Ramipril Prevents Endothelial Dysfunction Induced by Oxidized Low-Density Lipoproteins
A Bradykinin-Dependent Mechanism
Abstract We wished to determine whether the acute toxic effects of oxidized LDL are attenuated in aortas isolated from rats chronically treated with an angiotensin-converting enzyme (ACE) inhibitor. In aortic rings incubated with human oxidized LDL (300 μg/mL), the endothelium-dependent relaxations to acetylcholine were attenuated, but not those to A23187 and to nitroprusside. This toxic effect of oxidized LDL was completely prevented in preparations coincubated with oxidized LDL and the nitric oxide (NO) precursor l-arginine (0.3 mmol/L). In aortas isolated from rats orally treated for 6 weeks with 10 mg/kg ramipril (group 1) or 1 mg/kg ramipril (group 2), this toxic effect of oxidized LDL was also markedly attenuated. In contrast, in aortas isolated from rats cotreated with ramipril (10 mg/kg) for 6 weeks and subcutaneous injections of Hoe 140 (a B2 kinin antagonist), 500 μg/kg per day for the last 2 weeks (group 3) or from rats orally treated for 6 weeks with losartan (an AT1-type angiotensin II receptor antagonist), 20 mg/kg (group 4), the inhibitory effect of oxidized LDL on acetylcholine-induced relaxations was similar to that observed in the control group (group 5). Moreover, long-term treatment with ramipril increased relaxations to acetylcholine in groups 1 and 2 and also relaxations to A23187 and aortic cGMP content in group 1, suggesting an enhanced NO availability. Thus, the protective effect of long-term ACE inhibition against the acute vascular toxicity of oxidized LDL is bradykinin dependent and seems to involve a facilitation of NO release via endothelial B2 kinin receptors.
Attenuation of vascular endothelial dysfunction by ACE inhibitors has been demonstrated in experimental models of atherosclerosis1 2 and recently in patients with coronary artery disease.3 The mechanisms of this antiatherogenic effect remain controversial; reduction of angiotensin II levels should play a role, as this peptide not only stimulates the synthesis of growth and chemotactic factors4 5 but also promotes the generation of superoxide anion, leading to excessive degradation of NO.6 Nevertheless, several studies1 7 8 have shown that this vascular protection seems related to locally increased kinin by ACE inhibition. Indeed, inhibition of bradykinin breakdown seems to enhance formation of EDNO, which in addition to its vasodilator and antiaggregatory effects, has antiproliferative influence on vascular smooth muscle cells.9 This increase in EDNO availability has been demonstrated in acute conditions10 11 and also after chronic administration12 and seems to be mediated by endothelial B2 kinin receptors.
Therefore, we wished to determine whether aortas isolated from rats chronically treated with ramipril are protected against this oxLDL-induced endothelial dysfunction.
To assess whether the possible protective effect is related to bradykinin accumulation or angiotensin II reduction, some rats received ramipril and the B2 kinin receptor antagonist Hoe 140 (icatibant), whereas others were treated with the AT1-type angiotensin II receptor antagonist losartan.
Male Wistar rats weighing 100 to 150 g were housed in a controlled environment and given free access to standard rat chow and tap water. All animal procedures were approved by our university animal care and use committee.
Four groups of rats were treated during 6 weeks. This treatment duration was chosen according to our previous study demonstrating that only long-term administration of ramipril alters endothelial function.12 The first group orally received 10 mg/kg ramipril per day (in drinking water). In the second group, ramipril was given at a dose of 1 mg/kg per day for 6 weeks in drinking water. The third group received ramipril (10 mg/kg per day in drinking water) for 6 weeks and also, for the last 2 weeks, subcutaneous injections of Hoe 140 (250 μg/kg per day BID). According to our previous study,12 with this dose and method of administration, Hoe 140 has a selective antagonist property on B2 receptors and did not affect other endothelium-dependent responses. The fourth group orally received the AT1-type angiotensin II receptor antagonist losartan (20 mg/kg per day for 6 weeks). From a previous study15 and our preliminary experiments, this treatment is as potent as 10 mg/kg ramipril per day for reduction of blood pressure. The fifth group served as control and received no treatment except subcutaneous injections of saline solution (Hoe 140 solvent) during the last 2 weeks.
These various treatments did not alter the weight of the rats but decreased systolic blood pressure (measured by tail plethysmography by using an Apollo 179, IITC Inc).
At the end of these treatment periods, the rats were killed under ether anesthesia and their thoracic aorta was removed, cleaned of adhering fat, and cut into rings 3 to 4 mm long. During the dissection, utmost care was taken to protect the endothelial lining. In some preparations, the endothelium was removed by gently rubbing the intimal surface with forceps.
Preparation of LDL
LDL was isolated from human plasma collected in 1 mmol/L EDTA by sequential ultracentrifugation, with density adjustments by potassium bromide.14 16 To remove EDTA, LDL was passed through a PD 10 Sephadex G-25 M column (Pharmacia) and eluted in NH4HCO3 (5 μmol/L)/NaCl (0.15 mol/L). Protein concentration was determined by the method of Lowry et al.17 LDL was diluted in the elution buffer to yield a final concentration of 1 g protein per liter. LDL was then oxidized at a concentration of 5 μmol/L CuSO4 per 100 mg protein for 24 hours at room temperature.18 The reaction was stopped by addition of EDTA. OxLDL was passed through the PD 10 column and harvested in Krebs-Henseleit medium. Protein concentration was determined and LDL diluted to a final concentration of 300 μg protein per milliliter.
LDL samples prepared under similar conditions by simultaneous addition of copper and EDTA are referred to as non-oxLDL.
Studies of Vasomotor Responses
All rings were mounted under 1.5 g resting tension on stainless steel hooks in 20 mL organ baths. These organ chambers were filled with Krebs-Henseleit solution of the following composition (in mmol/L): NaCl 118.1, KCl 4.7, MgSO4 1.2, CaCl2 2.5, NaHCO3 25, and glucose 5, aerated with a mixture of 95% O2/5% CO2 and kept at 37°C. Tension was measured isometrically with a force transducer (Grass FT O3C) and recorded continuously with a transducer amplifier (Janssen Scientific Instruments) and a pen recorder. After 60 minutes of equilibration, the rings were progressively stretched and exposed to 40 mmol/L KCl at each level of stretch until the optimal point of the length-tension relation was reached.
The concentration-response curves to the various vasodilators were constructed on rings contracted with phenylephrine, and the concentration of this amine was adjusted to obtain equivalent plateaus in all preparations (1.3 to 1.8 g tension). After the plateau was reached, the relaxing agent was added cumulatively, each concentration being allowed to develop its maximal effect. Relaxations to nitroprusside were performed in vessels without endothelium to be in optimal conditions.19 The other experiments were performed in unrubbed vessels. To assess the effect of LDL on the responses to acetylcholine and nitroprusside, three preparations were studied in parallel; one served as control, and two concentration-response curves were constructed at 1-hour intervals to exclude an altered sensitivity of the preparation to these agonists. In the other two preparations, after a repeated washing of 30 minutes, the preparations were incubated for 30 minutes with LDL or oxLDL, and thereafter (with LDL left in the bath) the second concentration-response curve was constructed. As calcium ionophore induced a marked tachyphylaxis, only one concentration-response curve was constructed, and the responses in the absence or in the presence of LDL were compared in adjacent preparations.
To prevent foaming, Antifoam B (1 vol %) containing nonionic emulsifiers was added to the organ bath before addition of LDL. As pointed out in a previous study,20 this antifoam did not alter the physical integrity of LDL and did not modify the responses to the various agonists tested.
Measurement of Tissue Levels of GMP
Unrubbed rings were prepared as described above and incubated without tension in the Krebs solution aerated with a mixture of 95% O2/5% CO2 and kept at 37°C for 3 hours. Thereafter, tissues were snap frozen in liquid nitrogen, and cyclic nucleotide levels were assayed as previously described.12 22 Briefly, frozen tissues were homogenized in 6% ice-cold trichloroacetic acid and samples centrifuged. Supernatant fractions were extracted with ether and assayed for cGMP by using an enzyme immunoassay22 (Amersham). The acetylation technique was used. Each assay was performed in duplicate.
The following drugs were used: acetylcholine chloride (Sterop), phenylephrine hydrochloride, l-arginine, calcium ionophore A23187, Antifoam B, dextran sulfate (molecular weight, 1 000 000), and lysolecithin (all from Sigma Chemical Company), and nitroprusside (Roche). Ramipril, Hoe 140, and losartan were generously provided by Astra Pharma (Brussels, Belgium), Hoechst Marion Roussel (Frankfurt, Germany), and Merck Sharp & Dohme (Brussels, Belgium), respectively. Stock solutions of the drugs were prepared in distilled water, except phenylephrine, which was dissolved in distilled water containing ascorbic acid (1 mmol/L), and A23187, which was dissolved in DMSO. Ascorbic acid and DMSO did not alter smooth muscle tone in the concentrations resulting from drug additions.
Results are mean±SEM. The number of experiments is also the number of rats used. Relaxations are expressed as percentages of inhibition of tension developed to phenylephrine. For each concentration-response curve, the AUC was calculated and the concentration causing half-maximal relaxation was assessed and expressed as negative log molar concentration (pD2). A two-tailed paired t test was used to compare means of the same preparations and an unpaired t test to detect differences in the means of adjacent preparations. For multiple comparisons between groups, a one-way ANOVA followed by Fisher’s least significant difference test was used. Significance was accepted at P<.05.
Effect of LDL Incubation on Endothelium-Dependent and -Independent Relaxations in Control Group
OxLDL (300 μg/mL) significantly inhibited the relaxations to acetylcholine (Fig 1⇓ and Table 1⇓). This inhibition was completely prevented when the preparations were coincubated with oxLDL and l-arginine (0.3 mmol/L; Table 1⇓). The concentration-response curves to acetylcholine also remained unaltered when preparations were incubated with oxLDL in the presence of 10 μg/mL dextran sulfate, a specific blocker of the scavenger receptor (Table 1⇓).
In contrast to acetylcholine-induced responses, the relaxations to A23187 were unaffected by oxLDL (Fig 2⇓): the maximal relaxation was 76±4% (n=7) versus 81±3 (n=7) in the absence of oxLDL; the AUC and pD2 values were 181±6 and 8.1±0.2 (n=7) versus 172±11 and 8.0±0.1 (n=7, NS) in the absence of oxLDL.
Nitroprusside-induced relaxations were also unaltered by oxLDL (Fig 3⇓): the maximal response was 95±6 (n=7) versus 100±0 (n=7); the AUC and pD2 values were 176±16 and 7.7±0.2 (n=7) versus 163±21 and 7.9±0.3 (n=7) in the absence of oxLDL.
Ramipril Treatment and Endothelium-Dependent Relaxations
In the group treated with 10 mg/kg ramipril per day, acetylcholine-induced relaxations were significantly potentiated in comparison with the control group (Table 2⇓), whereas in the group treated with 1 mg/kg ramipril per day, only the maximal response to acetylcholine was increased. This potentiation was completely abolished in the group cotreated with 10 mg/kg ramipril per day and Hoe 140 (NS versus control group).
A23187-induced relaxations were also potentiated in the group treated with 10 mg/kg ramipril per day, but not in the other groups (Table 3⇓).
Ramipril Treatment and the NO–Guanylate Cyclase Pathway
The various in vivo treatments with ramipril did not modify the nitroprusside-induced relaxations (Table 4⇓).
In contrast, aortic cGMP content was significantly increased in the group treated with 10 mg/kg ramipril per day: 15.1±1.9 pmol/g wet wt versus 8.6±1.2 pmol/g wet wt in the control group (P<.05). In the groups treated with 1 mg/kg ramipril per day or 10 mg/kg ramipril per day and Hoe 140, aortic cGMP contents were similar to those of the control group: 8.0±1.1 pmol/g wet wt and 8.2±1.5 pmol/g wet wt, respectively (NS versus control group).
LDL Incubation and Relaxations to Acetylcholine in Ramipril-Treated Groups
In preparations isolated from the group treated with 10 mg/kg ramipril per day for 6 weeks (Fig 4⇓), oxLDL (300 μg/mL) did not modify the concentration-response curves to acetylcholine: the maximal relaxation was 93±5% (n=8) versus 97±2% (n=8) in the absence of oxLDL; the AUC and pD2 values were 131±22 (n=8) and 7.5±0.2 (n=8) versus 142±19 (n=8) and 7.2±0.1 (n=8) in the absence of oxLDL. In preparations isolated from the group treated with 1 mg/kg ramipril per day for 6 weeks (Fig 4⇓), oxLDL (300 μg/mL) also did not modify the AUC values: 251±32 (n=12) versus 211±20 (n=12) in the absence of oxLDL. The pD2 values were also similar: 6.3±0.4 (n=12) versus 6.7±0.2 (n=12). The maximal relaxation was, however, significantly (P<.05) attenuated: 76±7% (n=12) versus 91±4% (n=12) in the absence of oxLDL. Interestingly, ramiprilat incubation (10 μmol/L for 30 minutes) of control preparations or in vivo 24-hour ramipril (10 mg/kg) treatment did not attenuate the oxLDL-induced inhibition of the relaxations to acetylcholine (data not shown).
In the group treated with 10 mg/kg ramipril per day and Hoe 140 (Fig 4⇑), oxLDL also inhibited the responses to acetylcholine in a similar way to that observed in the control group: AUC values were 335±9 (n=10) versus 216±10 (n=10) in the absence of oxLDL (P<.001); the maximal relaxation was 50±3% (n=10) versus 82±4% (n=10) in the absence of oxLDL (P<.01).
LDL Incubation and Relaxations to Acetylcholine in Losartan-Treated Group
In preparations from the losartan-treated group (n=6), concentration-response curves to acetylcholine were similar to those of the control group; there was no attenuation of oxLDL-induced inhibition of the relaxations to acetylcholine (Table 5⇓).
Effects of the Various Treatments on Blood Pressure
After 6 weeks of treatment, systolic blood pressure decreased from 130±4 to 110±5 mm Hg (n=8, P<.01), from 129±3 to 115±4 mm Hg (n=12, P<.05), from 131±5 to 116±4 mm Hg (n=10, P<.05), and from 126±6 to 106±6 mm Hg (n=6, P<.01) in the groups treated with 10 mg ramipril, 1 mg ramipril , ramipril and Hoe 140, and losartan, respectively. In the control group (n=9), systolic blood pressure did not change significantly (from 128±5 to 131±4 mm Hg).
Our results show that human oxLDL but not native LDL inhibits the acetylcholine-induced responses in preparations isolated from control rats but not from those of ramipril-treated groups. The absence of LDL effects contrasts with the results observed in rabbit aorta, in which native LDL is as potent as oxLDL in inhibiting acetylcholine-induced relaxations.23 Therefore, rather than interacting with LDL receptor, in the current study, oxLDL seems to interfere with the endothelial function by activating the scavenger receptor. In line with this hypothesis, dextran sulfate completely prevented the toxic effect of oxLDL; this polymer is indeed considered as a competitive antagonist of acetylated LDL for the scavenger receptor and does not seem to directly interact with oxLDL.24 25 The discrepancy between rat and rabbit preparations may be related to species differences and also to the protocol used, as 10 times higher concentrations of LDL were used in the rabbit aorta. Although these high concentrations are similar to those found in plasma of hypercholesterolemic patients, the real concentrations of oxLDL are unknown, as oxidation of LDL seems to occur mainly in the subintima of blood vessels.26 According to Simon et al,20 an oxLDL concentration of 100 μg/mL can be easily reached in vivo within the vascular wall. Our aortic rat model, in fact, is very similar to that described in isolated pig coronary artery, in which similar concentrations of oxLDL were used.14 In this latter study,14 as in the current study, the relaxations to calcium ionophore, inducing a receptor-independent EDNO-mediated relaxation, and to nitroprusside, a direct donor of NO, were unaffected by the presence of oxLDL. Therefore, these data exclude a reduced responsiveness of vascular smooth muscle to NO and also suggest that the latter does not directly interact with oxLDL. As previously pointed out,14 oxLDL seems to interfere with the l-arginine pathway, since the effect of oxLDL is completely prevented in the presence of this NO synthase substrate. This finding is in agreement with clinical studies showing that l-arginine supplementation in hypercholesterolemic patients improves acetylcholine-induced vasodilation in peripheral and coronary circulations.27 28
In agreement with previous studies,12 29 ACE inhibitors administered at high doses for several weeks potentiate all endothelium-dependent relaxations. In the current study, not only were receptor-independent endothelial responses enhanced but also the aortic cGMP content, which reflects the amount of EDNO released under basal conditions. Moreover, as ramipril treatments did not modify the nitroprusside-induced relaxations, an alteration of the guanylate cyclase activity may be excluded. Therefore, long-term ACE inhibition increases EDNO availability. This effect seems to depend on the doses of ramipril used, since in the group treated with 1 mg/kg per day, only the maximal relaxation to acetylcholine was potentiated. This increase in EDNO may be related not only to an upregulation of the NO synthase pathway but also to a downregulation of oxidative enzyme activity, thereby decreasing NO breakdown. Both mechanisms may indeed act in concert, as NO reduces intracellular oxidative stress.30
Interestingly, the protective effect of ramipril against oxLDL-induced toxicity seems directly related to this increased EDNO availability. Indeed, oxLDL did not inhibit acetylcholine-induced relaxation in the preparations of the group treated with 10 mg/kg per day, whereas it slightly attenuated the maximal response to this agonist in preparations isolated from rats treated with 1 mg/kg ramipril per day. Furthermore, in vivo administration of Hoe 140 abolished not only this enhanced EDNO availability but also the protective effect of ramipril. Of note, these vascular effects of long-term administration of ramipril are highly sensitive to this B2 kinin antagonist, as they were abolished by the coadministration of Hoe 140 for only 2 weeks. Hence, long-term ACE inhibition probably counteracts the toxic effect of oxLDL via this enhanced EDNO availability. Nevertheless, long-term ACE inhibition and oxLDL could interfere at different levels in the NO synthase pathway. Indeed, oxLDL seems to alter receptor-operated NO formation, whereas ramipril (at least at high doses) interacts more directly with NO synthase activity, as it potentiates all EDNO-mediated effects, including the basal release of NO.
In cultured endothelial cells acutely exposed to ramiprilat (the active metabolite of ramipril), the enhanced formation of EDNO has been demonstrated.31 In contrast, in our experiments,12 acute effects of ramipril in vitro or in vivo did not affect the endothelium-dependent responses. In line with this finding, experimental endothelial dysfunction induced by hypercholesterolemia or perivascular manipulation (producing neointima formation) is attenuated by chronic but not acute ACE inhibition.2 32 According to Holtz and Goetz,33 this endothelial protection may be mainly the result of an upregulation of constitutive NO synthase induced by the chronic exposure to bradykinin.
In the current study, inhibition of angiotensin II effects does not seem to play a major role, since in preparations isolated from losartan-treated rats, oxLDL inhibited the responses to acetylcholine in a similar way to that observed in the control group. Nevertheless, in atherosclerotic arteries, superoxide anion production is enhanced,34 and therefore inhibition of angiotensin II effects may be more relevant, since as mentioned above, angiotensin II seems to promote the generation of this oxygen-derived radical.6
Recently it has been shown that ACE inhibitors also potentiate responses mediated by the endothelium-derived hyperpolarizing factor.35 We might speculate that the protective mechanism observed in the present study is also related to this potentiation of endothelium-derived hyperpolarizing factor–mediated effects, especially since lysolecithin, produced during oxidation of LDL, seems to specifically inhibit endothelium-derived hyperpolarizing factor–mediated responses.36 However, this probability seems unlikely, since in the present study, the toxic effect of oxLDL is completely prevented by l-arginine incubation, suggesting a direct interference with the NO synthase pathway. Moreover, the effects of lysolecithin may not exactly mimic those of oxLDL. In contrast to oxLDL, lysolecithin inhibits calcium ionophore–induced relaxations in isolated vessels.14 36
In conclusion, our results indicate that human oxLDL impairs muscarinic receptor–operated NO formation and that long-term ACE inhibition protects the endothelium against this vascular toxicity. This beneficial action is related to an increase in EDNO availability via B2 kinin receptor.
Selected Abbreviations and Acronyms
|AUC||=||area under the curve|
This work was supported by research grants from Fonds National de la Recherche Scientifique (FNRS 9.4553.92), the Bekales Foundation, and Fondation pour la Chirurgie Cardiaque. We thank Astra Pharma, Hoechst Marion Roussel, and Merck Sharp & Dohme for the supply of ramipril, Hoe 140, and losartan. We also thank Micheline Lambermont, MD, for supplying human plasma and Patricia D’hondt, PhD, for preparing the LDL.
- Received November 5, 1996.
- Revision received December 11, 1996.
- Accepted January 23, 1997.
Becker RHA, Wiemer G, Linz W. Preservation of endothelial function by ramipril in rabbits on a long-term atherogenic diet. J Cardiovasc Pharmacol. 1991;18(suppl 2):S110-S115.
Mancini GBJ, Henry GC, Macaya C, O’Neill BJ, Pucillo AL, Carere RG, Waegovich TJ, Mudra H, Lüscher TF, Klibaner MI, Haber HE, Uprichard ACG, Pepine CJ, Pitt B. Angiotensin-converting enzyme inhibition with quinapril improves endothelial vasomotor dysfunction in patients with coronary artery disease: the TREND (Trial on Reversing ENdothelial Dysfunction) study. Circulation. 1996;94:258-265.
Stouffer GA, Owens GK. Angiotensin II–induced mitogenesis of SHR-derived cultured smooth muscle cells is dependent on autocrine production of TGF-B. Circ Res. 1992;70:820-828.
Daemen MJAP, Lombardi DM, Bosman FT, Schwartz SM. Angiotensin II induces smooth muscle cell proliferation in the normal and injured rat arterial wall. Circ Res. 1991;68:450-456.
Linz W, Wiemer G, Schölkens BA. Contribution of bradykinin to the cardiovascular effects of ramipril. J Cardiovasc Pharmacol. 1993;22(suppl 9):S1-S8.
Wiemer G, Schölkens BA, Linz W. Endothelial protection by converting enzyme inhibitors. Cardiovasc Res. 1994;28:166-172.
Zanzinger J, Zheng X, Bassenge E. Endothelium-dependent vasomotor responses to endogenous agonists are potentiated following ACE inhibition by a bradykinin-dependent mechanism. Cardiovasc Res. 1994;28:209-214.
Berkenboom G, Brékine D, Unger P, Grosfils K, Staroukine M, Fontaine J. Chronic angiotensin-converting enzyme inhibition and endothelial function of rat aorta. Hypertension. 1995;26:738-743.
Quinn MT, Parthasarathy S, Fong LG, Steinberg D. Oxidatively modified low density lipoproteins: a potential role in recruitment and retention of monocyte/macrophages during atherogenesis. Proc Natl Acad Sci U S A. 1987;84:2995-2998.
Tanner FC, Noll G, Boulanger CM, Lüscher TF. Oxidized low density lipoproteins inhibit relaxations of porcine coronary arteries: role of scavenger receptor and endothelium-derived nitric oxide. Circulation. 1991;83:2012-2020.
Farhy RD, Carretero OA, Ho KL, Scicli AG. Role of kinins and nitric oxide in the effects of angiotensin converting enzyme inhibitors on neointima formation. Circ Res. 1993;72:1202-1210.
Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Protein measurement with the Folin phenol reagent. J Biol Chem. 1951;193:265-275.
Steinbrecher UP, Parthasarathy S, Leake DS, Witztum JL, Steinberg D. Modification of low density lipoprotein by endothelial cells involves lipid peroxidation and degradation of low density lipoprotein phospholipids. Proc Natl Acad Sci U S A. 1984;81:3883-3887.
Steiner AL, Parker CW, Kipnis DM. Radioimmunoassay for cyclic nucleotides. J Biol Chem. 1972;247:1106-1113.
Goldstein JL, Ho YK, Basu SK, Brown MS. Binding site on macrophages that mediates uptake and degradation of acetylated low density lipoprotein, producing massive cholesterol deposition. Proc Natl Acad Sci U S A. 1979;76:333-337.
Bossaller C, Auch-Schwelk W, Weber F, Götze S, Gräfe M, Graf K, Fleck E. Endothelium-dependent relaxations are augmented in rats chronically treated with the angiotensin converting enzyme inhibitor enalapril. J Cardiovasc Pharmacol. 1992;20(suppl 9):S91-S95.
Niu XF, Smith CW, Kubes P. Intracellular oxidative stress induced by nitric oxide synthesis inhibition increases endothelial cell adhesion to neutrophils. Circ Res. 1994;74:1133-1140.
Hecker M, Dambacher T, Busse R. Role of endothelium-derived bradykinin in the control of vascular tone. J Cardiovasc Pharmacol. 1992;20(suppl 9):S55-S61.
Holtz J, Goetz RM. Vascular renin-angiotensin-system, endothelial function and atherosclerosis? Basic Res Cardiol. 1994;89(suppl 1):71-86.
Mombouli JV, Illiano S, Nagao T, Vanhoutte PM. The potentiation of bradykinin-induced relaxations by perindoprilat in canine coronary arteries involves both nitric oxide and endothelium-derived hyperpolarizing factor. Circ Res. 1992;71:137-144.
Eizawa H, Yui Y, Inoue R, Kosuga K, Hattori R, Aoyama T, Sasayama S. Lysophosphatidylcholine inhibits endothelium-dependent hyperpolarization and Nω-nitro-l-arginine/indomethacin–resistant endothelium-dependent relaxation in the porcine coronary artery. Circulation. 1995;92:3520-3526.