Secondary Logo

Journal Logo


Volatile anaesthetics restore bradykinin and serotonin-induced coronary vasodilation after blocking nitric oxide synthase: lack of anaesthetic effects on KATP channels and prostaglandin pathways

Stowe, D. F.; Heisner, J. S.; Chung, W. W.*; Fujita, S.

Author Information
European Journal of Anaesthesiology: April 2001 - Volume 18 - Issue 4 - p 219-230



Release of nitric oxide (NO) and prostaglandins (PG) by the vascular endothelium and ATP-regulated K+ channel opening in vascular smooth muscle has been demonstrated to contribute importantly to coronary vasodilation elicited by a variety of physiological and pharmacological stimuli. There are conflicting reports whether volatile anaesthetics alter endotheliumdependent vasodilation by the nitric oxide synthase (NOS) [1, 2, 3, 4, 5, 6, 7, 8] or cyclo-oxygenase (COX)-PG synthase (PGS) pathways [9, 10, 11, 12]. Most studies have been undertaken in vitro in endothelial/vessel strip or coculture preparations in the absence and presence of COX [9,10] or NOS antagonism [1,2,13, 14, 15, 16, 17, 18, 19, 20]. Anaesthetics also have been proposed to cause vasodilation, at least in part, by opening sarcolemmal KATP channels [21, 22, 23, 24, 25, 26]. We reported recently that volatile anaesthetics did not interfere with bradykinin (BK)-induced coronary flow increases, NO release and L-citrulline production in the isolated guinea pig heart [27]. There remained the possibility that anaesthetics alter endothelium-mediated vascular responses after inhibiting NOS or COX or blocking KATP channels in the same model. To test this we examined further whether the anaesthetics sevoflurane (SEV), isoflurane (ISO) and halothane (HAL) alter bradykinin or serotonin [5-hydroxytryptamine (5-HT)]-induced increases in coronary artery flow and decreases in percentage oxygen extraction in guinea pig hearts after blocking KATP channels, or after inhibiting basal and stimulated release of nitric oxide or prostaglandin. Drugs used to test this were glibenclamide (GLIB) to block KATP channels, nitro-L-arginine methylester (L-NAME) to inhibit NOS and indomethacin (INDO) to inhibit the COX-PGS pathway.


Langendorff heart preparation

The investigation conformed to the US National Institutes of Health (NIH publication No. 85-23, revised 1995) laboratory animal care and use standards and was approved by the Medical College of Wisconsin animal studies committee. Description of the surgical preparation for this model has been reported in detail previously [27, 28, 29]. Hearts were harvested from albino English short-haired guinea pigs (250-300g). Each heart was perfused in retrograde fashion through the aorta at 55mmHg with a 95% oxygenated, modified Krebs-Ringer's (KR) solution as described previously [27, 28, 29]. Coronary sinus effluent was collected by placing a small, gas impermeable cannula into the right ventricle through the pulmonary artery after ligating the superior and inferior venae cavae. Coronary outflow (coronary sinus) PO2 was measured continuously on-line with a miniature thermostabile Clark O2 electrode (Instech Laboratories, Model 203B, Plymouth Meeting, PA, USA). Perfusate, bath and O2 electrode temperatures were maintained tightly at 37.2 ± 0.1°C using a thermostatically controlled water circulator to jacketed glass tubing, bath and aluminum heat exchangers. Coronary inflow and effluent pH and PO2 and PCO2 were measured during each manoeuvre off-line at 37°C with an intermittently self-calibrating gas analyser. Isovolumetric left ventricular pressure, coronary flow, percentage oxygen extraction and spontaneous heart rate were measured continuously as described previously [27, 28, 29]. Percentage oxygen extraction was utilized as an indicator of vascular tone independent of cardiac metabolism with the assumption that a change in oxygen need is met by a change in oxygen delivery in the absence of ischaemia. Percentage oxygen extraction was calculated as the inflow and outflow tension difference multiplied by 100 and divided by the inflow PO2. Inflow perfusate O2 tension was kept constant at 0.7 kPa above atmospheric pressure. Heart rate, inflow and outflow oxygen (kPa), coronary flow, isovolumetric systolic and diastolic left ventricular pressures (kPa) were displayed continuously an eight-channel chart recorder. Adenosine (0.2mL of 200 μM stock solution) was injected as a bolus to measure maximal coronary flow after the heart had initially stabilized. Adenosine was given again at the end of each experiment to observe any change in maximum flow reserve.

Anaesthetics were delivered by calibrated agent specific vaporizers and anaesthetic concentrations were measured in the effluent by gas chromatography as described previously [28,29]. Anaesthetics were delivered at two concentrations approximating 1-2 MAC (minimum alveolar concentration). Coronary sinus anaesthetic concentrations for sevoflurane (n = 47) were 0.31 ± 0.03mM and 0.63 ± 0.04mM, which, respectively, gave concentrations of 2.2 ± 0.2 and 4.5 ± 0.3 Vol.%, or approximately 1.0 and 2.2 MAC. For the low and high concentrations of isoflurane (n = 15) the corresponding values were 0.20 ± 0.02 and 0.35±0.03mM, 0.9±0.01 and 1.6±0.0 Vol.% and 1.3 MAC; values for halothane (n = 16) were 0.19 ± 0.02 and 0.38 ± 0.01 mM, 0.06 ± 0.0 and 1.3 ± 0.2 Vol.%, or 0.8 and 1.6 MAC.


The hypothesis was to test whether anaesthetics influence vasodilatory responses to endothelium dependent drugs after blocking one of three pathways that modulate vascular tone. After baseline control values were re-established, 84 hearts were randomized into three series of studies identified by the three blockers glibenclamide (4 μM), L-NAME (100μM) and indomethacin (100μM). Before and during blockade in the presence or absence of an anaesthetic, these hearts was infused with random ordered concentrations of 10−12 bradykinin [27,29] or 10−8-10−6 M 5-hydroxytryptamine (5-HT) [29,30] for 2min at each concentration with a 5 min intervening control period. After administering a given blocker, each heart was exposed to two randomized concentrations of halothane, isoflurane or sevoflurane. Thus, a range of bradykinin or 5-HT responses were tested 3-4 times in each heart. Because of the potential for long-lasting effects of glibenclamide, L-NAME and indomethacin, step increases in bradykinin or 5-HT were conducted first in the absence of these blockers. Responses to bradykinin or 5-HT alone did not change over time. Anaesthetic effects on endothelium-dependent vasodilation were not routinely conducted in the absence of blockers because we have reported previously that anaesthetics do not alter bradykinin-induced vasodilatory responses in this model [27] and because each experiment comprised three treatments and a control, each with 14 measurements, and lasted approximately 240 min. Ancillary experiments to test bradykinin and 5-hydroxytryptamine responses, also in the presence of an anaesthetic, confirmed results of the earlier study.

Statistical analysis

All data are expressed as means ± standard errors of the mean (SEM). Each heart served as its own control for either glibenclamide, L-NAME or indomethacin and two concentrations of one anaesthetic, as well as for the repeated effects of either bradykinin or 5-HT. No comparisons were made among different groups, i.e. different anaesthetics, treatments, or vasodilators. Effects of glibenclamide, L-NAME or indomethacin vs. no treatment (control), either anaesthetic concentration vs. control or glibenclamide, L-NAME or indomethacin, and effects of bradykinin or 5-hydroxytryptamine, on coronary flow, percentage oxygen extraction and isovolumetric left ventricular pressure (Figs 1, 2 and 3), were compared by Tukey's comparison of means tests after ANOVA for repeated measures (Super Anova 1.11® software for Macintosh®, Abacus Concepts, Inc, Berkeley, CA, USA).

Fig. 1.
Fig. 1.:
(a, b, c) Cardiac effects of 4 μM glibenclamide (GLIB) in the absence and presence of sevoflurane (SEV) on responses to increasing concentrations of bradykinin (BK). Bradykinin responses were measured first without treatment (Control) and repeated in the presence of glibenclamide alone, glibenclamide plus low sevoflurane (2.2%) and glibenclamide plus high sevoflurane (4.5%). Anaesthetic concentrations were randomized. See text and Table 1a-g for bradykinin slope analyses. Glibenclamide with high sevoflurane reduced percentage oxygen extraction (%O2E) more than glibenclamide alone or with low sevoflurane, but bradykinin-induced changes in coronary flow and percentage oxygen extraction slopes were not altered significantly by glibenclamide, or by glibenclamide with sevoflurane. Left ventricular pressure (LVP) decreased slowly but more so in the presence of sevoflurane. G1, first control; G2, second control.
Fig. 2.
Fig. 2.:
(a, b, c) Cardiac effects of 100 μM L-NAME in the absence and presence of sevoflurane (SEV) on responses to increasing concentrations of bradykinin. Bradykinin responses were measured first without treatment (Control) and repeated in the presence of L-NAME alone, L-NAME plus low sevoflurane (2.2%) and L-NAME plus high sevoflurane (4.5%). L-NAME alone decreased coronary flow and increased percentage oxygen extraction (%O2E), and markedly blunted the bradykinin-induced increase in coronary flow and decrease in percentage oxygen extraction. However, in the presence of sevoflurane, these responses were improved. Isovolumetric left ventricular pressure decreased slowly but more so in the presence of sevoflurane. L1, first control; L2, second control.
Fig. 3.
Fig. 3.:
(a, b, c) Cardiac effects of 100 μM indomethacin (INDO) in the absence and presence of sevoflurane (SEV) on responses to increasing concentrations of bradykinin. Bradykinin responses were measured first without treatment (Control) and repeated in the presence of indomethacin alone, indomethacin plus low sevoflurane (2.2%) and indomethacin plus high sevoflurane (4.5%). Indomethacin alone increased coronary flow and decreased percentage oxygen extraction. High sevoflurane reduced percentage oxygen extraction more than indomethacin alone or with low sevoflurane, but bradykinin-induced changes in coronary flow and percentage oxygen extraction slopes were not altered significantly by indomethacin, or by indomethacin with sevoflurane. Isovolumetric left ventricular pressure decreased slowly but more so in the presence of sevoflurane. I1, first control; I2, second control.

Concentration response curves to bradykinin or 5-hydroxytryptamine in the presence or absence of glibenclamide, L-NAME, or indomethacin, and in the presence or absence of sevoflurane, isoflurane, or halothane, were analysed by logarithmic regression (Statview®, Abacus Concepts, Calabasas, CA, USA) in the form of coronary flow, or percentage O2 extraction, = m x + B (where m = slope, B = Y-intercept at a given log concentration x = bradykinin or 5-HT and r2 = the regression coefficient). In the equations, bradykinin values are in nM and 5-hydroxytryptamine values are in μM. Regression coefficients and slope and intercept confidence intervals were plotted using multivariate scattergrams. Regression curves were compared by posteriori tests for differences among a set of regression coefficients by the simultaneous test procedure. Differences among means, regression coefficients, slopes and intercepts were considered statistically significant when P = 0.05.


Coronary flow at least doubled in each series after the initial bolus adenosine injection and the flow increase to that elicited initially was statistically similar (P < 0.05) to that after the final adenosine injection, respectively, for the glibenclamide series (Fig. 1a-c), 17.2 ± 0.8vs. 16.3 ± 1.2 mLg−1min−1, and for the indomethacin series (Fig. 3a-c), 16.3 ± 0.8 vs. 14.8 ± 0.9mLg−1min−1, but not for the L-NAME series (Fig. 2a-c), 14.0 ± 0.8 and 10.4 ± 0.8mLg−1min−1 (P > 0.1). The three figures display coronary flow, percentage oxygen extraction and isovolumetric left ventricular pressure data for each of the three series treated with bradykinin and sevoflurane only. These data include responses to a vasodilator with and without a blocking agent alone and with two concentrations of an anaesthetic. Other results not displayed graphically derive from additional glibenclamide and L-NAME series data (Table 1d-g) in which halothane or isoflurane was substituted for sevoflurane and 5-hydroxytryptamine was substituted for bradykinin. Table 1a-g summarizes regression analyses of bradykinin or 5-HT responses in all eight study groups before or after inhibition with glibenclamide, L-NAME, or indomethacin and two concentrations of one of the three anaesthetics.

Table 1
Table 1:
Regression analyses of blockers with and without volatile anaesthetics on responses to bradykinin (BK) and 5-hydroxytryptamine (HT) in isolated hearts.

Figures 1a, 2a and 3a show effects of sevoflurane on bradykinin-induced increases in coronary flow for the glibenclamide, L-NAME and indomethacin series. Glibenclamide (Fig. 1a) alone and sevoflurane + glibenclamide did not change coronary flow. The concentration dependent rise (slope) in flow elicited by 10−12-10−8 M bradykinin (Table 1a) was not altered significantly by glibenclamide or by low and high sevoflurane + glibenclamide. L-NAME (Fig. 2a) alone decreased flow but sevoflurane + L-NAME did not further change flow. The rise in coronary flow by 10−11-10−8 M bradykinin (Table 1b) was shifted significantly downward (lower Y-intercept) by L-NAME and the slope was not different from zero. However, the bradykinin slope was again significantly increased in the presence of low or high sevoflurane + L-NAME, whereas the Y-intercept remained lower. Indomethacin (Fig. 3a) slightly increased flow but low and high sevoflurane + indomethacin did not further change flow. The rise in coronary flow by 10−13-10−8 M bradykinin (Table 1c) was not altered significantly by indomethacin or by low and high sevoflurane + indomethacin.

Figures 1b, 2b and 3b show effects of sevoflurane on bradykinin-induced decreases in percentage O2 extraction for each series. Glibenclamide (Fig. 1b) alone did not change percentage oxygen extraction, but high sevoflurane + glibenclamide did decrease it. The concentration-dependent fall (negative slope) in percentage O2 extraction by 10−12-10−8 M bradykinin (Table 1a) was not altered significantly by glibenclamide or by low or high sevoflurane + glibenclamide, but the Y-intercept was decreased by high sevoflurane + glibenclamide. L-NAME (Fig. 2b) alone increased percentage oxygen extraction, but low and high sevoflurane + L-NAME decreased the extraction back to control values. The slope fall in the percentage oxygen extraction by 10−11-10−8 M bradykinin (Table 1b) was not different from zero whereas the slope was shifted significantly upward (higher Y-inter-cept) after L-NAME. However, in the presence of low or high sevoflurane + L-NAME, respectively, the slopes and Y-intercepts for percentage oxygen extraction were restored to control values. Indomethacin (Fig. 3b) slightly decreased the percentage oxygen extraction but low and high sevoflurane + indomethacin did not further decrease it compared with indomethacin alone. The slope fall in the percentage oxygen extraction by 10−13-10−8 M bradykinin (Table 1c) was not altered significantly by indomethacin alone or by low and high sevoflurane + indomethacin.

Figures 1c, 2c and 3c show a gradual 17-23% fall in isovolumetric left ventricular pressure over 240 min for each series, irrespective of drug treatment. Isovolumetric left ventricular pressure was not apparently decreased by glibenclamide independent of the time effect (Fig. 1c); low and high sevoflurane equivalently decreased isovolumetric left ventricular pressure and bradykinin had no additional effect on it. Isovolumetric left ventricular pressure was decreased after L-NAME (Fig. 2c), per se, and more so after high sevoflurane, but bradykinin had no additional effect on it. Isovolumetric left ventricular pressure was not apparently decreased by indomethacin independent of the time effect (Fig. 3c); low and high sevoflurane equivalently decreased isovolumetric left ventricular pressure, and bradykinin had no additional effect.

Four additional experimental groups of the glibenclamide and L-NAME series were conducted but not displayed graphically. Table 1d-g displays the following relationships when 10−8-10−6 M 5-hydroxytryptamine was given in place of 10−11-10−8 M bradykinin, or when 0.8 and 1.6% halothane, or 0.9 and 1.6% isoflurane, was administered in place of sevoflurane. In glibenclamide + 5-hydroxytryptamine + sevoflurane hearts (Table 1d), glibenclamide alone did not alter coronary flow or percentage oxygen extraction. Glibenclamide + low and high sevoflurane did not alter the 5-hydroxytryptamine-induced increase in the coronary flow slope or the 5-hydroxytryptamineinduced decrease in the percentage oxygen extraction slope, but glibenclamide + high sevoflurane decreased the Y-intercept. In L-NAME + 5-HT + sevoflurane hearts (Table 1e), L-NAME alone significantly decreased coronary flow by 1.5 ± 0.1mLg−1min−1 and increased the percentage oxygen extraction by 15 ± 3% (P = 0.05). L-NAME + low sevoflurane did not alter the coronary flow slope but increased the percentage oxygen extraction slope compared with their control, whereas L-NAME + high sevoflurane increased both the coronary flow and percentage oxygen extraction slopes and reduced the Y-intercepts.

In L-NAME + bradykinin + halothane hearts (Table 1f), L-NAME alone decreased coronary flow by 2.1 ± 0.2mLg−1min−1 and increased the percentage oxygen extraction by 22 ± 3% (P = 0.05). L-NAME + low and high halothane did not alter the slopes for coronary flow compared with their control but decreased the Y-intercepts, whereas L-NAME + high halothane increased the slopes for the percentage oxygen extraction and did not alter the Y-intercepts. In L-NAME + bradykinin + isoflurane hearts (Table 1g), L-NAME alone decreased coronary flow by 2.2 ± 0.2mLg−1min−1 and increased the percentage oxygen extraction by 18 ± 3% (P = 0.05). L-NAME + low and high isoflurane did not alter the slopes for coronary flow compared with their control but decreased the Y-intercepts, whereas L-NAME + high isoflurane increased the slopes for the percentage oxygen extraction and did not alter the Y-intercepts.


The purpose of this study was to determine whether volatile anaesthetics alter endothelium-dependent coronary vascular tone after NOS or COX, prostaglandin synthase (PGS) inhibition, and to determine whether they alter this tone after KATP channel blockade. The results demonstrate that both bradykinin and 5-hydroxytryptamine increased coronary flow and decreased percentage oxygen extraction, generally in a concentration-dependent manner in this model. Important to the aims of this study was the negative finding that sevoflurane in the presence of glibenclamide or indomethacin (Figs 1,3 and Table 1a, c) did not alter the bradykinin-induced increases in the positive coronary flow slope and the negative slope for the percentage oxygen extraction. This was also found for 5-hydroxytryptamine responses in the glibenclamide + sevoflurane group (Table 1d). As may be expected, L-NAME alone attenuated basal coronary flow and increased basal percentage oxygen extraction and greatly attenuated the coronary flow and the percentage oxygen extraction slope responses to bradykinin; only 1 nM bradykinin increased flow and reduced the percentage oxygen extraction (Fig. 2). The most interesting finding for all subsets of the L-NAME series (Fig. 2 and Table 1b, e-g) was that the high, or both low and high, concentrations of sevoflurane, halothane and isoflurane either restored, or accentuated, the bradykinin or 5-HT-induced effects on flow and percentage oxygen extraction slopes even though flow remained lower after L-NAME.

Anaesthetic effects on products of nitric oxide synthase and cyclo-oxygenase prostaglandin synthase

The results from the L-NAME series suggest that anaesthetics not only do not inhibit NOS, but also that they in large part restore or augment the reduction in bradykinin and 5-HT-induced vasodilation caused by NOS inhibition. Possibilities for these unexpected results are that (a) anaesthetics elicit greater endothelium-mediated vasodilation by augmenting release of nitric oxide to overcome the blocking effect of L-NAME (i.e. by upregulation), (b) they interfere with the blocking effect of L-NAME, e.g. by reducing the effectiveness of L-NAME, (c) they enhance nitric oxide effects on soluble guanylyl cyclase, e.g. by reducing cGMP metabolism by phosphodiesterases, (d) they shift the response of bradykinin and 5-hydroxytryptamine from a NOS-dependent endothelial dependent pathway to a non NOS-dependent pathway, possibly a endothelium derived hyperpolarizing factor, or (e) they exert a more effective vasorelaxant effect via vascular receptors or intracellular messengers due to the relative vasoconstrictor effect of L-NAME caused by NOS inhibition. For example, NOS inhibition is known to alleviate the inhibition of endothelium derived hyperpolarizing factor release by nitric oxide [31], and to increase adenosine production [32]. In any case, reduced bradykinin or 5-hydroxytryptamine-induced coronary vascular responsiveness in the intact heart after NOS inhibition can, at least in part, be overridden by a volatile anaesthetic. By whatever means these anaesthetics interfere with NOS inhibition, whether an effect of nitric oxide on soluble gyanylyl cyclase or cGMP, endotheliumderived hyperpolarizing factor, or intracellular signalling, the net effect is improved vascular responsiveness in the intact, L-NAME-blocked, coronary circulation.

Volatile anaesthetics differentially depresses mechanical activity and energy demands, so it was very important to measure not only coronary flow, but also the percentage oxygen extraction, to control for the metabolic and mechanical component of coronary flow autoregulation [28]. Blocking the endogenous vasodilatory effect of nitric oxide by L-NAME increases the percentage oxygen extraction as well as decreases coronary flow in a nearly inversely proportional manner, but the percentage oxygen extraction is probably a better measure of vascular tone in the intact heart. Similar to coronary flow after L-NAME, the anaesthetics restored or enhanced the responses to bradykinin and 5-hydroxytryptamine induced decreases in percentage oxygen extraction after L-NAME.

In a recent article, we reported that perfusion pressure-induced changes in coronary flow were associated with proportional changes in coronary effluent nitric oxide concentration and that these effects were attenuated by L-NAME [29]. In a related study [27], we found a significant correlation between increases in effluent nitric oxide concentration and coronary flow induced by bradykinin, showing a NOS-dependent effect. We showed that anaesthetics did not alter this relationship; moreover, they did not alter bradykinin-induced increases in effluent L-citrulline concentration. The finding of no significant influence of anaesthetics on bradykinin-induced coronary flow, nitric oxide release and L-citrulline concentration suggested that they do not interfere significantly in the pathway between endothelial release of nitric oxide and coronary smooth muscle relaxation. In fact, isoflurane enhanced the bradykinin-induced decrease in the percentage oxygen extraction, which suggested that after normalizing for autoregulatory influence on coronary flow, vascular tone is additively reduced by bradykinin and isoflurane. This present study wherein NOS was inhibited, supports our earlier study and additionally suggests that under NOS inhibition and reduced basal and stimulated release of nitric oxide, volatile anaesthetics may directly or indirectly restore or enhance endothelium-dependent vasodilatory responses.

Coronary flow is locally regulated in part by nitric oxide in the guinea pig heart [33] as we have also observed [29]. Volatile anaesthetics could alter specific endothelial receptor responses (e.g. to thrombin, muscarinic, Ca2+ ionophores and BK2 receptors) [34], that modulate nitric oxide production, or alter effects of nitric oxide on the guanylyl cyclase, cGMP system that promotes vascular relaxation [5,6]. But several studies suggest anaesthetics do not stimulate release of endothelium-dependent relaxing factors and that they have little or no effect on endothelium-independent vasodilation induced by nitro-vasodilators [1,4,14]. On the other hand, several indirect studies [2,8,16, 17, 18,35] suggest anaesthetics modify responses to endothelium-dependent vasodilators. Halothane was reported to inhibit vasodilation induced by nitric oxide and nitroglycerin [20], but halothane did not inhibit measured soluble guanylyl cyclase [36]. More than 2% halothane or 1% isoflurane was seen to depress cGMP formation independent of guanylyl cyclase activation when endothelial and vascular smooth muscle cell cocultures weretreated with bradykinin [2]. In intact dogs treated with L-NAME, it was reported that 1.2% halothane interfered with NOS activity in carotid, mesenteric and renal vascular beds, but not in the coronary vascular bed [35]. One suggestion was that halothane impairs nitric oxide mediated regulation of some vascular beds, but not the coronary vascular bed, because it largely eliminated systemic responses of NOS inhibition by L-NAME in rats [8].

Although anaesthetic-induced vasodilator effects are probably independent of soluble guanylyl cyclase, cGMP-induced vasorelaxation, anaesthetics could attenuate endothelium-dependent relaxation by one factor, e.g. nitric oxide, while enhancing relaxation by another, e.g. endothelium-derived hyperpolarizing factor. If one system is blocked they could alter activation of endothelial receptor sites, modify second messengers, or affect intracellular Ca2+ flux in the cascade of events leading to vasodilation. Anaesthetics may reduce endothelial Ca2+ entry or release of Ca2+ from intracellular stores, or inhibit phospholipase C (PLC) and IP3-mediated Ca2+ release [19]. Anaesthetics have been suggested to attenuate non-receptor-mediated nitric oxide production, an effect that bypasses receptors [16], or by reducing nitric oxide [18] efficacy once released from NOS. Based on a bioassay technique, or with relaxation induced by nitroprusside, it was proposed that halothane attenuates nitric oxide mediated relaxation, not by disrupting endothelial cell release of nitric oxide, but by interfering with nitric oxide stability after endothelial release [18]. In isolated preconstricted rat coronary vessels, isoflurane attenuated relaxation due to acetylcholine, to a calcium ionophore, and to a high concentration of nitroprusside; the interpretation was that isoflurane blocks endothelial Ca2+ and guanylyl cyclase [37].

In contrast, there are reports that anaesthetics do not alter vasodilation induced by nitric oxide [1,2,4,14,27]. Isoflurane relaxed isolated cerebral arteries rings independently of the endothelium [14] and isoflurane-induced relaxation of isolated aortic rings was independent of nitric oxide as well as cGMP-mediated mechanisms [1]. Coronary vasodilation induced by isoflurane in open-chest dogs was not affected by L-NAME [4]. Anaesthetics were found not to stimulate or inhibit cGMP production in endothelialvascular co-cultures suggesting that they did not activate either NOS or guanylyl cyclase [2]. We showed that anaesthetics neither alter nitric oxide release nor change coronary flow, nitric oxide and L-citrulline release induced by bradykinin [27]. The present results agree with many of the above studies that anaesthetics do not significantly decrease relaxation mediated by nitric oxide. Moreover, our results demonstrate that attenuated bradykinin-induced vasodilation by L-NAME is significantly reversed by clinical levels of anaesthetics in the intact heart. A key factor may be the unmasking of another vasodilatory pathway after the relative vasoconstriction caused by L-NAME.

We examined vasodilatory effects of both bradykinin and 5-hydroxytryptamine. In our related study [29], we found that indomethacin did not alter a bradykinininduced increase in coronary flow or release of nitric oxide, but that L-NAME (given in the presence of indomethacin and tetraethylammonium, a non-specific K+ channel blocker) inhibited increases in effluent coronary flow and nitric oxide induced by up to 10−9 M bradykinin. Although L-NAME completely blocked release of nitric oxide to 10−8 M bradykinin, it did not completely block the increase of coronary flow. This suggested the major mechanism of vasodilation by up to 10−8 M bradykinin is dependent on release of nitric oxide but that higher bradykinin concentrations exert additional mechanisms of endothelium-induced vasodilation, or a direct effect on vascular smooth muscle via BK2 receptors [34]. Thus in addition to stimulating endothelial cell Ca2+ influx to trigger production of nitric oxide, bradykinin and 5-hydroxytryptamine may also cause vasodilation via prostacyclin production [9,10,34,38] or via production of a endothelium-derived hyperpolarizing factor [39], of which a major class is derived from P450 metabolites of arachidonic acid called epoxyeicosatrienoic acids [40]. This effect is evidenced by smooth muscle cell hyperpolarization via Ca2+-activated K+ channels in the presence of inhibited prostaglandins and NOS to rule out these two effector pathways [41]. From our study, it appears that low to middle concentrations of bradykinin affect primarily the NOS pathway because L-NAME blocked most of the vasodilatory effect of bradykinin.

In most studies the relative contributions of nitric oxide, prostaglandins, endothelium-derived hyperpolarizing factor and KATP channel opening to anaesthetic modulated vasodilation were not tested. It is, thus, interesting that indomethacin, an inhibitor of the COX-PGS pathway, did not alter bradykinin-induced vasodilation in this study, and that the presence of a volatile anaesthetic had no effect on this relationship. This suggests anaesthetics do no interfere significantly with the COX-PGS pathway in the intact heart. But it remains possible that anaesthetics stimulate or upregulate the pathway of the epoxyeicosatrienoic acids after NOS inhibition because inhibiting PGS indirectly via COX with indomethacin did not blunt bradykinin-induced vasodilation.

Anaesthetic effects on KATP channels

In contrast to the findings with L-NAME, and like the findings with indomethacin, glibenclamide did not block vasodilation elicited by bradykinin or 5-hydroxytryptamine in the presence or absence of volatile anaesthetics. This indicates that bradykinin-induced vasodilation in the intact crystalloid perfused heart, whether endothelium dependent or independent, is not mediated by KATP channel opening in the absence of ischaemia. Moreover, volatile anaesthetics do not significantly affect this pathway.

Similar to the NO pathway, the magnitude and importance of the reported attenuating effect on volatile anaesthetic-induced vasodilation after KATP channel blockade is controversial. Several studies suggest that anaesthetics produce vasodilation, at least in part, by opening KATP channels because blocking these channels results in attenuated vasodilation and because dilatation induced by KATP channel agonists may be reduced by anaesthetics. It was reported initially that glibenclamide reduced halothane-induced vasodilation in tetrodotoxin-arrested crystalloid perfused rat hearts by about 56% [21]. In the regionally perfused swine heart it was demonstrated that glibenclamide abolished vasodilation induced by 1.5% and 3% isoflurane [22]; however, because glibenclamide alone decreased regional blood flow while decreasing regional oxygen consumption, the region of perfusion may have been ischaemic so that endogenous KATP channels were opened. In dogs, glibenclamide blunted increases in coronary blood flow induced by isoflurane, sevoflurane and enflurane [23]. But in another study [42], sevoflurane-induced increases in coronary collateral blood flow in dogs were not blocked by glibenclamide. In pig isolated, preconstricted coronary vessels, glibenclamide attenuated isoflurane-induced vasodilation but this effect waned over time [24]. In the pulmonary circulation of intact dogs, the attenuating effect of isoflurane on hypoxic pulmonary vasoconstriction was abolished by indomethacin [11]. In preconstricted canine pulmonary vascular rings, isoflurane attenuated relaxation of rings induced by both bradykinin and by a KATP channel opener, but not by a nitric oxide donor; but it was suggested that isoflurane interfered with a synergistic interaction between nitric oxide and prostaglandin, possibly via KATP channels [10]. It is possible that anaesthetic effects on these pathways differ between the pulmonary and coronary circulations.

It is important to know whether volatile anaesthetics inhibit, modify, or enhance coronary vasodilation in patients who are taking drugs that affect the above pathways, or who have abnormal vascular endothelial function. This study in an intact animal heart model suggests that volatile anaesthetics not only do not attenuate coronary vasodilatation induced by two endothelium-dependent vasodilators after KATP channel blockade, or COX-PGS inhibition, but that anaesthetics also abrogate the reduced responsiveness of these vasodilators after coronary vasoconstriction caused by NOS inhibition.


This research has been reported in preliminary form: Anaesth Analg 86: SCA33S, 1998 and was was supported by the National Institutes of Health (NIGMS 5T32GM 08377) and by the American Heart Association (Wl 95-GS-85). The authors wish especially to thank Enis Novalija, Richard L. Paulsen and Mark Polenski for their valuable assistance to this project.


1 Brendel JK, Johns RA. Isoflurane does not vasodilate rat thoracic aortic rings by endothelium-derived relaxing factor or other cyclic GMP-mediated mechanisms. Anesthesiology 1992; 77: 126-131.
2 Johns RA, Tichotsky A, Muro M, Spaeth JP, Le Cras TD, Rengasamy A. Halothane and isoflurane inhibit endothelium-derived relaxing factor-dependent cyclic guanosine monophosphate accumulation in endothelial cell-vascular smooth muscle co-cultures independent of an effect on guanylyl cyclase activation. Anesthesiology 1995; 83: 823-834.
3 Greenblatt EP, Loeb AL, Longnecker DE. Endothelium-dependent circulatory control -a mechanism for the differing peripheral vascular effects of isoflurane versus halothane. Anesthesiology 1992; 77: 1178-1185.
4 Crystal GJ, Kim SJ, Salem MR, Khoury E, Gurevicius J. Nitric oxide does not mediate coronary vasodilation by isoflurane. Anesthesiology 1994; 81: 209-220.
5 Johns RA. Endothelium, anesthetics, and vascular control. Anesthesiology 1993; 79: 138-1391.
6 Stowe DF, EbertTE. Neural and endothelial control of the peripheral circulation: implications for anesthesia: Part II. Endothelium mediated effects in the normal and diseased circulation. J Cardiovasc Anesthesiol 1996; 10: 159-171.
7 Wang YX, Zhou T, Chua TC, Pang CC. Effects of inhalation and intravenous anaesthetic agents on pressor response to NG-nitro-L-arginine. Eur J Pharmacol 1991; 198: 183-188.
8 Sigmon DH, Florentino-Pineda I, Van Dyke RA, Beierwaltes WH. Halothane impairs the hemodynamic influence of endothelium-derived nitric oxide. Anesthesiology 1995; 82: 135-143.
9 Loeb AL, O'Brien DK, Longnecker DE. Halothane inhibits bradykinin-stimulated prostacyclin production in endothelial cells. Anesthesiology 1994; 81: 931-938.
10 Gambone LM, Murray PA, Flavahan NA. Isoflurane anesthesia attenuates endothelium-dependent pulmonary vasorelaxation by inhibiting the synergistic interaction between nitric oxide and prostacyclin. Anesthesiology 1997; 86: 936-944.
11 Lennon PF, Murray PA. Attenuated hypoxic pulmonary vasoconstriction during isoflurane anesthesia is abolished by cyclooxygenase inhibition in chronically instrumented dogs. Anesthesiology 1996; 84: 404-414.
12 de Larminat V, Boczkowski J, Dureuil B et al. Role of prostaglandins and nitric oxide on halothane-induced arteriolar dilatation in rat diaphragm. Br J Anaesth 1996; 77: 232-237.
13 Stone DJ, Johns RA. Endothelium-dependent effects of halothane, enflurane, and isoflurane on isolated rat aortic vascular rings. Anesthesiology 1989; 71: 126-132.
14 Flynn NM, Buljubasic N, Bosnjak ZJ, Kampine JP. Isoflurane produces endothelium-independent relaxation in canine middle cerebral arteries. Anesthesiology 1992; 76: 461-467.
15 Greenblatt EP, Loeb AL, Longnecker DE. Marked regional heterogeneity in the magnitude of EDRF/NO-mediated vascular tone in awake rats. J Cardiovasc Pharmacol 1993; 21: 235-240.
16 Uggeri MJ, Proctor GJ, Johns RA. Halothane, enflurane, and isoflurane attenuate both receptor and non-receptormediated EDRF production in rat thoracic aorta. Anesthesiology 1992; 76: 1012-1017.
17 Nakamura K, Terasako K, Toda H et al. Mechanisms of inhibition of endothelium-dependent relaxation by halothane, isoflurane, and sevoflurane. Can J Anaesth 1994; 41: 340-346.
18 Blaise G, To Q, Parent M, Lagarde B, Asenjo F, Sauve R. Does halothane interfere with the release, action or stability of endothelium-derived relaxing factor/nitric oxide? Anesthesiology 1994; 80: 417-426.
19 Sill JC, Nelson OR, Uhl C. Isoflurane-, halothane- and agonist-evoked responses in pig coronary arteries and vascular smooth muscle cells. In: Blanck TJJ, Wheeler DF, eds. Mechanisms of Anaesthetic Action in Skeletal, Cardiac, and Smooth Muscle. New York: Plemum, 1991: 257-269.
20 Hart JL, Jing M, Bina S, Freas W, Van Dyle RA, Muldoon SM. Effects of halothane on EDRF/cGMP-mediated vascular smooth muscle relaxations. Anesthesiology 1993; 79: 323-331.
21 Larach D, Schuler H. Potassium channel blockade and halothane vasodilation in conduction and resistance coronary arteries. J Pharmacol Exp Ther 1993; 267: 72-81.
22 Cason BA, Shubayev I, Hickey RF. Blockade of adenosine triphosphate-sensitive potassium channels eliminates isoflurane-induced coronary artery vasodilation. Anesthesiology 1994; 81: 1245-1255.
23 Crystal GJ, Gurevicius J, Salem MR, Zhou X. Role of ATP-sensitive potassium channels in coronary vasodilation by halothane, isoflurane, and enflurane. Anesthesiology 1997; 86: 448-458.
24 Zhou X, Abboud W, Manabat NC, Salem MR, Crystal GJ. Isoflurane-induced dilation of porcine coronary arterioles is mediated by ATP-sensitive potassium channels. Anesthesiology 1998; 89: 182-189.
25 lida H, Ohata H, lida M, Watanabe Y, Dohi S. Isoflurane and sevoflurane induce vasodilation of cerebral vessels via ATP-sensitive K+ channel activation. Anesthesiology 1998; 89: 954-960.
26 Seki S, Horibe M, Murray PA. Halothane attenuates endothelium-dependent pulmonary vasorelaxant response to lemakalim, and adenosine triphosphate (ATP) -sensitive potassium channel agonist. Anesthesiology 1997; 87: 625-634.
27 Fujita S, Roerig DL, Chung WW, Bosnjak ZJ, Stowe DF. Volatile anesthetics do not alter bradykinin induced nitric oxide release and coronary vasodilation in isolated guinea pig hearts. Anesthesiology 1997; 87: 625-634.
28 Stowe DF, Monroe SM, Marijic J, Bosnjak ZJ, Kampine JP. Comparison of halothane,enflurane, and isoflurane with nitrous oxide on contractility and oxygen supply and demand in isolated hearts. Anesthesiology 1991; 75: 1062-1074.
29 Fujita S, Roerig DL, Bosnjak ZJ, Stowe DF. Effects of vasodilators and perfusion pressure on coronary flow and simultaneous release of nitric oxide from guinea pig isolated hearts. Cardiovasc Res 1998; 38: 655-667.
30 Curtis MJ, Ellwood AJ. Mechanism of 5-hydroxytryptamine-induced coronary vasodilatation assessed by direct detection of nitric oxide production in guinea-pig isolated heart. Br J Pharmacol 1996; 119: 721-729.
31 Bauersachs J, Popp R, Hecker M, Sauer E, Fleming I, Busse R. Nitric oxide attenuates the release of endothelium-derived hyperpolarizing factor. Circulation 1996; 94: 3341-3347.
32 Minamino T, Kitakaze M, Node K, Funaya H, Mori M. Inhibition of nitric oxide synthesis increases adenosine production via an extracellular pathway through activation of protein kinase C. Circulation 1997; 96: 1586-1592.
33 Ueeda M, Silvia SK, Olsson RA. Nitric oxide modulates coronary autoregulation in the guinea pig. Cir Res 1992; 70: 1296-1303.
34 Vanhoutte P, Boulanger C, Illiano S, Nagao T, Vidal M, Mombouli J. Endothelium-dependent effects of converting-enzyme inhibitors. J Cardiovasc Pharm 1993; 22 (Suppl. 5): S10-S16.
35 Chelly EJ, Doursout MF, Lechevalier T et al. Cardiac and regional hemodynamic interactions between halothane and nitric oxide synthase activity in dogs. Anesthesiology 1996; 85: 142-149.
36 Eskinder H, Hillard CJ, Flynn N, Bosnjak ZB, Kampine JP. Role of guanylate cyclase -cGMP systems in halothane-induced vasodilation in canine cerebral arteries. Anesthesiology 1992; 77: 482-427.
37 Park KW, Dai HB, Lowenstein E, Darvish A, Sellke FW. Isoflurane and halothane attenuate endothelium-dependent vasodilation in rat coronary microvessels. Anesth Analg 1997; 84: 278-284.
38 Lamontagne D, Koenig A, Bassenge E, Busse R. Prostacyclin and nitric oxide contribute to the vasodilator action of acetylcholine and bradykinin in the intact rabbit coronary bed. J Cardiovasc Pharmacol 1992; 20: 652-657.
39 Suzuki H, Chen G. Endothelium-derived hyperpolarizing factor (EDHF). An endogenous potassium-channel activator. Trends Pharmacol Sci 1990; 5: 212-215.
40 Nakashima M, Mombouli JV, Taylor AA, Vanhoutte PM. Endothelium-dependent hyperpolarization caused by bradykinin in human coronary arteries. J Clin Invest 1993; 92: 2867-2871.
41 Campbell WB, Harder DR. Endothelium-derived hyperpolarizing factors and vascular cytochrome P450 metabolites of arachidonic acid in the regulation of tone. Circ Res 1999; 84: 484-488.
42 Kersten JR, Schmeling T, Tessmer J, Hettrick DA, Pagel PS, Warltier DC. Sevoflurane selectively increases coronary collateral blood flow independent of KATP channels in vitro. Anesthesiology 1999; 90: 246-256.


© 2001 European Society of Anaesthesiology