Furosemide is a potent loop diuretic widely used in the treatment of acute left ventricular failure, pulmonary edema and, to a lesser degree, hypertension. The effect of furosemide was first attributed to inhibition of the Na+/K+/2Cl− cotransport in the kidney (1), which leads to natriuresis and volume depletion. However, several lines of evidence suggest that furosemide also may exert direct vascular effects (2). Indeed, in patients with pulmonary edema, furosemide induced systemic hemodynamic changes that appeared before natriuresis (3), and in conscious rats, furosemide induced renal hemodynamic effects even when volume loss was prevented by intravenous quantitative fluid replacement (4). Furthermore, plethysmography study on healthy patients showed that furosemide induced a dose-dependent direct venodilator effect of the dorsal hand vein (5).
However, the data on the vascular effect of furosemide in vitro are conflicting and fragmentary. Furosemide has been shown to relax precontracted rat perfused kidney artery (6), canine pulmonary and saphenous veins (7), and guinea-pig mesenteric arteries (8), whereas it failed to relax precontracted canine coronary artery (9), canine pulmonary artery, splenic artery, mesenteric artery, and tibial artery (7). These discrepancies suggest that the vasodilator effect of furosemide vary according to the vascular beds, the species investigated, and the experimental conditions.
In addition, the mechanism underlying the vascular effects of furosemide remains uncertain. Furosemide has been reported to induce the release from the kidney of an unidentified prostanoid-derived vasoactive compound (10) or to stimulate the biosynthesis and the release of relaxant prostaglandins from the vessel wall (5,11-13), which in turn mediate the vascular effect. However, the finding that furosemide was still able to relax rat vasculature after bilateral nephrectomy or indomethacin treatment (6) did not support the involvement of a prostanoid derived either from the kidney or from nonrenal sources in the vascular effect of furosemide, whereas the direct dilation of the dorsal hand vein in healthy subjects appeared to be mediated by local vascular prostaglandin synthesis (5).
It also has been reported that furosemide may directly act on the vascular smooth muscle because furosemide inhibited angiotensin II-induced contraction on human isolated internal mammary artery and saphenous vein by a prostaglandin-independent mechanism (14). Inhibition of the Na+/K+/2Cl− contransport may account in part for these effects of furosemide on isolated vessels (7,14).
In addition, on human bronchi, furosemide has been reported to inhibit thromboxane A2 (TXA2)-induced contraction in a competitive manner (15). Because TXA2 release is increased in various cardiovascular diseases such as hypertension (16), pulmonary edema (17), and heart failure (18), the main objective of our study was to assess the effect of furosemide on the contraction induced by the stable TXA2-mimetic U46619 on human vascular smooth muscle: the internal mammary artery (IMA) and the saphenous vein (SV).
Isolation and study of human IMA and SV
Human IMA segments (230) harvested from 70 patients undergoing coronary artery bypass grafting and 250 SV segments harvested from 62 patients undergoing surgical removal of varicose veins were used in this study. Samples were collected in accordance with the French Law on biological research. The discarded vessel segments during operation were immediately placed in oxygenated physiologic salt solution (Krebs solution) maintained at 4°C, and then transferred to the laboratory. The Krebs solution had the following composition: NaCl (118.0 mM), KCl (4.7 mM), CaCl2 (2.5 mM), MgSO4 (1.0 mM), KH2PO4 (1.0 mM), glucose (11.0 mM), NaHCO3 (25.0 mM), and was aerated with a gas mixture (5% CO2/95% O2). The experiments were carried out within 24 h. The vessels were dissected free of connective tissue, cut into 3-mm (IMA) or 5-mm (SV)-long rings. The number of rings taken from segments of artery and vein varied from two to six.
The rings were mounted between two stainless steel wires in a 10-ml organ bath containing the Krebs solution at 37°C continuously bubbled with the gas mixture; the lower wire was fixed to a micrometer (Mitutoyo, Japan), and the upper wire was attached to a force transducer (UF-1, Pioden), through which changes in isometric forces were continuously displayed on recorders (Linseis 200, Bioblock, France). At the beginning of the experiments, the rings were stretched to an initial tension of 5 g for the IMA and 2 g for the SV as previously described (14,19). This passive tension was maintained throughout the experiments. Resting segments were allowed to stabilize for 60 min, the Krebs solution being changed every 15 min. The rings were then challenged with KCl (90 mM) to stabilize the preparations. In some preparations, rings also were submaximally precontracted with 3 μM norepinephrine, and when the plateau of contraction was reached, vessels were exposed to acetylcholine (1 μM) to check endothelial function. A functional endothelium was absent in all tissues tested, as no relaxation was observed in response to acetylcholine. We conclude that the lack of endothelial relaxation in saphenous vein was related to the use of varicose vein, whereas failure to relaxation to acetylcholine in IMA rings was probably due either to inadvertent endothelial damage during experimental processing or loss of endothelial function during storage in Krebs solution, as previously shown in our laboratory (20).
Effect of furosemide on contractile responses to U46619
Concentration-response curves to U46619 (0.1 nM-3 or 10 μM) were made by cumulative addition of increasing concentrations (0.5-log increments) in the organ baths. The IMA and SV rings were incubated for 60 min with furosemide (100 μM, 300 μM, and 1 mM) or its vehicle (methanol, 0.2%). As far as possible, experiments were conducted on four rings, including one control ring, obtained from the same vascular segment mounted in parallel in different organ baths.
To study the role of vasorelaxant prostanoids in the effect of furosemide on U46619-induced contraction, indomethacin (1 μM) or its vehicle (ethanol, 0.01%) was added 30 min before the addition of furosemide (1 mM).
A further series of experiments was designed to study the selectivity of the effect of furosemide on U46619-induced contraction. Cumulative concentration-response curves to 5-HT (1 nM-30 μM) and to KCl (7.5-120 mM) on SV and to norepinephrine (1 nM-0.1 mM) and to endothelin-1 (1 nM-0.3 μM) on IMA were performed in the presence or the absence (vehicle) of furosemide (1 mM).
In an attempt to study whether the effect of furosemide on U46619-induced contraction was related to the blockade of the Na+/K+/2Cl− symporter, cumulative concentration-response curves to U46619 were made in human SV after incubation for 60 min with bumetanide (0.1 mM), a potent inhibitor of the Na+/K+/2Cl− symporter (21) or its vehicle (ethanol, 1%).
Relaxant effect of furosemide
In a separate series of experiments, the ability of furosemide to relax U46619-precontracted arteries and veins was studied. Vessel rings were contracted with a concentration of U46619 that produced 50% of the maximal contraction. After stabilization of the U46619-induced contraction, cumulative dose-response curves to furosemide (0.1 μM-1 mM) were recorded. Then sodium nitroprusside (SNP) at 0.1 mM was added to achieve complete relaxation.
In another set of experiments, the effect of indomethacin on furosemide-induced relaxation was studied. Concentration-response curves to furosemide were obtained on rings precontracted with U46619, in the absence (control) or the presence of indomethacin (1 μM) added 30 min before U46619-induced contraction. SNP, 0.1 mM, was then added to assess complete relaxation.
In all experiments, concentration-response curves were obtained by cumulative addition of the drug at time intervals sufficient to reach a stable response before the next addition was made. Only one dose-response curve was performed in each vascular ring.
The contractions were expressed as isometric force (grams) developed by the rings in response to the different agonists. The maximal effect (Emax) was the greatest response obtained with an agonist. The concentrations of agonist producing 50% of the maximal effect (EC50) were determined from each curve by a logistic curve-fitting equation. The pD2 value represents the negative logarithm of the EC50. The EC50 from each curve to U46619 was used to calculate the concentration ratio (CR) for the different concentrations of furosemide. The agonist concentration ratios (CR) were determined by dividing the EC50 of U46619 in the presence of furosemide by the EC50 of the agonist required to produce the same response in the absence of the furosemide. The pA2 defines the negative logarithm of the molar concentration of an antagonist that produces a twofold rightward shift of concentration-response curves. The pA2 value [negative logarithm of (CR-1)] was determined according to the Schild analysis (22). For each series of experiments, the pA2 value and the slope of the linear regression line were calculated. Slopes of the Schild plot that were not different from the unity were indicative that the drugs-receptor interaction was competitive.
For relaxation-response curves, results are expressed as percentage of the maximal relaxation achieved by SNP, 0.1 mM. IC40 defines the concentration of furosemide that induces 40% of the maximal relaxation elicited by SNP, 0.1 mM.
Results are presented as mean ± standard error of the mean (SEM) for the specified number of preparations tested. Statistical analysis were performed by using analysis of variance (ANOVA) for repeated measures followed by Bonferroni corrected t test. Individual comparisons were made by Student's t test for paired or unpaired data as appropriate. The p values <0.05 were considered to be significant.
Furosemide was purchased from Hoechst Marion Roussel (Puteaux, France), the stable thromboxane A2-mimetic (U46619: 9,11-dideoxy-11α, 9α-epoxy-methano-prostaglandin F2α methyl), SNP (sodium nitroferricyanide), 5-hydroxytryptamine hydrochloride, indomethacin, and bumetanide were from Sigma (L'lle d'Abeau, France), norepinephrine bitartrate was from Assistance Publique des Hôpitaux de Paris (France), and KCl from Prolabo (France). All drugs except furosemide, bumetanide, and indomethacin were dissolved in distilled water. Furosemide was dissolved in a mixture of water/methanol (0.8:0.2, vol/vol), and the maximal methanol concentration in the organ bath was 0.2%. Bumetanide and indomethacin were dissolved in ethanol, and the maximal ethanol concentrations in the organ bath were 1% and 0.01%, respectively. Drugs were stocked at −20°C and were freshly dissolved in distilled water to the appropriate concentrations expressed as final molar concentrations in the organ bath.
Effect of furosemide on contractile response to U46619
A slow decrease of the resting tone in both IMA and SV was observed after addition of furosemide (1 mM). This decrease in tone, measured after 60-min incubation, was 0.8 ± 0.1 g on IMA (n = 5) and 0.3 ± 0.1 g on SV (n = 5). The vehicle used (methanol) was devoid of effect on the resting tone at the maximal concentration (0.2%) to which the tissues were exposed during the experiments.
Pretreatment of IMA and SV rings with furosemide produced parallel, concentration-dependent rightward shifts of U46619 response curves, with no change in the maximal effects (Fig. 1). The shift was significant with 0.3 and 1 mM furosemide on IMA and with 1 mM furosemide on SV. However, the potency to U46619 was significantly reduced in the presence of furosemide 0.3 and 1 mM in both IMA and SV (Table 1). Schild plots were calculated from the rightward shifts of the concentration-response curves to U46619 caused by furosemide (Fig. 1a and b, inset). The pA2 values of furosemide were 5.1 ± 0.3 and 4.8 ± 0.4 M on IMA and SV, respectively, and the slopes of the Schild plot linear regression lines were 1.2 ± 0.3 and 1.1 ± 0.2 (not significantly different from unity) on IMA and SV, respectively.
Addition of indomethacin induced a decrease in the resting tone of 0.4 ± 0.1 g on both IMA and SV (n = 4 for each vessel type). Indomethacin did not significantly alter the response to U46619 in IMA and SV (Table 1). In addition, indomethacin had no influence on the inhibitory effect of furosemide on U46619-induced contractions in IMA and SV (Table 1). The pA2 values of furosemide in the presence of indomethacin (4.5 ± 0.6 M on IMA and 4.2 ± 0.2 M on SV) were not statistically different from the pA2 obtained in the absence of indomethacin.
Pretreatment with furosemide (1 mM) induced no significant change of the response to 5-HT and KCl on SV (Table 2) or to norepinephrine on IMA (Table 2). In contrast, furosemide significantly reduced endothelin-1-induced contraction in IMA (Fig. 2, Table 2).
Last, the addition of bumetanide at 0.1 mM, which by itself induced no change of the basal tone of SV (n = 5), did not modify U46619-induced contractions on SV in terms of potency [pD2: 8.03 ± 0.14 for control (n = 5) and 8.11 ± 0.26 for bumetanide (n = 5)] and efficacy [Emax: 2.8 ± 0.2 g for control (n = 5) and 2.6 ± 0.2 g for bumetanide (n = 5)].
Relaxant effect of furosemide
The cumulative addition of furosemide induced a dose-dependent relaxation of U46619-precontracted IMA and SV with −log IC40 values of 4.1 ± 0.2 on both IMA and SV. The maximal relaxations elicited by furosemide were 84.6 ± 5.8% (n = 7) and 80.3 ± 4.8% (n = 8) on IMA and SV, respectively. In addition, pretreatment with indomethacin (1 μM) had no significant effect on responses elicited by furosemide (Fig. 3).
This study provides compelling evidence of the direct effect of furosemide on human IMA and SV because furosemide inhibited U46619-induced contractions in a competitive way and relaxed U46619-precontracted IMA and SV. This inhibitory effect was not mediated by relaxant prostaglandins synthesis.
Furosemide inhibited U46619-induced contractions without affecting the maximal responses in both IMA and SV preparations. In addition, the slopes of the linear Schild plot regression lines were close to unity, suggesting a competitive antagonism. These data suggest that furosemide blocked the TXA2 receptor (TP receptor) in human IMA and SV with pA2 values greater than those determined in human bronchi preparation (4.01) (15). The inhibitory effect of furosemide on U46619-induced contraction was not caused by induction of relaxant prostaglandin synthesis because pretreatment with indomethacin did not alter the effect of furosemide. It was important to assess the possible mediation by prostanoids of these effects of furosemide because a prostanoid derived from the kidney has been suggested to mediate the anticonstrictor effect of furosemide (10), and because the direct venodilator effect of furosemide on the dorsal hand vein in healthy volunteers appears to be mediated by local vascular prostaglandin synthesis (5). In our study, the use of isolated human vessel preparations eliminates the possible induction by furosemide of renal-derived autacoids, and the use of indomethacin excluded the role of prostanoid synthesized by the vessels.
The inhibitory effect of furosemide on U46619-induced contraction appeared specific because the highest concentration of furosemide (1 mM) induced no significant change of the contraction elicited by 5-HT and KCl on human SV and on the contraction elicited by norepinephrine on human IMA. In contrast, furosemide decreased endothelin-1-induced contractions on human IMA. Endothelin-1 has been reported to induce the release of TXA2(23). Furthermore, on human IMA (19), rat aorta (24), and guinea-pig bronchi (25), TXA2 receptor antagonists inhibited endothelin-1-induced contraction. Therefore, the blockade of TP receptors by furosemide most likely explains the reduction of endothelin-1-induced contraction. Furthermore, we have previously shown on human IMA and SV that furosemide was capable of inhibiting angiotensin II-induced contraction by inhibition of the Na+/K+/2Cl− cotransport (14). To assess whether the inhibitory effect of furosemide on U46619-induced contraction was related to an inhibition of the Na+/K+/2Cl− cotransport, we tested the effect of bumetanide at a concentration (0.1 mM), which is considered to be at least as potent as furosemide (1 mM) on the cotransport (21). Bumetanide did not alter U46619-induced contraction on SV, suggesting that inhibition of the Na+/K+/2Cl− symport should not be involved in the inhibitory effect of furosemide. Moreover, furosemide (1 mM) induced a decrease in the resting tone of the vessels, whereas bumetanide had no effect, also suggesting that inhibition of the Na+/K+/2Cl− symport is not involved in the vascular effect of furosemide. In addition, furosemide was able to relax U46619-precontracted IMA and SV with the same potencies. This finding is consistent with the vasorelaxant effect of furosemide reported on PGF2α-precontracted rat kidney artery (6), and on U46619-precontracted canine pulmonary and splenic veins (7). In addition, the potencies of furosemide (expressed as −log IC40) were similar to those determined in the animal vessel preparations precontracted with U46619. In canine pulmonary and splenic vein, the IC50 of furosemide averaged 0.1-0.3 mM, respectively (7). Similarly, the IC50 value of furosemide was 0.1 mM in rat kidney artery (6).
The relaxations elicited by furosemide were not altered in the presence of indomethacin. These data are consistent with previous in vitro studies showing on animal isolated vessels (6,7,26) that the vasodilator effect of furosemide was not related to the release of vasodilator prostanoids within the vessel wall. However, as mentioned earlier, dilation of human dorsal hand veins by furosemide was inhibited by indomethacin (5). In addition, furosemide has been shown to enhance in vivo the secretion of prostacyclin in healthy persons and in patients with chronic heart failure (27). Thus modulation of prostanoid synthesis by blood circulating cells or the endothelial cell layer may explain the differences between the in vitro results and those reported in vivo.
The vasodilator effect of furosemide on human IMA and SV occurred in absence of a functional endothelium. These observations excluded participation of nitric oxide produced by endothelial cells in the in vitro effect of furosemide (28) and are in line with previous findings on human dorsal hand vein (5), on guinea-pig mesenteric arteries (8), and rat tracheal arterioles (26), showing that furosemide induced vasodilation by a nitric oxide-independent mechanism. In our experimental conditions, the release of kinins or prostacyclin from the endothelial cells in response to furosemide (28) is also excluded.
This study clearly demonstrates in two human vascular preparations that furosemide is able to inhibit U46619-induced contraction and to relax U46619-precontracted IMA and SV. However, these vascular effects occurred for in vitro concentrations of furosemide corresponding to plasma concentrations obtained during intravenous infusion of high doses (≤2 g/day) reported as effective in the treatment of severe congestive heart failure with minor and reversible ototoxic side effects (29,30). Because TXA2 release is increased in heart failure (18) and because TP-receptor antagonist has been shown to reverse U46619-induced pulmonary edema (17), TP-receptor blockade by furosemide could partly explain the early vasodilator effect of furosemide, which preceded the diuresis, in the treatment of acute pulmonary edema.
1. Puschett JB. Sites and mechanisms of action of diuretics in the kidney. J Clin Pharmacol
2. Dormans TPJ, Pickkers P, Russel FGM, Smits P. Vascular effects of loop diuretics. Cardiovasc Res
3. Biagi RW, Bapat BN. Frusemide in acute pulmonary oedema. Lancet
4. Christensen S, Petersen J. Effect of furosemide on renal haemodynamics and proximal tubular sodium reabsorption in conscious rats. Br J Pharmacol
5. Pickkers P, Dormans TPJ, Russel FGM, et al. Direct vascular effects of furosemide in humans. Circulation
6. Barthelmebs M, Stephan D, Fontaine C, Grima M, Imbs JL. Vascular effects of loop diuretics: an in vivo and in vitro study in the rat. Naunyn Schmiedebergs Arch Pharmacol
7. Greenberg S, McGowan C, Xie J, Summer WR. Selective pulmonary and venous smooth muscle relaxation by furosemide: a comparison with morphine. J Pharmacol Exp Ther
8. Freslon JL, Gonzalbes C, Marthan R. Direct relaxing effects of piretanide and furosemide in rat and guinea-pig mesenteric resistance arteries. Br J Pharmacol
9. Uchida T, Kido H, Yamanaga K, Okita M, Watanabe M. A novel loop diuretic, torasemide, inhibits thromboxane A2
-induced contraction in the isolated canine coronary artery. Prostaglandins Leukot Essent Fatty Acids
10. Gerkens JF, Smith AJ. Inhibition of vasoconstriction by frusemide in the rat. Br J Pharmacol
11. Ciabattoni G, Pugliese F, Cinotti GA, et al. Characterization of furosemide-induced activation of the renal prostaglandin system. Eur J Pharmacol
12. Katayama S, Attallah AA, Stahl RAK, Bloch DL, Lee JB. Mechanism of furosemide-induced natriuresis by direct stimulation of renal prostaglandin E2
. Am J Physiol
13. Miyanoshita A, Terada M, Endou H. Furosemide directly stimulates prostaglandin E2
production in the thick ascending limb of henle's loop. J Pharmacol Exp Ther
14. Stanke F, Devillier P, Bréant P, et al. Furosemide inhibits angiotensin II-induced contraction on human vascular smooth muscle. Br J Clin Pharmacol
15. Molimard M, Naline E, Hirsch A, Advenier C. Furosemide inhibits bradykinin-induced contraction of human bronchi: role of thromboxane A2
receptor antagonism. Eur J Pharmacol
16. Ishimitsu T, Uehara Y, Ishii M, Ikeda T, Matsuoka H, Sugimoto T. Thromboxane and vascular smooth muscle cell growth in genetically hypertensive rats. Hypertension
17. Yoshimura K, Tod ML, Pier KG, Rubin LJ. Role of venoconstriction in thromboxane-induced pulmonary hypertension and edema in lambs. J Appl Physiol
18. Castellani S, Paladini B, Paniccia, et al. Increase renal formation of thromboxane A2
and prostaglandin F2alpha
in heart failure. Am Heart J
19. Stanke F, Cracowski JL, Chavanon O, et al. Glibenclamide inhibits thromboxane A2
-induced contraction in human internal mammary artery and saphenous vein. Eur J Pharmacol
20. Stanke-Labesque F, Cracowski JL, Devillier P, Caron F, Bessard G. Functional assessment of rat aorta after cold storage in different media. Fund Clin Pharmacol
21. Jackson EK. Diuretics. In: Goodman & Gilman's The pharmacological basis of therapeutics.
9th ed. New York: McGraw-Hill, 1995:685-715.
22. Arunlakshana O, Schild HO. Some quantitative uses of drug antagonists. Br J Pharmacol Chemother
23. Takayasu-Okishio M, Terashita ZI, Kondo K. Endothelin-1 and platelet activating factor stimulate thromboxane A2
biosynthesis in rat vascular smooth muscle cells. Biochem Pharmacol
24. Reynolds EE, Mok LLS. Role of thromboxane A2
receptor in the vasoconstrictor response of rat aorta to endothelin. J Pharmacol Exp Ther
25. Lueddeckens G, Bigl H, Sperling J, Becker K, Braquet P, Forster W. Importance of secondary thromboxane A2
release in mediating of endothelin-1 induced bronchoconstriction and vasopressin in the guinea pig. Prostaglandins Leukot Essent Fatty Acids
26. Corboz MR, Ballard ST, Inglis SK, Taylor AE. Dilatory effect of furosemide on rat tracheal arterioles and venules. Am J Respir Crit Care Med
27. Liguori A, Casini A, Di Loreto M, Andreini I, Napoli C. Loop diuretics enhance the secretion of prostacyclin in vitro, in healthy persons, and in patients with chronic heart failure. Eur J Clin Pharmacol
28. Wiemer G, Fink E, Linz W, Hropot M, Schölkens BA, Wohlfart P. Furosemide enhances the release of endothelial kinins, nitric oxide and prostacyclin. J Pharmacol Exp Ther
29. Dormans TPJ, Van Meyel JJ, Gerlag PGG, Tan Y, Russel FGM, Smits P. Diuretic efficacy of high dose furosemide in severe heart failure: bolus injection version continuous infusion. J Am Coll Cardiol
30. Gerlag PGG, Van Meijel JJ. High-dose furosemide in the treatment of refractory congestive heart failure. Arch Intern Med