Endothelial cells release a labile relaxing substance

Transcription

1 VIII. Endothelial Mediation of Autonomic Control Nature of Endothelium-Derived Relaxing Factor: Are There Two Relaxing Mediators? Gabor M. Rubanyi and Paul M. Vanhoutte The role of arachidonic acid in forming endothelium-derived relaxing factor remains controversial. This controversy may be explained if more than one factor exists. To test this hypothesis, the effects of various inhibitors of arachidonic acid metabolism were studied. The perfusate from canine femoral arteries with endothelium was bioassayed with coronary artery rings without endothelium. Treatment of the perfused segment (but not the bioassay ring) with inhibitors of phospholipase A 2 (quinacrine) or cytochrome P 45 0 (metyrapone) had no effect on the basal relaxing activity of the effluent; treatment with the inhibitor of lipoxygenase, nordihydroguaiaretic acid, significantly depressed it. With increasing concentrations of acetylcholine, a biphasic concentration-relaxation curve was obtained; the enzyme inhibitors depressed or prevented thefirstphase but did not affect the second phase. Infusion of arachidonic acid or soybean llpoxldase directly on the bioassay ring did not cause relaxation; together, they evoked concentration-dependent relaxations. These data suggest that acetylcholine can trigger the release of two chemically different relaxing mediators from the endothelium of the canine femoral artery. One factor may be a product of lipoxygenase (or epoxigenase). The second factor is not a metabolite of arachidonic acid and may be released under basal conditions. The existence of two (or more) chemically different endothelium-derived mediators may help to explain the controversial data regarding the nature of the factor(s). (Circulation Research 1987;61(suppl II):II-61-II-67) Endothelial cells release a labile relaxing substance when stimulated by acetylcholine. '~ 7 To differentiate from other vasoactive substances, the term endothelium-derived relaxing factor 8 was introduced. Anoxia and various inhibitors of phospholipase A 2 (the enzyme facilitating the release of arachidonic acid from membrane phospholipids) and lipoxygenase prevent endothelium-dependent relaxation to acetylcholine in a variety of blood vessels from different species. 8 " 10 Thus, it has been postulated that endothelium-derived relaxing factor may be a product of lipoxygenase generated during the oxidative metabolism of arachidonic acid. 8 However, bioassay studies on the rabbit aorta have demonstrated that inhibitors of lipoxygenase do not prevent the intraluminal release of endothelium-derived relaxing factor but chemically inactivate the released mediator in the perfusate. 2 Furthermore, cultured endothelial cells release a relaxing factor that is hydrophylic 6 but not lipophylic as are the metabolites of arachidonic acid. These contradictory findings concerning the involvement of product(s) of the metabolism of arachidonic acid in endothelium-dependent inhibitory responses may be explained if there were multiple From the Department of Physiology and Biophysics, Mayo Clinic and Mayo Foundation, Rochester, Minn. Supported in part by National Heart, Lung and Blood Institute grants HL and HL Address for correspondence: Gabor M. Rubanyi, MD, PhD, Department of Pharmacology, Berlex Laboratories Inc., Cedar Knolls, NJ pathways for the formation of the relaxing mediator or if more than one endothelium-derived relaxing factor existed, as suggested by previous observations that prolonged infusion of acetylcholine through a femoral artery with endothelium evokes biphasic relaxations of superfused bioassay rings of coronary arteries. 3 To analyze these hypotheses, experiments were performed under bioassay conditions to study the effects of inhibitors of various enzymes involved in the metabolism of arachidonic acid (phospholipase A 2, lipoxygenase, and cytochrome P 4J0 ) on basal and acetylcholine-induced release of endothelium-derived relaxing factor into the lumen of perfused femoral arteries. Materials and Methods Experiments were performed on femoral and left circumflex coronary arteries isolated from mongrel dogs of either sex (18-28 kg) anesthetized with sodium pentobarbital (30 mg/kg i.v.). The blood vessels were studied in modified Krebs-Ringer bicarbonate solution of the following millimolar composition: NaCl 118.3, KC1 4.7, CaCl 2 2.5, MgSO < 1.2, KH 2 PO 4 1.2, NaHCO , calcium disodium EDTA 0.026, and glucose 11.1 (control solution). Bioassay Technique Side branches of segments (length cm) of the femoral artery were tied. The endothelium was kept intact as much as possible. The segments were fixed to stainless steel cannulae (1.5 mm i.d.) and placed into an organ chamber maintained at 37 C and filled with

2 11-62 Supplement II Circulation Research Vol 61, No 5, November ml aerated (95% O 2-5% CO 2 ) control solution. The segments were perfused at constant flow (2 ml/min) by means of a multichannel roller pump (Minipuls 2, Gilson, Middleton, Wis.) with control solution maintained at 37 C. A stainless steel tube was also placed in the organ chamber through which control solution was pumped at the same rate. A ring of coronary artery from 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 connected to an isometric force transducer (Grass FTO3D) to permit recording 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 supervised with the perfusate from either the femoral segment with endothelium (endothelial superfusion) or the stainless steel tube (direct superfusion). In some experiments, the endothelium of the femoral artery segment was removed mechanically (vascular superfusion). The transit time between the distal end of the perfused segment and the bioassay ring was 1 second. 5 Previously, the functional intactness of the endothelium in the donor vessel, under the conditions described, has been determined three different ways: (1) by the ability of the segment to release a nonprostanoid relaxing mediator (endothelium-derived relaxing factor) under basal conditions and when stimulated by acetylcholine, 5 (2) by the ability to release prostacyclin under basal conditions and when stimulated by acetylcholine," and (3) by the ability of acetylcholine to evoke endothelium-dependent relaxation in the donor segment in parallel with the intraluminal release of endothelium-derived relaxing factor Protocols To inhibit the synthesis of vasoactive products of cyclooxygenase, all experiments were performed in the presence of indomethacin (1O 5 M). The bioassay ring was first superfused directly with control solution for 60 minutes. During this interval, it was stretched in a stepwise manner until the basal tension reached approximately 10 g, the optimal basal tension for active contraction of rings of isolated canine coronary arteries To evoke sustained contraction of the bioassay ring, prostaglandin F^ (4 x 10" 6 M) was added to the perfusate. After the contraction reached steady level (approximately minutes), the bioassay ring was moved below the outlet from the artery segment with endothelium, and the relaxing activity of the effluent was determined (basal release of endothelium-derived relaxing factor). When the relaxations were stabilized (after approximately 10 minutes), increasing concentrations (10^ to 10" 6 M) of acetylcholine were infused into the perfusate upstream of the perfused segment. The determination of basal and acetylcholine-induced release of endothelium-derived relaxing factor was repeated after 60 minutes, during which the femoral artery segment was perfused in the absence (time control) or presence of an inhibitor of the metabolism of arachidonic acid while the bioassay ring was superfused through the direct line. To avoid an interaction of the inhibitor with either the released endothelium-derived relaxing factor in transit or its action on the smooth muscle of the bioassay tissue, the infusion of the inhibitors was stopped 3-5 minutes prior to starting endothelial superfusion of the bioassay ring. When the effects of exogenous arachidonic acid and lipoxidase (alone or in combination) were tested, the bioassay rings were superfused through the direct line, and the drugs were infused into the perfusate upstream of the heat exchanger to mix with the solution for approximately 3 minutes. Chemicals The following drugs (Sigma Chemical Co., St. Louis, Mo.) were used: acetylcholine hydrochloride, arachidonic acid (sodium salt), indomethacin, lipoxidase (liophylized from Soybean, Type I, linoleate: oxygen oxidoreductase, EC ), 2-methyll,2-di-3-pyridyl-l-propanone (metyrapone), nordihydroguaiaretic acid (NDGA), prostaglandin F^, and quinacrine (mepacrine). Stock solutions of the drugs were prepared daily and kept on ice during the experiments. All drugs were dissolved in distilled water, except indomethacin, which was dissolved in 1.5 X10" 3 M NajCC^ by sonication, and NDGA, which was dissolved in 0.1% dimethyl sulfoxide (DMSO). Drugs were either added to the control solution or infused into the perfusate by means of infusion pumps (model 901, Harvard Apparatus, S. Natick, Mass.) at a rate of ^0.1 ml/min. Their concentration is expressed as final concentration (molar [M], or /u.g/ml for lipoxidase) in the perfusate. Calculations and Statistical Analysis Relaxation of the bioassay ring was expressed as percent decrease of the initial contraction to prostaglandin F^. The inhibitory effects of various concentrations of acetylcholine were measured either as the nadir (for transient responses) or the steady state (for sustained responses). Unless otherwise noted, data are expressed as mean ± SEM for 6 experiments on blood vessels from different dogs. Statistical analysis was carried out by Student's / test for paired or unpaired observations. When p<0.05, differences were considered to be statistically significant. Results Basal Relaxing Activity The endothelial perfusate relaxed the bioassay rings (contracted with prostaglandin FjJ by 26.0 ±4.5% (Figure 1). Treatment of the perfused segment with the inhibitor of phospholipase A 2, quinacrine (10" 4 M), or the inhibitor of cytochrome P 4J0, metyrapone (2 X 10" M), had no significant effect on the relaxing activity of the perfusate, but after exposure to the inhibitor of lipoxygenase, NDGA (5X 10" 5 M), the basal relaxations were significantly depressed (Figure 1). The effluent from arterial segments without endothelium

3 Rubanyi and Vanhoutte Nature of Endothelium-Derived Relaxing Factor Control n- 20 Qulnacrlne, 1(T 4 M 6 NDGA, Metyrapone, 5x10" 5 2x10" 4 M FIGURE 1. Effect of inhibitors of phospholipase A 2 (quinacrine), lipoxidase (nordihydroguaiaretic acid, NDGA), and cytochrome P 450 (metyrapone) on basal relaxing activity of perfusate from canine femoral artery with endothelium. Relaxing activity was bioassay ed with superfused ring of canine coronary artery without endothelium, contracted with prostaglandin F 2a (4 X 10-6 M) in the presence of indomethacin (1O~ 5 M). Inhibitors were infused through artery for 40 minutes prior to test of relaxing activity (see text for details). Relaxing activity is expressed as percent inhibition of prostaglandin F 2a -induced contraction (% = control, 5.1 ±0.6 g; quinacrine, 6.2±0.8 g; NDGA, 6.5 ±0.9 g; metyrapone, 4.9±0.8 g) and shown as mean ± SEM of experiments (n) carried out on blood vessels from different dogs. *Difference from control is statistically significant (p<0.05). had no significant effect on the bioassay rings ( ±4.2%, «= 3; see also Rubanyi et al 5 ). Acetylcholine-Induced Relaxations Acetylcholine (10" 8 to 10" 6 M) did not cause relaxation in the bioassay rings when infused during direct or vascular supervision or when given below the arterial segments during endothelial superfusion (data not shown, see also Rubanyi et al 5 ). When infused upstream of the perfused femoral artery with endothelium, acetylcholine induced biphasic concentration-dependent relaxations in the bioassay ring (Figure 2, control curves). Transient relaxations with increasing magnitude were evoked in the concentration range of 10-8 to 5x 10-8 M acetylcholine (first phase); at 10~ 7 and 2x 10~ 7 M, acetylcholine did not evoke further relaxations (in some experiments, the isometric force even increased toward the initial level). Higher concentrations of acetylcholine (5 x 10~ 7 to 10" 6 M) caused secondary relaxations, which in contrast with those evoked by the lower concentrations were sustained for at least 20 minutes (second phase). In the absence of inhibitors, the second concentration-relaxation curve (repeated after a 60-minute interval) did not differ significantly from the first one (Figure 2, upper left panel). Exposure of the perfused segment to quinacrine (10^ M) or NDGA (5 x 10" 5 M) prevented the relaxations evoked by lower concentrations of acetylcholine but had no significant effect on the second phase of the concentration-relaxation curve (Figure 2, upper right and lower left panels). After treatment with metyrapone (2 x 10^ M), the relaxations in the first phase of the curve were significantly depressed (although not completely prevented as with the other two inhibitors), but the second phase was not affected (Figure 2, lower right panel). Arachidonic Acid and Lipoxidase During direct superfusion, infusion of increasing concentrations (5 x 10"* to 2 x 10 ~5 M) of arachidonic acid caused moderate but significant increases of isometric force in bioassay rings (contracted with prostaglandin F^ in the presence of indomethacin, Table 1). Soybean lipoxidase (10 and 20 /u.g/ml) caused no significant change in isometric force when infused alone (data not shown, n = 3). In contrast, infusion of lipoxidase together with arachidonic acid evoked relaxations of the bioassay rings, which were dependent on the concentration of both arachidonic acid and lipoxidase (i.e., the greatest inhibition was achieved by the mixture of 2 X 10"' M arachidonic acid and 20 /ng/ml lipoxidase) (Table 1). Incubation of lipoxidase (20 /ig/ml) with NDGA (5 x 10~ 5 M) prior to its infusion prevented relaxations (data not shown, n = 3). Discussion Endothelium-Derived Relaxing Factor and Metabolism of Arachidonic Acid The first hypothesis to explain the nature of endothelium-derived relaxing factor was that various stimulants (including acetylcholine) enhance the influx of calcium into endothelial cells, which activates phospholipases (A 2 or C) with consequent liberation of free arachidonic acid (or other fatty acids) and that the relaxing mediator is a free radical or an oxidative product of lipoxygenase. 16 The demonstration of 1) the generation of free radicals' 7 and lipoxygenase products in endothelial cells 18 " 20 ; 2) the prevention of endothelium-dependent relaxations by the removal of extracellular calcium, by calcium-antagonists, 21 free radical scavengers (see Furchgott 8 ), and inhibitors of lipoxygenase (see Furchgott 8 ); and 3) endothelium-dependent relaxations in response to exogenous arachidonic acid all seemed to confirm the original hypothesis. However, fatty acids, which are not metabolized, also evoke endothelium-dependent relaxation, 2425 suggesting that arachidonic acid is not a precursor but rather the promoter of cellular events (e.g., changes in cell membrane fluidity; see Furchgott 25 ) leading to the synthesis or release of endothelium-derived relaxing factor, which is not one of its metabolites. Although exogenous arachidonic acid stimulates the release of endothelium-derived relaxing factor from canine coronary arteries, it can be prevented by ouabain, while the endothelium-dependent relaxations induced by acetylcholine, adenosine diphosphate (ADP), thrombin, and

4 n-64 Supplement II Circulation Research Vol 61, No 5, November 1987 Control Quinacrine, 10" 4 M» Control (1st D/R) D = Control (2nd D/R) o> Inhibitor (2nd D/R) X 9 NDGA, 5x10" 5 M # O9 o o~^i 25r\ *\ 50 Metyrapone, 2x10 ~ 4 M FIGURE 2. Effect of inhibitors of metabolism of arachidonic acid on concentration-response (D/R) curve to acetylcholine causing the release of endothelium-derived relaxingfactorfs) from canine femoral arteries. Effects of relaxing mediators) were tested with superfused ring of canine coronary artery (without endothelium), contracted with prostaglandin F 2a in presence of indomethacin (see text for details). Relaxations are expressed as percent inhibition of the prostaglandin F 2a -induced contraction (% = control: 1st D/R, 5.6 ±0.8 g; 2nd D/R, 4.6 ±0.5 g; quinacrine: control, 5.7±0.5g;drug, g;NDGA:control, 4.9±0.4 g; drug, 5.4±0.5 g; metyrapone: control, 5.4±0.6 g; drug, 5.0±l.l g) and shown as mean ± SEM of experiments (n) carried out on blood vessels from different dogs. *Difference from control is statistically significant (p<0.05) Acetylcholine, -log M oleic and elaidic acids are not affected. 24 These findings suggest that either acetycholine and arachidonic acid stimulate the release of different endothelium-derived relaxing factors or that the same mediator is released via different cellular mechanisms. 24 Quinacrine, several inhibitors of lipoxidase (e.g., 5,8,11,14-eicosatetraynoic acid [ETYA], NDGA, and BW755C), scavengers of free radicals (e.g., hydroquinone, superoxide dismutase, and catalase), and the calcium channel blockers were ineffective as inhibitors Table 1. Effect of Arachidonic Acid and Soybean Lipoxidase on Isometric Force of Superfused Canine Coronary Arteries Without Endothelium* Arachidonic acid (M) 5xl0-«10-5 Initial contraction to prostaglandin F^a (g) 3.2± ± ±1.4 Effect of Arachidonic acid (%)t + 5.5± H H Effect of arachidonic acid + lipoxidaset 10 /xg 20 g (%) (%) -21.8± ± H -45.5± ±4.5 H 2X20" Rings of left circumflex coronary arteries without endothelium were superfused with Krebs-Ringer bicarbonate solution containing indomethacin (10 5 M) through the direct line (see "Materials and Methods"). Mean±SEM. n. Number of experiments on blood vessels isolated from different dogs. tpercent of initial contraction to prostaglandin F^ (+, contraction). ^Soybean lipoxidase infusion started after response to arachidonic acid reached steady state (approximately 5 minutes). Pcrcent of isometric force measured before addition of lipoxidase (-, relaxation). Changes in isometric force are statistically significant (p<0.05). HChanges in isometric force are significantly different from those obtained with 5x 10" 6 M arachidonic acid (p<0.05).

5 Rubanyi and Vanhoutte Nature of Endothelium-Derived Relaxing Factor of endothelium-dependent relaxation caused by several substances in various tissues. 26 " 31 This lack of effectiveness suggests that different vasodilators trigger the release of different mediators 18 or that the nature of endothelium-derived relaxing factor may be species specific. 3 Bioassay studies in the thoracic aorta of the rabbit 2 and the femoral artery of the dog 5 have demonstrated that inhibitors of lipoxidase can inactivate the released endothelium-derived relaxing factor in transit but that they do not prevent relaxation unless they possess antioxidant properties. These findings, which were confirmed in the rat aorta, 32 cast serious doubt on the potential role of product(s) of lipoxidase in endothelium-dependent relaxations evoked by acetylcholine. Alternate routes of metabolism have been suggested for the factor released from endothelial cells, such as oxidation by cytochrome P 4JO. 33~35 Endothelial cells contain cytochrome P 4J0 monoxygenases, which can oxidize arachidonic acid to labile epoxygenase metabolites. 20 Inhibitors of this enzyme (e.g., metyrapone) antagonize endothelium-dependent relaxation to muscarinic stimulation in the thoracic aorta of the rabbit 33 but not in the rat aorta. 26 Thus, it remains uncertain from earlier studies whether endothelium-derived relaxing factor is a derivative of arachidonic acid. Are There Two Chemically Different Endothelium- Derived Relaxing Factors? The present studies favor the hypothesis that the same vasodilator substance (acetylcholine) may trigger the release of two different relaxing mediators from the endothelial cells; however, alternative explanations (i.e., different cellular mechanisms leading to the production of the same factor) should also be considered. One of the factors may be an oxidized derivative of arachidonic acid. This tentative conclusion is based on the observations that 1) in the presence of indomethacin, acetylcholine produces a biphasic concentrationrelaxation curve in the bioassay rings, 2) treatment of the endothelium with various inhibitors of the metabolism of arachidonic acid prevents or depresses the relaxations caused by lower concentrations of acetylcholine but not those evoked by higher concentrations of the amine or those due to basal release of endotheliumderived relaxing factor, and 3) arachidonic acid and lipoxidase can produce a relaxing factors) in vitro. Only one, rather high, concentration of acetylcholine (or of other vasodilators; e.g., ADP, the calcium ionophore A23187, or bradykinin) has been used to demonstrate the intraluminal release of endotheliumderived relaxing factor ; in all reported bioassay studies, the agonists were infused for a relatively short time period. When the infusion of a high concentration (10~* M) of acetylcholine is maintained for a longer period, the initial rapid (and partially transient) relaxation is followed by a secondary slowly developing (but sustained) inhibition of the contraction in bioassay rings. 5 Infusion of catecholamines into the effluent (downstream of the perfused segment to avoid contact with the endothelium) altered the two phases differently: they prevented the secondary sustained response but only moderately depressed the initial transient phase. In contrast, treatment of the endothelium in the perfused segment with quinacrine prevents the first rapid phase but has no effect on the secondary relaxation (unpublished observations). These findings suggest the existence of two chemically different endothelium-derived relaxing factors. This interpretation may be confirmed by the present experiments, which demonstrated that three different inhibitors of the metabolism of arachidonic acid selectively prevented the first but not the second phase of the concentration-relaxation curve to acetylcholine. These inhibitors must have blocked the synthesis or the release of a relaxing mediator rather than inactivating it in transit or antagonizing its action on smooth muscle since the bioassay ring was never exposed to the antagonists and since they were no longer present when the endothelium was exposed to acetylcholine. The selectivity of the inhibitors regarding the two phases of the concentrationrelaxation curve is not likely to be due to their washout at the time the higher concentrations of acetylcholine were infused because quinacrine and metyrapone had no effect on basal release of endothelium-derived relaxing factor (which was tested immediately after the infusion of the inhibitors was stopped); and the effect of higher doses of acetylcholine was not affected when they were infused immediately after the cessation of the administration of the inhibitor (data not shown). The effectiveness of quinacrine suggests that stimulation of phospholipase A 2 and the liberation of free arachidonic acid from membrane phospholipids may be an important initial step in the synthesis of the relaxing mediator released by low concentrations of acetylcholine. However, the effect of quinacrine may be related to its potential antimuscarinic or calciumantagonistic properties. 836 The inhibition by the same concentration of quinacrine of acetylcholine-induced relaxation in femoral arterial rings studied in organ chambers 18 but not of the second phase of the relaxation in bioassay rings is hard to explain solely by the muscarinic antagonistic properties of the drug. Both relaxations are due to stimulation of the same muscarinic receptor subtype on the endothelial cells. 37 The lack of effects of the calcium-antagonist verapamil on acetylcholine-induced relaxations of femoral arteries in the organ chamber 38 or on the release of endothelium-derived relaxing factor under bioassay conditions 30 argues against the assumption that a possible calciumantagonistic property of quinacrine plays an important role in the inhibition. The finding that exogenous arachidonic acid restores the relaxation caused by lower concentrations of acetylcholine after it has been inhibited by quinacrine suggests that quinacrine prevents the first phase of the concentration-relaxation curve because it inhibits the liberation of endogenous arachidonic acid (or of other fatty acids). The effectiveness of NDGA and, to a lesser extent, metyrapone suggests that a product of lipoxidase or cytochrome P 4J o-dependent monooxygenase may be the relaxing mediator responsible for the first phase of the relaxation. The

6 n-66 Supplement II Circulation Research Vol 61, No 5, November 1987 possibility that the relaxing factor may be an oxygenderived free radical (which can also be produced by the activation of these enzyme systems 39 ) has been ruled out in earlier studies. 227 " 29 However, the present study does not allow determination of which of the two metabolic pathways plays a dominant role since NDGA can inhibit the metabolism of arachidonic acid mediated by cytochrome P 450 in anterior pituitary cells 40 and metyrapone may inhibit production of arachidonic acid metabolites catalyzed by lipoxidase. 33 Another alternative explanation for the results is that the same endothelium-derived relaxing factor is released, but different cellular mechanisms are involved in its synthesis and/ or release during stimulation by low and high concentrations of acetylcholine. As we suggested earlier, 5 the transient relaxation may represent the release of endothelium-derived relaxing factor from existing store(s), while the sustained responses reflect de novo synthesis of the relaxing mediator. The present study does not allow exclusion of the possibility that the inhibitors prevent the release of endothelium-derived relaxing factor from its stores in a nonspecific manner. The present study shows that in vitro interaction of lipoxidase with arachidonic acid can produce a factor (or factors) that relaxes coronary arterial smooth muscle. The assumption that the relaxant factors) is an enzymatic (lipoxidase) product of arachidonic acid may be supported by the dependency of the extent of relaxation on the concentration of both the substrate and the enzyme and the prevention of the relaxation by treatment of the enzyme with its inhibitor. Soybean lipoxidase, when acting on arachidonic acid, preferentially produces the labile metabolite 15-hydroperoxyeicosatetrienoic acid (15-HPETE), 41 which has an inhibitory action on vascular smooth muscle. 18 Although the production of an epoxide by the enzyme cannot be ruled out, it remains to be determined whether epoxides have vasoactive properties. The potential role of more stable lipoxidase products (like 15-hydroxy and dihydroxy metabolites or C-6-sulfodipeptide leukotrienes) can be ruled out since they do not relax vascular smooth muscle Thus, these data are consistent with the hypothesis that the relaxing mediator released in response to lower concentrations of acetylcholine from endothelial cells of canine femoral arteries may be a labile hydroperoxide or an epoxide. None of the inhibitors of the metabolism of arachidonic acid affected the second phase of the response of acetylcholine. With the exception of NDGA, they did not affect the basal relaxing activity of the endothelial perfusate. This suggests that higher concentrations of acetylcholine trigger the release of a relaxing factor from the endothelium of femoral arteries that is not an oxidized product of arachidonic acid and that may be similar to the factor released under basal conditions. 12 This latter suggestion is emphasized by the earlier finding that catecholamines inactivate the basally released relaxing mediator. 5 Although NDGA depressed the basal relaxation, this may be due to inactivation of the mediator in transit by the washout of the powerful antioxidant into the perfusate In conclusion, a survey of the literature and new data suggest that endothelium-derived relaxing factor may represent two (or more) chemically distinct substances. Acknowledgments We thank Mr. R.R. Lorenz and Mrs. H. Hendrickson for preparing the illustrations and Mrs. C. Camrud for secretarial assistance. References 1. Furchgott RF, Zawadzki JV: The obligatory role of endothelial cells in the relaxation of arterial smooth muscle by acetylcholine. Nature 1980;286: Griffith T, Edwards D, Lewis M, Newby A, Henderson A: The nature of endothelium-derived relaxing factor. Nature 1984; 308: Forstermann U, Trogisch G, Busse R: Species-dependent differences in the nature of endothelium-derived vascular relaxing factor. Eur 1 Pharmacol 1984; 106: Rubanyi GM, Vanhoutte PM: Hypoxia releases a vasoconstrictor substance from the canine vascular endothelium. J Physiol (Lond) 1985;364: Rubanyi GM, Lorenz RR, Vanhoutte PM: Bioassay of endothelium-derived relaxing factors): Inactivation by catecholamines. Am J Physiol 1985;249(//earf Circ Physiol): H95-H Cocks TM, Angus JA, Campbell JH, Campbell GR: Release and properties of endothelium-derived relaxing factor (EDRF) from endothelial cells in culture. J Cell Physiol 1985;123: Loeb AL, Owens GK, Peach MJ: Evidence for endotheliumderived relaxing factor in cultured cells. Hypertension 1985; 7: Furchgott RF: Role of the endothelium inresponsesof vascular smooth muscle. Circ Res 1983;53: Peach MJ, Loeb AL, Singer HA, Saye JA: Endothelium-derived vascular relaxing factor. Hypertension 1985;7(suppl 1): I-94-I- 10. Vanhoutte PM, Rubanyi GM, Miller VM, Houston DS: Modulation of vascular smooth muscle contraction by the endothelium. Ann Rev Physiol 1986;48: Rubanyi EL, Rubanyi GM, Romero JC, Vanhoutte PM: Release of prostacyclin and endothelium-derived relaxing factor by acetylcholine from canine femoral arteries (abstract). New Orleans, 37th Annual Fall Meeting of the American Physiological Society, Rubanyi GM, Romero JC, Vanhoutte PM: Flow-induced release of endothelium-derived relaxing factor. Am J Physiol l9&6;250(heart Circ Physiol):H\ 145-H Rubanyi GM: Potassium-induced release of endothelium-derived relaxing factor (abstract). Circulation 1986;74(suppl 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: Rubanyi GM, Vanhoutte PM: Endothelium-removal decreases relaxations of canine coronary arteries caused by beta-adrenergic agonists and adenosine. J Cardiovasc Pharmacol 1985;7: Furchgott RF, Zawadzki JV, Cherry PD: Role of endothelium in the vasodilator response to acetylcholine, in Vanhoutte PM, Leusen I (eds): Vasodilatation. New York, Raven Press, pp Rosen GM, Freeman BA: Detection of superoxide generated by endothelial cells. Proc Nad Acad Sci USA 1984;81: DeMey 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: Kuhn H, Ponicke K, Nalle W, Schewe T, Foster W: Evidence

7 Rubanyi and Vanhoutte Nature of Endothelium-Derived Relaxing Factor for the presence of lipoxygenase pathway in cultured endothelial cells. Biomed Biochim Ada 1983;42:K1-K4 20. Johnson A, Revtyak G, Campbell W: Arachidonic acid metabolites and endothelial injury: Studies with cultures of human endothelial cells. Fed Proc 1985;44: Singer HA, Peach MJ: Calcium- and endothelial-mediated vascular smooth muscle relaxation in rabbit aorta. Hypertension 1982;4(suppl 11) Long CJ, Stone TW: The release of endothelium-derived relaxant factor is calcium dependent. Blood Vessels 1985;22: Singer HA, Peach MJ: Endothelium-dependent relaxation of rabbit aorta. I. Relaxation stimulated by arachidonic acid. J Pharmacol Exp Ther 1983;226: Rubanyi GM, Vanhoutte PM: Ouabain inhibits endotheliumdependent relaxations to arachidonic acid in canine coronary arteries. J Pharmacol Exp Ther 1985;235: Furchgott RF, Jothianandan D, Cherry PD: Endothelium-dependent responses: The last three years, in Vanhoutte PM, Vatner SF (eds): Vasodilator Mechanisms. Basel, Karger, pp Rapoport RM, Murad F: Endothelium-dependent and nitrovasodilator-induced relaxation of vascular smooth muscle: Role of cyclic GMP. J Cyclic Nucleotide Protein Phosphor Res 1983;9: Silin PJ, Strulowitz JA, Wolin MS, Belloni FL: Absence of a role of superoxide anion, hydrogen peroxide and hydroxyl radical in endothelium-mediated relaxation of rabbit aorta. Blood Vessels 1985;22: Rubanyi GM, Vanhoutte PM: Oxygen-derived free radicals, endothelium, and responsiveness of vascular smooth muscle. Am J Physiol \9S6;250( Heart Circ /%«0//):H815-H Rubanyi GM, Vanhoutte PM: Superoxide anions and hyperoxia inactivate endothelium-derived relaxing factor. Am J Physiol l9s6;250(heart Circ Physiol):HS22-H Rubanyi GM, Schwartz A, Vanhoutte PM: The effect of diltiazem and verapamil on endothelium-dependent responses in canine blood vessels (abstract). Boston, Annual Meeting of the American Society Pharmacol. Exp. Therap., Winquist RJ, Bunting PB, Schofleld TL: Blockade of endothelium-dependent relaxation by the amiloride analog dichlorobenzamil: Possible role of Na + -Ca + + exchange in the release of endothelium-derived relaxant factor. J Pharmacol Exp Ther 1985;235: Davies JM, Hensby CN, Loginbuhl B, Williams Kl: Stimulation and inhibition of endothelial dependent relaxations of rat aortic strips (abstract). Br J Pharmacol 1985;85: Singer HA, Saye JA, Peach-MJ: Effects of cytochrome P 45O inhibitors on endothelium-dependent relaxation in rabbit aorta. Bloodvessels 1984;21: Abraham NG, Pinto A, Mullane KM, Levene RD, Spokas E: Presence of cytochrome P-450-dependent monooxygenase in intimal cells of the hog aorta. Hypertension 1985;7: Pinto A, Abraham N, Mullane K: Endothelial-dependent relaxation to arachidonic acid in canine coronary arteries: Mediation via a cytochrome P 450 -dependent monooxygenase (abstract). Circulation 1985;72:III Singer HA, Peach MJ: Endothelium-dependent relaxation of rabbit aorta. II. Inhibition of relaxation stimulated by methacholine and A23187 with antagonists of arachidonic acid metabolism. J Pharmacol Exp Ther 1983;226: Rubanyi GM, Vanhoutte PM: Muscarinic receptor subtypes mediating the release of endothelium-derived relaxing factor(s) in canine femoral arteries (abstract). Physiologist 1985;28: DeMey J, Vanhoutte PM: Interaction between Na +, K + exchanges and the direct inhibitory effect of acetylcholine on canine femoral arteries. Circ Res 1980;46: Freeman BA, Crapo JD: Biology of disease. Free radicals and tissue injury. Lab Invest 1982;47: Snyder GD, Capdevila J, Chacos N, Manna S, Falck JR: Action of luteinizing hormone-releasing hormone: Involvement of novel arachidonic acid metabolites. Proc Natl Acad Sci USA 1983;80: Hamberg M, Samuelson B: On the specificity of the oxygenation of unsaturated fatty acids catalyzed by soybean lipoxidase. J Biol Chem 1967;242: Forstermann U, Neufang B: The endothelium-dependent relaxation of rabbit aorta: Effects of antioxidants and hydroxylated eicosatetraenoic acids. Br J Pharmacol 1984;82: Forstermann U, Neufang B: C-6-Sulfidopeptide leukotrienes are unlikely to be involved in the endothelium dependent relaxation of rabbit aorta by acetylcholine. Prostaglandins 1984; 27: KEY WORDS acetylcholine arachidonic acid bioassay calcium coronary artery cytochrome P 450 endothelium-derived relaxing factor femoral artery flow lipoxygenase phospholipase A 2