Abstract

Cannabinoids, including the endogenous ligandarachidonyl ethanolamide (anandamide), elicit not only neurobehavioralbut also cardiovascular effects. Two cannabinoid receptors, CB1 andCB2, have been cloned, and studies with the selective CB1 receptorantagonist SR141716A have implicated peripherally located CB1 receptorsin the hypotensive action of cannabinoids. In rat mesenteric arteries,anandamide-induced vasodilation is inhibited by SR141716A, but otherpotent CB1 receptor agonists, such as HU-210, do not causevasodilation, which implicates an as-yet-unidentified receptor in thiseffect. Here we show that “abnormal cannabidiol” (Abn-cbd) is aneurobehaviorally inactive cannabinoid that does not bind to CB1receptors, yet causes SR141716A-sensitive hypotension and mesentericvasodilation in wild-type mice and in mice lacking CB1 receptors orboth CB1 and CB2 receptors. Hypotension by Abn-cbd is also inhibited bycannabidiol (20 μg/g), which does not influence anandamide- orHU-210-induced hypotension. In the rat mesenteric arterial bed,Abn-cbd-induced vasodilation is unaffected by blockade of endothelialNO synthase, cyclooxygenase, or capsaicin receptors, but it isabolished by endothelial denudation. Mesenteric vasodilation byAbn-cbd, but not by acetylcholine, sodium nitroprusside, or capsaicine,is blocked by SR141716A (1 μM) or by cannabidiol (10 μM).Abn-cbd-induced vasodilation is also blocked in the presence ofcharybdotoxin (100 nM) plus apamin (100 nM), a combination ofK⁺-channel toxins reported to block the release of anendothelium-derived hyperpolarizing factor (EDHF). These findingssuggest that Abn-cbd and cannabidiol are a selective agonist andantagonist, respectively, of an as-yet-unidentified endothelialreceptor for anandamide, activation of which elicits NO-independentmesenteric vasodilation, possibly by means of the release of EDHF.

Materials and Methods

Results

Fig.1 summarizes the changes in mean arterialpressure in pentobarbital-anesthetized CB1 receptor knockout mice(CB1^−/−) and their control littermates(CB1^+/+) in response to the potent syntheticcannabinoid HU-210 (A) and anandamide (B). Thedoses used were submaximal for hypotension, as deduced fromdose–response studies in wild-type ICR mice (not shown). InCB1^+/+ mice, both compounds elicited prolongedhypotension, blocked by pretreatment with SR141716A, whereas theseeffects were completely absent in CB1^−/− mice.Anesthetized CB1^−/− andCB1^+/+ mice had similar basal blood pressure(74 ± 2 vs. 75 ± 2 mmHg,n = 41 vs. 45),which is compatible with earlier findings of a lack of effect on thisparameter by SR141716A (10). Both of these findings suggest that CB1receptors involved in the control of blood pressure are not tonicallyactive.

Because recent findings suggest that anandamide-inducedmesenteric vasodilation is mediated by receptors other than CB1cannabinoid receptors (16), we tested whether such an effect is presentin CB1^−/− mice. Intra-arterial injection ofanandamide into the mesenteric vasculature caused dose-dependent,prolonged but reversible vasodilation, which was somewhat greater inCB1^−/− than in CB1^+/+mice and was inhibited in the presence of 1 or 5 μM SR141716A in theperfusion medium (Fig.2A). At the 5 μMconcentration, SR141716A did not inhibit the vasodilator response to 3μg of sodium nitroprusside (54% ± 4%, vs. 49% ± 3%,n = 10). Anandamide in tissues is rapidly degraded intoarachidonic acid and ethanolamine by an amidohydrolase (24). However,its mesenteric vasodilator effect could not be caused by a degradationproduct, because the metabolically stable (R)-methanandamide(25) also caused vasodilation (44% ± 5% at 10 μg,n = 5, and 50% ± 5% at 100 μg,n =4), which was inhibited by 5 μM SR141716A (10% ± 5% at 10 μg,n = 4,P < 0.005, and 23% ± 3% at100 μg,n = 4,P < 0.005). Thefinding that (R)-methanandamide caused similar mesentericvasodilation inCB1^−/−/CB2^−/−double-knockout mice (10 μg: 34% ± 4%,n = 3; 100μg: 48% ± 2%,n = 3) excluded the role of CB2receptors in this effect. These findings lend further support to theexistence of one or more as-yet-undefined receptors in the mesentericvasculature, for which anandamide and (R)-methanandamide areagonists and SR141716A is an antagonist. Although the inhibitorypotency of SR141716A appears somewhat lower than its potency for CB1receptors (26), significant inhibition was observed at a concentrationof 1 μM, which does not cause the direct inhibition ofCa²⁺ or K⁺ channelsobserved at higher concentrations of SR141716A (27).

Abn-cbd (see Fig.3D) is a synthetic cannabinoidanalog that was reported to lack cannabinoid-like behavioral effects intwo paradigms but to cause profound hypotension in dogs (28). Becausethese findings suggest that Abn-cbd may be a selective agonist at theputative vascular anandamide receptor, we synthesized Abn-cbd andtested its neurobehavioral and cardiovascular effects and its CB1receptor binding activity. Abn-cbd given i.v. to ICR mice in doses of15–60 μg/g was inactive in a tetrad of tests considered diagnosticfor cannabinoid action (22), whereas 3 μg/g THC, which causes thesame degree of hypotension in CB1^+/+ mice as 20μg/g Abn-cbd, was strongly active (Table1). Inligand binding assays using rat brain plasma membranes, Abn-cbd (0.03to 100 μM) did not displace [³H]CP-55,940, apotent cannabinoid agonist (displacement at 100 μM: 6% ± 9% ofspecific binding,n = 4). However, Abn-cbd eliciteddose-dependent hypotension (5–30 μg/g i.v.,ED₅₀: 7.6 ± 0.6 μg/g) without affectingheart rate in anesthetized CB1^+/+ andCB1^−/− mice, and the hypotension was blocked bypretreatment with 3 μg/g SR141716A in both groups (Fig.1C). Abn-cbd (20 μg/g) also elicited hypotension inCB1^−/−/CB2^−/− mice(−24 ± 5 mmHg,n = 4), which was similar inmagnitude and duration to that seen in CB1^−/−mice. In the buffer-perfused mesenteric vascular bed ofCB1^−/− mice, Abn-cbd elicited pronounced andlong lasting vasodilation, which was significantly inhibited by 1–5μM SR141716A (Fig.2B). In mesenteric preparationsfrom CB1^−/−/CB2^−/−mice, Abn-cbd (200 μg) caused vasodilation (75% ± 4%,n = 4) similar to that in preparations fromCB1^−/− mice. These findings indicate thatAbn-cbd acts at a vascular site other than CB1 or CB2 receptors.

The vasodilator response to Abn-cbd was analyzed in more detailby using the rat isolated mesenteric arterial bed preparation.Vasodilator mechanisms have been much more widely documented in thispreparation than in the mouse mesentery and, in our hands, reliableendothelial denudation could be achieved only in the rat preparation.Abn-cbd caused dose-dependent (ED₅₀: 2.46 ±0.15 mg), prolonged vasodilation (see Fig.3A). Vasodilationin response to the maximally effective dose of Abn-cbd was stronglyinhibited in the presence of 1 μM SR141716A and was absent whentested after endothelial denudation (see Fig.5). This observation isin agreement with earlier findings that the SR141716A-sensitivecomponent of anandamide-induced mesenteric vasodilation is endotheliumdependent (16). However, anandamide can also relax mesenteric vascularsmooth muscle by a noncannabinoid mechanism (16), which Abn-cbd cannot.

To begin to examine the structural requirements for thevasodilator activity of Abn-cbd, we synthesized O-1602, an analog ofAbn-cbd in which the pentyl side chain was shortened to a methyl group.Similar to Abn-cbd, O-1602 did not bind to CB1 receptors in ligandbinding assays. In the rat isolated mesenteric vascular bed, O-1602 wasa full agonist in causing dose-dependent vasodilation, and it was ≈80times more potent than Abn-cbd (ED₅₀: 30.5± 1.5 μg). Similar to Abn-cbd, the vasodilation induced by O-1602(54% ± 2% at 100 μg,n = 4) was markedly reducedin the presence of 1 μM SR141716A (15% ± 3%,n =3,P < 0.005) or after endothelial denudation (18% ±2%,n = 3,P < 0.005).

Cannabidiol, the parent compound of Abn-cbd (see Fig.3D), is a naturally occurring constituent of cannabis devoidof both neurobehavioral (29) and hypotensive (28) activity. We testedcannabidiol for potential antagonism of the mesenteric vasodilatoreffect of Abn-cbd. The inclusion of 10 μM cannabidiol in theperfusion medium did not significantly alter perfusion pressure, butprevented the dilator response to the subsequent intraarterialinjection of 4 mg of Abn-cbd (Fig.3C), or caused its rapidreversal when administered during the plateau phase of the dilatorresponse (Fig.3B). The same concentration of cannabidiolalso significantly attenuated the vasodilator response to anandamide(Fig.4), but not the responses to 10 μg ofacetylcholine (88% ± 9% vs. 78% ± 5% in controls,n = 4), 10 μg of bradykinin (77% ± 6% vs. 70% ±4%,n = 4) or 10 μg of sodium nitroprusside (76% ±2% vs. 78% ± 7%,n = 5). In anesthetized mice, thei.v. administration of 20 μg/g cannabidiol did not influence theCB1 receptor-mediated hypotensive response to submaximal hypotensivedoses of anandamide or HU-210, but significantly inhibitedAbn-cbd-induced hypotension (Δ mean arterial pressure: −20 ± 2vs. −8 ± 2 mmHg,n = 6,P <0.01). These findings suggest that cannabidiol may be a selectiveantagonist of the endothelial anandamide receptor.

We also analyzed the respective role of endothelial nitric oxide(NO), cyclooxygenase, and potassium channels in Abn-cbd-inducedmesenteric vasodilation. The vasodilator response to 4 mg of Abn-cbdremained unchanged in the presence of 100 μMN^G-nitro-l-argininemethyl ester (l-NAME) and 10 μM indomethacin inthe perfusion buffer, which indicates that endothelial NO andcyclooxygenase products such as prostacyclin do not contribute to theresponse (Fig.5). The individual presence of 100 nMapamin or 100 nM charybdotoxin, inhibitors of differentCa²⁺-activated K⁺ channels,also did not inhibit Abn-cbd-induced mesenteric vasodilation (Fig.5),although in the presence of apamin the effect was shorter lasting thanin its absence. However, the combined presence of these two toxinsstrongly inhibited the response to Abn-cbd (Fig.5), similar to theirreported ability to block anandamide-induced mesenteric vasodilation(30), whereas the combination of apamin plus iberiotoxin wasineffective (Fig.5).

Because of structural similarities between anandamide and somesynthetic agonists of capsaicine-sensitive vanilloid receptors (31), weexamined the effect of the vanilloid receptor antagonist capsazepine onAbn-cbd-induced vasodilation in the endothelium-intact rat mesentericarterial preparation. In the presence of 5 μM capsazepine(K_d at vanilloid VR1 receptors: 285nM), the bolus injection of 4 mg of Abn-cbd elicited the same longlasting vasodilation as in its absence (Fig.5). As illustrated in Fig.6, the same concentration of capsazepine caused theexpected significant right-shift of the dose–response curve forcapsaicin-induced vasodilation, which confirms that VR1 vanilloidreceptors were indeed inhibited. As also shown in Fig.6, the effect ofcapsaicine remained unchanged in the presence of 1 μM SR141716A,which indicates that inhibition of the effect of Abn-cbd by 1 μMSR141716A (see Fig.5) cannot be attributed to nonspecific blockade ofVR1 receptors.

Discussion

On the basis of classical pharmacological criteria, such as therank order of agonist potencies and inhibition by a selectiveantagonist, the hypotensive effect of cannabinoids, including that ofthe endogenous ligand anandamide, has been attributed tothe activation of the CB1 subtype of cannabinoid receptors (10). Asimilar pharmacological approach led to the unexpected conclusion thatCB1 receptors are not involved in the mesenteric vasodilator effect ofanandamide (16), even though the effect could be inhibited by the CB1receptor antagonist SR141716A (16,30,32). Through the use of micedeficient in CB1 receptors, the present studies provide definitiveevidence for both hypotheses: the hypotensive effect of anandamide islost in CB1^−/− mice, whereas itsSR141716A-sensitive mesenteric vasodilator effect is maintained or evenincreased compared with responses in CB1^+/+ mice.Thus, anandamide is able to elicit cardiovascular depressor effects byboth CB1 receptor-dependent and -independent mechanisms. These findingsalso indicate that the non-CB1 receptor-mediated mesenteric vasodilatoreffect of anandamide does not significantly contribute to itshypotensive actionin vivo, at least in the case ofexogenous anandamide. Anandamide is known to have a non-CB1receptor-mediated pressor effect (refs.5 and10; see also Fig.1B), which may counteract decreases in vascular resistancelimited to certain vascular beds. Also, the compensatory increase insympathetic tone normally triggered by vasodilation would be limited bythe simultaneous action of anandamide at presynaptic CB1 receptors inwild-type animals (6–9) but not in CB1^−/−mice.

Our previous findings highlighted an additional complexity in themesenteric vasodilator action of anandamide, which is, to a largedegree, endothelium-independent, and only a relatively minorendothelium-dependent component is sensitive to inhibition by lowconcentrations of SR141716A (16). The evidence presented here indicatesthat Abn-cbd may be a selective agonist at this latter site. Similar toanandamide, Abn-cbd elicits SR141716A-sensitive mesentericvasodilation, but it is neurobehaviorally inactive and does notinteract with CB1 receptors even at very high concentrations. Infurther contrast to the effect of anandamide, mesenteric vasodilationby Abn-cbd is abolished by endothelial denudation. The persistence ofthe SR141716A-sensitive mesenteric vasodilator effect of Abn-cbd inCB1^−/− mice rules out the involvement of CB1receptors and implicates a non-CB1 site of action of SR141716A.

A recent report has demonstrated that micromolar concentrationsof anandamide bind to vanilloid receptors, whereas much lowerconcentrations trigger the release of calcitonin gene-related peptidein certain isolated blood vessels, and that inhibition of this processantagonizes anandamide-induced vasodilation (33). However, theendothelium-dependent vasodilator effect of Abn-cbd and anandamidecannot be due to such a mechanism. The vanilloid receptor antagonistcapsazepine did not influence the effect of Abn-cbd at a concentrationat which it significantly inhibited vasodilation by capsaicin and,conversely, SR141716A did not influence the response to capsaicin at aconcentration (1 μM) at which it significantly inhibited the effectof Abn-cbd. Because SR141716A does not inhibit vanilloid receptors(Fig.6), a role of such receptors is likely limited to theendothelium-independent, SR141716A-resistant component of the effect ofanandamide (16). Indeed, in preliminary experiments we found that inendothelium-denuded rat mesenteric arteries, in which Abn-cbd does notcause vasodilation (Fig.5), the SR141716A-resistant vasodilatorresponse to anandamide was inhibited by 5 μM capsazepine. Thus, themechanism of the vasodilator effect of anandamide is complex: inaddition to the demonstrated involvement of CB1 receptors in some bloodvessels (34), it has an SR141716A-sensitive component mediated by anas-yet-undefined endothelial receptor, and an endothelium-independent,SR141176A-resistant component likely mediated by vanilloid receptors.

Nonspecific blockade of cation channels is also unlikely toaccount for the inhibitory effect of SR141716A. SR141716A, at aconcentration of 10 μM but not at 1 μM, has been reported toinhibit vasorelaxation caused by direct activation ofK⁺ channels (27) and to nonselectively inhibitagonist-induced Ca²⁺ mobilization in endothelialcells at 5 μM but not at 1 μM (35). In the present experiments, 1μM SR141716A nearly completely blocked the vasodilator effect of amaximally effective dose of Abn-cbd.

Unlike SR141716A, an inhibitor of both CB1 receptors and theendothelial site of action of Abn-cbd and anandamide, cannabidiol maybe a selective inhibitor of the latter. Cannabidiol, a naturalconstituent of the marijuana plant, is similar to Abn-cbd in that it isneurobehaviorally inactive (29,36). Unlike Abn-cbd, cannabidiol doesnot elicit mesenteric vasodilation, but it is able to prevent orreverse the effect of Abn-cbd (Fig.3) and also to partially inhibitthe vasodilator effect of anandamide. Cannabidiol also inhibits thein vivo hypotensive effect of Abn-cbd, but not the CB1receptor-mediated hypotensive effects of anandamide or HU-210. Thisspectrum of activity suggests that cannabidiol is a selectiveantagonist of the endothelial receptor for Abn-cbd and anandamidedescribed here, and also supports the idea that thein vivoandin vitro effects of Abn-cbd are mediated by similarmechanisms. The apparent selectivity of cannabidiol as an antagonistand the indications of a structure–activity relationship for Abn-cbdand its analog O-1602 strongly suggest that the site of action of thesecompounds is a receptor.

The unchanged vasodilator response to Abn-cbd in the presence ofinhibitors of endothelial NO synthase and cyclooxygenase discounts therole of NO and cyclooxygenase products in this effect. However, it ispossible that a compensatory increase in NO-independent pathwaystriggered by the inhibitors may mask an NO-mediated component presentin untreated preparations, by analogy to the reported up-regulation ofEDHF effects in the presence of inhibitors of NO synthase (37).Blockade of Abn-cbd-induced vasodilation by a combination of apaminplus charybdotoxin but not apamin plus iberiotoxin is reminiscent ofblockade of the effects of EDHF by the same combination of toxins (30,38,39), where the observed inhibition was attributed to blockade ofthe release of EDHF from the endothelium (39). Therefore, thesefindings suggest that Abn-cbd causes NO-independent vasodilation in ratmesenteric arteries by triggering the release of an EDHF, such asK⁺ itself (39) or an epoxygenase product (40).Vascular endothelial cells contain Ca²⁺-activatedK⁺ channels, which are opened by a rise inintracellular calcium (41), and anandamide has been shown to inducecalcium transients in vascular endothelial cells by the activation ofan SR141716A-sensitive mechanism (35,42). It has been proposed thatEDHF itself may be an endocannabinoid acting on CB1-like receptors invascular smooth muscle, based on the ability of SR141716A to inhibitEDHF-induced mesenteric vasodilation (30). The present findings are notcompatible with this proposal, because Abn-cbd loses its vasodilatoreffect after endothelial denudation, and the residual dilator responseto anandamide is no longer sensitive to inhibition by SR141716A (16).However, endocannabinoids have been identified in vascular endothelium(43,44), and it is possible that in vessels in which SR141716A wasfound to inhibit the effect of EDHF, acetylcholine may cause theluminal release of an endocannabinoid, which then acts on anendothelial receptor to release EDHF into the myoendothelial junction.

Regardless of the cellular mechanisms involved, the availabilityof a selective, nonpsychoactive agonist and antagonist forcannabinoid-induced vasodilation may have therapeutic implications invascular diseases, and such compounds may be used as tools foridentifying their molecular site of action.

Acknowledgments

We thank Dr. Francis Barth for providing SR141716A and Lei Wang fortechnical assistance. This work was supported by grants from theNational Institutes of Health. Z.J. and J.A.W. were supported byfellowships from Sanofi Research, Inc., and the DeutscheForschungsgemeinschaft, respectively.

Abbreviations

THC

Δ⁹-tetrahydrocannabinol

EDHF

endothelium-derived hyperpolarizing factor

Abn-cbd

abnormalcannabidiol

Footnotes

References