Distortion of KB estimates of endothelin‐1 ETA and ETB receptor antagonists in pulmonary arteries: Possible role of an endothelin‐1 clearance mechanism

Abstract Dual endothelin ETA and ETB receptor antagonists are approved therapy for pulmonary artery hypertension (PAH). We hypothesized that ETB receptor‐mediated clearance of endothelin‐1 at specific vascular sites may compromise this targeted therapy. Concentration‐response curves (CRC) to endothelin‐1 or the ETB agonist sarafotoxin S6c were constructed, with endothelin receptor antagonists, in various rat and mouse isolated arteries using wire myography or in rat isolated trachea. In rat small mesenteric arteries, bosentan displaced endothelin‐1 CRC competitively indicative of ETA receptor antagonism. In rat small pulmonary arteries, bosentan 10 μmol L−1 left‐shifted the endothelin‐1 CRC, demonstrating potentiation consistent with antagonism of an ETB receptor‐mediated endothelin‐1 clearance mechanism. Removal of endothelium or L‐NAME did not alter the EC 50 or Emax of endothelin‐1 nor increase the antagonism by BQ788. In the presence of BQ788 and L‐NAME, bosentan displayed ETA receptor antagonism. In rat trachea (ETB), bosentan was a competitive ETB antagonist against endothelin‐1 or sarafotoxin S6c. Modeling showed the importance of dual receptor antagonism where the potency ratio of ETA to ETB antagonism is close to unity. In conclusion, the rat pulmonary artery is an example of a special vascular bed where the resistance to antagonism of endothelin‐1 constriction by ET dual antagonists, such as bosentan or the ETB antagonist BQ788, is possibly due to the competition of potentiation of endothelin‐1 by blockade of ETB‐mediated endothelin‐1 clearance located on smooth muscle and antagonism of ETA‐ and ETB‐mediated contraction. This conclusion may have direct application for the efficacy of endothelin‐1 antagonists for treating PAH.


| INTRODUCTION
In rats, 1 rabbits, 2 and nonhuman primates, 3 dual ET A and ET B receptor antagonists or ET B -selective endothelin-1 antagonists increased the immunoreactive endothelin-1 plasma level acutely by 3-to 10-fold. After chronic oral dosing in rats with A-182086, a dual ET A and ET B antagonist, the endothelin-1 plasma levels rose by more than 24-fold after 9 days. 4 Micro positron emission tomography using 18 F-labeled endothelin-1 in anesthetized rats confirmed that endothelin-1 rapidly binds to rat lung and is cleared from the circulation (t 0.5 0.43 minutes). 5 Pretreatment with the ET B -selective antagonist BQ788 decreased the endothelin-1 clearance by 85%.
While this intriguing mechanism of endothelin-1 clearance by ET B receptors was first determined in vivo, we asked, could this mechanism affect the pharmacodynamics of endothelin-1 interactions with ET A and ET B receptors mediating smooth muscle contraction in isolated tissue assays when determining the pK B of endothelin-1 receptor antagonists? The impact of sites of loss of agonist or antagonist concentrations on pK B estimations has been observed in the acid-secreting mouse stomach (figure 1 in Angus and Black 6 ) and further developed by Kenakin. 7 Indeed, we have previously reported that endothelin-1 concentration-contraction curves in rat small interlobar pulmonary arteries were surprisingly LEFT-shifted; ie, endothelin-1 contractions were "potentiated" in the presence of the dual ET A and ET B antagonist bosentan 10 lmol L À1 , 8 an observation that is consistent with blockade of a site of loss of endothelin-1.
Here, we report the pharmacodynamic interactions and analyses of endothelin-1 receptor antagonists in a range of isolated arteries and tracheal smooth muscle preparations with endothelin-1 and the selective ET B receptor agonist venom peptide sarafotoxin S6c. Some arteries were treated with L-NAME or had the endothelial cell layer removed. Our results show that the localized ET B clearance mechanism for endothelin-1 on smooth muscle cells could explain the dramatic effect on the estimation of the dissociation constant for ET A and ET B antagonists when endothelin-1 is used as the agonist and the endothelin-1 clearance mechanism is present.
The conclusions provide a theoretical framework to test for the "ideal" dual ET A and ET B receptor antagonist if significant antagonism is to occur at ET A or ET B constrictor receptors and the ET B receptor-mediated clearance of endothelin-1 is blocked which potentiates the potency of endothelin-1. This clearance mechanism, thus, joins other well-known mechanisms of ET B -mediated endothelin-1 release of thromboxane A 2 , prostacyclin, and nitric oxide that would either enhance or functionally antagonize ET A -or ET B -mediated vasoconstriction. 9 Animal studies are reported in compliance with the ARRIVE guidelines. 13,14 Male Sprague-Dawley rats (280-320 g; Biomedical Sciences Animal Facility, University of Melbourne, Australia) and male Swiss mice (30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40) g; Animal Resources Centre, Murdoch, WA, Australia) were used in this study. Animals were housed (3-4 per high-topped cage with shredded paper bedding) at 22°C on a 12-hour light/dark cycle with access to food and water ad libitum. Rats and mice were individually placed in a secure chamber and deeply anesthetized by inhalation of 5% isoflurane in oxygen, then killed by rapid excision of the heart. The rat and mouse tissues were rapidly excised and placed Group sizes were equal by design; however, variations due to predetermined criteria (described in the methodology) are explained in the figure legends. Animal tissues were randomized to treatment groups.

| Arteries
As previously described, 15  Hg was used (D 20 ). The micrometer was then adjusted to decrease the passive stretch to an equivalent diameter of 90% of D 100 (or 90% of D 20 , as applicable) and the artery remained at that setting of passive stretch for the remainder of the experiment. 15,16 Thirty minutes later the arteries were exposed for 2 minutes to potassium depolarizing solution (K + replacing Na + in PSS, ie, 124 mmol L À1 ; termed KPSS), before replacing with PSS. Subsequent responses were expressed as a % of this KPSS reference contraction in each artery.
Rat or mouse mesenteric arteries that contracted to KPSS with <3 mN force, mouse tail arteries that contracted to KPSS with <20 mN force, or rat and mouse pulmonary arteries that contracted to KPSS with <1 mN force were considered as violations of predetermined criteria. As a further test of viability of the artery, a single 2-minute exposure to 10 lmol L À1 noradrenaline was performed and then replaced with drug-free PSS. To test the integrity of the endothelium, arteries were precontracted with noradrenaline 1 lmol L À1 (which contracts to about 80% of KPSS) and acetylcholine 1 lmol L À1 was added which would normally completely relax the artery in <30 seconds to the baseline force. Some arteries were equilibrated for 30 minutes with L-NAME (N x -nitro-L-arginine methyl ester; 100 lmol L À1 ) and one concentration of an endothelin antagonist (bosentan 1, 10, or 100 lmol L À1 ; BQ788 0.3, 1, or 3 lmol L À1 ). In one study, BQ788 3 lmol L À1 and 0, 1, or 10 lmol L À1 bosentan were equilibrated before the concentrationresponse curve was constructed to endothelin-1. In each study, the artery was then exposed to a single cumulative concentration-contraction curve (0.1 nmol L À1 to 3 lmol L À1 , depending on agonist, tissue, and treatment) to either sarafotoxin S6c or endothelin-1, added in half-log 10 M increments allowing time for the contraction to reach a plateau before raising the concentration.

| Endothelium removal
In rat small pulmonary artery, the normalization procedure was completed before testing the relaxation to acetylcholine 1 lmol L À1 in arteries contracted by U46619 (0.1 lmol L À1 ). The artery passive force was then relaxed, and a human black hair was inserted into the artery lumen. Lateral movement of the hair and careful rotation of the artery loosely suspended on the 2 wires removed the endothelial cells. The passive force was reapplied to the level prior to the endothelial cell removal and the acetylcholine (1 lmol L À1 ) test repeated in the presence of U46619 (0.1 lmol L À1 ). Failure to relax to acetylcholine was considered the functional test of endothelial cell removal. The endothelium-denuded arteries can still deliver a full relaxation response to sodium nitroprusside 1 lmol L À1 .

| Trachea
The main trachea (10 mm long) was dissected free from the rat, cut into 2-to 3-mm-long ring segments, and mounted on wires in 15-mL organ baths (see figure 1 in Angus and Wright), 15 used for large diameter ring segments. In some trachea ring segments, the epithelial cell layer was removed by using a splinter of wood and gently rubbing the lumen for 1 minute. The rings were stretched to a passive F I G U R E 1 Average single exposure concentration-contraction curves to endothelin-1 in rat (A) mesenteric artery (n = 15) and (B) pulmonary artery (n = 15), pretreated with L-NAME 100 lmol L À1 , in the absence Control, (0 lmol L À1 ) or presence of bosentan 1, 10 or 100 lmol L À1 . Data are expressed as % KPSS maximum contraction (y axis). Subsequent responses were expressed as a % of this KPSS reference contraction in each tracheal ring. Tracheae that contracted to KPSS with <1 g force were considered as violations of predetermined criteria. The resting force was readjusted to 1 g and the trachea left to equilibrate for 30 minutes in the absence or presence of bosentan (3, 10, or 30 lmol L À1 ). A single concentration-contraction curve to sarafotoxin S6c or endothelin-1 was constructed up to a maximum concentration of 0.3 lmol L À1 for sarafotoxin S6c or 3 lmol L À1 for endothelin-1. (Auspep, Parkville, Victoria, Australia). All drugs were dissolved in MilliQ water except for endothelin-1 which was dissolved in 10% dimethylformamide to 10 À4 mol L À1 , then diluted in MilliQ water, macitentan which was dissolved in DMSO to 10 À3 mol L À1 , then diluted in MilliQ water, and BQ788 which was dissolved in DMSO to 10 À4 mol L À1 .

| Statistics and analyses
All data are expressed as mean AE SEM from n experiments. The data and analyses comply with the recommendations on experimental design and analysis in pharmacology. 18 All contraction responses to endothelin-1 or sarafotoxin S6c were measured as a % of the Emax

| Clark plot and analyses
Endothelin-1 rapidly activates the respective ET A or ET B receptors before being internalized for recycling (ET A ) or destruction (ET B ) (see Bremnes et al 19 and Paasche et al 20 ). This phenomenon makes it particularly difficult to establish multiple concentration-response curves within a particular artery. In practical terms, the ET A or ET B receptor may be rapidly activated, but the resultant calcium mobilization and contraction takes a considerable time to develop even in small arteries <200 lm diameter. Thus, we routinely designed our experiments around a single cumulative concentration-response curve in the presence or absence of an antagonist concentration.
Our chosen experimental design of only one concentrationresponse curve per tissue does not allow for Schild plot analyses or determination of concentration ratios within tissue. By preference, we used the Clark plot and global fit analysis with its robust advantages. 21 To determine the antagonist dissociation constant (K B ) for each endothelin antagonist, we applied the global regression method 22 that was simplified from that developed originally by Stone and Angus. 21 A computer-based nonlinear regression was performed to solve for K B (pK B = Àlog K B ) by iterative approximation for ALL the endothelin-1 (or sarafotoxin S6c) pEC 50 values in the absence or presence of antagonist (B) concentrations thus: where n is a "power departure" equivalent to allowing the slope of a Schild plot to vary from unity (see Lew and Angus 22 ).
Having solved pK B , the relationship between the mean pEC 50 values of the actual data were plotted against the antagonist concentra- Statistical significance was taken when P < .05.

| Rat mesenteric and pulmonary small arteries
In rat small mesenteric arteries (i.d. 352 AE 6 lm), single endothelin-1 concentration-response curves had a pEC 50 of 8.12 AE 0.02 and an Emax of 108 AE 5% KPSS (n = 4; data not shown). In the presence of L-NAME (100 lmol L À1 ), the pEC 50 for endothelin-1 was 9.35 AE 0.13 (n = 6), significantly higher (17-fold more potent) than in the absence of L-NAME, and the Emax was 123 AE 9% KPSS (Figure 1A). In rat small second-order pulmonary arteries (524 AE 20 lm i.d.), the pEC 50 for endothelin-1 was 7.55 AE 0.20 with an Emax of 124 AE 4% KPSS (n = 5; data not shown). In the presence of L-NAME (100 lmol L À1 ), the pEC 50 was 7.91 AE 0.10, not significantly different from control, and the Emax was 135 AE 7% KPSS (n = 5; Figure 1B). In the presence of L-NAME and bosentan 1 and 10 lmol L À1 , the endothelin-1 concentration-response curves were right-shifted in a competitive manner in the rat mesenteric artery ( Figure 1A), but significantly left-shifted with bosentan 10 lmol L À1 in the rat pulmonary artery ( Figure 1B). In the presence of bosentan 100 lmol L À1 , the endothelin-1 curve was located not significantly different to the control in the presence of L-NAME ( Figure 1B Evidence that L-NAME or endothelial cell removal had abolished the relaxation to acetylcholine 1 lmol L À1 was shown by the result that before treatment with L-NAME or endothelial removal the relaxation to acetylcholine 1 lmol L À1 as a % of the precontractile tone was À54 AE 4% (n = 19) or À56 AE 2% (n = 18), respectively, and after treatment was À2 AE 2% or À1 AE 2%, respectively (data not shown).

| Other arteries
In the mouse, we examined 3 different arteries to determine if the responses to bosentan and endothelin-1 in the small pulmonary artery of the rat could be replicated. In the main pulmonary artery

| Macitentan
In the rat small mesenteric artery, macitentan (0.3 and 1 lmol L À1 ) was a potent competitive endothelin-1 receptor antagonist (Figure 5A). The Clark plot and analyses gave a pK B of 7.05 AE 0.10 (n = 15 points) and fitted the competitive model. The endothelin-1 concentration-contraction curves in the rat small pulmonary artery were completely unaffected by 1 and 10 lmol L À1 macitentan, as shown in Figure 5B.

| Modeling
To model the interaction between endothelin-1 clearance and ET B and ET A receptor antagonism of the contraction response in small pulmonary arteries, we set the following criteria: 1. The ET B receptor-sensitive endothelin-1 clearance mechanism (C ETB ) can decrease the endothelin-1 concentration at the ET A or ET B receptor environment by a maximum of 10-fold (1 pEC 50 unit).

2.
The theoretical dual ET A and ET B receptor antagonist has the same pK B value (8.5) at the "clearance ET B receptor" as at the ET B and ET A receptor modulating contraction.

The efficiency of endothelin-1 at ET A and ET B constrictor recep-
tors is the same.
In Figure 7A, we set the control pEC 50 for the sarafotoxin S6c Àlog M) would rise as shown in Figure 7A and the Schild plot would show competitive antagonism (slope = 1) and pK B 8.5 ( Figure 7B). If We also present the Schild plot for compound A-182086 which was developed with just threefold ET A to ET B selectivity (pK B at ET A 8.5 and at ET B 8; Figure 8C). Thus, in the presence of clearance, the plasma levels would need to be about 1 lmol L À1 (ie, 6 Àlog M) for a 10-fold shift in ET B receptor constrictor activity, while there would be at least 300-fold shift for ET A receptors. Indeed, these peak plasma levels of 4.3 lmol L À1 were achieved in rats given A-182086 10 mgÁkg À1 oral or even greater in dogs (34.5 lmol L À1 ), but significantly less in monkeys (0.16 lmol L À1 ) as the bioavailability varied from 54%, 71%, and 11%, respectively. 24 Finally, we present the theoretical Schild plot for a selective ET B antagonist Ro 46-8443 where the pK B at ET B is 7.1 and 5.7 at ET A receptors ( Figure 8D). This ET B to ET A selectivity of 25 shows an important effect that when even with clearance in operation there is still more ET B constrictor antagonism than ET A .

| DISCUSSION
Our work supports the hypothesis that in very specific vascular beds, the local clearance of endothelin-1 lowers the endothelin-1 concentration that would activate ET A or ET B endothelin receptors. The pK B estimate for ET A , ET B , or mixed ET A and ET B receptor antagonists will be confounded by 2 competing processes: one to potentiate the agonist endothelin-1 and the second to antagonize its action at ET A and/or ET B receptors.
The tissue assays reported here confirm that there are special defined locations in some vascular beds and tracheal tissue that have a major population of ET B receptors on smooth muscle. Functional ET B receptors were defined by the substantial contraction up to the tissue maximum by the potent ET B -selective agonist sarafotoxin S6c.
This agonist is not a substrate for the ET B receptor-sensitive clearance mechanism specifically shown for endothelin-1 and blocked by ET B antagonists. Thus, the rat tracheal ring with agonist sarafotoxin S6c and epithelium intact proved to be a robust assay to define the pK B for ET B antagonists. We calculated the pK B for bosentan as 5.76 AE 0.23 for ET B receptors with sarafotoxin S6c and similarly 5.41 AE 0.28 with endothelin-1. Importantly, the pK B for bosentan and sarafotoxin S6c was the same whether the epithelium was present or absent (pK B 5.76 AE 0.23 and 6.06 AE 0.18, respectively). In the original bosentan report, in rat tracheal rings, the pA 2 was reported as 5.94 AE 0.04 with Schild slope 0.90 AE 0.18. 17 Thus, tracheal smooth muscle ET B receptors mediate contraction, but there is no evidence of clearance of endothelin-1 in this assay.
For the ET A receptor, the analysis is less certain as there is no selective ET A receptor agonist. 25 The main assay used to determine F I G U R E 7 (A) Hypothetical relationship between the pEC 50 values for endothelin-1 or sarafotoxin S6c concentration-contraction curves and the concentration of a theoretical dual ET A and ET B receptor antagonist with a pK B of 8.5 at both receptors. For simplicity, the control (0 antagonist) pEC 50 for sarafotoxin S6c ( ) was set 1 log unit higher (10-fold more potent) than for endothelin-1 ( ). In the presence of ET B receptors and the clearance (C) mechanism for endothelin-1, the maximum clearance was set at 10-fold (1 log unit) so that the pEC 50 in the presence of no antagonist (0) rises 1 log unit (• or ▲). As the ET B antagonism starts to block the endothelin-1 clearance, so the pEC 50 rises (•) but just as does the ET B and ET A antagonism so that the resultant shows the actual pEC 50 is not altered. (B) The Schild plot for endothelin-1 (or sarafotoxin S6c) as the agonist and the dual ET A and ET B antagonist with pK B of 8.5 is shown. Separate theoretical lines are shown for ET A ( ; eg, rat aorta) and ET B ( ; eg, trachea). In the presence of ET B -mediated clearance (C) that removes endothelin-1, as in pulmonary artery, the Schild plot points (▲) move parallel 1 log unit to decrease the potency of the dual antagonist by 10-fold (ie, the pK B of 8.5 becomes 7.5). The y axis is the agonist log(concentration ratio-1) and the x axis shows the concentration of dual ET A and ET B antagonist (Àlog M) the pK B (7.28) for bosentan at ET A receptors was the contraction to endothelin-1 of rat aortic rings, with endothelium removed. 26 Our competitive pK B values for bosentan and endothelin-1 in human large diameter arteries such as pulmonary (i.d. 5.5 mm) and radial First to the role of nitric oxide, L-NAME made no significant difference to the pK B estimation for bosentan in rat small pulmonary artery ( Figure 2). L-NAME (100 lmol L À1 ) was effective in eliminating the release of NO as demonstrated by the complete abolition of the relaxation to acetylcholine (1 lmol L À1 ) in U46619-precontracted arteries. Second, despite L-NAME being present, endothelin-1 was much less potent (lower pEC 50 ) in rat pulmonary small artery than in rat mesenteric artery. Third, in the presence of bosentan 10 lmol L À1 , the pEC 50 for endothelin-1 was right-shifted and F I G U R E 8 (A) Schild plot for a theoretical endothelin-1 antagonist that is 30-fold more selective at ET A vs ET B receptors (pK B : ET A 8.5 and ET B 7.0). Note that in the presence of ET B -mediated clearance, the plasma concentration of the dual antagonist must rise to 10 lmol L À1 to give a 10fold antagonism at ET B constrictor receptors and 3000-fold antagonism at ET A constrictor receptors. (B) Schild plot for a theoretical endothelin-1 antagonist that is 10-fold more selective at ET B vs ET A receptors (pK B : ET B 8.5 and ET A 7.5). In the presence of ET B -mediated clearance, the plasma concentration of the dual antagonist must rise to 0.3 lmol L À1 to give a 10-fold antagonism at both ET B and ET A receptors. (C) Schild plot for endothelin-1 antagonist A-182086 that is threefold more selective for ET A vs ET B receptors (pK B : ET A 8.5 and ET B 8.0; see Table 1). In the presence of ET B -mediated clearance, the plasma concentration of the dual antagonist must rise to 1 lmol L À1 to give a 10-fold antagonism at ET B constrictor receptors and 300-fold antagonism at ET A constrictor receptors. (D) Schild plot for endothelin-1 antagonist Ro 46-8443 that is 25-fold more selective at ET B vs ET A receptors (pK B : ET B 7.1 and ET A 5.7; see Table 1). In the presence of ET B -mediated clearance, the plasma concentration of the dual antagonist must rise to 7.9 lmol L À1 to give a 10-fold antagonism at ET B receptors, with a fivefold antagonism at ET A receptors. The y axis is the agonist log(concentration ratio-1) and the x axis shows the concentration of dual ET A and ET B antagonist (Àlog M) lowered to 7.2 (Àlog M) in the mesenteric artery, while in contrast, it was left-shifted and raised to a pEC 50 of 8.7 compared with control in the pulmonary artery ( Figure 1A,B). We suggest that this anomalous result and inability to determine a pK B with bosentan in rat pulmonary artery is explained by the continuous removal of endothelin-1 by the ET B receptor-sensitive clearance mechanism found in this particular artery, but not in the rat mesenteric artery or aorta, nor human large pulmonary or radial artery. 27 Further, direct functional evidence of the clearance of endothelin-1 in rat pulmonary artery comes from the selective ET B antagonist BQ788 assay. With the agonist sarafotoxin S6c, and L-NAME present, the pattern of BQ788 competitive antagonism shows right- arteries, the ET A receptor antagonist BMS182874 was ineffective against low concentrations of endothelin-1. 28 The finding that endothelium removal did not affect the EC 50 nor Emax to endothelin-1 or change the action of BQ788 in the rat small pulmonary artery compared to endothelium-intact tissues suggests that the arterial smooth muscle cells are the primary location of the proposed clearance mechanism (Figure 9).
In earlier work, Hay et al 29 reported that in rabbit pulmonary artery, sarafotoxin S6c gave a pK B of 7.7 for the mixed ET A and ET B receptor antagonist SB209670, but 6.7 when endothelin-1 was the agonist. In human small pulmonary arteries (150-200 lm i.d.) sarafotoxin S6c was more than 100-fold more potent than endothelin-1 and the authors concluded that both ET A and ET B receptor antagonists are required to antagonize endothelin-1. 28 We also found that the apparently weak antagonism of endothelin-1 by bosentan in rat pulmonary arteries is shared with ambrisentan and macitentan (Figure 5). Indeed, these latter 2 endothelin-1 antagonists are more ET A than ET B receptor selective (  (Figures 1 and 8A) suggests that to obtain a 10-fold antagonism of the ET B constrictor receptor in the presence of clearance, then a plasma concentration of 3000 times higher than the pK B at ET A receptors would be required. For bosentan, for example, plasma levels would need to be in the range of 200 lmol L À1 ! If these levels are not obtained, the antagonist would generally behave only as an effective ET A antagonist in the clinic.
Our modeling suggests that a 10-fold selective ET B vs ET A antagonist might be ideal in antagonizing the pulmonary artery ET B receptors in the presence of CLEARANCE ( Figure 8B). Ro 46-8443 35 is 25-fold selective for ET B vs ET A , and modeling would suggest that with a pK B of 7.1 at ET B receptors ( Figure 8D), a plasma level would be required of nearly 10 lmol L À1 to give a 10-fold antagonism of ET B receptors, but ET A antagonism would not be sufficient if inhibition of clearance presented a higher level of endothelin-1. Another nonselective and potent ET A and ET B antagonist with selectivity ratio of just 3, A-182086 (Table 1), has been used in animals and shows that effective ET A and ET B receptor antagonism was achieved after oral dosing. 24 Given that any antagonism of clearance will raise plasma endothelin-1 levels, there must be sufficient ET A receptor antagonism present to obviate vasoconstriction from this raised endothelin-1 concentration. Theoretically then, an ET A vs ET B selectivity of 10fold would be sufficient, provided a high plasma concentration is achieved for ET B antagonism. From Table 1, we predict that given clearance of endothelin-1 in important tissues such as pulmonary artery, the effective ET B antagonism is 10-fold weaker so that the ET A to ET B + clearance selectivity ratio increases by 10-fold. In effect, this suggests that the 3 antagonists in the clinic for pulmonary artery hypertension are principally ET A -selective agents. The  Rat aorta (without endothelium) and agonist endothelin-1. i Rat trachea (without epithelium) and agonist sarafotoxin S6c. j pK B for ET B receptors under the influence of endothelin-1 clearance theoretically taken to be 10-fold (ie, 1 log unit). k ET A to ET B selectivity ratio calculated as antilog (pK B ET A À pK B ET B ). l ET A to ET B + C selectivity ratio calculated as antilog (pK B ET A À pK B ET B + C). m With endothelium. n Rabbit pulmonary artery (without endothelium).