Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
D. Harris: Department of Pharmacology, Quintiles Ltd, Heriot-Watt University Research Park, Edinburgh, UK.
The endogenous cannabinoid anandamide has recently been identified as a vasorelaxant but the underlying mechanisms are controversial. The vasorelaxant responses to anandamide have now been examined in the rat mesenteric arterial bed. Anandamide caused potent vasorelaxations (pD2= 6.24 ± 0.06; Rmax= 89.4 ± 2.2 %) which were unaffected by inhibition of nitric oxide synthase with NG-nitro-l-arginine methyl ester (l-NAME; 300 μm). The responses were also predominantly endothelium independent and were unaffected by the cannabinoid CB1 receptor antagonist SR141716A (1 μm), although at higher concentrations (3 and 10 μm) SR141716A was inhibitory. Both 1 mm ouabain (pD2= 5.90 ± 0.07; Rmax= 50.4 ± 6.5 %) and 100 μm 18α-glycyrrhetinic acid (pD2= 6.04 ± 0.14; Rmax= 40.9 ± 5.8 %) opposed anandamide-induced vasorelaxation. However, the gap junction inhibitors carbenoxolone (100 μm) and palmitoleic acid (50 μm) did not affect vasorelaxation to anandamide. Relaxation to anandamide was significantly attenuated by both capsaicin pretreatment to deplete the sensory nerves of neurotransmitters (pD2= 5.86 ± 0.18; Rmax= 56.3 ± 5.2 %) and the vanilloid antagonist ruthenium red (10 μm; pD2= 5.64 ± 0.09; Rmax= 33.7 ± 3.9 %). However, these inhibitory effects were prevented by the additional presence of l-NAME, when the relaxation to anandamide was unaffected (pD2= 6.19 ± 0.07; Rmax= 81.9 ± 2.8 %). The inhibitor of neuronal nitric oxide synthase, 7-nitroindazole, also prevented capsaicin from inhibiting the responses to anandamide. The results of this study point to anandamide acting via several mechanisms, which include the involvement of sensory nerves, but only in the presence of nitric oxide.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Recent interest has been focused on the cardiovascular actions of the endogenous cannabinoids (Randall & Kendall, 1998a). In this respect, anandamide is a potent vasodilator (Randall et al. 1996; White & Hiley, 1997; Wagner et al. 1999), but its mechanism of action is unclear. It was originally reported that relaxation to anandamide was antagonized by the cannabinoid receptor antagonist SR141716A in the rat isolated mesentery, thus implicating the involvement of cannabinoid (CB) receptors in this response (Randall et al. 1996). This was confirmed by White & Hiley (1997) in mesenteric arterial segments, but opposed by Plane et al. (1997) and Chataigneau et al. (1998) using the same preparation. In addition, anandamide was reported to induce endothelium-independent vasorelaxations (Randall et al. 1996; White & Hiley, 1997; Zygmunt et al. 2000). However, Zygmunt et al. (1997) and Chataigneau et al. (1998) have shown that anandamide induces endothelium-dependent hyperpolarization, suggesting that anandamide acts via the release of endothelial autacoids. Furthermore, Deutsch et al. (1997) demonstrated that vasorelaxation to anandamide in rat renal arterioles occurs via endothelium-derived nitric oxide (NO). It has also been shown that relaxations to anandamide were sensitive to indomethacin in the rat cerebral vasculature, therefore suggesting that cannabinoids may act via the release of prostanoids (Ellis et al. 1995). However, Pratt et al. (1998) showed that cytochrome P450 inhibitors attenuated relaxation to anandamide, leading to the proposal that anandamide was metabolized to vasoactive arachidonic acid metabolites. Zygmunt et al. (1997) also showed that anandamide acted via inhibition of intracellular calcium mobilization in vascular smooth muscle, while Gebremedhin et al. (1999) provided evidence that anandamide directly blocks vascular smooth muscle calcium channels.
The possibility of cannabinoid receptors mediating the vasorelaxant effects of anandamide is uncertain. However, it has been shown that the hypotensive action of anandamide is absent in mice lacking CB1 receptors (Ledent et al. 1999). The involvement of CB1 receptors has also been implicated following the detection of CB1 receptor messenger RNA in sympathetic nerves, vascular endothelium and smooth muscle (Deutsch et al. 1997; Sugiura et al. 1998; Darker et al. 1998).
Recently, Chaytor et al. (1999) demonstrated that anandamide acts in both an endothelium-dependent and -independent manner in rabbit mesenteric arteries. The endothelium-dependent component was sensitive to inhibition of myoendothelial gap junctions. In this respect, high concentrations of SR141716A inhibited the endothelium-dependent relaxations to anandamide via inhibition of gap junctional communication. From these observations, they concluded that part of the vasorelaxation to anandamide was due to an action on the endothelium which was communicated to the vascular smooth muscle via gap junctions. This is consistent with Wagner et al. (1999) who showed, in rat mesenteric vessels, that a small endothelium-dependent component of vasorelaxation to anandamide was SR141716A sensitive but not mediated by CB1 receptors. This led them to propose that there is a novel cannabinoid receptor present on the endothelium. Moreover, a recent study by Jarai et al. (1999) reported that abnormal cannabidiol caused SR141716A-sensitive vasodilatation, which is abolished by cannabidiol. It was therefore proposed that cannabidiol is an antagonist of a novel endothelial cannabinoid receptor.
In 1999 Zygmunt et al. proposed that anandamide induces vasodilatation by activating vanilloid receptors on perivascular sensory nerves, causing release of calcitonin gene-related peptide (CGRP). They demonstrated that the vasodilator effects of anandamide were sensitive to pretreatment with capsaicin (to deplete sensory nerves of neurotransmitters), the vanilloid receptor antagonist capsazepine and also the CGRP receptor antagonist, CGRP (8–37), but not the CB1 receptor antagonist SR141716A. These effects were exclusive to anandamide and were not mimicked by other endogenous or synthetic cannabinoid agonists. Radioimmunoassay studies showed that anandamide produced an increase in tissue CGRP levels which was sensitive to both capsaicin and capsazepine. Similar observations have been made with the analogue of anandamide, methanandamide (Ralevic et al. 2000). More recently, Zygmunt et al. (2000), demonstrated, in the guinea-pig basilar artery, that relaxation to anandamide did not affect membrane potential but was sensitive to capsaicin pretreatment and also to resiniferatoxin. The Na+,K+-ATPase and gap junction inhibitor ouabain also inhibited relaxations to both anandamide and capsaicin, whereas 18α-glycyrrhetinic acid had no effect. By contrast, in rat or mouse lumbar spinal chord, anandamide does not induce a typical receptor-mediated capsaicin-like response (Richardson et al. 1998a,b).
In this study, we have attempted to characterize the mechanisms underlying anandamide-induced vasorelaxation in the rat mesenteric arterial bed. Specifically we have investigated the role of sensory nerves in relaxation to anandamide by using capsaicin (both acutely and neonatal pretreatment) to deplete the sensory nerves of neurotransmitters, and ruthenium red as an antagonist of sensory nerve function. A comparison has been made with the relaxant responses to capsaicin. In view of the proposed involvement of NO, the endothelium, EDHF, gap junctions and CB receptors in relaxation to anandamide, we have investigated their relative contributions.
Rat isolated, perfused superior mesenteric arterial bed
Male Wistar rats (250–350 g) were anaesthetized with sodium pentobarbitone (60 mg kg−1, i.p.) and killed by exsanguination. Following a mid-line incision, the superior mesenteric artery was cannulated. The arterial vasculature was dissected away from the intestines and placed in a jacketed organ bath as previously described (Randall et al. 1997) and perfused at 5 ml min−1 with oxygenated (95 % O2-5 % CO2) Krebs-Henseleit buffer (composition, mm: NaCl 118, KCl 4.7, MgSO4 1.2, KH2PO4 1.2, NaHCO3 25, CaCl2 2, d-glucose 10; 37 °C) containing the cyclooxygenase inhibitor indomethacin (10 μm). In some cases, NG-nitro-l-arginine methyl ester (l-NAME; 300μm) was added to the buffer to inhibit synthesis of nitric oxide (NO). The perfusion pressure was continuously monitored by means of a pressure transducer coupled to a MacLab 4e recording system (ADInstruments, Castle Hill, New South Wales, Australia). All procedures were carried out in accordance with regulations of the United Kingdom's Animals (Scientific Procedures) Act 1986.
Following 30 min equilibration, methoxamine (1–2 μm) was added to the buffer to increase perfusion pressure by ca 100 mmHg. Once stable tone had been established, the vasorelaxant effects of anandamide (10 nm to 10 μm) or the vanilloid agonist capsaicin (1 nm to 10 μm) were assessed as cumulative concentration- response curves. Both anandamide and capsaicin were added directly to the Krebs-Henseleit buffer to achieve the desired concentrations. In preparations receiving 1 mm ouabain (an inhibitor of Na+,K+-ATPases and gap junctions; Harris et al. 2000), l-NAME (a non-selective inhibitor of nitric oxide synthase; 300μm), 7-nitroindazole (an inhibitor of neuronal nitric oxide synthase; 100 μm; Moore et al. 1993), LY320135 (CB1 receptor antagonist; 10 μm; Felder et al. 1998) or SR141716A (a CB1 receptor antagonist, which also blocks gap junctions; Chaytor et al. 1999) the drugs were added to the buffer to achieve the desired concentration and allowed to equilibrate for 30 min before addition of vasorelaxants. The ‘pure’ gap junction inhibitors carbenoxolone (100 μm; Davidson et al. 1986) and palmitoleic acid (50 μm; Harris et al. 2000), and 18α-glycyrrhetinic acid (an inhibitor of Na+,K+-ATPases and gap junctions; 100 μm; Terasawa et al. 1992; Chaytor et al. 1999), clotrimazole (a cytochrome P450 inhibitor; which also inhibits gap junctional communication; 10 μm; Harris et al. 2000) and ruthenium red (vanilloid antagonist; 10 μm) were allowed to incubate for 1 h. The vanilloid agonist capsaicin (10 μm) was perfused for 1 h (or 2 h), with a 20 min washout, to allow depletion of neurotransmitters from sensory nerves (Zygmunt et al. 1999) before induction of relaxant responses to anandamide or capsaicin. Some arterial beds were bathed in a capsaicin-containing Krebs-Henseleit buffer for 1 h. Any loss of established tone following addition of these agents was restored by further addition of methoxamine. In endothelium-denuded preparations, the endothelium was removed by 3 min perfusion of distilled water (Wagner et al. 1999). Preparations were considered endothelium denuded when relaxation to 5.48 μmol carbachol was less than 10 %.
A group of rats was neonatally treated with capsaicin (50 mg kg−1, s.c.) under ice anaesthesia (Ralevic et al. 2000). These experiments were carried out in accordance with regulations of the United Kingdom's Animals (Scientific Procedures) Act 1986 as specified in project licence number 40/1955. After 6 weeks, the rats were anaesthetized with sodium pentobarbitone (60 mg kg−1, i.p.) and killed by exsanguination, and the mesenteric arterial bed was prepared as described above. In order to assess the effects of neonatal capsaicin treatment on the destruction of the sensory nerves, vasorelaxation to a bolus dose of capsaicin (500 nmol) was assessed in each mesenteric preparation.
In order to control for non-selective actions of the inhibitors, their effects were also investigated against the vasorelaxant responses to levcromakalim, an endothelium-independent agent which acts via ATP-sensitive potassium channels and hyperpolarization.
Data and statistical analysis
The dose-response curves were fitted to a logistic equation (Randall et al. 1997), and the -log ED50 (pD2) and maximal relaxation (Rmax) values obtained were compared by analysis of variance with Bonferroni's post hoc test.
All drugs were supplied by Sigma Chemical Co. (Poole, UK) except where stated. Carbenoxolone, carbachol, ouabain, and l-NAME were dissolved in saline; palmitoleic acid was dissolved in saline by sonication; indomethacin, clotrimazole and capsaicin were dissolved in ethanol. The capsaicin that was injected into neonatal rats was dissolved in a Tween, ethanol and saline mixture (1:1:18). Ruthenium red, 7-nitroindazole and 18α-glycyrrhetinic acid (18α-GA) were obtained from Sigma and dissolved in DMSO. Anandamide was synthesized from arachidonoyl chloride and ethanolamine and dissolved in an inert oil-water emulsion by Dr P. DeBank, School of Pharmaceutical Sciences, University of Nottingham. Levcromakalim (dissolved in ethanol) was obtained from SmithKline Beecham (Harlow, UK). SR141716A (N-piperidino-5-(4-chlorophenyl)-1-(2-dichlorophenyl)-4-methyl-3-pyrazole-carbaoxamide) was obtained from Tocris Cookson (Bristol, UK) and dissolved in ethanol. LY320135 ([6-methoxy-2-(4-methoxyphenyl)benzo[b]thien-3-yl][4-cyanophenyl]methanone) was a kind gift of Dr K. Fahey (Eli Lilly, Indianapolis, IN, USA) and was dissolved in DMSO. The final concentration of DMSO in the buffer was 0.01 % (v/v), a concentration which does not influence vascular responses (Harris et al. 2000).
Characterization of vasorelaxant responses to anandamide
Anandamide (10 nm to 10 μm) caused concentration-dependent vasorelaxation (pD2= 6.22 ± 0.08; Rmax= 93.7 ± 3.4 %; n= 6). Addition of indomethacin (10 μm) did not affect relaxation to anandamide (pD2= 6.24 ± 0.06; Rmax= 89.4 ± 2.2 %; n= 8; Fig. 1A), nor did the additional presence of l-NAME (300 μm; pD2= 6.23 ± 0.08; Rmax= 91.1 ± 1.5 %; n= 6; Fig. 1A). Clotrimazole (10 μm) caused a rightward shift of the anandamide concentration- response curve (pD2= 5.55 ± 0.05, P < 0.001; Rmax= 76.1 ± 4.5 %, P < 0.05; n= 6; Fig. 1A). Removal of the endothelium had a slight inhibitory effect on vasorelaxation to anandamide (pD2= 5.89 ± 0.04, P < 0.01; Rmax= 82.6 ± 2.5 %, P < 0.05; n= 5; Fig. 1A).
In some additional experiments designed to examine the site of action of anandamide, the bed was bathed in buffer and preconstricted with methoxamine and anandamide was applied to the bath (i.e. extraluminally). Under these conditions abluminal application of anandamide did not induce significant vasorelaxation of the arterial bed (at 10 μm the relaxation was 5.2 ± 1.3 %; n= 3).
Effects of EDHF inhibitors and gap junction inhibitors on vasorelaxation to anandamide
Addition of the gap junction and Na+,K+-ATPase inhibitor ouabain (1 mm) or 18α-GA (100 μm) significantly inhibited anandamide-induced vasorelaxation (Fig. 1B). The gap junction inhibitors carbenoxolone (100 μm; Fig. 1B) and palmitoleic acid (50 μm; Fig. 1B) had no effect on vasorelaxation to anandamide. In endothelium-denuded preparations, 18α-GA (100 μm) also inhibited anandamide-induced vasorelaxation and these responses were no different from those in the presence of 18α-GA in endothelium-intact preparations (Fig. 1B). Vasorelaxation to levcromakalim (pD2= 8.91 ± 0.10; Rmax= 96.5 ± 4.5 %; n= 4) was unaffected by either 18α-GA (pD2= 9.16 ± 0.03; Rmax= 95.2 ± 1.4 %; n= 4) or palmitoleic acid (pD2= 8.96 ± 0.06; Rmax= 89.2 ± 2.7 %; n= 6).
Effects of cannabinoid receptor antagonists on vasorelaxation to anandamide
In the presence of indomethacin (10 μm) and l-NAME (300 μm) relaxation to anandamide was described by pD2= 6.05 ± 0.04 and Rmax= 88.5 ± 2.1 % (n= 8). The cannabinoid receptor antagonist SR141716A (1 μm) did not affect these relaxations (pD2= 6.16 ± 0.03; Rmax= 88.7 ± 2.6 %; n= 3; Fig. 1C). However, at 3 μm SR141716A caused a significant (P < 0.05) rightward shift in the concentration-response curve for relaxation to anandamide (pD2= 5.81 ± 0.14; Rmax= 84.8 ± 8.2 %; n= 3; Fig. 1C). SR141716A (10 μm) further attenuated these relaxations (pD2= 5.72 ± 0.16, P < 0.05; Rmax= 77.3 ± 6.9 %, P < 0.05; n= 3; Fig. 1C). The CB1 receptor antagonist LY320135 (10 μm) did not affect the concentration-response curve to anandamide (pD2= 6.00 ± 0.10; Rmax= 81.6 ± 3.6 %; n= 6; Fig. 1C).
Vasorelaxation to levcromakalim was attenuated by 10 μm SR141716A (pD2= 5.60 ± 0.06, P < 0.01; Rmax= 43.7 ± 1.9 %, P < 0.001; n= 4). In a previous study LY320135 (10 μm) did not affect responses to levcromakalim (Harris et al. 1999).
Effects of capsaicin treatment and ruthenium red on vasorelaxation to anandamide in the presence of indomethacin
In the presence of 10 μm indomethacin, anandamide caused concentration-dependent vasorelaxations (Fig. 2A). Following 1 h capsaicin pretreatment (10 μm), relaxations to anandamide were significantly inhibited (Fig. 2A), with a 2-fold rightwards shift in the concentration-response curve. Bathing the entire arterial bed for 1 h in a capsaicin-containing buffer, whilst perfusing with equivalent buffer, had no additional inhibitory effects (pD2= 5.58 ± 0.06; Rmax= 49.9 ± 5.1 %; n= 6).
In the presence of ruthenium red (10 μm), relaxations to anandamide were also significantly attenuated (Fig. 2A).
Effects of capsaicin pretreatment and ruthenium red on vasorelaxation to anandamide in the presence of nitric oxide synthase inhibitors
In the presence of 10 μm indomethacin and 300 μm l-NAME, anandamide caused concentration-dependent vasorelaxations (Fig. 2B). Addition of capsaicin (10 μm) caused a significant decrease of established tone (95.3 ± 3.1 %; n= 36). Following capsaicin pretreatment (1 h with 20 min washout), vasorelaxation to anandamide was unaffected in the presence of indomethacin and l-NAME (Fig. 2B). Capsaicin (10 μm) treatment for 2 h slightly reduced the maximum relaxation to anandamide (pD2= 6.26 ± 0.05; Rmax= 78.7 ± 4.5 %, P < 0.05; n= 4). Capsaicin pretreatment (1 h with 20 min washout) with the entire arterial bed also bathed in a 10 μm capsaicin-containing buffer, whilst also being perfused, again had no effect on anandamide-induced relaxation (pD2= 6.14 ± 0.07; Rmax= 82.9 ± 2.0 %; n= 7) nor did capsaicin pretreatment following removal of the endothelium (pD2= 6.29 ± 0.11, P < 0.05; Rmax= 85.8 ± 2.0 %; n= 6). However, the addition of 10 μm clotrimazole following capsaicin pretreatment caused a marked attenuation of anandamide relaxation (pD2= 5.44 ± 0.06, P < 0.001; Rmax= 43.1 ± 8.6 %, P < 0.001; n= 6). Following pretreatment with capsaicin, ouabain opposed the relaxations to anandamide (pD2= 5.76 ± 0.01; Rmax= 55.3 ± 6.2 %; n= 4) to the same extent as its effects in untreated preparations.
In the presence of l-NAME and indomethacin the addition of the vanilloid antagonist ruthenium red (10 μm) did not affect vasorelaxation to anandamide (Fig. 2B).
The neuronal NOS inhibitor 7-nitroindazole (100 μm) caused a rightward shift of the concentration-response curve to anandamide (pD2= 5.92 ± 0.06, P < 0.01; Rmax= 81.6 ± 4.6 %; n= 5; Fig. 2C). Pretreatment with capsaicin had no additional inhibitory effects on the responses to anandamide (pD2= 5.88 ± 0.14, P < 0.05; Rmax= 80.7 ± 1.6 %, P < 0.01; n= 8; Fig. 2C).
Relaxation to levcromakalim (pD2= 8.45 ± 0.07; Rmax= 101 ± 3 %; n= 5) was enhanced by capsaicin pretreatment (pD2= 9.10 ± 0.03, P < 0.001; Rmax= 102 ± 2; n= 3) but unaffected by the presence of ruthenium red (pD2= 8.60 ± 0.05; Rmax= 98.9 ± 1.5 %; n= 5). 7-Nitroindazole (100 μm) modestly enhanced the relaxations to levcromakalim, with the pD2 value being increased from 8.45 ± 0.07 to 8.79 ± 0.09 (n= 5) and respective Rmax values of 101 ± 3 % and 99 ± 1 %.
Effects of capsaicin pretreatment, ruthenium red and gap junction inhibitors on vasorelaxation to capsaicin
In the presence of 10 μm indomethacin, capsaicin (1 nm to 10 μm) caused concentration-dependent vasorelaxations (pD2= 7.64 ± 0.18; Rmax= 85.4 ± 2.8 %; n= 6; Fig. 3). Following capsaicin (10 μm) pretreatment, relaxations to capsaicin were markedly attenuated (pD2= 5.41 ± 0.06, P < 0.001; Rmax= 59.4 ± 12.2 %, P < 0.05; n= 4), with a 112-fold rightward shift in the concentration-response curve (Fig. 3). Ruthenium red (10 μm) also caused a rightward shift in the concentration-response curve for capsaicin (pD2= 6.26 ± 0.18, P < 0.001; Rmax= 54.6 ± 6.8 %, P < 0.01; n= 4)
The presence of 300 μm l-NAME did not affect vasorelaxation to capsaicin (pD2= 7.45 ± 0.19; Rmax= 88.3 ± 3.8 %; n= 6; Fig. 3). Following capsaicin pretreatment, relaxation to capsaicin was also significantly reduced (Fig. 3). In the presence of 300 μm l-NAME, ruthenium red (10 μm) also significantly attenuated relaxation to capsaicin (Fig. 3).
In the presence of 10 μm indomethacin alone, 7-nitroindazole (100 μm) significantly attenuated relaxation to capsaicin (pD2= 6.73 ± 0.22, P < 0.001; Rmax= 85.9 ± 4.3 %; n= 6).
The gap junction and Na+,K+-ATPase inhibitors 18α-GA (100 μm) and ouabain (1 mm) also significantly (P < 0.01) attenuated capsaicin-induced relaxation (with respective values of Rmax= 64.7 ± 6.4, pD2= 6.68 ± 0.61, n= 5; Rmax= 35.5 ± 7.4, pD2= 6.90 ± 0.36, n= 5) whereas clotrimazole did not significantly affect the responses (pD2= 7.20 ± 0.22; Rmax= 71.4 ± 7.7 %; n= 5).
Effects of neonatal capsaicin treatment on vasorelaxation to anandamide in the absence and presence of l-NAME
In view of the limited effects of capsaicin on the responses to anandamide, which were negated by the removal of endogenous NO, it was considered important to carry out further experiments in which rats were treated as neonates with capsaicin to deplete their sensory nerves. In these experiments, neonatal capsaicin pretreatment had no effect on vasorelaxation to anandamide in the absence or presence of 300 μm l-NAME (Fig. 4; n= 4). In the presence of l-NAME, relaxation to anandamide was significantly enhanced in capsaicin-treated rats only (Rmax= 84.5 ± 9.5 % in mesenteric arterial beds from capsaicin-treated rats and 113 ± 9 % in l-NAME perfused mesenteries from capsaicin-treated rats (P < 0.05; Fig. 4)).
Effects of neonatal capsaicin treatment on vasorelaxation to capsaicin
Neonatal capsaicin pretreatment significantly (P < 0.01) reduced the maximum relaxation to capsaicin from 65.3 ± 9.0 % to 13.8 ± 7.6 % (n= 4). In the presence of l-NAME, the maximum relaxation to capsaicin was 31.6 ± 12.8 % (n= 4). In preparations from control rats, the presence of 300 μm l-NAME also enhanced the vasorelaxation to capsaicin (from 65.3 ± 9.0 % to 83.5 ± 8.7 %).
The principal aim of the present study was to characterize the mechanisms underlying anandamide-induced vasorelaxation. The results point to anandamide acting in a predominantly endothelium-independent manner, via several mechanisms, including the partial involvement of sensory nerves, but only in the presence of NO, and also a ouabain-sensitive mechanism.
The involvement of cannabinoid receptors in the vasorelaxation to anandamide was examined using both SR141716A and LY320135. In these experiments SR141716A only inhibited relaxations at the higher concentrations used and LY320135 had no effect. The lack of effect of SR141716A at 1 μm would probably exclude anandamide acting via CB1 receptors as previously reported by ourselves and others (Randall et al. 1996; White & Hiley, 1997). However, there was modest inhibition at higher concentrations of the antagonist and these findings might suggest the participation of a novel CB receptor with low affinity of SR141716A or that only a very small component of the responses to anandamide is mediated via CB1 receptors (Wagner et al. 1999; Chaytor et al. 1999; Jaria et al. 1999). Alternatively, the inhibitory effects of SR141716A at high concentrations might reflect actions of this agent at additional, non-cannabinoid receptor sites such as myoendothelial gap junctions (Chaytor et al. 1999) or as shown here against potassium channels, as indicated by the reduced responses to levcromakalim and also reported by White & Hiley (1998). SR141716A, at low micromolar concentrations, also opposes the action of vanilloids (De Petrocellis et al. 2001).
In 1999 Chaytor et al. showed that relatively high concentrations (10 μm) of the CB1 receptor antagonist SR141716A inhibited relaxation to anandamide via inhibition of gap junctional communication. In view of this finding we have investigated the effects of the gap junction inhibitors 18α-GA (Davidson et al. 1986), ouabain (Harris et al. 2000), carbenoxolone (Davidson et al. 1986) and palmitoleic acid (Harris et al. 2000) against anandamide-induced vasorelaxation in the presence of indomethacin and l-NAME. 18α-GA and ouabain significantly inhibited vasorelaxation to anandamide, whereas carbenoxolone and palmitoleic acid had no effect. In view of evidence that 18α-GA and ouabain also inhibit Na+,K+-ATPases at the concentrations used (Terasawa et al. 1992), whereas palmitoleic acid and carbenoxolone do not, it would seem that the responses are sensitive to inhibition of the sodium pump. This is partially consistent with the recent finding that ouabain, but not 18α-GA, blocked the responses to anandamide in the guinea-pig basilar artery (Zygmunt et al. 2000). One obvious possibility is that anandamide is acting via activation of the sodium pump, which would lead to hyperpolarization and relaxation. However, inhibition of the sodium pump has widespread effects on a range of neuronal and smooth muscle activities within the vascular bed. Hence the actions of ouabain and 18α-GA reported here might reflect non-specific oppositions to vasorelaxation to anandamide or functional antagonism through tissue depolarization. In this respect 1 mm ouabain substantially impairs both sensory nerve-mediated vasorelaxation and vasorelaxant responses to exogenous CGRP (V. Ralevic, personal communication).
In view of the proposal of Zygmunt et al. (1999) that anandamide acted as a vanilloid agonist, stimulating the releasing of CGRP from sensory nerves, we attempted to replicate their work in the entire rat mesenteric arterial bed. Pretreatment with capsaicin to deplete the sensory nerves of neuropeptides caused a substantial reduction in the responses to capsaicin, thus confirming the effectiveness of this treatment. However, following this treatment vasorelaxation to anandamide was affected far less, with a only 2-fold rightward shift in the concentration- response curve for anandamide compared to the 112-fold reduction in potency seen with capsaicin. Similarly, the vanilloid antagonist ruthenium red also opposed relaxation to capsaicin and anandamide, but once again the effects against capsaicin were substantially greater. Taken together these findings would suggest that anandamide acts, to a limited extent, via activation of sensory nerves but that this does not explain the full actions of anandamide, and so anandamide should not solely be regarded as an endogenous capsaicin-like substance. Indeed, White et al. (2001) have now also reported that vasorelaxation to anandamide in the rat coronary artery is unaffected by either capsaicin pretreatment or the vanilloid antagonist capsazepine. One possibility for the apparent differences between anandamide and capsaicin could be that capsaicin acts purely as a vanilloid agonist on sensory nerves, whereas anandamide acts in a complex manner via both stimulatory vanilloid receptors and inhibitory cannabinoid receptors on these nerves (Richardson et al. 1998a,b). Our findings are in contrast to those found with the stable analogue of anandamide, methanandamide, which is far more sensitive to both capsaicin treatment and ruthenium red (Ralevic et al. 2000). One possibility is that anandamide and methanandamide may exhibit different pharmacological actions, with the latter acting predominantly as a vanilloid agonist.
In additional experiments designed to investigate the putative involvement of sensory nerves, rats were neonatally treated with capsaicin to chronically deplete the sensory nerves. The efficacy of this treatment was confirmed by reduced relaxations to capsaicin but once again the responses to anandamide were unaffected, in both the absence and presence of l-NAME. This would suggest that, even if sensory nerves mediated relaxation to endocannabinoids, the effects of chronic chemical denervation may be fully compensated by upregulation of other vasorelaxant mechanisms.
In the context of sensory nerves, depolarization by ouabain would be expected to cause the release of transmitters, with potential for depletion. This could explain the inhibitory actions of ouabain and 18α-GA against responses to capsaicin. However, in the presence of l-NAME, ouabain opposed responses to anandamide but pretreatment with capsaicin did not, indicating that this explanation cannot account for the sensitivity of the responses to anandamide.
In Zygmunt et al.'s (1999) study the experiments with anandamide were carried out in the presence of l-NAME and indomethacin. When we repeated our experiments under these conditions we found that, following capsaicin pre-treatment, although responses to capsaicin were inhibited (thus confirming the effectiveness of the treatment), the responses to anandamide were not. This was in contrast to the findings of Zygmunt et al. (1999), and it is possible that the lack of effect of the capsaicin treatment on responses to anandamide (but not capsaicin) might reflect methodological differences. In our experiments, capsaicin was perfused luminally, unlike the isolated segments bathed fully in capsaicin-containing buffer used by Zygmunt et al. Thus lumenal perfusion may have limited access of capsaicin to sensory nerves. However, when the perfused mesenteric arterial bed was additionally bathed in a Krebs-Henseleit buffer containing capsaicin (1 h; 20 min washout) there were still no inhibitory effects on anandamide-induced relaxation. However, relaxation to capsaicin was substantially reduced following capsaicin pretreatment. To investigate the possibility that sensory nerves were not fully depleted of neurotransmitters, capsaicin pretreatment was also performed for 2 h, and once again relaxation to anandamide was largely unaffected. In view of the difference in site of application, we also applied anandamide extraluminally to non-capsaicin-treated preparations, where it would have had greater access to sensory nerves and not be constrained by any barrier effects of the endothelium. Under these conditions anandamide was without vascular effects, indicating that it acts luminally.
One of the intriguing findings from the present study was that relaxations to anandamide were only sensitive to capsaicin pretreatment, albeit to a limited extent, in the presence of a functional NO system. These findings may suggest that NO has a modulatory role in sensory nerve function. It has been previously reported that CGRP release is modulated by NO in the rabbit coronary artery (Mitchell et al. 1997). We therefore examined the responses to anandamide in the presence of the neuronal NOS inhibitor 7-nitroindazole (Moore et al. 1993). 7-Nitroindazole caused a rightward shift of the anandamide concentration-response curve suggesting that neuronal NO may be involved in anandamide-induced vasorelaxation. Moreover, following capsaicin treatment, 7-nitroindazole had no additional effects perhaps suggesting that capsaicin was interfering with nitrergic nerves. Similar observations have also been made in the rat skin microvasculature (Merhi et al. 1998). The possibility of NO being involved in the sensory neurotransmission is suggested by the ability of 7-nitroindazole to inhibit capsaicin-induced relaxation. This suggests that neuronal NO has a role in sensory nerve-mediated vasorelaxation, consistent with the findings that NO is involved in the endotoxin-induced release of CGRP in the rat mesenteric arterial bed (Wang et al. 1996). Moreover, Mitchell et al. (1997) reported that, in rabbit isolated coronary circulation, relaxations to capsaicin and CGRP were inhibited by the CGRP antagonist CGRP (8–37), whereas l-NAME inhibited the effects of capsaicin and substance P but not CGRP. This suggests that CGRP release following capsaicin-induced sensory nerve stimulation is modulated by NO. The role of NO in sensory nerve-dependent vasorelaxation clearly warrants further study. However, what is not clear is why 7-nitroindazole but not l-NAME opposed both capsaicin- and anandamide-induced relaxations, as l-NAME acting as a non-selective inhibitor should also impair nitrergic responses. It is possible that l-NAME, by inhibiting other NO synthases, such as the endothelial form, upregulates alternative vasorelaxant mechanisms, thus negating any effect on neuronal NO synthase.
The apparent dependence of sensory nerves on endogenous NO requires further explanation. In this context, De Petrocellis et al. (2001) have recently reported that anandamide acts on sensory nerves from an intracellular site and so is dependent on cellular uptake via a cannabinoid transporter. The cannabinoid transporter is activated by NO and so it is entirely possible that basal NO acts to facilitate carrier-mediated uptake of anandamide; accordingly in the absence of endogenous basal NO the uptake is impaired and anandamide can no longer act as an intracellular vanilloid agonist. Furthermore, the actions of intracellular fatty acid amide hydrolase (FAAH), which metabolizes anandamide, terminate this action. Hence, methanandamide, which is metabolically more stable, might be more effective as an intracellular vanilloid agonist and this might explain the greater effects of capsaicin against responses to methanandamide (Ralevic et al. 2000), compared to those to anandamide.
Another possibility for the additional mechanisms of actions of anandamide is that it acts via the release of EDHF. Obviously the very small endothelium-dependent component to the anandamide responses suggests that if this is the case then it is only likely to make a marginal contribution. In the present study the cytochrome P450 inhibitor, gap junction inhibitor (Harris et al. 2000), and potassium channel inhibitor (Zygmunt et al. 1996) clotrimazole only had modest inhibitory effects on vasorelaxation to anandamide. Clearly, the lack of the effect of the gap junction inhibitors reported above rules out a role for EDHF but the partial sensitivity to clotrimazole might be explained by anandamide acting in part via potassium channel activation (Randall et al. 1997; Randall & Kendall, 1998b). Once again capsaicin and anandamide exhibited different pharmacological profiles, with capsaicin acting in a clotrimazole-insensitive manner. It is interesting to note that the combination of capsaicin pretreatment and clotrimazole resulted in a far greater attenuation of relaxant responses to anandamide than these compounds alone. This provides direct evidence that at least two mechanisms of action exist for anandamide-induced vasorelaxation.
In summary, anandamide-induced vasorelaxation appears to occur due to activation of a number of mechanisms in the rat mesenteric arterial bed. Gap junctions do not appear to have a role in anandamide-induced vasorelaxation in this tissue as only 18α-GA and ouabain, which also block Na+,K+-ATPases, had any inhibitory effects. The fact that anandamide relaxations were sensitive to 18α-GA and ouabain suggests that the sodium pump might be a specific target or reflect functional antagonism of the responses to anandamide. The relaxations to anandamide were only modestly sensitive to capsaicin pretreatment or ruthenium red in the absence of the NOS inhibitor l-NAME. This is in contrast with the findings of Zygmunt et al. (1999) and suggests that a functional NO system is required for sensory nerves to have a role in anandamide-induced vasorelaxation, perhaps by facilitating its cellular uptake. The relaxation to anandamide was partly sensitive to clotrimazole, consistent with the partial involvement of potassium channels. Differences in pharmacology between capsaicin and anandamide suggest that anandamide does not act solely as a vanilloid agonist in this preparation. It thus appears likely that anandamide induces vasorelaxation via multiple mechanisms, which include sensory nerves but only in the presence of NO.
D.H. held an MRC studentship. This work was funded by the British Heart Foundation. We should like to thank Dr Vera Ralevic for her comments on both the study and the manuscript.