The site where volatile anesthetics inhibit endothelium-dependent, nitric oxide-mediated vasodilation is unclear. To determine whether anesthetics could limit endothelium-dependent nitric oxide production by inhibiting receptor-mediated increases in cytosolic Ca2+, experiments were performed to see if the inhalational anesthetics halothane, isoflurane, and enflurane affect intracellular Ca2+ ([Ca2+]i) transients induced by the agonists bradykinin and adenosine triphosphate in cultured bovine aortic endothelial cells.
Bovine aortic endothelial cells, which had been loaded with the fluorescent Ca2+ indicator Fura-2, were added to medium preequilibrated with volatile anesthetic (1.25% and 2.5% for isoflurane, 1.755 and 3.5% for enflurane, and 0.75% and 1.5% for halothane). In Ca(2+)-containing medium, intracellular Ca2+ transients were elicited in response to bradykinin (10 nM and 1 microM) or adenosine triphosphate (1 microM and 100 microM).
Both bradykinin and adenosine triphosphate triggered a rapid rise to peak [Ca2+]i followed by a gradual decline to a plateau above the resting level. Although basal [Ca2+]i was unaltered by the anesthetics, both halothane and enflurane, in a dose-dependent manner, depressed the peak and plateau of the [Ca2+]i transient elicited by 10 nM bradykinin, whereas isoflurane had no effect. When [Ca2+]i transients were elicited by 1 microM bradykinin, halothane (1% and 5%) did not alter peak and plateau levels. Halothane and enflurane also decreased [Ca2+]i transients evoked by 1 microM and 100 microM adenosine triphosphate, whereas isoflurane also had no effect in this setting.
Halothane and enflurane, but not isoflurane, inhibit bradykinin- and adenosine triphosphate-stimulated Ca2+ transients in endothelial cells. Limitations of Ca2+ availability to activate constitutive endothelial nitric oxide synthase could allow for part, but not all, of the inhibition of endothelium-dependent nitric oxide-mediated vasodilation by inhalational anesthetics.
Key words: Adenosine triphosphate. Anesthetics, volatile: enflurane; halothane; isoflurane. Bradykinin. Calcium, intracellular. Endothelial cells. Fura-2.
Vascular endothelium is actively involved in regulating blood flow and blood pressure under basal conditions and during adaptive responses to various physiologic and pathologic conditions, including the classical barrier function, macromolecular transport, growth, phagocytosis, free-radical production, thrombosis, vasospasm, and hypoxic dysfunction. Most of these functions are associated with endothelial [Ca2+]i. The endothelium modulates vascular tone by releasing vasodilating substances in response to various physiologic and pharmacologic stimuli. Several agonists, including bradykinin, adenosine triphosphate (ATP), substance P, histamine, thrombin, acetylcholine, and serotonin, can increase [Ca2+]iin endothelial cells by receptor-mediated mechanisms. [2,3]The endothelial cell agonists bradykinin and ATP stimulate phospholipase C metabolism, resulting in the dose-dependent production of inositol trisphosphate (IP3), which is thought to initiate mobilization of intracellular calcium stores. These endothelial cell agonists also produce a sustained increase in [Ca2+]ithat is due to the influx from extracellular pools. [5,6]
Endothelium-derived relaxing factor/nitric oxide (EDRF/NO), first discovered in the vascular endothelium, is now recognized as the primary signal transduction mechanism for activating soluble guanylyl cyclase. It is a potent vasodilating agent released by endothelium in response to agonist stimulation and acts as a major modulator of basal vascular tone. In the vasculature, EDRF/NO is produced from L-arginine by the endothelial isoform of nitric oxide synthase (eNOS), an enzyme activated by increased intracellular Calcium sup 2+ via Calcium2+ calmodulin. Once produced, NO diffuses to the vascular smooth muscle, where it activates soluble guanylyl cyclase-enhancing cyclic guanosine monophosphate production, which subsequently leads to vascular relaxation. Researchers in several laboratories have shown that volatile anesthetics inhibit endothelium-dependent NO-mediated vasodilation. [8-11]Recent studies suggest that only stimulated, but not basal EDRF/NO production is inhibited by volatile anesthetics. We have provided evidence that this inhibitory effect occurred before the steps of NO synthase and guanylyl cyclase [8,14,15]activation. In addition, volatile anesthetics did not alter NOS activity or directly interfere with activation of guanylyl cyclase by either NO itself or NO donors. Inhalational anesthetics have been shown to have profound and specific effects on Calcium2+ homeostasis in various cell types, including endothelial, neuronal, and cardiac. [16-19]An effect of inhalational anesthetics on Calcium2+ availability is a likely site of anesthetic interaction with NO generation. Although the inhibition of calcium ionophore A23187-stimulated endothelial relaxation by anesthetics suggests that anesthetic inhibition of receptor-activated Calcium2+ release cannot account for all available aspects of inhibition of the NO pathway, some effect on receptor activation could not be dismissed. 
To determine whether anesthetic inhibition of receptor-mediated increases of cytosolic Calcium2+ could contribute to the inhibitory action of these volatile anesthetics on NO production, we investigated the dose-dependent effect of halothane, enflurane, and isoflurane on bradykinin- and ATP-stimulated calcium transients in cultured bovine aortic endothelial cells.
Materials and Methods
Bovine aortae were obtained from a slaughterhouse and the aortic endothelial cells were prepared for culture by lightly scraping the intimal surface of the aortae. The cells adhering to the blade were rinsed into a 35-mm tissue culture dish and grown in Waymouth's medium supplemented with 10% fetal bovine serum, penicillin (100 IU/ml), and streptomycin (100 micro gram/ml). A pure culture of endothelial cells was obtained by fluorescent-activated cell sorting using Dil-acyl-low-density lipoprotein as the fluorescent marker. Identity of the endothelial cells was further confirmed by immunostaining for Factor VIII antigen and for endothelial cell actin, by demonstrating a single mRNA for actin in endothelium using Northern blot analysis and by observing a characteristic cobblestone appearance. Confluent cells were split in a ratio 1:10 and subcultured up to 12 passages. The following experiments were performed with cells in passages from 9 to 12 using four separate subcultured cell lines.
Determination of Cytosolic-free Calcium Concentrations in Isolated Endothelial Cells
Monolayer cultures of endothelial cells were grown in 75-cm sup 2 flasks until 2 or 3 days after confluence. The flasks were rinsed with 10 ml Calcium2+ - and Magnesium2+ -free phosphate-buffered saline, followed by dispersion of the endothelial cells by incubation with 0.05% trypsin and 0.53 mM ethylenediamine tetraacetic acid for 2 min at 37 degrees Celsius. The trypsin was inactivated by adding the cell suspension to a 0.1-mg/ml trypsin inhibitor in a 50-ml centrifuge tube. The flask was rinsed twice with 9 ml HEPES buffer (NaCl, 140 mM; KCl, 5 mM, MgSO4, 1 mM; CaCl2, 1.5 mM; glucose, 10 mM; HEPES, 15 mM; bovine serum albumin, 0.1%) and the rinses were added to the centrifuge tube. Endothelial cells were then formed into pellets by centrifugation at 150g for 2 min, followed by a wash with 10 ml HEPES buffer and a subsequent centrifugation. HEPES buffer was used to resuspend the cells at a concentration of 1.5 x 10 sup 7 cells/ml and the cells were maintained at 37 degrees Celsius in room air. After removal of an aliquot of cells to serve as an autofluorescence control, the remaining cells were incubated with 5 micro Meter Fura-2/AM in HEPES buffer for 20 min at 37 degrees Celsius. The cells were protected from light during this time and gently stirred every 5 min during incubation. The cell suspension was diluted tenfold and incubated at 37 degrees Celsius for 10 min more. Cells were formed into pellets by low-speed centrifugation (at 150g for 2 min) and resuspended to a concentration of 1.5 x 106cells/ml. Fura-2-loaded cells were stored at 5 degrees Celsius until they were needed. Two-milliliter aliquots of cells were used in each experiment. To ensure removal of any extracellular Fura-2, the cell suspensions were gently centrifuged and the supernatants withdrawn, followed by resuspension of the cells in warm (37 degrees Celsius) HEPES buffer. The cells were subjected to centrifugation one final time, resuspended in warm buffer, and placed in cuvettes. Cells were allowed to equilibrate for 5 min at 37 degrees Celsius before each experiment. From preliminary experiments, the intracellular concentration of Fura-2 under these conditions was estimated to be 125 micro Meter, and the dye remained confined to the cytosolic compartment. 
Fluorescence measurements were made using a Perkin-Elmer LS50 fluorometer with excitation wavelengths alternating between 345 nm and 380 nm every 500 ms, and emission fluorescence was recorded at 500 nm. Excitation and emission slit widths were generally set at 2 and 16 nm, respectively. The emission intensities at excitation wavelengths of 345 nm (F345) and 380 nm (F380) were corrected automatically for autofluorescence and the fluorescence ratio (R = F345/F380) was calculated. The maximum fluorescence ratio, Rmax, and the minimum fluorescence ratio, Rmin, were determined by sequentially adding 75 micro Meter digitonin and 20 mM TRIS/5 mM ethyleneglycol-bis-(beta-aminoethylether)-N,N,N',N'-tetraacetic acid. The [Ca2+]iwas calculated by Equation 1where the Kdfor Fura-2 is 224 nm at 37 degrees Celsius, and F380Dand F380Eare the emission intensities at an excitation wavelength of 380 nm after adding 75 micro Meter digitonin and 20 mM TRIS/5 mM ethyleneglycol-bis-(beta-aminoethylether)-N,N,N',N'-tetraacetic acid, respectively. 
Anesthetic Application and Agonist Stimulation
The incubation medium was preequilibrated in a closed container at 37 degrees Celsius for at least 30 min by bubbling with anesthetic gas mixture in air, which was delivered from a calibrated vaporizer. Fura-2-loaded endothelial cells were gently added to anesthetic-preequilibrated medium in a quartz cuvette. The endothelial cells were incubated for 10 min more while the anesthetic gas-air mixture was maintained over the solution to ensure a constant anesthetic level before measuring [Ca2+]i. Endothelial cells used in control experiments were added to medium treated in an identical manner, but in the absence of anesthetic. Aliquots of media were drawn up into gas-tight syringes to determine anesthetic concentrations using gas chromatography. This analysis confirmed the concentrations delivered from the calibrated vaporizers.
The cells were stimulated by directly adding bradykinin (10 nM and 1 micro Meter, final concentrations) or ATP (1 and 100 micro Meter, final concentrations) into the cell suspension, and [Ca2+] sub i was determined as described previously. Each concentration of anesthetic and agonist was repeated at least four separate times, with experiments (control and anesthetic-exposed cells) performed on aliquots obtained from the same pool of endothelial cells. The control and anesthetic-exposed experiments were performed in a randomized manner. The peak transient was defined as the difference between the peak and the basal levels. The plateau was defined as the difference between the plateau and the basal levels. The measurement for the plateau was taken at the end of the experiment. In limited preliminary experiments, the reversibility of the effect of volatile anesthetics on calcium transients was determined by exposing the endothelial cell suspension to the volatile anesthetic and then washing out the anesthetic by further gassing the cell suspension with air alone. Stimulation of the cells with agonist revealed no difference in the calcium transient when compared with cells that had not been exposed to volatile anesthetic.
Results are expressed as mean +/- SEM. Comparisons between the groups were made using Fisher's protected least-significant difference test (Statview 4.0). An effect was considered statistically significant at P < 0.05.
Under control conditions, both bradykinin and ATP, as shown in Figure 1, triggered a rapid rise to peak [Ca2+]ifollowed by a gradual decline to a plateau above the resting level, as described previously. Overall, the peak and plateau of the [Ca2+]itransient elicited by 10 nM bradykinin were depressed by both halothane and enflurane in a dose-dependent manner. However, each of the anesthetics failed to affect the basal [Ca2+]ilevel, which ranged from 70 to 110 nM,
Both agonists induced transient increases in [Ca2+]i, and successive applications of the same agonist produced similar [Ca sup 2+]itransients in the cell suspension. It was notable that when the two agonists were given in succession to the same cell suspension, altering the order of addition produced contrasting responses. As shown in Figure 2, preceding the addition of ATP (100 micro Meter) with an addition of bradykinin (10 nM) resulted in the production of [Ca2+]itransients after each agonist application, whereas preceding the addition of bradykinin with ATP application produced a [Ca2+]itransient only after the ATP application.
The concentration dependence of volatile anesthetics on agonist-induced changes in [Ca2+]iwas subsequently investigated (Figure 3, Figure 4, Figure 5). As indicated in Table 1, halothane (0.75% and 1.5%) decreased the peak of the bradykinin-induced [Ca2+]itransient by 22 +/- 8% and 44 +/- 5% from the control level. The [Ca2+]iplateau was reduced by 38 +/- 8% and 38 +/- 5% for the 0.75% and 1.5% concentrations of halothane, respectively. Enflurane at a concentration of 1.75% had no effect on the bradykinin-induced [Ca2+]itransient peak, but it resulted in a 20 +/- 6% reduction in the bradykinin-induced [Ca2+]iplateau. At the higher concentration of 3.5% enflurane, the bradykinin-induced [Ca2+]itransient peak and plateau were reduced by 35 +/- 6% and 19 +/- 5%, respectively. In contrast, isoflurane at either concentration (1.25% and 2.5%) had no effect on the bradykinin-induced [Ca2+]itransient peak and plateau phases. Examination of the [Ca2+]itransients elicited by bradykinin at a much higher concentration, 1 micro Meter, revealed that halothane at 1% and 5% had no effect on the peak and plateau levels.
[Ca2+]itransients evoked by ATP were less sensitive to modulation by halothane and enflurane than were those evoked by bradykinin, as shown in Table 2. Halothane (1.5%) decreased the peak of the [Ca2+]itransient induced by 1 micro Meter ATP by only 17 +/- 3% from its control level, along with a reduction in the [Ca2+]iplateau by 30 +/- 14%. Enflurane (3.5%) decreased the peak of the [Ca2+]itransient induced by 1 micro Meter ATP by only 19 +/- 4% from its control level, with no change in the [Ca2+] sub i plateau.
At the higher dose of 100 micro Meter ATP, the effects of halothane and isoflurane were similar to those seen with the lower concentration. Halothane (1.5%) decreased the peak of the [Calcium2+]itransient by 14 +/- 3%, with a reduction in the [Calcium2+]iplateau by 25 +/- 4%. Similarly, enflurane (3.5%) decreased the peak of the [Calcium2+]itransient induced by 100 micro Meter ATP by 20 +/- 3% from its control level, along with a reduction in the [Calcium2+]iplateau by 31 +/- 2%. Again, isoflurane had no effect on the ATP-evoked (1 micro Meter and 100 micro Meter) [Calcium sup 2+]itransient.
Although enflurane, halothane, and isoflurane did not change the fluorometrically determined level of endothelial cell Calcium2+, enflurane and halothane depressed bradykinin- and ATP-induced Calcium2+ transients. Each of the anesthetics failed to affect the basal [Calcium2+]ilevels. After bradykinin (10 nM) stimulation, we observed that halothane decreased the peak and plateau [Calcium2+]itransients in a dose-dependent manner, whereas enflurane only affected the peak and plateau [Calcium2+]itransients at the higher (3.5%) concentration. Isoflurane, in contrast, did not affect either the peak or plateau [Calcium2+]itransient after bradykinin stimulation, even at the higher concentration.
The [Calcium2+]itransients evoked by ATP were less sensitive to modulation by halothane and enflurane. At both ATP concentrations (1 and 100 micro Meter), halothane decreased the peak and plateau [Calcium2+]itransients, whereas enflurane only affected the [Calcium2+]iplateau induced by 100 micro Meter ATP. Again, isoflurane had no effect on the ATP-evoked (1 and 100 micro Meter) peak and plateau [Calcium2+]itransients. It is noteworthy that the greater depression of 10 nM bradykinin-stimulated Calcium2+ -transients could be overcome by increasing bradykinin to 1 micro Meter, whereas the lesser depression of ATP-stimulated Calcium sup 2+ transients was present in either concentration.
Changing the order of agonist addition resulted in the [Calcium2+]iresponses, which differed from those of Weintraub and associates. In contrast to our experiments, in which the plateau phase remained above baseline, Weintraub and associates reported that even with continued exposure to an agonist (e.g., bradykinin) the [Calcium2+]iresponse returned to baseline. They showed that successive doses of agonist resulted in a decrease in the amplitude of the [Calcium2+]iresponse for bradykinin, but not for ATP. In addition, they noted that the order of presentation of the agonists did not alter the response patterns. In our experiments, both agonists induced transient increases in [Calcium2+]i, and successive applications of the same agonist produced similar [Calcium2+]itransients in the cell suspension. When the two agonists were given in succession to the same cell suspension, changing the order of addition produced contrasting responses. We noted that preceding the addition of ATP (100 micro Meter) with an addition of bradykinin (10 nM) resulted in the production of [Calcium2+]itransients after each agonist application, whereas preceding the addition of bradykinin with ATP application produced a [Calcium2+]itransient only after the ATP application. Whether the difference in responses was due to differences in the concentration of agonists or other experimental conditions is uncertain.
Current evidence suggests that at least two pathways exist through which endothelial cell surface receptor stimulation can mediate a signal that results in an increase in cell cytosolic [Calcium2+] sub i. First, the interaction of agonist (bradykinin and ATP) with the endothelial cell causes the release of [Calcium2+]ifrom intracellular stores and results in a transient elevation of cytosolic [Calcium2+]i. This correlates with our observed peak [Calcium sup 2+]iresponses. This initial increase in [Calcium2+]iin endothelial cells is very rapid and occurs in the absence of external Calcium2+, indicating that Calcium2+ is released from intracellular storage sites. Second, the agonist stimulates Calcium2+ influx across the surface membrane of the endothelial cell, which is responsible for the plateau phase of the calcium transient. [4,6,24,25]This influx results in a sustained elevation of cytosolic [Calcium2+]iabove the resting level, an elevation that could be abolished when the cells were superfused with Calcium2+ -free buffer. 
This agonist-stimulated Calcium2+ influx responsible for the plateau phase may occur by Calcium2+ depletion-activated cation channels (IodineCRAC), as seen in human leukemic T lymphocytes or activation of a channel via an internal messenger other than Calcium2+, such as IP4. [28,29]These two pathways could help explain the maintained increase in cytosolic Calcium2+ in response to agonist stimulation of endothelial cells.
The available evidence supports the observation that inhalational anesthetics are potent inhibitors of endothelium-dependent vascular relaxation at clinically relevant doses. [8,10,30]Isoflurane, halothane, and enflurane can inhibit both receptor- and nonreceptor-mediated EDRF/NO-dependent vasodilation. These anesthetics inhibited the endothelium-dependent vasodilation induced by the receptor-mediated agent methacholine and that by the receptor-independent agent calcium ionophore A23187 but had no effect on the endothelium-independent vasodilation induced by sodium nitroprusside. Although this study showed that the anesthetic inhibition of EDRF/NO vasodilation occurred distal to receptor activation of the endothelial cell, it did not rule out an additional effect on receptor mechanisms, as has been demonstrated in other cell types. [8,31-35]Toda and colleagues reported that both isoflurane and halothane inhibited acetylcholine-induced endothelium-dependent relaxation of rat aorta and simultaneously prevented acetylcholine-induced increases in cyclic guanosine monophosphate. They noted that isoflurane was more potent than halothane in this regard, which is consistent with the findings of Uggeri and associates. However, Toda and colleagues did not find an inhibition of A23187-induced vasodilation.
Studies using isolated vascular rings have not resolved the question of whether inhalational anesthetics can interfere with NO signaling through an action in vascular smooth muscle involving guanylyl cyclase. The effects of inhalational anesthetics on guanylyl cyclase activation in the vascular smooth muscle were studied in the context of EDRF-dependent vasodilation. Conflicting data exist on whether NO and NO donor (sodium nitroprusside and nitroglycerin) compound-induced vasodilation was inhibited by inhalational anesthetics. [8,10,11,36]The possibility that inhalational anesthetics affect NO signaling through interaction with guanylyl cyclase was investigated using partially isolated soluble and particulate guanylyl cyclase from the rat brain. Inhalational anesthetics, at wide-ranging concentrations, did not affect either the basal or the agonist-stimulated activity. 
Although anesthetic inhibition of muscarinic receptor (or other receptor) activation may be a component of the mechanism by which anesthetics inhibit EDRF/NO production in response to those specific agonists, receptor activation is not likely to be the major site of anesthetic inhibition of NO signaling. We know this from the findings of Uggeri and associates, who found significant anesthetic inhibition of calcium ionophore A23187-stimulated EDRF/NO production that bypasses the receptor effects. However, it could be argued that due to subcellular localization of Calcium2+, release of intracellular Calcium2+ may play a greater role in NO synthase activation than that mediated by ionophore.
We have shown that halothane and enflurane, but not isoflurane, inhibit bradykinin- and ATP-stimulated Calcium2+ transients in endothelial cells. Limitations of Calcium2+ availability to activate constitutive endothelial NO synthase thus could account for part, but not all of the inhibition of endothelium-dependent NO-mediated vasodilation by inhalational anesthetics, because isoflurane, which inhibits the NO pathway more potently than halothane or enflurane, had no effect on Calcium2+ transients induced by NO synthase-activating concentrations of bradykinin and ATP.
Our findings are consistent with those of Loeb and coworkers, who investigated the effects of halothane and isoflurane on agonist-induced [Ca2+]itransient peaks. Although both studies showed a decrease in the [Ca2+]itransient elicited by bradykinin in the presence of halothane, but not in the presence of isoflurane, Loeb and coworkers reported that neither halothane nor isoflurane, at concentrations as high as 2 mM, altered the [Ca2+] sub i response to ATP (10 micro Meter). In addition, they did not observe an increase in agonist-induced [Ca2+]iplateau levels, possibly due to the transient application of the agonists.
Activation of a small conductance Calcium2+ -activated Potassium sup + channel, perhaps via a G-protein, appears to be critical for maintaining a driving force for Calcium2+ entry during the plateau phases. In fact, a blocker of Calcium2+ -dependent Potassium sup + current, tetrabutylammonium chloride (5 mM), completely blunts the plateau phase with little or no effect on the peak Calcium sup 2+ transient elicited by bradykinin (Miao and Lynch, unpublished data). Our findings indicate that the reduction of bradykinin- and ATP-induced Calcium2+ plateau phases by halothane and enflurane could be explained by anesthetic effects at any of the putative mechanisms of Calcium2+ entry or by the function of Calcium2+ -activated Potassium sup + channels. Another difference between the anesthetics is their ability to deplete intracellular Calcium2+ stores by activating Calcium2+ release channels (ryanodine receptors). Such channels, which have been histologically and physiologically demonstrated in endothelial cells, are activated by halothane and possibly enflurane, but not isoflurane, [42,43]which is consistent with the effects we observed.
Our data show that the [Ca2+]itransients evoked by ATP (1 and 100 micro Meter) were far less sensitive to modulation by halothane and enflurane than those evoked by bradykinin. The ability of halothane and enflurane (at the higher dose) to attenuate responses to bradykinin and, to a lesser extent, ATP may result from the action of the volatile anesthetic on receptor-agonist interactions or on specific differences in receptor G-protein-effector coupling that mediate the responses to agonists. There are clearly differential effects of halothane and enflurane when compared with isoflurane in their ability to inhibit the mobilization of [Ca2+]iby receptor-operated mechanisms in endothelial cells. The data of Loeb and colleagues showed that halothane, but not isoflurane, inhibited bradykinin-stimulated prostaglandin I2(PGI2) production by vascular endothelial cells, suggesting that halothane-mediated inhibition of bradykinin-stimulated prostacyclin production does not involve a pertussis toxin-sensitive G protein and may result from an interaction of halothane at some other step in the signal transduction pathway, including the inhibition of protein kinase C. Given the importance of intracellular [Ca2+]imobilization in activating NO synthase to produce NO, with its physiologically critical role in modulating vascular tone, such anesthetic actions have considerable implications. The property of halothane and enflurane to decrease intracellular [Ca2+]imobilization may potentially interfere with physiologic [Ca2+]imechanisms mediated by the endothelium in various organs. In contrast, we might anticipate that such regulation would be better maintained by isoflurane. Our findings, consistent with those of Loeb and colleagues, [37,44]show that halothane and enflurane, but not isoflurane, can interfere with the receptor-mediated Calcium2+ mobilization in vascular endothelial cells. How the mechanisms by which the differential effects of the volatile anesthetics act on the signal transduction pathway remains to be clarified.
The authors thank Dr. Joseph J. Pancrazio for his contributions to this article.