Endothelial dysfunction is known to occur in a number

Transcription

1 Endothelium-Derived Hyperpolarizing Factor Identification and Mechanisms of Action in Human Subcutaneous Resistance Arteries Paul Coats, PhD, BSc; Fiona Johnston, BSc; John MacDonald, MD, MRCP, BSc; John J.V. McMurray, MD, FRCP, FESC, BSc; Chris Hillier, PhD, BSc Background Both a vascular endothelial cytochrome P450 (CYP450) product of arachidonic acid metabolism and the potassium ion (K ) have been identified as endothelium-derived hyperpolarizing factors (EDHFs) in animal vascular tissues. We studied the relative importance of EDHF, nitric oxide (NO), and prostacyclin (PGI 2 ) as vasodilators in human subcutaneous arteries. We also examined the mechanisms underlying the vasodilator action of EDHF to elucidate its identity. Methods and Results Subcutaneous resistance arteries were obtained from 41 healthy volunteers. The contribution of EDHF to the vasodilation induced by acetylcholine was assessed by inhibiting production of NO, PGI 2, and membrane hyperpolarization. The mechanisms underlying the relaxation evoked by K and EDHF were elucidated. EDHF was found to account for 80% of acetylcholine-mediated vasorelaxation. Its action was insensitive to the combination of barium and ouabain, whereas barium and ouabain reversed K -mediated vasorelaxation. EDHF-mediated vasorelaxation, however, was sensitive to the phospholipase A 2 inhibitor oleyloxyethyl phosphorylcholine and the CYP450 inhibitor ketoconazole. Conclusions EDHF is the major contributor to endothelium-dependent vasorelaxation in human subcutaneous resistance arteries. A product of phospholipase A 2 /CYP450 dependent metabolism of arachidonic acid and not K is the likely identity of EDHF in human subcutaneous resistance arteries. (Circulation. 2001;103: ) Key Words: endothelium-derived factors nitric oxide Endothelial dysfunction is known to occur in a number of cardiovascular diseases, including atherosclerosis, hypertension, and heart failure. 1 6 Therefore, considerable interest is attached to understanding the mechanisms underlying endothelium-dependent vasorelaxation. It is hoped that therapeutic strategies may be developed to counter vascular endothelial dysfunction in these and other diseases. 7 The vascular endothelium modulates local blood flow via the dynamic release of numerous vasoactive factors. Receptor-dependent agonists, such as acetylcholine, bradykinin, and substance P, indirectly relax vascular smooth muscle by inducing the release of the known endothelium-derived relaxing factors (EDRFs) nitric oxide (NO), prostacyclin (PGI 2 ), and endothelium-derived hyperpolarizing factor (EDHF) At present, there is no clear consensus on the identity of EDHF or the exact mechanisms by which EDHF relaxes vascular smooth muscle. Some evidence, however, demonstrates that the action of EDHF involves calcium-sensitive K channels (K Ca ), which are sensitive to the combined effects of the toxins apamin and charybdotoxin To date, numerous studies have aimed to identify the nature of EDHF and its mechanisms of action. Experimental evidence suggests that the most likely candidates are the P450 metabolite of arachidonic acid metabolism, 11,12- epoxyeicosatrienoic acid (11,12-EET), and the potassium ion (K ) K has long been known to possess vasodilator properties Recently, Edwards and colleagues 15 proposed that K, as an EDHF, was extruded from the endothelium into the myoendothelial gap, leading to activation of vascular smooth muscle inwardly rectifying potassium channels and Na 2,K -ATPase pumps. This in turn resulted in efflux of K from the vascular smooth muscle, inducing hyperpolarization and vasorelaxation. Other recent and compelling evidence, however, suggests that the CYP450 enzyme product 11,12- EET, derived from the endothelium-dependent metabolism of arachidonic acid, is an EDHF in porcine coronary arteries and hamster gracilis muscle artery. 19,23 The evidence that K or a P450 enzyme product is an EDHF, however, remains controversial Therefore, real uncertainty exists regarding the identification of EDHF. Received July 13, 2000; revision received October 4, 2000; accepted October 16, From the Department of Biological and Biomedical Sciences, Glasgow Caledonian University (P.C., F.J., C.H.), and the Department of Medicine and Therapeutics, Western Infirmary (J.M., J.J.V.M), Glasgow, Scotland. Correspondence to Dr Paul Coats, School of Biological and Biomedical Sciences, Glasgow Caledonian University, Cowcaddens Road, Glasgow, Scotland, UK. p.coats@gcal.ac.uk 2001 American Heart Association, Inc. Circulation is available at

2 Coats et al Endothelium-Derived Hyperpolarizing Factor 1703 TABLE 1. Basic Clinical Details of Volunteers Who Provided Biopsies Age, y Creatinine, mol/l Urea, mmol/l Cholesterol, mmol/l Glucose, mmol/l Systolic BP, mm Hg Diastolic BP, mm Hg BP indicates blood pressure There have been few studies of EDHF in human arteries. Nonetheless, it appears that an endothelium-dependent non- NO, nonprostanoid mechanism of vasorelaxation predominates in these arteries To date, there has been no specific study of EDHF in human subcutaneous resistance arteries. Therefore, in this study we aimed to identify the mechanisms of action and the likely identity of EDHF in human small subcutaneous resistance arteries. Methods Patients and Vessel Preparation The hospital ethics review committee approved the study, and each subject gave informed written consent. Clinical characteristics of the volunteers are given in Table 1. Healthy volunteers with no history of vascular disease, diabetes, hypertension, or renal impairment attended the Clinical Investigations Research Unit at the Western Infirmary, Glasgow. Subcutaneous gluteal fat biopsies ( cm) were excised under local anesthesia with 1% lidocaine and immediately transferred to cold physiological saline solution (PSS). Forty-six subcutaneous resistance-size arteries (lumen diameter m) were isolated from 41 biopsies and cleaned of any adherent tissue under a dissection microscope (Zeiss Stemi 2000, magnification 6 to 45). Isolated arteries were mounted on a pressure myograph (Danish MyoTech P100) and studied pharmacologically at an intraluminal pressure of 40 mm Hg in a no-flow state by methods described previously. 35 Briefly, PSS was gassed with 95% O 2 /5% CO 2, with ph maintained at 7.4 at 37 C. Functional viability was assessed by maximum vasoconstriction to 60 mmol/l K PSS and norepinephrine (10 mol/l) and vasorelaxation ( 80%) to acetylcholine. All arteries fulfilled these criteria, and none were discarded. Pharmacological Protocols All studies were carried out in arteries preconstricted with norepinephrine to 80% of maximum response. In experiments in which K -modified PSS was used, the concentration of norepinephrine was reduced to maintain a preconstriction diameter similar to previous cumulative concentration-response curves (CCRCs). Relative Importance of EDHF CCRCs were constructed with acetylcholine alone and after cumulative incubation with 100 mol/l N G -nitro-l-arginine (L-NOARG) for 1 hour, 30 mol/l indomethacin for 30 minutes, and 25 mmol/l K PSS Identification of K Channels All subsequent acetylcholine experiments were carried out with arteries preincubated with 100 mol/l L-NOARG 30 mol/l indomethacin. CCRCs were then constructed after intraluminal incubation with either (1) glibenclamide (10 mol/l); (2) apamin (100 nmol/l); (3) charybdotoxin (100 nmol/l); (4) apamin charybdotoxin; (5) apamin iberiotoxin; (6) barium (Ba 2,30 mol/l); (7) ouabain (1 mmol/l); and (8) Ba 2 ouabain. Role of Arachidonic Acid Acetylcholine CCRCs were constructed after intraluminal incubation with oleyloxyethyl phosphorylcholine (OOPC) and ketoconazole. Endothelial specificity of these inhibitors was assessed by repeating CCRCs with sodium nitroprusside, pinacidil, and 1-ethyl-2- benzimidazoline (EBIO) in the presence of maximal concentrations of OOPC and ketoconazole. Mechanisms of K -Mediated Vasorelaxation In standard PSS, the extracellular K concentration ([KCl] o ) was elevated from 4.6 to 20 mmol/l in 2-mmol/L steps. This protocol was then repeated in the presence of Ba 2 or ouabain and Ba 2 ouabain. Determination of cgmp Content Human subcutaneous artery segments ( 3 mm) were isolated from 3 subcutaneous biopsies and incubated in PSS in the presence of either (1) norepinephrine (1 mol/l), (2) norepinephrine acetylcholine (10 mol/l), (3) norepinephrine acetylcholine L-NOARG (100 mol/l), or (4) norepinephrine acetylcholine oxadiazoloquinoxalin (ODQ, 10 mol/l). All experiments were in the presence of isobutylmethylxanthine (10 mol/l). At the end of the incubation period, the tissues were immersed in liquid nitrogen, homogenized in 95% ethanol, and centrifuged at 3000 rpm for 15 minutes. The supernatant was extracted and cgmp determined by radioimmunoassay. 39 The amount of protein in the centrifuged pellet was determined by Bradford s assay. 40 Drugs and Solutions Acetylcholine, isobutylmethylxanthine, indomethacin, L-NOARG, norepinephrine, pinacidil, EBIO, and sodium nitroprusside were purchased from Sigma Chemical Co; apamin from Calbiochem; charybdotoxin from Bachem; ketoconazole from Tocris Cookson Ltd; and OOPC from Affinity Research. Anti-cGMP was a gift from Dr David Bunton (Glasgow Caledonian University, UK). PSS composition (in mmol) was NaCl 119, KCl 4.5, NaHCO 3 25, KH 2 PO 4 1.0, MgSO 4 7H 2 O 1.0, glucose 11.0, and CaCl K PSS composition (25 and 60 mmol/l) was equimolar substitution of NaCl with KCl. Data and Statistical Analysis Relaxation data are represented as relaxation relative to the preconstricted diameter of the artery. Values are presented as mean SEM. Statistical comparisons of pec 50 (concentration required to produce 50% of the maximum response) and maximum response were performed with Student s paired t test followed by multiple comparisons by Bonferroni s test where appropriate. Not calculable (NC) appears where the pec 50 could not be determined. Comparison of Figure 1. Effect of cumulative incubation with L-NOARG (100 mol/l), indomethacin (INDO; 30 mol/l), and 25 mmol/l K PSS on acetylcholine (ACh) vasorelaxation (ANOVA for repeated measures, comparison of curves: *P 0.05, response vs ACh, n 12).

3 1704 Circulation March 27, 2001 Figure 2. Measurement of cgmp by radioimmunoassay in human subcutaneous resistance arteries after (I) application of norepinephrine (NE, 1 mol/l), (II) NE plus acetylcholine (10 mol/l), (III) NE plus acetylcholine in presence of L-NOARG (100 mol/l), and (IV) NE plus acetylcholine in presence of ODQ (10 mol/l). n 3. *P 0.05 response vs previous response, Bonferroni s t test for multiple comparisons. CCRCs was by one-way ANOVA for repeated measures. Statistical significance was assumed at a value of P Results The effects of L-NOARG, indomethacin, and 25 mmol/l K PSS on acetylcholine-mediated vasorelaxation are shown in Figure 1. Incubation with L-NOARG significantly reduced both the maximum relaxation and pec 50 ( versus nmol/l, P 0.05) to acetylcholine. Subsequent incubation with indomethacin failed to modify this response further. Exchange of PSS with 25 mmol/l K PSS, however, abolished the relaxation response to acetylcholine (Figure 1). Figure 2 summarizes the results of the cgmp radioimmunoassay. As expected, acetylcholine substantially increased cgmp generation. Incubation with L-NOARG or ODQ abolished the acetylcholine-dependent liberation of cgmp. Both L-NOARG (100 mol/l) and ODQ (10 mol/l) were equipotent at inhibiting the generation of cgmp. Incubation with 10 mol/l glibenclamide had no effect on the acetylcholine-dependent response (Figure 3). Apamin or charybdotoxin alone also had little effect on the response to acetylcholine (Figure 4). Incubation with the combination of Figure 4. Vasorelaxation to acetylcholine (ACh) in presence of after luminal incubation with either apamin (100 nmol/l) or charybdotoxin (ChTX, 100 nmol/l) and apamin ChTX (n 6). apamin and charybdotoxin, however, abolished the relaxation to acetylcholine. Substitution of charybdotoxin with iberiotoxin had no effect on the relaxation response (Figure 5). Cumulative addition of KCl on preconstricted arteries initially produced a concentration-dependent relaxation (4.6 to 14 mmol/l). This was reversed at higher concentrations (16 to 20 mmol/l) to a vasoconstriction response (Figure 6). Ba 2 (30 mol/l) or ouabain (1 mmol/l) alone was unable to significantly modify the K -mediated responses (data not presented). When combined, however, Ba 2 and ouabain reversed K -induced vasorelaxation, resulting in vasoconstriction at all concentrations of KCl (Figure 6). In contrast, Ba 2 and ouabain either alone (data not shown) or combined, at concentrations that reversed K -mediated vasorelaxation, had no effect on acetylcholine-mediated vasorelaxation (Figure 7). The phospholipase A 2 inhibitor OOPC reduced both sensitivity and maximum relaxation responses to the L-NOARG/ Figure 3. Vasorelaxation to acetylcholine (ACh) in presence of after luminal incubation with glibenclamide (10 mol/l, n 6). Figure 5. Vasorelaxation to acetylcholine (ACh) in presence of after luminal incubation with apamin and iberiotoxin (IbTX, 100 nmol/l, n 3).

4 Coats et al Endothelium-Derived Hyperpolarizing Factor 1705 Figure 6. Effect of increasing [K o] from 4.6 to 20 mmol/l in small steps before and after incubation with combination of Ba 2 (30 mol/l) and ouabain (1 mmol/l, n 6). indomethacin-insensitive component of acetylcholinemediated vasorelaxation in a concentration-dependent manner (Figure 8; EC 50 : ACh, mol/l; 30 mol/l OOPC, mol/l; 100 mol/l OOPC, NC; P 0.05, 30 mol/l OOPC versus ACh). OOPC had no effect on the sensitivity to norepinephrine, because the concentration used for preconstriction remained relatively constant. Furthermore, OOPC failed to inhibit the response to cumulative addition of sodium nitroprusside (Table 2). Incubation with the highest OOPC concentration (100 mol/l) failed to modify the relaxation response to the ATP-sensitive K channel opener pinacidil and the Ca 2 -sensitive K channel opener EBIO (Table 2). The cytochrome P450 inhibitor ketoconazole also resulted in a concentration-dependent inhibition of the L-NOARG/ indomethacin-insensitive relaxation to acetylcholine (Figure 9; EC 50 : ACh mol/l; 1 mol/l ketoconazole, mol/l; 10 mol/l ketoconazole, mol/l; 30 mol/l ketoconazole, NC; 100 mol/l ketoconazole, NC; P 0.05, 1 mol/l ketoconazole versus Figure 7. Vasorelaxation to acetylcholine (ACh) in presence of after incubation with Ba 2 and ouabain (n 6). Figure 8. Vasorelaxation to acetylcholine (ACh) in presence of 100 mol/l L-NOARG 30 mol/l indomethacin after luminal incubation with phospholipase A 2 inhibitor OOPC. ANOVA for repeated measures, comparison of curves: *P 0.05, ACh response vs previous ACh response (n 6). ACh and 10 mol/l ketoconazole versus 1 mol/l ketoconazole). Again, ketoconazole did not modify the sensitivity of the tissue to the preconstricting agonist norepinephrine. Also, sodium nitroprusside, EBIO-, and pinacidil-dependent responses were unaffected by the highest concentration of ketoconazole (100 mol/l, Table 3). Discussion This study has 2 important findings. First, it demonstrates that EDHF in human subcutaneous resistance arteries is the major component of endothelium-dependent vasorelaxation. Second, a product of arachidonic acid metabolism, probably a cytochrome P450 product rather than the potassium ion, is likely to be an EDHF in these arteries. A number of studies have suggested that a substance independent of NO and PGI 2 is the major EDRF(s) in resistance arteries. Our results demonstrate that the L-NOARG/indomethacin-insensitive component of acetylcholine-mediated relaxation is the major component of endothelium-dependent relaxation in human subcutaneous resistance arteries. The NO component, ie, the L-NOARG sensitive component, accounted for only 20% of the maximum relaxation response to acetylcholine. Furthermore, indomethacin had no effect on the endothelium-dependent relaxation, indicating that PGI 2 plays no role in the relaxation response in these ex vivo conditions. In contrast, the major component of the endothelium-dependent relaxation was sensitive to 25 mmol/l K PSS. This sensitivity to high TABLE 2. Sensitivity (pec 50 ) and Maximum Relaxation Response to Pinacidil (n 3), Sodium Nitroprusside (SNP, n 3), and EBIO (n 3) Before and After Incubation With 100 mol/l OOPC Maximum Relaxation, pec 50, nmol/l % Pinacidil Pinacidil 100 mol/l OOPC SNP SNP 100 mol/l OOPC EBIO NC 86 4 EBIO 100 mol/l OOPC NC 84 4

5 1706 Circulation March 27, 2001 Figure 9. Vasorelaxation to acetylcholine (ACh) in presence of 100 mol/l L-NOARG 30 mol/l indomethacin after luminal incubation with P450 enzyme inhibitor ketoconazole (KETO). ANOVA for repeated measures, comparison of curves: *P 0.05, ACh response vs previous ACh response (n 6). concentrations of K PSS, which has been shown to be a characteristic of EDHF, accounted for 75% of the maximum response to acetylcholine. 15,36,38,41,42 This study confirms the relative importance of EDHF and that NO and PGI 2 play relatively minor roles in endothelium-dependent relaxation of human resistance arteries. 11,34,43,44 In vessels in which NO synthase and cyclooxygenase-1 are inhibited, glibenclamide, a K ATP channel inhibitor previously shown to inhibit EDHF-mediated vasorelaxation in rabbit cerebral arteries, had no measurable effect on EDHFmediated relaxation to acetylcholine in human subcutaneous resistance arteries. 45 This indicates that K ATP channels have no role in this response. One of the few points of consensus to emerge in EDHF research has been that preincubation of arteries with either apamin, an inhibitor of large-conductance calcium-sensitive K channels (BK Ca ), or charybdotoxin, an inhibitor of intermediate-conductance calcium-sensitive K channels (IK Ca ), alone has little effect on the EDHF-mediated relaxation to acetylcholine; when combined, however, apamin and charybdotoxin together abolish the EDHF-mediated relaxation. 13,15,46 Our results confirm that the EDHF-mediated relaxation in human subcutaneous resistance arteries is similarly resistant to apamin or charybdotoxin alone but is abolished by the combination of the 2 toxins. Interestingly, substitution of charybdotoxin with iberiotoxin, a selective inhibitor of BK Ca, in the presence of apamin had no effect on the relaxation response. 47 Interpretation of these findings is TABLE 3. Sensitivity (pec 50 ) and Maximum Relaxation Response to Pinacidil (n 3), Sodium Nitroprusside (SNP, n 3), and EBIO (n 3) Before and After Incubation with 100 mol/l Ketoconazole (KETO) Maximum Relaxation, pec 50, nmol/l % Pinacidil Pinacidil 100 mol/l KETO SNP SNP 100 mol/l KETO EBIO NC 87 4 EBIO 100 mol/l KETO NC 89 3 difficult. Inhibition of IK Ca or BK Ca alone clearly does not inhibit the actions of EDHF, suggesting that there is a BK Ca /IK Ca channel codependency involved in this mechanism of vasorelaxation. Alternatively, there may be an as yet unidentified BK Ca /IK Ca -like heteromultimeric K channel associated with an EDHF-mediated response. 48 Charybdotoxin has inhibitory action at BK Ca and a number of voltagesensitive K channels. 49 The fact that charybdotoxin, as opposed to iberiotoxin, is required in combination with apamin to block the relaxation response insensitive to L-NOARG and indomethacin suggests that Ca 2 -sensitive K channels and voltage-sensitive K channels are involved in this response. Elevation of the extracellular potassium concentration, [K ] o, has been shown to dilate isolated arteries from rat cerebral, coronary, hepatic, and mesenteric vascular beds. 15,21 The mechanistic basis of this response involves K efflux from vascular smooth muscle via the inwardly rectifying potassium channel (K IR ) and the Na,K -ATPase pump, leading to hyperpolarization and vasorelaxation. Using the protocols established by Edwards and colleagues, we mimicked the effects of EDHF by increasing [K ] o. In our study, elevation of [K ] o resulted in a small vasorelaxation of preconstricted arteries. The maximum relaxation response occurred at 14 to 16 mmol/l [K ] o ; increasing [K ] o further resulted in vasoconstriction. The [K ] o -dependent vasodilation observed was sensitive to the combination of Ba 2,an inhibitor of K IR, plus ouabain, an inhibitor of the Na,K - ATPase pump, but not Ba 2 or ouabain alone. In fact, elevation of [K ] o in the presence of Ba 2 plus ouabain resulted in a concentration-dependent vasoconstriction at all increments of [K ] o and in all arteries studied. This observation confirms K as having vasodilator properties and confirms the role of the K IR and the Na,K -ATPase pump in K -dependent vasodilation in these vessels. Ba 2 plus ouabain, however, had no effect on the EDHF-mediated relaxation to acetylcholine. This observation shows that EDHF is insensitive to inhibition of the K IR and the Na,K - ATPase pump. These results demonstrate that the EDHF component of the acetylcholine-mediated endothelium-dependent relaxation is mediated via mechanisms different from those through which K mediates vasorelaxation. If K were an EDHF in human subcutaneous resistance arteries, this response should have been sensitive to the combination of Ba 2 plus ouabain. Phospholipase A 2 dependent metabolism of membranebound phospholipids is a primary source of arachidonic acid and contributes to EDHF-mediated vasorelaxation in rabbit mesenteric arteries. 50 In turn, the metabolism of arachidonic acid by P450 enzymes into vasoactive substances is a commonly identified phenomenon in vascular physiology In the present study, luminal incubation with OOPC, a specific inhibitor of phospholipase A 2, had a profound effect on endothelium-dependent relaxation to EDHF. Likewise, luminal incubation with the P450 enzyme inhibitor ketoconazole resulted in a concentration-dependent inhibition. The OOPC and ketoconazole results provide strong evidence that the endothelium-dependent relaxation to EDHF is a product

6 Coats et al Endothelium-Derived Hyperpolarizing Factor 1707 of phospholipase A 2 /arachidonic acid/p450 enzyme metabolism in human subcutaneous resistance arteries. We are confident that these results reflect specifically endothelium-dependent mechanisms, because the vasorelaxation response to the endothelium-independent vasodilator sodium nitroprusside was unaffected by the highest concentrations of OOPC and ketoconazole. Moreover, the endothelium-independent mechanisms of hyperpolarization were unaffected by this treatment, because the K ATP channel opener pinacidil and the Ca 2 -sensitive K channel opener EBIO produced potent vasorelaxation unaffected by either OOPC or ketoconazole. In this study, we were unable to measure smooth muscle membrane potential and therefore have no direct evidence that the acetylcholine-mediated relaxation insensitive to L-NOARG and indomethacin is definitely the result of hyperpolarization. Raising [K ] o to 25 mmol/l, however, completely blocked the L-NOARG/indomethacin-insensitive response to acetylcholine. Previous studies have shown that raising [K ] o has little effect on NO and PGI 2 -mediated vasorelaxation, but rather it specifically antagonizes the actions of EDHF by counterbalancing smooth muscle cell membrane potential. Experimental evidence suggests that the commonly used concentrations of NO synthase inhibitors may not be completely effective in blocking all NO production. 56,57 This is important, because NO has been shown to have a hyperpolarizing effect via cgmp in vascular smooth muscle and the response insensitive to L-NOARG may be as a consequence of residual NO production. L-NOARG (100 mol/l) in this tissue, however, abolished the acetylcholine-dependent liberation of cgmp. Therefore, we are confident that there was no residual NO and that the relaxation responses observed were independent of NO. Moreover, NO has been shown to hyperpolarize vascular tissue via cgmp- and campdependent mechanisms sensitive to iberiotoxin and glibenclamide, respectively. Both iberiotoxin and glibenclamide had no effect on the relaxation response insensitive to L-NOARG and indomethacin. Therefore, we are confident that the acetylcholine-mediated relaxation insensitive to L-NOARG and indomethacin is truly an NO/prostanoidindependent phenomenon and is likely to be mediated by an EDHF. This study shows that endothelium-dependent relaxation to EDHF is mediated via mechanisms different from those mediating K vasorelaxation; therefore, it is unlikely that K is an EDHF in human subcutaneous resistance arteries. Moreover, this mechanism is sensitive to phospholipase A 2 and P450 enzyme inhibition, thus indicating that EDHF in human subcutaneous resistance arteries is likely to be a P450 enzyme dependent product of arachidonic acid metabolism. Also, vasodilation depends largely on K channels, suggesting that selective use of appropriate potassium channel drugs may provide a useful therapeutic approach for the treatment of vascular pathologies. Acknowledgments This study was supported by the Barnwood House Trust. Dr Coats was supported by a Glasgow Caledonian University (UK) Research Studentship. The authors thank Dr David Bunton for his assistance with cgmp radioimmunoassay. References 1. Busse R, Fleming I. Endothelial dysfunction in atherosclerosis. J Vasc Res. 1996;33: Lindsay DC, Holdright DR, Clarke D, et al. Endothelial control of lower limb blood flow in chronic heart failure. Heart. 1996;75: Nakamura M, Ishikawa M, Funakoshi T, et al. Attenuated endotheliumdependent peripheral vasodilation and clinical characteristics in patients with chronic heart failure. Am Heart J. 1994;128: Panza JA, Casino PR, Kilcoyne CM, et al. Role of endothelium-derived nitric oxide in the abnormal endothelium-dependent vascular relaxation of patients with essential hypertension. Circulation. 1993;87: Zhang X, Zhao S, Li X, et al. Endothelium-dependent and -independent functions are impaired in patients with coronary heart disease. Atherosclerosis. 2000;149: De Meyer GR, Herman AG. Vascular endothelial dysfunction. Prog Cardiovasc Dis. 1997;39: Cooke JP. The endothelium: a new target for therapy. Vasc Med. 2000; 5: Furchgott RF, Zawadzki JV. The obligatory role of endothelial cells in the relaxation of arterial smooth muscle by acetylcholine. Nature. 1980;288: Moncada S, Vane JR. Pharmacology and endogenous roles of prostaglandins, endoperoxides, thromboxane A 2 and prostacyclin. Pharmacol Rev. 1979;30: Feletou M, Vanhoutte PM. Endothelium-dependent hyperpolarization of canine coronary smooth muscle. Br J Pharmacol. 1988;93: Taylor SG, Weston AH. Endothelium-derived hyperpolarizing factor: a new endogenous inhibitor from the vascular endothelium. Trends Pharmacol Sci. 1988;9: Hecker M, Bara AT, Bauersachs J, et al. Characterization of endothelium-derived hyperpolarizing factor as a cytochrome P450-derived arachidonic acid metabolite in mammals. J Physiol. 1994;481: Petersson J, Zygmunt PM, Hogestatt ED. Characterisation of potassium channels involved in EDHF-mediated relaxation in cerebral arteries. Br J Pharmacol. 1997;120: Zygmunt PM, Edwards G, Weston AH, et al. Involvement of voltagedependent potassium channels in the EDHF-mediated relaxation of rat hepatic artery. Br J Pharmacol. 1997;121: Edwards G, Dora KA, Gardener MJ, et al. K is an endothelium-derived hyperpolarizing factor in rat arteries. Nature. 1998;396: Komori K, Vanhoutte PM. Endothelium-derived hyperpolarizing factor. Blood Vessels. 1990;27: Campbell WB, Gebremedhin D, Pratt PF, et al. Identification of epoxyeicosatrienoic acids as endothelium-derived hyperpolarizing factors. Circ Res. 1996;78: Chen G, Cheung DW. Modulation of endothelium-dependent hyperpolarization and relaxation to acetylcholine in rat mesenteric artery by cytochrome P450 enzyme activity. Circ Res. 1996;79: Fisslthaler B, Popp R, Kiss L, et al. Cytochrome P450 2C is an EDHF synthase in coronary arteries. Nature. 1999;401: McCarron JG, Halpern W. Potassium dilates rat cerebral arteries by two independent mechanisms. Am J Physiol. 1990;259:H902 H Knot HJ, Zimmermann PA, Nelson MT. Extracellular K -induced hyperpolarization and dilations of rat coronary and cerebral arteries involve inward rectifier K channels. J Physiol. 1996;492: Dong H, Waldron GJ, Cole WC, et al. Roles of calcium-activated and voltage-gated delayed rectifier potassium channels in endotheliumdependent vasorelaxation of the rabbit middle cerebral artery. Br J Pharmacol. 1998;123: Bolz SB, Fisslthaler B, Pieperhoff S, et al. Antisense oligonucleotide against cytochrome P450 2C8 attenuates EDHF-mediated Ca 2 changes and dilation in isolated resistance arteries. FASEB J. 2000;14: Chataigneau T, Feletou M, Duhault J, et al. Epoxyeicosatrienoic acids, potassium channel blockers and endothelium-dependent hyperpolarization in the guinea-pig carotid artery. Br J Pharmacol. 1998;123: Quignard JF, Feletou M, Thollon C, et al. Potassium ion and endothelium-derived hyperpolarizing factor in guinea-pig and porcine coronary arteries. Br J Pharmacol. 1999;127: Ding H, Kubes P, Triggle C. Potassium- and acetylcholine-induced vasorelaxation in mice lacking endothelial nitric oxide. Br J Pharmacol. 2000;129:

7 1708 Circulation March 27, Drummond GR, Selemidis S, Cocks TM. Apamin-sensitive, non-nitric oxide (NO) endothelium-dependent relaxations to bradykinin in the bovine isolated coronary artery: no role for cytochrome P450 and K. Br J Pharmacol. 2000;129: Edwards G, Thollon C, Gardener MJ, et al. Role of gap junctions and EETs in endothelium-dependent hyperpolarization of porcine coronary artery. Br J Pharmacol. 2000;129: Lacy PS, Pilkington G, Hanvesakul R, et al. Evidence against potassium as an endothelium-derived hyperpolarizing factor in rat mesenteric small arteries. Br J Pharmacol. 2000;129: Nakashima M, Mombouli JV, Taylor AA, et al. Endothelium-dependent hyperpolarization caused by bradykinin in human coronary arteries. J Clin Invest. 1993;92: Pascoal IF, Umans JU. Effects of pregnancy on mechanisms of relaxation in human omental microvessels. Hypertension. 1996;28: Wallerstedt SM, Bodelsson M. Endothelium-dependent relaxation by substance P in human isolated omental arteries and veins: relative contribution of prostanoids, nitric oxide and hyperpolarization. Br J Pharmacol. 1997;120: Harasawa LM, Shimokawa H, Nakashima M, et al. Importance of endothelium-derived hyperpolarizing factor in human arteries. J Clin Invest. 1997;100: Buss NH, Simonsen U, Pilegaard HK, et al. Nitric oxide, prostanoid and non-no, non-prostanoid involvement in acetylcholine relaxation of isolated human arteries. Br J Pharmacol. 2000;129: Coats P, Hillier C. Determination of an optimal axial-length tension for the study of isolated resistance arteries on a pressure myograph. Exp Physiol. 1999;84: Adeagbo ASO, Triggle CR. Varying extracellular [K ]: a functional approach to separating EDHF- and EDNO-related mechanisms in perfused rat mesenteric arterial bed. J Cardiol Pharmacol. 1993;21: Petersson J, Zygmunt PM, Brandt L, et al. Substance P-induced relaxation and hyperpolarization in human cerebral arteries. Br J Pharmacol. 1995; 115: Gerber RT, Holemans K, O Brien-Coker I, et al. Cholesterol-independent endothelial dysfunction in virgin and pregnant rats fed a diet high in saturated fats. J Physiol. 1999;517: Harper JF, Brooker G. Femtomole sensitive radioimmunoassay for cyclic AMP and cyclic GMP after 2 0 acetylation by acetic anhydride in aqueous solution. J Cyclic Nucleotide Res. 1975;1: Bradford MM. A rapid and sensitive method for quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem. 1976;7: Nishikawa Y, Stepp DW, Chilian WM. In vivo location and mechanism of EDHF-mediated vasodilation in canine coronary microcirculation. Am J Physiol. 1999;277:H1252 H McCulloch AI, Bottrill FE, Randall MD, et al. Characterization and modulation of EDHF-mediated relaxations in the rat isolated superior mesenteric artery. Br J Pharmacol. 1997;20: Kilpatrick EV, Cocks TM. Evidence for differential roles of nitric oxide (NO) and hyperpolarization in endothelium-dependent relaxation of pig isolated coronary artery. Br J Pharmacol. 1994;112: Garland CJ, Plane F, Kemp BK, et al. Endothelium-dependent hyperpolarization: a role in the control of vascular tone. Trends Pharmacol Sci. 1995;16: Brayden JE. Membrane hyperpolarization is a mechanism of endothelium-dependent cerebral vasodilation. Am J Physiol. 1990;259: H668 H Zygmunt PM, Hogestatt ED. Role of potassium channels in endothelium-dependent relaxation resistant to nitroarginine in rat hepatic artery. Br J Pharmacol. 1996;121: Garcia ML, Galvez A, Calvo MG, et al. Use of toxins to study potassium channels. J Bioenerg Biomembr. 1991;23: Zygmunt PM, Edwards G, Weston AH, et al. Involvement of voltagedependent potassium channels in the EDHF-mediated relaxation of rat hepatic artery. Br J Pharmacol. 1997;121: Garcia ML, Knaus HG, Munujos P, et al. Charybdotoxin and its effects on potassium channels. Am J Physiol. 1995;269:C1 C Hutcheson IR, Chaytor AT, Evans WH, et al. Nitric oxide-independent relaxations to acetylcholine and A23187 involve different routes of heterocellular communication: role of gap junctions and phospholipase A2. Circ Res. 1999;84: Harder DR, Cambell WB, Roman RJ. Role of P-450 enzymes and metabolites of arachidonic acid in the control of vascular tone. J Vasc Res. 1995;32: Imig JDZ, Oritz de Montellano PR, Sui Z, et al. Cytochrome P-450 inhibitors alter afferent arteriolar responses to elevations in pressure. Am J Physiol. 1994;266:H1879 H Miura H, Gutterman DD. Human coronary arteriolar dilation to arachidonic acid depends on cytochrome P450 monooxygenase and Ca 2 -activated K channels. Circulation. 1998;83: Widmann MD, Weintraub NL, Fudge JL, et al. Cytochrome P-450 pathway in acetylcholine-induced canine coronary microvascular vasodilation in vivo. Am J Physiol. 1998;274:H283 H Bakker EN, Sipkema P. Components of acetylcholine-induced dilation in isolated rat arterioles. Am J Physiol. 1997;273:H1848 H Cohen RA, Plane F, Najibi S, et al. Nitric oxide is the mediator for both endothelium-dependent relaxation and hyperpolarization of rabbit carotid artery. Proc Natl Acad Sci U S A. 1997;94: Simonsen U, Wadsworth RM, Buss NH, et al. Simultaneous nitric oxide release and relaxation of the rat superior mesenteric artery. J Physiol. 1999;516: