INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

IN ENDOTHELIAL CELLS, nitric oxide (NO) is synthesized by Ca²⁺/calmodulin-dependent NO synthase (NOS) [endothelial NOS (eNOS)]. The activity of the enzyme is tightly controlled by its interaction with caveolin-1 in a specialized domain of the membrane, namely the caveolae. An increase in intracellular free Ca²⁺ concentration ([Ca²⁺]_i) dissociated the NOS III-caveolin 1 complex and induced action of the NOS III-Ca²⁺/calmodulin complex (9, 13, 21). Numerous intermediates are involved in the main signal transduction pathway of agonist-stimulated eNOS activation. For example, NO synthesis is regulated by serine/threonine kinases, including cAMP-dependent protein kinase (3, 4), protein kinase C (26, 30), Ca²⁺/calmodulin protein kinase II (38), and protein kinase B/Akt (6, 8). In addition to phosphorylation, Ca²⁺-dependent regulation of eNOS also involves an increase in subplasmalemmal [Ca²⁺]_i, supporting the role of local distribution of Ca²⁺ in caveolaes (10, 28). The subplasmalemmal [Ca²⁺]_i is mainly dependent on plasma membrane Ca²⁺-ATPase (23, 34), which has a high affinity for Ca²⁺ but a low turnover rate, and Na⁺/Ca²⁺ exchanger (NCX), which transports Ca²⁺ outside the cell in forward mode and inside the cell in reverse mode with lower affinity but a higher turnover rate. In forward mode, NCX operates in parallel with the plasma membrane Ca²⁺-ATPase pump to maintain Ca²⁺ homeostasis. Na⁺/Ca²⁺ exchange current has been reported in heart cells (16), arterial smooth muscle cells (14, 29), and more recently in endothelial cells by Kaye and Kelly (15), where NCX is thought to contribute to endothelium-dependent control of contraction and relaxation (1, 7). The NCX inhibitor dichlorobenzamil decreased endothelium-dependent relaxation of vascular smooth muscle without affecting endothelium-independent relaxation caused by the NO donor sodium nitroprusside (33, 41). These results supported the involvement of the reverse mode of Na⁺/Ca²⁺ exchange in the release of endothelium-derived NO. Recently, Teubl et al. (37) demonstrated the effect of intracellular Na⁺ load in the Ca²⁺-dependent activation of eNOS and suggested the involvement of NCX in the caveolae-associated cell signaling. However, endothelium-dependent relaxation involves not only NO but also endothelium-derived hyperpolarizing factor (EDHF) (5, 24), suggesting that real-time measurement of endothelial cell NO release might confirm the involvement of NCX in Ca²⁺-dependent NO synthesis. Direct measurement of NO release may be evaluated by chemiluminescence (27) and electrochemistry with polarographic electrode (20). Recently, new generations of amperometric sensors have allowed high-sensitivity real-time detection of NO. The present study was designed to specify the role of monensin-induced Na⁺ loading in endothelium-dependent relaxation of the rat aorta. NCX involvement was also confirmed by direct measurement of NO release in cultured porcine aortic endothelial cells (PAECs).

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Drugs. Reagents were from Sigma and RBI, distributed by Sigma (Saint-Quentin Fallavier, France) unless otherwise stated. The Na⁺ ionophore monensin and the selective blocker of the Na⁺/Ca²⁺ channel benzamil hydrochloride (N-benzylamidino-3,5-diamino-6-chloropyrazinecarboxamide) were dissolved in ethanol. Phenylephrine hydrochloride (₁-adrenergic agonist), acetylcholine chloride (ACh; muscarinic agonist), calcium ionophore A-23187, sodium nitroprusside (NO donor), and N-nitro-L-arginine methyl ester hydrochloride (L-NAME; analog of arginine and competitive, isozyme nonselective, NOS inhibitor) were made up in bidistilled deionized water. Each was made freshly every day and protected from light. All other drugs were dissolved in bidistilled deionized water except indomethacin, which was dissolved in ethanol. Further dilutions were made in Krebs solution (see Tissue preparation and tension measurement for composition) for in vitro experiments. The drug solutions were prepared each day from dry powder.

Tissue preparation and tension measurement. All animal procedures were applied in accordance with the European directives for animal experiments 86/609 (Centre National de la Recherche Scientifique, Paris, France). At the time of experimentation, adult male Sprague-Dawley rats (251-275 g, Charles River; Saint-Aubin Les Elbeufs, France) were anesthetized with thiopental sodium (Nesdonal; 80 mg/kg ip) and heparinized (100 IU ic). After a thoracotomy, the aorta was excised en bloc in cold Krebs balanced salt solution [composed of (in mM) 118 NaCl, 5.9 KCl, 1.2 MgSO₄ · 7H₂O, 1.2 NaH₂PO₄ · 2H₂O, 2.5 CaCl₂ · 2H₂O, 25.5 NaHCO₃, and 5.6 D-glucose] containing 1 mM indomethacin. The thoracic aorta of internal diameter 2.5 ± 0.2 mm was isolated clean of adherent fat and connective tissue and cut into rings of 3.0 mm in length. Two L-shaped stainless steel wires were inserted into the arterial lumen, and the rings were suspended in a 20-ml tissue bath. One stainless steel holder was attached to the chamber, and the other to an isometric force-displacement transducer (Emka Technologie; Paris, France). Temperature of the organ chambers was kept constant at 37°C in Krebs buffer solution gassed with 95% O₂-5% CO₂ (Air Liquide santé; Paris, France). The tension was recorded with the isometric force transducer connected to a data processing system (MacLab /8e, ADInstruments; Paris, France). The rings were set at an initial resting tension of 1.5 g (the resting tension had been previously determined to be the optimal tension for length development in response to 60 mM KCl) and allowed to equilibrate for 60 min, with the rings being repeatedly washed every 15 min. Verification of endothelium integrity was performed by testing the vascular relaxation produced in phenylephrine (0.1 µM)-precontracted rings by the endothelium-dependent vasodilator ACh (0.1 µM). In the experiments with endothelium-denuded rat aortic rings, the endothelial cell layer was removed by rubbing the luminal surface of the vessel with a cotton swab. After equilibration, the vascular rings were submaximally precontracted with 0.1-1 µM phenylephrine, and stepwise pharmacological sequence was initiated.

Characterization of the pharmacological responsiveness of aortic rings. At the plateau phase, concentration-relaxation curves to ACh (0.001-10 µM) were constructed by increasing the concentration in the organ chamber in cumulative increments after a steady-state response had been reached with each increment on precontracted rings with phenylephrine (1 µM). After the highest concentration of ACh, the rings were totally relaxed by sodium nitroprusside (0.1 mM). In preincubated experiments, monensin (final concentrations: 1 and 10 µM) or benzamil (1 µM) + monensin (1 µM) were added to the bath at the plateau phase, and relaxation was proceeded after equilibration with the tissue for 15 min. Controls were determined by the addition of vehicle/Krebs balanced salt solution or 0.01% ethanol.

Histology. Endothelium integrity was assessed by postpharmacological challenge histology analysis. Tissues were harvested, fixed in 10% neutral buffered formaldehyde, embedded in paraffin, and sectioned. Histological sections (5 µm thick) were stained with hematoxylin and eosin (Sigma) and examined.

Endothelial cell culture. All the reagents used for cell culture were from GIBCO-BRL (Cergi-Pontoise, France) if not otherwise specified. The PAEC line established by Malassagne et al. (1998) (19) was a gift from B. Weill (Laboratoire d'Immunologie, Paris, France). PAECs were cultured into 25-cm² Primaria dishes (Polylabo; Paris, France) in RPMI 1640 medium-glutamax-1 supplemented with 100 U/ml penicillin, 100 µg/ml streptomycin, 50 µg/ml amphotericine B, and 10% fetal bovine serum. The cultures were incubated at 37°C in a humid atmosphere with 5% CO₂. When cells reached confluence, they were detached by incubation with 0.05% trypsin-EDTA in RPMI 1640 medium-glutamax-1 for 3 min at 37°C, centrifuged at 300 g for 10 min, suspended in fresh complete medium, and further cultured under the same conditions. Cells were characterized as endothelial cells by their morphology, their ability to take up acetylated low-density lipoproteins (25), the detection of von Willebrand factor (27), and expression of E-selectin (19).

Cell stimulation. Four sets of transfected PAECs (between passages 25 and 28) were tested. The cells were subcultured in 24-well flat-bottom culture plates (10⁵ cells/well) during 24 h to reach confluence. The culture medium was removed, and the cells were equilibrated in isotonic phosphate buffer (pH 7.4) containing 8 mM Na₂HPO₄, 1.5 mM KH₂PO₄, 137 mM NaCl, 2.7 mM KCl, 0.9 mM CaCl₂, 1 mM MgCl₂, and 1 mM indomethacin and oxygenated with 95% O₂-5% CO₂ (Air Liquide santé). The hypotonic low-Na⁺ phosphate buffer contained 100 mM NaCl. After equilibration, the sensor probe was inserted vertically into a well with confluent cells, the sensor membrane was positioned 50 µM above the monolayer using a manual micromanipulator, and the well was sealed. The distance between the sensor membrane and the monolayer was determined after a distance-response curve; it corresponded to the maximal signal without leak current. To investigate the response to the agonists, the cells were stimulated with either A-23187 (10⁹-10⁴ M) or bradykinin (10¹⁰-10⁵ M), and the concentration-response curve was plotted as the control. The effect of NCX was evaluated on the concentration-response curve for A-23187 or bradykinin of cells previously incubated with monensin (1 and 10 µM), benzamil (1 mM), or benzamil (1 mM) + monensin (10 µM). The endothelial cells were incubated for 30 min and stimulated by A-23187 (10⁹-10⁴ M) or bradykinin (10¹⁰-10⁵ M) with the same experimental procedure as the control.

Electrochemical detection of NO. S-nitroso-N-acetyl-D,L-penicillamine (SNAP), CuCl, EDTA, and NaOH were obtained from Sigma and Fluka. All of the solutions were dissolved in deoxygenated bidistilled water. Amperometric measurements of released NO were quantified with a Clark-type electrode (2-mm platinum disk NO sensor: Iso-NOP, World Precision Instruments; Stevenage, UK) connected to a Iso Mark II NO meter (World Precision Instruments). The newly developed electrode has a high selectivity for NO and a detection sensitivity of 1 nM. The principle of measurement has previously been described (35). NO diffusing through the gas-permeable membrane is oxidized at the working platinum electrode. The resulting redox current is proportional to the concentration of NO gas in aqueous solution. The analyzer was connected to an Apple Macintosh computer-based data-acquisition system via an analog-to-digital converter (MacLab /8e, ADInstruments). The output current was recorded at a constant laboratory temperature (23°C). The calibration of the electrode was performed daily according to the procedure described by Zhang et al. (42).

Data analysis. All responses to phenylephrine are expressed as force (in g). Responses to vasodilator agents are expressed as a percentage of maximal relaxation. In the endothelium-intact vessels, effects of NCX inhibitors were repeatedly measured on two or three rings for each aorta, and the values were averaged. For all experiments of relaxation, n represents the number of animals. EC₅₀ represents the concentration (in M) of agonist required for half-maximal relaxation. Sensitivity to the agonists in different conditions is expressed in term of pD₂ [pD₂ = log(EC₅₀)]. Data are expressed as means ± SE of n experiments; sample size is systematically indicated. Results were analyzed using one-way ANOVA to compare the effects of inhibitors versus control. When the F value for an effect was globally significant, comparisons were made using a Mann-Whitney U-test. In stimulated PAECs, NO concentration was expressed in nanomolars. Effects of monensin and NCX inhibitors were repeatedly measured on two or three wells, and the values were averaged; n represents the number of cultured plates. The baseline was automatically established by the Iso Mark II NO meter. Statistical evaluation of the difference between pretreated PAECs and control was assessed with Student's two-tailed t-test. All statistical tests were considered as significant for an -level below 0.05. Computation was performed with the Statistica software (Statsoft; Maison-Alfort, France).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Effect of the Na+ ionophore monensin on endothelium-dependent relaxation to acetylcholine. In phenylephrine-precontracted endothelium-intact rings, ACh (10⁹-10⁵ M) induced concentration-dependent relaxation (pD₂ = 6.52 ± 0.38, n = 5) (Fig. 1). L-NAME (0.05 mM) totally inhibited the ACh-evoked relaxation. Pretreatment of the rat aortic rings with monensin (1 and 10 µM) resulted in a leftward shift of the concentration-response curves (pD₂ = 6.92 ± 0.21, n = 4, and pD₂ = 6.89 ± 0.18, n = 5, respectively) (Fig. 1). Monensin (10 µM) potentiated the ACh-induced relaxation of phenylephrine-contracted rat aortic rings by 4 ± 1%, 9 ± 2%, and 19 ± 4% at 10⁹, 10⁸, and 10⁷ M ACh concentration, respectively (P < 0.05) (Fig. 1). The NO donor sodium nitroprusside (0.05 mM) totally relaxed the aortic rings (98 ± 2%). Sodium nitroprusside-induced relaxation was affected neither by the Na⁺ ionophore monensin nor by the NCX inhibitors benzamil or KB-R7943 (data not shown). The vehicle ethanol (0.01%) had no effect on both phenylephrine-induced contraction and ACh-stimulated relaxation.

View larger version (15K): [in this window] [in a new window]   Fig. 1.   Potentiation of endothelium-dependent relaxation to acetylcholine by the Na+ ionophore monensin. Concentration-response curves for the vasorelaxation effect of acetylcholine (109-105 M) on phenylephrine-contracted aortic rings with endothelium are shown. Effects of monensin (1 and 10 µM) compared with vehicle (0.01% ethanol) as the control (n = 6) are also shown. Monensin (1 and 10 µM) was added at the plateau evoked by the 1-adrenergic agonist and incubated 15 min before acetylcholine challenge. Results are expressed as percentages of maximal relaxation and are presented as means ± SE. **P < 0.01 and *P < 0.05, significant differences between monensin- and vehicle-treated preparations.

Effect of the NCX inhibitors benzamil or KB-R7943 on monensin potentiation of ACh-induced relaxation. Benzamil (1 mM) slightly decreased the ACh-induced relaxation of rat aortic rings. Preincubation with benzamil (1 mM) inhibited the monensin-dependent potentiation of ACh-induced relaxation [pD₂ = 6.53 ± 0.18, n = 5 (P < 0.05); Fig. 2A]. Neither monensin nor benzamil had an additional contracting effect on phenylephrine-precontracted rings (data not shown). Furthermore, neither monensin nor benzamil affected the response of endothelium-denuded aortic rings. Indeed, KB-R7943 (1 µM) had no significant effect on ACh-induced relaxation of rat aortic rings (Fig. 2B). However, preincubation with KB-R7943 (1 µM) inhibited the monensin-dependent potentiation of ACh-induced relaxation (pD₂ = 6.56 ± 0.04, n = 4).

View larger version (36K): [in this window] [in a new window]   Fig. 2.   Inhibitory effect of benzamil or KB-R7943 on monensin-dependent potentiation of endothelium-dependent relaxation to acetylcholine. A: effects of benzamil (1 mM) and monensin (10 µM) + benzamil (1 mM) compared with vehicle (0.01% ethanol) as the control for the vasorelaxation effect of acetylcholine (109-105 M) on phenylephrine-contracted aortic rings (n = 6). Benzamil was preincubated 5 min before monensin. Monensin was added at the plateau evoked by the 1-adrenergic agonist and incubated 15 min before acetylcholine challenge. B: effects of KB-R7943 (1 µM) and monensin (10 µM) + KB-R7943 (1 µM) in the same conditions. Results are expressed as percentages of maximal relaxation and are presented as means ± SE. *P < 0.05.

Calibration of the NO sensor. The electrode was calibrated by decomposition of SNAP (Fig. 3). The NO sensor was immersed in saturated CuCl solution. After stabilization, known volumes of the SNAP solution (for final concentrations of 10, 20, 50, 100, and 200 nM) were then added, and the response was monitored. Measurements of NO were performed under constant stirring in glass vials sealed with a septum. Linear calibration curves were obtained from the resulting calibration plot. The NO sensor responded straight away to SNAP addition. Steady state was reached in <3 s. The output current of the probe correlated linearly with the concentration of NO, current (in pA) = 1.29 NO concentration + 0.13 (P < 0.001).

View larger version (11K): [in this window] [in a new window]   Fig. 3.   Calibration of the nitric oxide (NO)-sensitive electrode. Linear regression analysis of the relationship between NO concentration and output current for the NO sensor (r2 = 0.98, 1 pA = 1.29 nM) is shown. The NO-sensitive electrode calibration is based on the decomposition of S-nitroso-N-acetyl-D,L-penicillamine in the presence of Cu+. The amperometric response was recorded during successive additions of 10, 20, 50, 100 and 200 nM NO to the stirred saturated CuCl solution.

Effect of the Na+ ionophore monensin on NO production. In PAECs, A-23187 (0.01-100 µM) and bradykinin (0.001-10 µM) induced NO release [pD₂ = 5.21 ± 0.09 (Fig. 4A) and 6.17 ± 0.30 (Fig. 4B), respectively]. Preincubation with L-NAME (0.05 mM) totally inhibited the agonist-induced NO release. Monensin (1 and 10 µM) shifted the concentration response curve of A-23187- and bradykinin-induced NO release to the right [pD₂ = 5.66 ± 0.15 (Fig. 4A) and 6.69 ± 0.27 (Fig. 4B), respectively, for 1 µM monensin and pD₂ = 5.67 ± 0.19 (Fig. 4A) and 6.72 ± 0.24 (Fig. 4B), respectively, for 10 µM monensin]. Monensin (1 and 10 µM) increased by 8.9 ± 0.5% and 26 ± 5% the A-23187 (10 µM)-evoked NO production and increased by 7.8 ± 0.4% and 24 ± 6% the bradykinin (1 µM)-evoked NO production. Low concentration of Na⁺ (119 vs. 151 mM) did not modify the bradykinin-induced NO release; only Na⁺-free buffer could activate NCX in reverse mode and potentiate NO release (data not shown).

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Leftward shift of concentration-response curves for endothelial cell NO release by the Na+ ionophore monensin. Monensin (1 and 10 µM) was incubated in the culture medium 30 min before A-23187 (109-104 M) or bradykinin (1010-105 M) stimulation. Concentration-response curves for the stimulated endothelial cell NO release were plotted by repeated measures of NO concentration (in nM) and are presented as means ± SE (n = 4). A and B: effects of monensin (1 and 10 µM) were compared with vehicle (0.01% ethanol) as the control (n = 5) and N-nitro-L-arginine methyl ester (L-NAME; 10 µM)-incubated endothelial cells (n = 3) in A-23187- (A) and bradykinin-stimulated endothelial cells (B).

Effect of benzamil and KB-R7943 on monensin-dependent NO production potentiation. Benzamil (1 mM) slightly decreased A-23187- and bradykinin-induced NO release (pD₂ = 5.19 ± 0.14 and 6.13 ± 0.23, respectively; Fig. 5). Interestingly, pretreatment of endothelial cells with benzamil (1 mM) resulted in the inhibition of the monensin (10 µM)-dependent potentiation of NO release induced by A-23187 and bradykinin [pD₂ = 5.24 ± 0.21 (P < 0.05) and 6.31 ± 0.18 (P < 0.05); Fig. 5, A and B, respectively]. Indeed, KB-R7943 (1 µM) slightly decreased the bradykinin-induced NO release (pD₂ = 6.11 ± 0.14) and inhibited the monensin (10 µM)-dependent potentiation of the bradykinin-induced NO release (pD₂ = 6.18 ± 0.11; Fig. 5C).

View larger version (32K): [in this window] [in a new window]   Fig. 5.   Effect of benzamil or KB-R7943 on increase in NO release induced by monensin-dependent Na+ load. Benzamil (1 mM) and KB-R7943 (1 µM) were preincubated in the culture medium alone 30 or 5 min before monensin (10 µM) (30 min). The effects of benzamil and monensin + benzamil were compared with that of the vehicle (0.01% ethanol) as the control for A-23187 (109-104 M)- and bradykinin (1010-105 M)-induced NO release on porcine aortic endothelial cells (n = 4). Results are expressed as NO concentration (in nM) and are presented as means ± SE. A and B: effects of benzamil on A-23187- (A) and bradykinin-stimulated (B) cells. C: effect of KB-R7943 on bradykinin-stimulated cells.

View larger version (12K): [in this window] [in a new window]   Fig. 6.   Effect of monensin and benzamil on NO release by bradykinin-stimulated endothelial cells. Porcine aortic endothelial cells were stimulated with bradykinin (1 µM) (A). In preincubated cells, monensin (10 µM) was added in the culture medium 30 min before bradykinin (1 µM) (B). In benzamil + monensin incubation, benzamil (1 mM) was added 5 min before monensin (C). NO release was recorded by an amperometric NO sensor positioned at 50 µm from the endothelial surface.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

According to the hypothesis that endothelial subplasmalemmal [Ca²⁺]_i was controlled by NCX (10), we studied the involvement of NCX in Ca²⁺-calmodulin-dependent eNOS activation. The role of NCX was investigated though modification of Na⁺ intracellular concentration with the Na⁺ ionophore monensin or through inhibition of NCX with the selective inhibitors benzamil and KB-R7943 on rat aortic rings or cultured PAECs. Our results are consistent with the involvement of NCX in endothelium-dependent relaxation. Na⁺ load induced by monensin potentiated ACh-evoked relaxation of the rat aorta and increased the Ca²⁺ ionophore A-23187- and bradykinin-induced NO release by PAECs. Both NCX inhibitors antagonized these effects. NCX inhibitors alone slightly decreased agonist-induced endothelium-dependent relaxation and endothelial cell NO release.

Involvement of NCX in NO release. NO release measured with the amperometric electrode in response to A-23187 or bradykinin was in the nanomolar range. These low concentrations are in accordance with those of previous studies using cultured endothelial cells or endothelium-intact vessels (36, 39, 40). The short half-life may be explained by the oxygenation of the buffer. Superoxide anions generated in oxygenated buffer may react with NO and decrease NO concentration (2, 12, 27). The specificity of the NO sensor was documented by the inhibition of NO release by the NOS inhibitor L-NAME. The inhibitory effect of benzamil or KB-R7943 on the leftward shift of the NO concentration-response curve induced by monensin confirms the contribution of NCX in cultured endothelial cell NO release. These findings are consistent with those of Teubl et al. (37), suggesting that Na⁺/Ca²⁺ exchange promotes Ca²⁺-dependent activation of eNOS. Furthermore, the NCX-dependent NO synthesis pathway was corroborated by the inhibition of the monensin-induced potentiation of NO release by the NCX inhibitors benzamil and KB-R7943. The potential involvement of Na⁺/H⁺ exchange in increased intracellular Na⁺ concentration in endothelial cells had been ruled out by the lack of effect of Na⁺/H⁺ exchanger inhibitor on the [Ca²⁺]_i increase induced by Na⁺ loading (18). Therefore, following previous results suggesting that both ryanodine-sensitive Ca²⁺ release and Na⁺/Ca²⁺ exchange are involved in capacitative Ca²⁺ entry and eNOS activation (28, 34), our results further support the hypothesis that subplasmalemmal Ca²⁺ control may regulate NO synthesis.

Potentiation of NO-dependent relaxation by the Na+ ionophore monensin. In rat aortic rings, Na⁺ load induced by monensin caused a leftward shift of the concentration-response curve to ACh. This result might relate to 1) direct potentiation of eNOS by an increase in intracellular Na⁺ concentration, 2) induction of the Ca²⁺ release from intracellular Ca²⁺ stores by monensin-induced Na⁺ entry, and 3) promotion of Ca²⁺ entry in caveolae via NCX in reverse mode. eNOS has not been described to be Na⁺ dependent. However, Na⁺ load might relax rat aortic rings by the Ca²⁺-independent eNOS pathway or EDHF. Involvement of EDHF is unlikely, because endothelial Na⁺ entry is more likely to depolarize than to hyperpolarize the cells. Furthermore, Teubl et al. (37) have previously shown that membrane depolarization was not involved in eNOS activation. Alternatively, Laskey et al. (17) and Groschner et al. (11) have shown that Na⁺ loading decreased store-operated Ca²⁺ entry via decreased electrochemical gradient. These results suggest that potentiation of eNOS activation by the Na⁺ ionophore monensin was independent from intracellular Ca²⁺ stores release. Therefore, the main hypothesis remains the involvement of NCX. Na⁺ loading allowed for Ca²⁺ entry via NCX in reverse mode (18, 32). Because eNOS is mainly dependent from Ca²⁺/calmodulin, our results suggest that monensin potentiates NO-dependent relaxation by increasing the caveolae Ca²⁺ concentration. These findings are consistent with previous results showing that NCX increase Ca²⁺-dependent activation of eNOS (37). The authors hypothesized that NCX might increase subplasmalemmal [Ca²⁺]_i, because eNOS and NCX were codistributed in caveolae; they concluded that Na⁺ load promotes NOS III activation without increasing overall [Ca²⁺]_i. The role of NCX in reverse mode was corroborated by the effect of the specific NCX inhibitors benzamil and KB-R7943 on endothelium-dependent relaxation. A previous study has reported that amiloride, an inhibitor of Na⁺/Ca²⁺ exchange, blunted ACh-stimulated increases in tissue cGMP levels. The presence of amiloride had no effect on increases in cGMP levels evoked by the NO donor sodium nitroprusside (33). Additionally, increasing concentrations of dichlorobenzamil functionally antagonized the relaxation elicited by ACh and A-23187 but had no effect on the sodium nitroprusside-dependent relaxation (41). All in all, NCX in reverse mode potentiated NO-dependent relaxation. In physiological conditions, NCX seems to extrude Ca²⁺ to prevent Ca²⁺ overload, whereas in pathological conditions, NCX increases Ca²⁺ in caveolae and promotes NO synthesis. Our results with low-Na⁺ buffer, in concert with those of Teubl et al. (37), suggest that only hyponatremia down to 100 mM may activate eNOS via NCX in reverse mode. The role of NCX functioning in forward mode remains, however, to be investigated.