Endothelium-derived nitric oxide causes vasodilation in part by increasing the dilator activity of other endothelium-derived mediators, including prostacyclin and a K+ATP channel-dependent hyperpolarizing factor. Although previous studies have proposed that isoflurane (ISO) depresses endothelium-dependent vasorelaxation by inhibiting endothelium-derived nitric oxide activity, the effects of ISO on the interactions among endothelium-derived dilators have not been characterized. The aim of this study was to determine the mechanisms underlying the inhibitory effect of ISO on endothelium-dependent relaxation in canine pulmonary arteries. Specifically, the goal was to assess the effects of ISO on the individual actions and on the synergistic interactions of these endothelium-derived mediators.
Canine pulmonary arterial rings were suspended for isometric tension recording. The effects of 1 minimum alveolar concentration ISO (0.4 mM) on vasorelaxation responses to bradykinin, A23187, acetylcholine, cromakalim, and SIN-1 were assessed in phenylephrine-precontracted rings with and without pretreatment with a nitric oxide synthase inhibitor (N omega-nitro-L-arginine methyl ester; L-NAME), a cyclooxygenase inhibitor (indomethacin), or a K+ATP channel inhibitor (glybenclamide).
Isofluane attenuated pulmonary vasorelaxation induced by bradykinin, A23187, and cromakalim but had no effect on relaxation induced by acetylcholine or SIN-1. Neither the nitric oxide-mediated nor the prostacyclin-mediated components of relaxation induced by bradykinin and A23187 were altered by ISO. However, ISO abolished the K+ATP-mediated component of relaxation and the K+ATP-dependent synergistic interaction between nitric oxide and prostacyclin.
These results suggest that ISO selectively attenuates endothelium-dependent relaxation in canine pulmonary arteries. It exerts its inhibitory effect by interfering with a synergistic interaction between nitric oxide and prostacyclin, possibly via an effect on K+ATP channels.
Inhalational anesthetics are known to selectively inhibit endothelium-dependent relaxation in isolated systemic arteries, [1–4] as well as in the intact canine pulmonary circulation. [5,6] However, the cellular mechanism underlying this inhibition has not been fully elucidated. Characterizing the cellular pathways by which inhalational anesthetics exert their effects will provide fundamental information about the mechanism of action of these anesthetic agents. The inhibitory effects of inhalational anesthetics have been attributed primarily to a reduction in the synthesis, release, or activity of nitric oxide (NO), the predominant endothelium-derived relaxing factor. [1–3] However, endothelium-dependent relaxation results from the release of multiple mediators from the endothelium, including NO, prostacyclin, and endothelium-derived hyperpolarizing factor, that act and interact on vascular smooth muscle cells.  We recently reported that a synergistic interaction between NO and prostacyclin occurs in isolated canine pulmonary arteries during relaxation induced by certain endothelial cell activators.  This synergy between NO and prostacyclin appears to be mediated by the activation of adenosine triphosphate (ATP)-sensitive potassium (K sup +ATP) channels, suggesting the possible involvement of an endothelium-derived hyperpolarizing factor. Thus several different endothelium-derived mediators, acting alone or in conjunction with other mediators, may be the target for the inhibitory influences of anesthetic agents. In the present study, our goal was to evaluate the extent to which isoflurane inhibits endothelium-dependent relaxation in isolated canine pulmonary arteries. Furthermore, we assessed the effects of isoflurane on the individual actions of endothelium-derived NO and prostacyclin, as well as their interactions, in response to endothelial activators. Our results indicate that isoflurane selectively attenuates the pulmonary vasorelaxant response to specific endothelial activators by inhibiting the K sup +ATP-dependent synergistic interaction between NO and prostacyclin.
Materials and Methods
All experimental procedures and protocols were approved by the Institutional Animal Care and Use Committee.
Organ Chamber Experiments
The present in vitro studies were performed using canine pulmonary arterial rings harvested from healthy male mongrel dogs that weighed 20–25 kg. The dogs were anesthetized with pentobarbital sodium (30 mg/kg given intravenously) and fentanyl citrate (15 micro gram/kg given intravenously). After tracheal intubation, the lungs were mechanically ventilated (Harvard respirator, Natick, MA). The chest was opened via a left lateral thoracotomy through the fifth intercostal space. A catheter was positioned in the left atrium through the left atrial appendage. The dogs were exsanguinated by controlled hemorrhage and killed with a bolus of saturated potassium chloride injected into the left atrium. Injection at this site arrests the heart before the potassium chloride perfuses the pulmonary circulation. The heart and lungs were removed from the thorax en bloc, and the lower right and lower left lung lobes were dissected free. Intralobar pulmonary arteries were carefully dissected and immersed in cold modified Krebs-Ringer bicarbonate solution composed of 118.3 mM NaCl, 4.7 mM KCl, 1.2 mM MgS04, 1.2 mM KH2PO4, 2.5 mM CaCl2, 25 mM NaHCO3, 0.016 mM Ca-EDTA, and 11.1 mM glucose. The arteries were cut into 0.5-cm length rings with care taken not to damage the endothelium. The rings were suspended between two stainless steel stirrups in organ chambers filled with 25 ml modified Krebs-Ringer bicarbonate solution (37 degrees Celsius) gassed with 95% oxygen and 5% carbon dioxide. One of the stirrups was anchored and the other was connected to a strain gauge (Grass FT03, Quincy, MA) to measure isometric force (using a Grass polygraph, model 7E).
Pulmonary arterial rings were stretched at 10-min intervals in increments of 0.5 g to reach optimal resting tension. Optimal resting tension was defined as the minimal amount of stretch required to achieve the largest contractile response to 20 mM KCl, and was determined in preliminary experiments to be 4 g for the arteries used in these studies (2–4 mm inner diameter). After the arterial rings had been stretched to their optimal resting tension, the contractile response to 60 mM KCl was assessed. After washout of KCl from the organ chamber and the return of isometric tension to prestimulation values, a concentration-response curve to phenylephrine was performed in each ring. This was achieved by increasing the concentration of phenylephrine in half-log increments (from 10 sup -8 to 3 x 10 sup -5 M) after the response to each preceding concentration had reached a steady state. Preliminary experiments indicated that phenylephrine caused sympathetic beta adrenergic relaxation in addition to alpha adrenergic contraction in these arteries. Thus the rings were pretreated with propranolol (5 x 10 sup -6 M; incubated for 30 min) before phenylephrine administration in all protocols. After washout of phenylephrine from the organ chamber and return of isometric tension to baseline values, the rings were again pretreated with propranolol and contracted to 50% of their maximal response to phenylephrine (ED50level of tension). At steady-state tension, concentration-response curves to the endothelial cell activators bradykinin, calcium ionophore (A23187), and acetylcholine were generated. Some vascular rings were exposed to two endothelial activators separated by thorough washouts. For these arterial rings, the sequence was either bradykinin followed by A23187 or acetylcholine followed by A23187. In a separate series, a concentration-response curve to the K sup +ATPchannel activator, cromakalim, was generated in endothelium-intact vascular rings. Concentration-response curves to the endothelium-independent vasorelaxant, SIN-1, were generated in endothelium-denuded vascular rings. This vasorelaxant is a NO donor that directly activates vascular smooth muscle-soluble guanylate cyclase. 
To identify the specific endothelium-derived mediators involved in the relaxation responses to bradykinin, A23187, and acetylcholine, endothelium-intact vascular rings obtained from the same lung lobe were incubated with one or more of the following pharmacologic inhibitors: N sup omega -nitro-L-arginine methyl ester (L-NAME, 3 x 10 sup -5 M), an inhibitor of NO synthase; indomethacin (10 sup -5 M), an inhibitor of cyclooxygenase; and glybenclamide (10 sup -6 M), an inhibitor of K sup + sub ATP channels. Pulmonary arterial rings were pretreated with these inhibitors for 30 min before phenylephrine contraction to the ED50level of tension. The inhibitors remained in the bath solution for the duration of the exposure to the endothelial activators. Vascular responses to bradykinin, A23187, and acetylcholine in inhibitor-treated rings were compared with responses measured in untreated rings stimulated by these endothelial activators. None of the inhibitors had an effect on baseline tension.
The effects of a clinically relevant concentration of isoflurane anesthesia (0.4 mM; equivalent to approximately 1 minimum alveolar concentration for a dog) on the vasorelaxation responses to bradykinin, A23187, acetylcholine, cromakalim, and SIN-1 were compared with control responses in size- and position-matched pulmonary arterial rings. The anesthetic was delivered to the organ chambers at a constant gas flow rate (1 l/min) via a vaporizer (Isotec 3; Ohmeda, Madison, WI). The concentration of isoflurane in the organ chamber solution was measured using high-performance liquid chromatography, and calculated using the method described by Adams and coworkers.*A steady-state equilibrium concentration of 0.4 mM isoflurane was measured after 30 min of exposure. At this point, concentration-effect curves to bradykinin, A23187, acetylcholine, cromakalim, and SIN-1 were generated.
Drugs and Solutions
A23187, acetylcholine chloride, cromakalim, indomethacin, L-NAME HCl, glybenclamide, propranolol HCl, and phenylephrine HCl were obtained from Sigma Chemical Company (St. Louis, MO). Bradykinin was obtained from Bachem (King of Prussia, PA), SIN-1 from Casella AG (Frankfurt, Germany), and isoflurane from Anaquest (Madison, WI). All concentrations are expressed as the final molar concentration in the organ chamber bath. Agonist and inhibitor solutions were prepared each day and were kept on ice during the experiment. All drugs were dissolved in distilled water with the exception of indomethacin (which was dissolved in NaHCO3and diluted in distilled water; final bath concentration of NaHCO3was 0.2 mM), A23187 (which was dissolved in DMSO and diluted in distilled water; final bath concentration of DMSO ranged from 0.00004–0.013% v/v depending on the concentration of A23187), and glybenclamide (which was dissolved in methanol and diluted in distilled water; final bath concentration of methanol was 0.16% v/v). These vehicles have no effect on relaxation responses at the concentrations used for drug preparation. 
Results are expressed as means +/- SEM. Responses to phenylephrine are expressed as grams of tension and responses to the vasorelaxants are expressed as a percentage of the contraction to phenylephrine. One hundred percent of phenylephrine-induced contraction in the figures represents 1.97 +/- 0.07 g absolute tension across all protocols. The effect of isoflurane on the concentration-effect curves for each vasorelaxant was evaluated by comparing the concentration of agonist (inhibitory concentration [IC]) causing 50% relaxation of the contraction to phenylephrine. This value was interpolated from the linear portion of the concentration-effect curve by regression analysis and is presented as the log IC50. The Student's t test for paired samples was used to compare the log IC50values. Differences were considered to be statistically significant when P < 0.05.
The method used to quantitate the contribution of individual mediators (e.g., NO and prostacyclin) to the total relaxation response to an agonist involved calculating the areas above the individual concentration-effect curves.  The area above the agonist concentration-effect curve representing the total response to the agonist (i.e., in the absence of inhibitors or isoflurane) was assigned a value of 100%. The area above the concentration-effect curve for an agonist after pretreatment with an inhibitor (e.g. area above the bradykinin concentration-effect curve after pretreatment with indomethacin) was calculated as a percentage of the total relaxation response to that agonist. Areas are expressed as means +/- SEM and were compared using Student's t test for paired samples.
Effect of Isoflurane on Phenylephrine-induced Contraction
In canine pulmonary arteries, isoflurane potentiated (P < 0.05) the contractile response to phenylephrine, causing a leftward shift in the concentration-effect curve (IC50(ISO)=-7.51 +/- 0.06 vs. IC50(cont)=-7. 18 +/- 0.10), with no change in the maximal response to the alpha agonist.
Effect of Isoflurane on Pulmonary Vasorelaxation
Isoflurane inhibited (P < 0.05) endothelium-dependent relaxation induced by bradykinin, causing a rightward shift in the concentration-effect curve (IC50(ISO)=-8.58 +/- 0.14 vs. IC50(cont)=- 9.00 +/- 0.03), with no change in the maximal response to the endothelial activator (Figure 1, top). A similar inhibitory effect was observed with endothelium-dependent relaxation to A23187, with isoflurane causing a significant (P < 0.05) rightward shift in the concentration-effect curve (IC50(ISO)=-7.35 +/- 0.05 vs. IC50(cont)-7.72 +/- 0.12;Figure 1, bottom). In contrast, isoflurane had no significant effect on the endothelium-dependent relaxation induced by acetylcholine (IC50(ISO)=-7.60 +/- 0.11 vs. IC50(cont)=-7.77 +/- 0.10;Figure 2, top). Similarly, the vasorelaxant response to the NO donor, SIN-1, was unaltered by isoflurane (IC50(ISO)=- 7.51 +/- 0.13 vs. IC50(cont)=-7.68 +/- 0.15;Figure 2, bottom).
Effect of Isoflurane on the Interaction between NO and Prostacyclin
Bradykinin-induced Pulmonary Vasorelaxation. The interaction between endothelium-derived NO and prostacyclin can be analyzed most effectively at low concentrations of endothelial cell activators.  Combined inhibition of NO synthase and cyclooxygenase (with L-NAME, 3 x 10 sup -5 M, and indomethacin, 10 sup -5 M, respectively) abolished relaxation induced by low concentrations of bradykinin (10 sup -50 to 3 x 10 sup -9 M;Figure 3(A), L-NAME + INDO curve), suggesting that the vasorelaxation was mediated by NO and prostacyclin. To observe the interactions between NO and prostacyclin, the action of each mediator was assessed when the response to the other mediator was inhibited. The area above the control curve (total relaxation response) was assigned a value of 100%, and the individual components of bradykinin-induced vasorelaxation were quantified and expressed as a percentage of the total relaxation response. In control arteries, the prostacyclin-mediated component of relaxation (L-NAME-resistant but indomethacin-sensitive component; stippled area of Figure 3(A)) accounted for 16%+/- 9% of the total relaxation response to bradykinin, whereas the NO-mediated component of relaxation (indomethacin-resistant but L-NAME-sensitive component; lined area of Figure 3(C)) accounted for 16%+/- 5% of the total relaxation response to bradykinin. Although isoflurane attenuated the total relaxation response to bradykinin (control vs. isoflurane curves;Figure 3), the anesthetic had no effect on either the prostacyclin-mediated component of relaxation (18%+/- 5%; stippled area of Figure 3(B)) or the NO-mediated component of relaxation (36%+/- 11%; lined area of Figure 3(D)). Values for areas above the concentration-effect curves for bradykinin are summarized in Figure 4(A). The total relaxation response to bradykinin was greater than the sum of the individual NO-mediated and prostacyclin-mediated components (Figure 4(A)), indicating a synergistic interaction between these mediators. In the presence of isoflurane, the total relaxation response to bradykinin was no longer greater than the sum of the individual vasorelaxant activities of NO and prostacyclin; that is, isoflurane abolished their synergistic interaction (Figure 4(A)).
A23187-induced Pulmonary Vasorelaxation. Combined inhibition with L-NAME and indomethacin also abolished relaxation induced by low concentrations of A23187 (Figure 5(A), L-NAME + INDO curve), suggesting that the vasorelaxation was mediated by NO and prostacyclin. In control arteries, the prostacyclin-mediated component of relaxation (L-NAME-resistant but indomethacin-sensitive component; stippled area of Figure 5(A)) accounted for 20%+/- 9% of the total relaxation response to A23187, whereas the NO-mediated component of relaxation (indomethacin-resistant but L-NAME-sensitive component; lined area of Figure 5(C)) accounted for 46%+/- 9%. Although isoflurane attenuated the overall relaxation response to A23187 (control vs. isoflurane curves;Figure 5), the anesthetic had no effect on the individual prostacyclin-mediated (11%+/- 6%;Figure 5(B)) or NO-mediated (50%+/- 3%;Figure 5(D)) components of relaxation. Values for areas above the concentration-effect curves for A23187 are summarized in Figure 4(B). In control rings, the total relaxation response to A23187 was greater than the sum of the individual vasorelaxant components mediated by NO and prostacyclin (Figure 4(B)), indicating a synergistic interaction between these mediators. In the presence of isoflurane, the total relaxation response to A23187 was no longer greater than the sum of the individual vasorelaxant activities of NO and prostacyclin; that is, isoflurane abolished their synergistic interaction (Figure 4(B)).
Effect of Isoflurane on K sup + sub ATP -mediated Pulmonary Vasorelaxation
In control rings, inhibition of K sup +ATPchannels with glybenclamide (10 sup -6 M) attenuated (P < 0.05) endothelium-dependent relaxation to bradykinin (IC50(GLYB)=-8.51 +/- 0.17 vs. IC50(cont)=-9.00 +/- 0.03;Figure 6(A)). In contrast, in isoflurane-treated arteries, inhibition of K sup +ATPchannels had no effect on the response to bradykinin (IC50(ISO)=-8.58 +/- 0.14 vs. IC50(GLYB + ISO)=-8.56 +/- 0.11;Figure 6(B)). Glybenclamide also attenuated the relaxation response to A23187 in control rings (IC50(GLYB)=-7.02 +/- 0.09 vs. IC50(cont)=-7.72 +/- 0.12;Figure 6(C)). In contrast, in isoflurane-treated rings K sup +ATPchannel inhibition had no effect on the response to A23187 (IC50(ISO)=-7.35 +/- 0.05 vs. IC50(GLYB + ISO)=-7.26 +/- 0.18;Figure 6(D)).
The component of endothelium-dependent relaxation mediated solely by the K sup +ATP-dependent endothelium-derived hyperpolarizing factor can be observed after inhibition of NO and prostacyclin using L-NAME and indomethacin. In the presence of combined NO synthase inhibition and cyclooxygenase inhibition, glybenclamide attenuated (P < 0.05) the vasorelaxant response to bradykinin (area between the L-NAME + INDO curve and the GLYB + L-NAME + INDO curve;Figure 7(A)). Isoflurane abolished this K sup +ATP-dependent component of bradykinin-induced relaxation (Figure 7(B)). Isoflurane also attenuated (P < 0.05) the relaxation response to the K sup +ATPchannel activator, cromakalim, causing a rightward shift in the concentration-effect curve (IC50(ISO)=-7.46 +/- 0.24 vs. IC50(cont)-7.77 +/- 0.01;Figure 8).
Isoflurane selectively inhibits endothelium-dependent relaxation in isolated systemic arteries, [2–4] as well as in the intact pulmonary circulation.  One of the primary molecular targets for the depressant effect of isoflurane, and other inhalational anesthetics, is believed to be endothelium-derived NO. [1–3] Nitric oxide can induce vasodilatation either directly by increasing vascular smooth muscle cyclic guanosine monophosphate or K sup + currents, or indirectly by amplifying responses to other endothelium-derived mediators. [11–13] The effect of isoflurane on the interactions among endothelium-derived mediators has not been explored before. In the present study, isoflurane selectively attenuated endothelium-dependent pulmonary vasorelaxation to bradykinin and A23187, but not the vasorelaxant response to acetylcholine. Isoflurane did not reduce the individual NO- or prostacyclin-mediated components of relaxation but rather abolished their synergistic interaction. Furthermore, isoflurane attenuated the vasorelaxant response to a K sup +ATPagonist, cromakalim, and exerted an inhibitory effect on the K sup +ATP-mediated component of bradykinin and A23187 vasorelaxation. These results suggest that isoflurane depresses endothelium-dependent relaxation in pulmonary arteries by inhibiting the synergistic interaction between endothelium-derived NO and prostacyclin. This action, in turn, appears to be mediated by an inhibitory effect of isoflurane on the activity of K sup +ATPchannels.
Although isoflurane inhibited endothelium-dependent relaxation induced by bradykinin and A23187, it did not attenuate endothelium-dependent relaxation in response to acetylcholine. This rules out the possibility that isoflurane exerts a nonselective depressant effect on the endothelium. Furthermore, isoflurane did not impair the direct vascular smooth muscle relaxation response induced by the NO donor, SIN-1, indicating that isoflurane does not cause an impairment in vascular smooth muscle dilator function. Isoflurane inhibits some aspect of endothelium-dependent relaxation/signaling that is used by bradykinin and A23187, but not by acetylcholine. A23187 is a calcium ionophore that stimulates endothelium-dependent vasorelaxation in a receptor-independent manner. Because the vasorelaxant response to A23187 was also attenuated by isoflurane, the signaling mechanism inhibited by the anesthetic is unlikely to involve endothelial receptor/G-protein coupling.
To assess the individual components of endothelium-dependent relaxation, the action of each mediator (NO or prostacyclin) was assessed when the response to the other mediator was inhibited. In the presence of NO synthase inhibition, isoflurane had no effect on the indomethacin-sensitive component of relaxation. Similarly, in the presence of cyclooxygenase inhibition, isoflurane did not alter the L-NAME-sensitive component of relaxation. Thus isoflurane did not attenuate either the NO- or the prostacyclin-mediated components of relaxation to bradykinin or A23187. Because isoflurane inhibited the overall responses to these endothelial activators, our results suggest that isoflurane acted by inhibiting the synergistic interaction between NO and prostacyclin. In control rings, the synergistic interaction between NO and prostacyclin depends on the activation of K sup +ATPchannels.  When this synergistic interaction was inhibited by glybenclamide in the present study, isoflurane no longer attenuated bradykinin or A23187 relaxation. This suggests that the anesthetic may be inhibiting the production or activity of a K sup +ATP-dependent, endothelium-derived hyperpolarizing factor. We also assessed the effects of isoflurane on the vasorelaxant response to the K sup +ATPchannel agonist, cromakalim. Isoflurane attenuated cromakalim-induced relaxation. We previously observed that isoflurane also inhibits the pulmonary vasodilator response to the K sup +ATPchannel activator, lemakalim, in the intact pulmonary circulation.  Together these results support the concept that isoflurane inhibits endothelium-dependent relaxation by impairing a K sup +ATPchannel-dependent synergy between NO and prostacyclin.
Interactions between inhalational anesthetics and various ion channels have been reported by other investigators. [15–17] There is some evidence to suggest that inhalational anesthetics bind specifically to proteins, such as ion channels, and alter their function directly. [16,18,19] Other evidence suggests that anesthetics may alter the function of integral membrane proteins indirectly by modifying the physical properties of the membrane. [19,20] Additional studies will be required to determine the exact molecular mechanism by which isoflurane inhibits K sup +ATPchannel activity.
Our proposed mechanism for the inhibitory action of isoflurane on endothelium-dependent relaxation is consistent with the lack of effect of isoflurane on acetylcholine-induced relaxation. In control rings, acetylcholine-induced relaxation is mediated predominantly by NO, with contributions from a K sup +ATP-independent, endothelium-derived hyperpolarizing factor at higher concentrations.  Unlike bradykinin and A23187, relaxation induced by acetylcholine does not involve a synergistic interaction between NO and prostacyclin, nor does it involve the activation of K sup +ATPchannels.  Thus it is likely that isoflurane did not attenuate acetylcholine-induced relaxation because the response to this endothelial activator does not involve a K sup +ATP-dependent synergy between NO and prostacyclin. Some investigators have found an inhibitory effect of isoflurane on acetylcholine-induced vasorelaxation in systemic vessels. [2,3] These differential effects of isoflurane on acetylcholine-induced vasorelaxation may be due to differences in species or vessel type; for example different levels of synergy may occur in different blood vessels.
In summary, a clinically relevant concentration of isoflurane (1 minimum alveolar concentration) attenuated endothelium-dependent relaxation induced by bradykinin and A23187 but had no effect on relaxation induced by acetylcholine or SIN-1. Bradykinin- and A23187-induced relaxation involves a synergistic interaction between NO and prostacyclin. Isoflurane had no effect on the individual vasorelaxant activities of either NO or prostacyclin but abolished their synergistic interaction. Isoflurane also inhibited the vasorelaxant response to a K sup +ATPchannel agonist. Thus, in isolated canine pulmonary arteries, isoflurane attenuates endothelium-dependent vasorelaxation by inhibiting K sup +ATPchannel activity, which regulates a synergistic interaction between two primary endothelium-derived relaxing factors, NO and prostacyclin.
The authors thank Tammy Aleskowitch and Rosie Cousins for technical assistance and Ronnie Sanders for preparing the manuscript.
*Adams RJ, Blanck TJJ, Johns RA, Merin RG, Muldoon SM, Nelson TE, Rusy BF: Correlation of experimental volatile anesthetic concentrations to in vivo situations. Subcellular mechanisms of anesthetic action in muscle. Medical College of Georgia, 1988, pp 1–17.