Intravenous infusion of prostacyclin has been used in humans to induce dilatation of the pulmonary vasculature in severe pulmonary hypertension (1,2). However, systemic vasodilatation resulting in hypotension is an undesirable side effect. Greater pulmonary/systemic selectivity has been obtained by administering prostacyclin by inhalation (3), although inhaled nitric oxide is claimed to be superior to prostacyclin for pulmonary hypertension after cardiac operations (4). We now describe the persistent relaxant actions of several nonprostanoid prostacyclin mimetics on human isolated pulmonary artery preparations, which may indicate that acceptable pulmonary/systemic vasodilator specificity could be achieved by inhalation of this type of agent. The pulmonary relaxant potencies of a number of stable prostacyclin analogs also have been measured to provide a more detailed pharmacologic profile of the IP receptor on the human pulmonary artery. Information on the classification of prostanoid receptors may be found in reference 5.
Cicaprost was chosen as the standard IP receptor agonist in our study because of its high potency and selectivity (6,7) (see Table 1 for structures). Of other available stable analogs of prostacyclin, iloprost has potency similar to that of cicaprost but also possesses marked agonist activity on the EP1 subtype of the prostaglandin E receptor (6-8). The isocarbacyclin 15-deoxy-16α-hydroxy-16β, 20-dimethyl-Δ6,6a-6a-carba PGI1 (TEI-3356) has an ω-chain identical to that of misoprostol, a prostaglandin E1 (PGE1) analog with potent EP2 and EP3 agonist actions (9). Indeed it has been reported that 20-dimethyl-Δ6,6a-6a-carba PGI1 (TEI-3356) is a highly selective EP3 agonist, based on radioligand binding studies on prostanoid receptors expressed in Chinese hamster ovary cells (10). However, TEI-3356 was only 100 times less potent than iloprost in competing with [3H]iloprost for the expressed IP receptor. In view of this and the recently reported presence of a sensitive EP3 contractile system in the human pulmonary artery (11), we thought it of interest to examine the activity of TEI-3356. In taprostene (originally CG 4203), the vinyl ether moiety, which is highly acid-labile in prostacyclin, has been stabilized by conjugation with an aromatic ring system; taprostene has inhibitory actions on platelets both in vitro and in vivo and hypotensive activity in the rat (12,13). Complete chemical stability has been achieved in benzodioxane prostacyclin by incorporating the vinyl ether into an aromatic ether unit; this agent is a weak inhibitor of aggregation in rabbit platelets (14), but no information is available on its activity on isolated blood vessels.
During investigation of a series of thromboxane (TP) receptor antagonists based on stable PGH2 ring systems, Armstrong et al. (15) found that a diphenylmethoximino ω-chain (-CH=N-O-CHPhe2) also bestowed agonist activity at IP receptors. Later the finding of platelet inhibition by octimibate (see Table 2 for structures) showed that the presence of a prostanoid ring is not essential for IP receptor agonism in this type of molecule (16). Further studies established the requirement for a diphenyl-substituted heteroatomic group situated at a critical distance from a carboxylate group (see refs. 17 and 18 for reviews). Of the nonprostanoid prostacyclin mimetics examined so far, 3-[4-(4,5-diphenyl-2-oxazolyl)-5-oxazolyl]phenoxy] acetic acid (BMY 45778) is the most potent, as measured by its ability to inhibit aggregation of human platelets in vitro (19). However, no information is available on the vasodilator potency of BMY 45778; hence the reason for including it in our study. Octimibate and the related analog BMY 42393 are of special interest, because both are claimed to be partial agonists at the IP receptor. Although octimibate abolished aggregation of human platelets in vitro (20), it showed only weak relaxant activity on isolated systemic arteries from humans and monkey and opposed the action of iloprost (21). BMY 42393 also abolished aggregation of human platelets in vitro but failed to increase platelet cyclic adenosine monophosphate (AMP) levels to the same maximum level as did iloprost (22).
We also examined two other nonprostanoids synthesized in our laboratory (Table 2). CU 602 has the diphenyloxazole ring system present in BMY 45778 and BMY 42393, whereas CU 23 contains a diphenylindole ring and is a novel addition to the nonprostanoid prostacyclin mimetic series.
Human pulmonary artery preparations
Lung tissue was obtained from patients undergoing surgery, mainly for carcinoma of the lung, at the Prince of Wales Hospital, Shatin; all were Chinese, aged 45-74 years, except for two aged 4 and 8 years, and none had received either α-adrenoceptor antagonists or Ca2+ channel blockers during the preceding 2-3 weeks. Lobar and segmental pulmonary arteries were dissected from macroscopically healthy areas of lung lobe within 1 h of removal. Adjacent rings (3 mm in length) were cut from a section of vessel and suspended in 10-ml organ baths by using stainless-steel hooks. In some experiments, the endothelium was removed by gently scraping the inner surface of the artery ring with a wooden toothpick. The bathing fluid was Krebs-Henseleit solution (NaCl, 118; KCl, 4.7; MgSO4, 1.18; CaCl2, 2.5; KH2PO4, 1.18; NaHCO3, 25; and glucose, 10 mM; pH 7.32), aerated with 95% O2/5% CO2, maintained at 37°C, and containing 1 μM indomethacin. Tension changes were recorded with isometric force-displacement transducers (Grass FT03) linked to MacLab recording systems (Chart software version 3.3; data-sampling rate, 40 per min).
Experimental protocols for pulmonary artery preparations
The resting tensions of lobar and segmental vessels were adjusted to 0.8 and 1.0 g, respectively, over a 30-min equilibration period. Two responses to 40 mM KCl were obtained; the second exposure gave mean tension increases of 1.29 g (range, 0.67-1.97; n = 73) and 0.78 g (0.47-1.33; n = 39) in lobar and segmental vessels, respectively. Tone was induced with either phenylephrine or KCl in the presence of 200 nM GR 32191, a TP receptor antagonist with a pA2 of 8.2-8.3 (11, 23) on human pulmonary artery. GR 32191 was included to block possible TP agonist activity of the prostacyclin analogs, as has been reported for isocarbacyclin (24). Tone was also induced with U-46619 (in the presence of 200 nM nifedipine to block L-type Ca2+ channels). The first exposure to cicaprost involved the cumulative addition of 1-, 3-, and (sometimes) 10-ng doses; this procedure always produced 50-90% relaxation and did not suppress subsequent responses to the contractile agent. About 50 min later, a cumulative dose-response relation was obtained for either cicaprost (second exposure) or the test compound against the same contractile agent. The nonprostanoids showed persistent relaxant effects, and in many cases, preparations were abandoned, and new preparations, which had been stored in Krebs-Henseleit solution at 4°C for ∼4 h, were set up; there were no obvious differences in the contractile or relaxant responses of these two sets of preparations. Relaxation was expressed as the percentage reduction of the tension increase elicited by the contractile agent.
Guinea-pig trachea preparations
Tracheal rings from guinea pigs (400- to 700-g males, killed by stunning and exsanguination) were set up for isometric tension recording as described for the human pulmonary artery preparations. The resting tension was set at 0.8-1.0 g. Cumulative doses of 17-phenyl-ω-trinor PGE2 (0.5-20 nM), the standard EP1 receptor agonist, were added, followed after washout and a rest period of 50 min by the test compound. For detecting possible TP receptor antagonism, a single dose of U-46619 (5 nM) was added to give a stable level of contraction, and then cumulative doses of the nonprostanoid were added. Guinea pig tracheal smooth muscle appears to lack IP receptors (15).
The method is essentially that of Merritt et al. (20), except that platelet-rich plasma was obtained from human blood after an initial centrifugation at 230 g for 15 min. In brief, 150-200 μg platelet membranes were incubated with and without test compound in 50 mM TRIS-HCl buffer (pH 7.4), containing 4 mM MgCl2, 40 μM EDTA, and 10 nM [3H]iloprost (specific activity, 12.7 Ci/mmole−1; final volume, 200 μl). After 30 min at 37°C, samples were filtered through Whatman GF/B filters by using a Brandel cell harvester and washed 3 times with ice-cold 50 mM TRIS-HCl buffer containing 5 mM MgCl2. Specific binding determined with 20 μM iloprost was 90% of total binding. Saturation experiments gave a Kd of 12.9 ± 1.7 nM and Bmax of 943 ± 49 fmol/mg protein for [3H]iloprost binding. In the competition experiments, each concentration of competing ligand was tested in triplicate in three separate experiments.
For isolated smooth-muscle data, all n values relate to single preparations obtained from different human donors or animals. EC50 or IC50 values for log concentration-response relations were derived by linear interpolation of graphic plots. Equieffective molar ratios (EMR) were calculated by dividing the EC50 (or IC50) for the test compound by the corresponding EC50 (or IC50) for first exposure to the standard agonist (within-preparations comparisons). Slope parameters for individual inhibitory log concentration-response curves were calculated as (log IC90-log IC25) values, and subjected to two-factor analysis of variance (ANOVA): factor 1, IP agonist (all compounds shown in Tables 1 and 2 except TEI-3356/phenylephrine and CU 602/phenylephrine); factor 2, contractile agent (U-46619 or phenylephrine). Comparisons of individual cells in the ANOVA (e.g., cica/U-46619 versus BMY 45778/U-46619) were made by testing contrasts of nonweighted means (SuperANOVA; Abacus Concepts, Inc., Berkeley, CA, U.S.A.). Two-factor ANOVA with means contrasts also was applied to the endothelium experiments: factor 1, cicaprost concentration; factor 2, ±endothelium. Significance was assumed when p < 0.05. Radioligand-binding inhibition data were fitted to a logistic function by using the GraphPad Prism program; Ki values were calculated from IC50 values by using the Cheng-Prusoff relation.
CU 23 (8-[2,3-diphenyl-1-indolyl]octanoic acid) and CU 602 (9-[4,5-diphenyl-2-oxazolyl]-8-oxa-nonanoic acid) were synthesised in the laboratory of Prof. Henry N. C. Wong, Department of Chemistry, Chinese University of Hong Kong. The following compounds were gifts: cicaprost, iloprost, and sulprostone (Schering AG, Berlin, Germany); TEI-9063 (17α, 20-dimethyl-Δ6,6a-6a-carba PGI1) and TEI-3356 (15-deoxy-16α-hydroxy-16β,20-dimethyl-Δ6,6a-6a-carba PGI1; Teijin Institute for Biomedical Research, Tokyo, Japan); taprostene (CG 4203; Grunenthal GmbH, Aachen, Germany); benzodioxane prostacyclin (Shionogi Research Laboratories, Osaka, Japan); BMY 45778 (3-[4-(4,5-diphenyl-2-oxazolyl)-5-oxazolyl]phenoxy] acetic acid), BMY 42393 (2-[3-[2-(4,5-diphenyl-2-oxazolyl)ethyl]phenoxy]acetic acid), and octimibate (Bristol-Myers Squibb Pharmaceutical Research Institute, Wallingford, CT, U.S.A.); GR 32191 (9α-(biphenylyl)methoxy-1 1β-hydroxy-12β-(N-piperidinyl)-ω-octanor-prost-4Z-enoic acid; Glaxo Group Research, Ware, U.K.). 17-Phenyl-ω-trinor PGE2 and U-46619 were purchased from Cayman Chemicals, Ann Arbor, MI, U.S.A., prostacyclin sodium salt from Calbiochem, San Diego, CA, U.S.A, CGRP (rat) from Cambridge Research Biochemicals, Northwich, U.K., and acetylcholine chloride, cromokalim, indomethacin, (±)-isoprenaline, L-phenylephrine hydrochloride, and nifedipine from Sigma Chemical Co., St. Louis, MO, U.S.A. [3H]Iloprost was purchased from Amersham Life Science, Little Chalfont, U.K.
Solutions (10 mM) of prostacyclin analogs in pure ethanol were diluted 50-fold with 0.9% NaCl solution (containing a trace of NaHCO3 to ensure solubility) to produce a primary aqueous stock, which was stored at −20°C when not in use. The nonprostanoid prostacyclin mimetics were treated similarly, except that dimethylsulphoxide was used for initial solubilization, because they are poorly soluble in ethanol. Prostacyclin sodium salt was dissolved in 0.05 M TRIS-HCl buffer (pH 9.0) to give a 10−4 stock solution, which was stored in aliquots at −20°C. Immediately before addition of prostacyclin to the organ bath, the TRIS stock was diluted with 0.9% NaCl solution and kept on ice.
Defining conditions for testing IP receptor agonists on the human pulmonary artery
Log concentration-response curves for the experiments described in this section are shown in Fig. 1. Potassium chloride (K+), the TP receptor agonist U-46619, and the α1-adrenoceptor agonist phenylephrine were used to increase tension in the pulmonary artery so that relaxation could be studied. Our previous study of prostanoid action on the human pulmonary artery (11) showed that the contractile strength of individual ring preparations varied considerably, partly because of the use of vessels of different sizes and partly because of the greater thickness of smooth muscle in vessels from some patients. Consequently it was decided to relate contractile activity to initial responses elicited by 40 mM K+; these responses remained reproducible over a period of 5 h and decayed rapidly to baseline on washout of the organ bath. The option of obtaining the maximal response to K+ on each preparation was rejected, because on some preparations, contractions elicited by 80-120 mM K+ showed an initial rapid decay on washout but thereafter took a considerable time to reach the resting tension and occasionally returned to a higher resting tensicn. Subsequent results of using K+ or phenylephrine as contractile agents were obtained on preparations treated with 200 nM GR 32191, a potent TP receptor antagonist that had no effect on K+ or phenylephrine responses.
Cicaprost (0.25-3.5 nM) relaxed preparations precontracted by 20 mM K+ with the maximum effect below the resting tension, whereas against 40 and 60 mM K+, maximum relaxations of 30-55% were obtained. The picture was different with the β-agonist isoprenaline and the potassium channel opener cromokalim; ∼85% relaxation of 20 mM K+ tone was obtained but only 10% relaxation of 60 mM K+ tone. In one experiment, calcitonin gene-related peptide (CGRP, 0.1-100 nM) induced <5% relaxation of 40 mM K+ tone (data not shown). Nifedipine at 200 nM abolished 20 mM K+ responses (n = 3) and inhibited 40 mM K+ responses by 90-95% (n = 4; data not shown). In subsequent experiments on the relaxant properties of the nonprostanoid prostacyclin mimetics, induction of tone with 40 mM K+ was adopted as one of the standard procedures.
U-46619 at 2 nM induced contractions similar in size to those elicited by 40 mM K+. Cicaprost relaxed the preparations to below the baseline tension and was active over the same concentration range as was found in the 20 mM K+ experiments. Responsiveness to cicaprost remained almost identical on second exposure; the ratio of second IC50 to first IC50 was 1.04 ± 0.03 (SEM, n = 9). In four different experiments, 5-10 times higher concentrations of cicaprost were required to produce equivalent relaxations of the higher tone induced by 10 nM U-46619. On two of the four preparations, cicaprost was retested against 10 nM U-46619, and its relaxant sensitivity was unchanged (IC50 ratios, 0.99 and 1.03). This protocol was repeated on the other two preparations in the presence of 1 μM tetrodotoxin; IC50 ratios of 1.08 and 1.32 were obtained. All four preparations relaxed to below the resting tone level when a cicaprost concentration of 100 nM was attained. Nifedipine at 200 nM reduced U-46619 responses to only a small extent (7-10%; n = 3; data not shown). Induction of a U-46619 response approximately matching the 40 mM K+ response, in the presence of 200 nM nifedipine, was adopted as the standard procedure for examining the nonprostanoid prostacyclin mimetics. In practice, 4 nM U46619 was usually used, with 2 or 6 nM in other cases, and responses between 80 and 100% of the standard K+ tone were obtained.
Against phenylephrine-induced tone, cicaprost was again able to relax the preparation to below the resting level. The sensitivity to cicaprost and the slope of its log concentration-response curve (Table 2) were slightly higher than those found in the U-46619 experiments (p < 0.05; two-factor ANOVA performed on all slope factors). The IC50 ratio was 1.07 ± 0.04 (n = 6) for second/first exposures to cicaprost. Isoprenaline was a relatively low potency relaxant, producing a 70% reduction in phenylephrine tone at 800 nM. In a single experiment, CGRP was active over the same concentration range as cicaprost but produced only 40% relaxation. Nifedipine (200 nM) inhibited 2 μM phenylephrine responses by 30-35% (IC20 = 24-40 nM; n = 3; data not shown). The standard procedure for assessing the relaxant potency of the prostacyclin mimetics involved the use of 2 μM phenylephrine. However, on the larger vessels in particular, phenylephrine concentrations of 5 or 10 μM were sometimes required; this reflected the lower maximal response of phenylephrine relative to either U-46619 or K+. The sensitivity to cicaprost was not dependent on the concentration of phenylephrine used.
The effect of endothelium removal on relaxations induced by acetylcholine and cicaprost was studied on preparations precontracted by phenylephrine. In three experiments on nondenuded preparations, acetylcholine induced only relaxation; threshold effects were obtained at 0.1 μM, and 96 (Fig. 1, expt A), 83, and 50% relaxations were obtained at 14.4 μM. On corresponding endothelium-denuded preparations, acetylcholine over the same concentration range respectively induced minimal relaxation (7 and 2% at 14.4 μM) and contraction (EC50 = 1.6 μM; maximum = 52% above phenylephrine response). In a fourth experiment, acetylcholine at 0.1 μM induced a distinct relaxation of the nondenuded preparation, but this converted to contraction as the acetylcholine concentration was increased (Fig. 1, expt B). On the corresponding denuded preparation, acetylcholine induced contraction (EC50 = 0.22 μM; maximum = 52%). These contractions to acetylcholine were quite large compared with those recently seen on denuded rabbit and guinea pig aortic rings (unpublished observations). In all four experiments, cicaprost at 3.5 nM induced near-maximal relaxation of both denuded and nondenuded preparations. Although it appeared that responses to 1 nM cicaprost were smaller in the endothelium-denuded than in the nondenuded preparations, this did not achieve statistical significance (two-factor ANOVA). The endothelium was not removed in other experiments in the study, and cicaprost invariably showed high relaxant potency: IC50 range, 0.29-1.17 nM (phenylephrine, 56 preparations).
Potencies of prostacyclin analogs on preparations contracted with phenylephrine
Four prostacyclin analogs, TEI-9063, iloprost, taprostene, and benzodioxane prostacyclin, completely inhibited phenylephrine-induced tone, and the slopes of their log concentration-response curves (Table 1) were not significantly different from that of cicaprost. Potencies relative to cicaprost (EMR) also are given in Table 1. The fifth prostacyclin analog, TEI-3356, behaved somewhat differently, producing only relaxation in two experiments (concentration range, 100-4,000 nM) and small contractile effects (25-500 nM) that progressed to relaxant effects (1-7 μM) in two other experiments (Fig. 2). On a second matched preparation from one of the latter experiments (in the absence of phenylephrine, but with GR 32191 still present), TEI-3356 induced a small sustained contraction at a concentration of 50 nM, with a slightly greater effect at 300 nM. The 50 nM TEI-3356 response was matched by 0.5 nM of the EP3 receptor agonist sulprostone.
In two experiments, prostacyclin was ∼10 times less potent than cicaprost (IC50, 3.6 and 5.2 nM). Responses to prostacyclin were not sustained; and 85% relaxant response had almost completely waned 15 min after its peak effect. By reference to its concentration-response relation, the biologic half-life of prostacyclin was estimated as 3-4 min.
Relaxant activities of nonprostanoid prostacyclin mimetics
Four nonprostanoids, BMY 45778, BMY 42393, octimibate, and CU 23 induced complete or almost complete relaxation against phenylephrine-induced tone. The log concentration-response curves for BMY 45778, CU 23 (Fig. 3), and BMY 42393 were shallower than those of cicaprost, whereas that of octimibate was not significantly different (Table 2; two-factor ANOVA performed on all slope factors). The relaxant action of CU 602 varied in different experiments: in one case, it induced 95% relaxation at 2 μM and had a slightly shallower slope than cicaprost; in three other experiments, its log concentration-response curves were much shallower and only 30, 65, and 87% relaxations were obtained at 8.5 μM(Fig. 3).
All the nonprostanoids, including CU 602, produced complete relaxation against U-46619-induced tone, and none of their slopes was significantly different from that of cicaprost (Table 2). Against 40 mM K+ tone, BMY 45778, BMY 42393, and octimibate showed maximal relaxations similar to cicaprost (Table 2); CU 602 and CU 23 were somewhat less effective relaxants, and the value for CU 23 (20-25%) may not represent its true maximal effect; we were reluctant to exceed a concentration of 10 μM with the nonprostanoids to avoid exposing the preparation to a high concentration of DMSO. EMRs (Table 2) for the three contractile situations agree closely, the ranking being cicaprost > BMY 45778 >> BMY 42393 > octimibate > CU 23 ≥ CU 602.
Rates of reversal of IP receptor-mediated relaxation
It was consistently observed that the onset of relaxation and its reversal after washout were much slower for all the nonprostanoid prostacyclin mimetics than for cicaprost and its analogs, irrespective of the contractile agent used. The exception was TEI-9063, which was slightly slower than cicaprost. However, unlike that of the nonprostanoids, its rate of reversal was fast enough for the preparation to be used for other tests within 40 min of agonist washout. The slow onset and persistent action of BMY 42393 is shown in Fig. 4 on a preparation that had a higher than average level of inherent tone. On a matching preparation, full sensitivity to U-46619 had returned within 20 min of washout of 3 nM cicaprost, which fully relaxed the preparation. Although maintenance of contractile sensitivity was assured through the use of a control preparation, eliciting a 40 mM K+ contraction (which is only partially inhibited by IP receptor agonists) during the long recovery phase of the nonprostanoid showed that the test preparation was not generally deteriorating with time.
Radioliogand binding to human platelet membranes
All 11 prostacyclin mimetics reduced specific [3H]iloprost binding to the IP receptor on human platelet membranes by >95%. Log Ki values and Hill slopes are given in Table 3. In Fig. 5, it can be seen that there is a good correlation between log IC50 values for relaxation of phenylephrine-induced tone on the human pulmonary artery and the log Ki values from the human platelet membrane experiments.
Activities of prostacyclin mimetics on guinea pig trachea
The high and low agonist potencies of iloprost and cicaprost, respectively, on the EP1 contractile system of the guinea pig isolated trachea were reported previously (6) and are confirmed here (Table 3). The EP1 activity of the other prostacyclin mimetics used in this study have also been investigated by using 17-phenyl-ω-trinor PGE2 as the standard agonist. TEI-9063 was a highly potent agonist, whereas TEI-3356 was much less potent, although still a full agonist. Taprostene and benzodioxane prostacyclin elicited small contractions (5-10% of maximum) at the highest concentration tested of 1 μM; the nonprostanoids elicited no contractile effects up to a concentration of 5 μM.
The nonprostanoid prostacyclin mimetics were also tested for their ability to inhibit contractions of the trachea induced by 5 nM U-46619; IC50 values are shown in Table 3. A concentration of nonprostanoid at one fifth of its IC50 value induced <10% inhibition of the U-46619 response.
Our experiments have thrown some light on the cellular mechanisms underlying the relaxation of human pulmonary artery by IP receptor agonists, although this was not the primary aim of the study. First, the lack of effect of the Na+-channel blocker tetrodotoxin on the relaxant action of cicaprost indicates that release of a relaxant agent from neural elements in the ring preparation is unlikely to be involved. It was important to establish this early in the study, because we recently showed (25) that IP receptor agonists inhibit the spontaneous contractility of the rat isolated colon through release of nitric oxide and an unidentified transmitter from nonadrenergic noncholinergic (NANC) enteric nerves. NANC neural elements capable of releasing relaxant agents during electrical field stimulation have been reported in isolated pulmonary arteries from cat (26) and guinea pig (27,28).
Second, cicaprost-induced relaxation was not dependent on the presence of the endothelium. Other workers reported similar findings for prostacyclin acting on isolated pulmonary arteries of human and dog (29) and for isocarbacyclin acting on isolated cerebral and mesenteric arteries of monkey (24). Acetylcholine's endothelium-dependent relaxation of the human pulmonary artery is mediated via M1 and M3 receptors and has both cyclooxygenase (inhibited by indomethacin) and nitric oxide synthase (inhibited by L-NOARG) components (30). In the presence of indomethacin, 10-30 μM acetylcholine elicited 60 ± 9% relaxation, which is similar to that seen in our three experiments in which acetylcholine produced only relaxation. In our fourth experiment (B in Fig. 2), the presence of both excitatory and inhibitory components in the acetylcholine dose-response curve may be due to some loss of endothelial-dependent relaxant function or the presence of a more effective direct contractile action or both; the latter is due to activation of M3 receptors on the smooth-muscle cells (30).
The third aspect relates to incomplete relaxation of K+ contraction by IP agonists. Any agent that relaxes arterial smooth muscle by opening plasma membrane K+-channels will become less effective as the tone of the preparation is raised by increasing concentrations of K+. Two mechanisms are likely to be involved. First, the simple elevation of tone means that higher concentrations of relaxant will be required to induce matching degrees of relaxation. Second, increasing the external K+ concentration will move the membrane potential (EM) toward the equilibrium potential for K+ (EK), so that opening of potassium channels will be of lesser consequence. In our studies, cromokalim at 2.5 μM induced 60% relaxation of phenylephrine-induced tone, 85% relaxation of 25 mM K+ tone, and minimal relaxation of 65 mM K+ tone (we refer to the total K+ in bathing fluid subsequently). These findings are consistent with a single mode of action for cromokalim: opening of plasma membrane K+ channels (31,32). Results obtained with high concentrations of cromokalim must be interpreted with caution, because it has been reported that >3 μM cromokalim blocks voltage-dependent Ca2+ channels in single muscle cells from rat portal vein (33). Shallow relaxation curves were also found for both isoprenaline and CGRP (see also 34) against phenylephrine, and minimal relaxation was seen against high potassium contractions; these results may indicate a high reliance on K+-channel opening. However, in the case of cicaprost, the relaxation mechanism is likely to be more complex. Cicaprost has steep log concentration-response curves against 20 mM K+, U-46619, and phenylephrine and relaxed the preparations to below the resting tension. With 65 mM K+ in the bathing fluid, cicaprost still induced relaxation over the 1-10 nM concentration range, with 35% reduction in tone at 50 nM. The simplest explanation of these findings is that cicaprost has two relaxant mechanisms operating over the same concentration range, only one of which involves the opening of plasma membrane K+ channels. From 42K+ and 24Na+ flux measurements in dog coronary artery strips and patch-clamp studies on single smooth-muscle cells from the rat portal vein, it has been shown that iloprost induces membrane hyperpolarisation and smooth-muscle relaxation by opening of both voltage-dependent and calcium-activated K+ channels (35). In more recent experiments (36), the KATP channel blocker gliben-clamide was shown partially to inhibit the vasodilator actions of PGI2 and iloprost on the rabbit coronary circulation, indicating the involvement of KATP channels. Further studies are in progress to investigate which K+ channels are involved in relaxation of human pulmonary vessels induced by IP receptor agonists.
We still consider cicaprost to be the best standard IP agonist because of its high selectivity and also by reference to Fig. 5 its high efficacy. TEI-9063 was slightly more potent than cicaprost, in agreement with the first report (37) of its high IP agonist potency in activating adenylate cyclase in mastocytoma P-815 cells. Its IC50 value of 3 nM for competition for [3H]iloprost binding in mastocytoma cell membranes is in close agreement with the IC50 of 4.8 nM found by us in human platelet membranes. However, it also has high EP1 agonist potency on the guinea pig trachea. Indeed, in this respect, it is one of the most potent of the prostacyclin analogs that we have investigated for EP1 agonism. The EP3 agonist potency of TEI-9063 is not known, because on the guinea pig vas deferens (the archetypal EP3 preparation; 5), its IP agonist activity to enhance electrical field-stimulation twitches overshadows any potential EP3 inhibitory activity (38). Taprostene, although it has only moderate IP potency, may be quite useful for characterizing IP receptors because it has little agonist activity at EP1 receptors (Table 3) and EP3 receptors (unpublished observations). Although benzodioxane prostacyclin is a rather weak IP agonist on the pulmonary artery, which agrees with its low potency as an inhibitor of rabbit platelet aggregation (IC50 = 1.35 μM; 14), it may still be useful for characterizing IP receptors.
TEI-3356 has been put forward as a highly specific EP3 receptor agonist (10). On some pulmonary artery preparations, we observed contractile activity that, in the presence of TP receptor blockade, can be tentatively assigned to activation of EP3 receptors. However, its IP relaxant activity, although quite low relative to cicaprost and iloprost, was dominant. This indicates that it may not be specific enough to be of real use in characterizing EP3 receptors. In addition, TEI-3356 has moderate EP1 agonist potency as judged by its ability to contract the guinea pig trachea (in the presence of TP receptor blockade).
The nonprostanoid nature of BMY 45778 and its relatives, and the fact that their terminal diphenyl-substituted heteroatomic units are structurally close to pharmacophores present in many drugs, prompted us to examine their relaxant actions on the human pulmonary artery more rigorously than those of the prostacyclin analogs. Relaxant activity was compared with cicaprost under three contractile situations: against 45 mM K+, which relies predominantly on opening of L-type voltage-dependent Ca2+ channels; against phenylephrine, which is partially dependent on L-type Ca2+ channels; and against U-46619 in the presence of nifedipine, where L-type Ca2+ channel function (which was a minor component anyway) has been abolished. BMY 45778, BMY 42393, octimibate, and CU 23 mimicked the actions of cicaprost and had similar relative potencies in all three contractile situations, implying that activation of IP receptors is responsible for their relaxant activities. BMY 45778 was by far the most potent agent, being only ∼3 times less active then cicaprost. Similarly, as an inhibitor of human platelet aggregation, it was one of the most potent members (IC50 = 27 nM) of a large group of related agents (18,19). However, an exact comparison of potencies is not possible because the platelet studies were carried out (as is usually the case) on platelet-rich plasma, in which the highly lipophilic nonprostanoids bind to plasma proteins to a much greater extent than do the relatively water-soluble prostacyclin analogs.
The shallower log concentration-response curves for BMY 45778, BMY 42393, and CU 23 against phenylephrine-induced tone may reflect their lower agonist efficacies relative to cicaprost and the other prostacyclin analogs. On the assumption that the Ki values for the human platelet receptor are applicable to the human pulmonary artery, it can be calculated that the IP receptor occupancies corresponding to IC100 would be ∼12% for cicaprost, ∼20-40% for the other prostacyclin analogs, and ∼70-80% for the three nonprostanoids. Thus for the nonprostanoids, complete relaxation would require close to full receptor occupancy, and this would result in a slightly shallower curve than a higher efficacy agonist (39) such as cicaprost. Octimibate behaved as a full agonist and did not show a shallower curve than cicaprost. However, inspection of the experimental tracings showed that of the nonprostanoids octimibate had the slowest rate of onset of relaxation, and we believe that 20-30 min was probably insufficient time for the lowest concentrations to achieve their true effects. In previous studies on 45 mM K+-contracted human vessels, octimibate appeared to behave as a partial agonist (21); it induced only 15% relaxation of coronary artery at 10 μM and only 10% and 40% relaxations of mesenteric and skin/subcutaneous tissue arteries at 3 μM; iloprost (standard agonist) produced ∼60% maximal relaxation and had IC30 values of 15, "not determined", and 30 nM, respectively. On the coronary and mesenteric artery, the relaxant action of iloprost was very effectively opposed by 3 μM octimibate (Kp = 200 nM). The reason for the differences in octimibate's relaxant profiles may relate to the higher sensitivity of our human pulmonary artery preparations to IP-agonist action (Fig. 1). This may reflect a higher receptor density in human pulmonary as opposed to human large peripheral vessels. A similar argument was proposed by Merritt et al. (21) to account for the higher maximal response to octimibate on human platelets compared with human large peripheral vessels. However, the degree of contractile tone is a confounding factor that can markedly influence the relaxant profile. This is well illustrated in our studies, in which 5-10 times more cicaprost is required for relaxation against 75% as opposed to 40% maximal contractions induced by U-46619 (Fig. 1). Again in the early studies of Brink et al. (40) on the human pulmonary artery, relatively low sensitivities to IP agonists were observed against maximal histamine contractions: IC40 values for cicaprost, iloprost, and prostacyclin were 21, 17, and 290 nM respectively, with maximal relaxations of 80%. It would be of interest directly to compare human pulmonary and peripheral arteries to determine if there are genuine differences in the IP agonist efficacy of octimibate.
The fifth nonprostanoid CU 602 appeared to behave as an IP partial agonist against phenylephrine-induced and perhaps against K+-induced tone. However, it completely relaxed U-46619-induced tone (IC50 = 835 nM), with a slope factor that was not significantly different from that of cicaprost. At >300 nM, CU 602 blocked U-46619 contractions of the guinea pig trachea (Table 3), and therefore its IP (partial) agonist activity on the pulmonary artery may be augmented by TP receptor antagonism. It is unlikely that TP receptor block contributes to the relaxant actions of the other nonprostanoids against U-46619 on the pulmonary artery, because their IC50s on the pulmonary artery are much lower than on the guinea pig trachea.
The data we collected on the analogs and nonprostanoid mimetics of prostacyclin should be of value in characterizing potential IP receptor subtypes, particularly now that cloning of the human IP receptor has been achieved (41).
The slow onset and offset of the relaxant action of the nonprostanoid prostacyclin mimetics is of interest from both mechanistic and clinical aspects. According to the "exact diffusion model" proposed by Colquhoun et al. (42), binding of a high-affinity ligand to its plasma membrane receptors delays the diffusion of the compound through the extracellular fluid pathways in a densely packed cellular system. Consequently the rate of increase of agonist concentration in the center of the muscle fiber mass is slowed, perhaps by as much as 1000-fold, depending on the ligand affinity and the density of receptors on the plasma membrane. On washout, the reservoir of bound ligand maintains the extracellular fluid concentration, and thus the action persists for a long time. Therefore at low concentrations of a high-affinity ligand, slow onset is matched by slow offset. We have suggested that such a mechanism may account for the very slow onset and offset of contraction of certain high-affinity TP receptor agonists (e.g., EP 171, Kd ∼1 nM) (43). In our experiments, this mechanism may also account for the slightly slower kinetics of the highest affinity IP agonist TEI-9063 (Ki ∼3 nM). However, in the nonprostanoid prostacyclin mimetics, IP receptor affinity may not be the dominant factor, because all these agents showed slow-onset/slow-offset kinetics irrespective of their affinities (Ki = 10-500 nM). As mentioned previously, a notable difference between the prostacyclin analogs and the nonprostanoid mimetics is the much greater lipophilicity of the latter. Colquhoun et al. (42) pointed out that the model of reduced diffusion through the extracellular space could still apply to a highly lipophilic agent partitioning into a cellular lipid phase. From a therapeutic viewpoint, it would be useful to know whether nonprostanoid prostacyclin mimetics show high retention in lung tissue after aerosol administration. Such a pharmacokinetic profile could enhance pulmonary/systemic vasodilator selectivity and thus clinical utility.
Acknowledgment: These studies were supported by a project grant from The Research Grants Committee of Hong Kong (CUHK 21/92M). Mr. Kevin Chow Bing-shui and Mr. Ng Kaon are thanked for their excellent technical assistance. Gifts of compounds from pharmaceutical companies indicated in the Methods section are gratefully acknowledged.
1. Cremona G, Higenbottam T. Role of prostacyclin in the treatment of pulmonary hypertension. Am J Cardiol
2. Barst RJ, Rubin LJ, Long WA, et al. A comparison of continuous intravenous epoprostenol (prostacyclin) with conventional therapy for primary pulmonary hypertension. N Engl J Med
3. Walmrath D, Schneider T, Pilch J, Grimminger F, Seeger W. Aerolised prostacyclin in adult respiratory distress syndrome. Lancet
4. Goldman AP, Delius RE, Deanfield JE, Macrae DJ. Nitric oxide is superior to prostacyclin for pulmonary hypertension after cardiac operations. Ann Thorac Surg
5. Coleman RA, Smith WL, Narumiya S. VIII International Union of Pharmacology Classification of prostanoid receptors: properties, distribution, and structure of the receptors and their subtypes. Pharmacol Rev
6. Dong YJ, Jones RL, Wilson NH. Prostaglandin E receptor subtypes in smooth muscle: agonist activities of stable prostacyclin analogues. Br J Pharmacol
7. Lawrence RA, Jones RL, Wilson NH. Characterization of receptors involved in the direct and indirect actions of prostaglandins E and I on the guinea pig ileum. Br J Pharmacol
8. Dong YJ, Jones RL. Effects of prostaglandin and thromboxane analogues on bullock and dog iris sphincter preparations. Br J Pharmacol
9. Coleman RA, Humphray JM, Sheldrick RLG, White BP. Gastric antisecretory prostanoids: actions at different prostanoid receptors. Br J Pharmacol
10. Negishi M, Harazono A, Sugimoto Y, Hazato A, Kurozumi S, Ichikawa A. TEI-3356, a highly selective agonist for the prostaglandin EP3
11. Qian YM, Jones RL, Chan KM, Stock AI, Ho JKS. Potent contractile actions of prostanoid EP3
-receptor agonists on human isolated pulmonary artery. Br J Pharmacol
12. Michel G, Seipp U. In vitro studies with the stabilized epoprostenol analog taprostene. Arzneimittelforschung
13. Flohe L, Bohlke H, Frankus E, et al. Designing prostacyclin analogues. Arzneimittelforschung 1983;33:1240-8.
14. Mori S, Takechi S. Synthesis of benzodioxane prostacyclin analogue. Heterocycles
15. Armstrong RA, Jones RL, MacDermot J, Wilson NH. Prostaglandin endoperoxide analogues which are both thromboxane receptor antagonists and prostacyclin mimetics. Br J Pharmacol
16. Seiler SM, Brassard CL, Arnold AJ, Meanwell NA, Fleming JS, Keely SL Jr. Octimibate inhibition of platelet aggregation: stimulation of adenylate cyclase through prostacyclin receptor activation. J Pharmacol Exp Ther
17. Jones RL, Wilson NH, Marr CG, Muir G, Armstrong RA. Diphenylmethylazine prostanoids with prostacyclin-like actions on human platelets. J Lipid Res
18. Meanwell NA, Romine JL, Seiler SM. Non-prostanoid prostacyclin mimetics. Drugs Future
19. Meanwell NA, Romine JL, Rosenfeld MJ, et al. Non-prostanoid prostacyclin mimetics. 5. Structure-activity relationships associated with [3-[4-(4,5-diphenyl-2-oxazolyl)-5-oxazolyl]phenoxy]acetic acid. J Med Chem
20. Merritt JE, Hallam TJ, Brown AM, et al. Octimibate, a potent non-prostanoid inhibitor of platelet aggregation acts via the prostacyclin receptor. Br J Pharmacol
21. Merritt JE, Brown AM, Bund S, et al. Primate vascular responses to octimibate, a non-prostanoid agonist at the prostacyclin receptor. Br J Pharmacol
22. Seiler SM, Brassard CL, Federici ME, et al. 2-[3-[2-(4,5-Diphenyl-2-oxazoly)ethyl]phenoxy]acetic acid (BMY 42393): a new structurally novel prostacyclin partial agonist: 1. Inhibition of platelet aggregation and mechanism of action. Thromb Res
23. Lumley P, White BP, Humphrey PPA. GR 32191, a highly potent and specific thromboxane A2
receptor blocking drug on platelets and vascular airways smooth muscle. Br J Pharmacol
24. Kawai Y, Ohnashi T. Effects of isocarbacyclin, a stable prostacyclin analogue, on monkey isolated cerebral and peripheral arteries. Br J Pharmacol
25. Qian YM, Jones RL. Inhibition of rat colon contractility by prostacyclin (IP-) receptor agonists: involvement of NANC neurotransmission. Br J Pharmacol
26. Hamasaki Y, Saga T, Said SI. Autonomic innervation of pulmonary artery: evidence for non-adrenergic non-cholinergic relaxation. Am Rev Respir Dis
27. Kubota E, Hamasaki Y, Sata T, Saga T, Said SI. Autonomic innervation of pulmonary artery: evidence for nonadrenergic noncholinergic inhibitory system. Exp Lung Res
28. Maggi CA, Patacchini R, Perretti F, et al. Sensory nerves, vascular endothelium and neurogenic relaxation of the guinea pig isolated pulmonary artery. Naunyn Schmiedebergs Arch Pharmacol
29. Hadházy P, Malomvolgyi B, Magyar K, Debreczeni LA, Hutás I. Species dependent relaxation of intrapulmonary arteries (IPA) of rabbits, dogs and humans by prostacyclin. Prostaglandins
30. Norel X, Walch L, Costantino M, et al. M1
muscarinic receptors in human pulmonary arteries. Br J Pharmacol
31. Hamilton TC, Weir SW, Weston AH. Comparison of the effects of BRL 34915 and verapamil on electrical and mechanical activity in rat portal vein. Br J Pharmacol
32. Yamashita T, Masuda Y, Tanaka S. Inhibitory properties of NIP-121, a potassium channel opener, on high potassium and norepinephrine-induced contraction and calcium mobilisation in rat aorta. J Cardiovasc Pharmacol
33. Okabe K, Kajioka S, Nako K, Kitamura K, Kuriyama H, Weston AH. Actions of cromakalim on ionic currents recorded from single smooth muscle cells of the rat portal vein. J Pharmacol Exp Ther
34. McCormack DG, Mak JCW, Coupe MO, Barnes PJ. Calcitonin gene-related peptide vasodilation of human pulmonary vessels. J Appl Physiol
35. Siegel G, Emden J, Wenzel K, Mironneau J, Stock G. Potassium channel activation in vascular smooth muscle. In: Frank GB, Bianchi P, Keurs HEDJ ten, et al., eds. Excitation-contraction coupling in skeletal, cardiac, and smooth muscle. Adv Exp Med Biol;
New York: Plenum Press, 1992:53-72.
36. Jackson WF, Konig A, Dambacher T, Busse R. Prostacyclin-induced vasodilation in rabbit heart is mediated by ATP-sensitive potassium channels. Am J Physiol
37. Negishi M, Hashimoto H, Yatsunami K, Kurozumi S, Ichikawa A. TEI-9063, a stable and highly specific prostacyclin analogue for the prostacyclin receptor in mastocytoma P-815 cells. Prostaglandins
38. Jones RL. Prostacyclin analogues potentiate contraction of the guinea pig vas deferens to electrical field stimulation. Br J Pharmacol
39. Stephenson RP. A modification of receptor theory. Br J Pharmacol
40. Haye-Legrand I, Bourdillat B, Labat C, et al. Relaxation of isolated human pulmonary muscle preparations with prostacyclin (PGI2
) and its analogs. Prostaglandins
41. Nakagawa O, Tanaka I, Usui T, et al. Molecular cloning of human prostacyclin receptor cDNA and its gene expression in cardiovascular system. Circulation
42. Colquhoun D, Henderson R, Ritchie JM. The binding of labelled tetrodotoxin to non-myelinated nerve fibres. J Physiol
43. Jones RL, Wilson NH, Lawrence RA. EP 171: a high affinity thromboxane A2
-mimetic, the actions of which are slowly reversed by receptor blockade. Br J Pharmacol