of indomethacin (an inhibitor of cyclooxygenase that prevents the synthesis of prostacyclin and other prostanoids in canine femoral arteries 1718 ), they cannot be mediated by vasodilator prostaglandins. Thus, the present observations are the first demonstration that K + can trigger the release of a nonprostanoid relaxing factor from the endothelium of canine femoral arteries. Similar to EDRF released by acetylcholine, 12 " prolongation of the transit time from 1 to 6 seconds significantly depressed the biological activity of the endothelial relaxing factor released by K +. Superoxide dismutase restored the biological activity of the mediator, probably by pre-

1102 Circulation Research Vol 62, No 6, June 1988 venting its destruction during the prolonged transit, which is similar to that observed with EDRF. 20 Therefore, it can be postulated that K + and acetylcholine stimulate the release of a similar labile nonprostaglandin relaxing factor from the endothelium of canine femoral arteries. In parallel with the intraluminal release of a relaxing factor (as assessed by relaxation of the bioassay ring), increases in K + concentration induced relaxations in some of the perfused segments, indicating that the relaxing factor was released abluminally (toward the underlying vascular smooth muscle cells) as well. Although the relaxation of the segments occurred after a delay during which they first contracted in response to potassium chloride, the release of the relaxing factor cannot be the consequence of a nonspecific squeezing action due to the contraction since other contractile agonists (phenylephrine and PGF^) did not evoke it. It is unlikely that elevation of extracellular K + concentration stimulates the release of the relaxing mediator by a direct action on endothelial cells because the intraluminal infusion of the ion did not evoke relaxations. The absence of relaxations of the bioassay ring during intraluminal infusion of potassium may be the consequence of simultaneous depolarization of smooth muscle cells by potassium that depresses the responsiveness of vascular smooth muscle to EDRF in various blood vessel preparations. 2122 However, during contractions evoked by intraluminal infusion of 12 mm KC1, the superfused bioassay ring relaxed after simultaneous extraluminal addition of 15 mm KC1. Thus, the ineffectiveness of intraluminal infusion of potassium to evokerelaxationcannot be explained by a simultaneous depolarization of the bioassay ring, and therefore, the release of the relaxing mediator is not likely to be the consequence of a direct depolarizing action" of the K on the endothelial cells. These findings indicate that the calcium channels that exist in endothelial cells are not sensitive to membrane depolarization evoked by increased extracellular K concentration; their properties must differ from those of the voltage-operated calcium channels in vascular smooth muscle. It is likely that K + stimulates the release of a substance (or substances) in the vessel wall that diffuses to the endothelial cells and triggers the release of EDRF. The relatively long latency period between the application of potassium chloride and the onset of relaxation may underly this assumption. Although the identity of this endogenous substance(s) is still unknown, the present experiments ruled out acetylcholine, norepinephrine, adenosine diphosphate and triphosphate, and products of cyclooxygenase by the use of pharmacological antagonists. Since K + triggered the release of a relaxing factor from the endothelium only when added extraluminally, the source of the mediator substance is presumably a structure(s) at or near the adventitial site of the vessel wall (e.g., perivascular nerve endings). Hence, the prevention of K-induced release of EDRF by removal of extracellular calcium could be the consequence of the inhibition of the release of this intermediary mediator and/or that of the endothelial relaxing factor. Irrespective of the exact nature of K + -induced release of EDRF, it may represent a novel mechanism for potassium-induced relaxation of blood vessels. Under the experimental conditions of the present study, the presence of endothelium was essential to observe K + -induced relaxations. The response was not affected by propranolol, indicating that a release of norepinephrinefromadrenergic nerves did not contribute to it either by a direct action on /3-adrenoceptors of the bioassay coronary artery rings 1423 or by facilitating the basal release of EDRF. 22 In the absence of endothelium, moderate increases of extracellular K + concentration lead to vasoconstriction in the bioassay preparations, indicating that under the given conditions a direct relaxant action of K + on the smooth muscle can also be ruled out. Potassium was effective in evoking relaxation only when added extraluminally, which is indicative that such a mechanism may play a role in vivo where the dominant source of elevated extracellular K + is the surrounding parenchymal tissue. Acknowledgments We wish to express our gratitude to Mr. Robert R. Lorenz for the preparation of illustrations and to Mrs. Susan Packie for secretarial assistance. References 1. Furchgott RF: Role of the endothelium in response to vascular smooth muscle. Circ Res 1983;53:557-573 2. Vanhoutte PM, Rubanyi GM, Miller VM, Houston DS: Modulation of vascular smooth muscle contraction by the endothelium. Annu Rev Physiol 1986;48:307-320 3. Singer HA, Peach MJ: Calcium- and endothelial-mediated vascular smooth muscle relaxation in rabbit aorta. Hypertension 1982;4(suppl Il):II-19-II-25 4. Rapoport RM, Murad F: Agonist-induced endothelium-dependent relaxation in rat thoracic aorta may be mediated through cgmp. Circ Res 1983;52:352-357 5. Winquist RF, Bunting PB, Schofield TL: Blockade of endothelium-dependent relaxation by the amiloride analog dichlorobenzamil: Possible role of Na + /Ca + + exchange in the release of endothelium-derived relaxing factor. J Pharmacol Exp Ther 1985;235:644-650 6. Long CJ, Stone TW: Thereleaseof endothelium-derived relaxing factor is calcium dependent. Blood Vessels 1985;22:2O5-208 7. Miller RC, Schoeffter P, Stoclet JC: Insensitivity of calciumdependent endothelium stimulation in rat isolated aorta to the calcium entry blocker, flunarizine. Br J Pharmacol 1985; 85:481-487 8. Schramm M, Thomas G, Towart R, Franckowiak G: Novel dihydropyridine with positive inotropic action through activation of Ca ++ channels. Nature 1983;303:535-537 9. Hof RP, Ruegg UT, Hof A, Vogel A: Stereoselectivity at the calcium channel; opposite action of the enantiomers of a 1,4- dihydropyridine. J Cardiovasc Pharmacol 1985;7:689-697 10. Rubanyi GM, Schwartz A, Vanhoutte PM: The calcium agonists Bay K 8644 and (+ )202,791 stimulate thereleaseof endothelium relaxing factor from canine femoral arteries. Eur J Pharmacol 1985;117:143-144 11. Northover BJ: The membrane potential of vascular endothelial cells. Adv Microcirc 1980;9:135-160 12. Rubanyi GM, Lorenz RR, Vanhoutte PM: Bioassay of endothelium-derived relaxing factors). Inactivation by catecholamines. Am J Physiol 1985;249:H95-H101 13. De Mey JG, Vanhoutte PM: Heterogeneous behavior of the

Rubanyi and Vanhoutte K + -Induced Release of EDRF 1103 canine arterial and venous wall: Importance of the endothelium. Circ Rcs 1982;51:439-447 14. Cohen RA, Shepherd JT, Vanhoutte PM: Prejunctional and postjunctional actions of endogenous norepinephrine at the sympathetic neuroeffector junction in canine coronary arteries. Circ Res 1983 ;52:16-25 15. Rubanyi GM, McKinney M, Vanhoutte PM: Biphasic release of endothelium-derived relaxing factors) by acetylcholine from perfused canine femoral arteries. Characterization of muscarinic receptors. / Pharmacol Exp Ther 1987;24O:802-808 16. Houston DS, Shepherd JT, Vanhoutte PM: Adenine nucleotides, serotonin and endothelium-mediated relaxation to platelets (abstract). Fed Proc 1985;44:1562 17. De Mey JG, Claeys M, Vanhoutte PM: Endothelium-dependent inhibitory effects of acetylcholine, adenosine triphosphate, thrombin and arachidonic acid in the canine femoral artery. J Pharmacol Exp Ther 1982;222:166-173 18. Rubanyi GM, Romero JC, Vanhoutte PM: Flow-induced release of endothelium-derived relaxing factor. Am J Physiol 1986;250:H1145-H1149 19. Griffith TM, Edwards DH, Lewis MJ, Newby AC, Henderson AH: The nature of endothelium-derived vascular relaxant factor. Nature (Lond) 1984;308:645-647 20. Rubanyi GM, Vanhoutte PM: Superoxide anions and hyperoxia inactivate endothelium-derived relaxing factor. Am J Physiol 1986;250:H822-H827 21. Furchgott RF, Zawadzki JV: The obligatory role of endothelial cells in the relaxation of arterial smooth muscle by acetylcholine. Nature 1980;288:373-376 22. Rubanyi GM, Vanhoutte PM: Endothelium-removal decreases relaxations of canine coronary arteries caused by beta-adrenergic agonists and adenosine. J Cardiovasc Pharmacol 1985; 7:139-144 23. Rubanyi G, Vanhoutte PM: Inhibitors of prostaglandin synthesis augment /3-adrenergic responsiveness in canine coronary arteries. Circ Res 1985;56:117-125 KEY WORDS catecholamines adenine nucleotides acetylcholine coronary artery depolarization endothelium EDRF potassium prostaglandins vasodilatation Downloaded from http://ahajournals.org by on January 13, 2019