Phytocannabinoids beyond the Cannabis plant – do they exist?

Today we perceive the endocannabinoid system (ECS) as a rather complex lipid signalling network in which different proteins play distinct roles in the control or in the modulation of numerous physiological and pathophysiological processes (Pertwee, 2005; Di Marzo, 2008). The ECS comprises classical cannabinoid receptors (CB₁ and CB₂), potentially also the orphan receptor GPR55, and arachidonic acid-derived ligands, which, however, also promiscuously target other receptors like, e.g. TRPV1 and PPAR-gamma (O'Sullivan, 2007; De Petrocellis and Di Marzo, 2010; Ross, 2009; Pertwee, 2010). Importantly, the enzymes degrading the endocannabinoids anandamide and 2-arachidonoyl glycerol, namely fatty acid amide hydrolase (FAAH) and monoacylglycerol lipase (MAGL), have been shown to be promising therapeutic targets (Di Marzo, 2008). Finally, there appears to be an anandamide cellular reuptake mechanism that can be blocked by specific inhibitors (Di Marzo, 2008). Both cannabinoid receptor agonists and antagonists have actual or potential therapeutic applications (Di Marzo, 2008; Oesch and Gertsch, 2009; Pertwee, 2009). Cannabinoids are defined as the terpenophenolic constituents of Cannabis sativa L and until recently, the phenylterpenoid Δ⁹-THC and some of its naturally occurring derivatives were the only plant natural products known to directly interact with cannabinoid receptors. However, in the last few years, several non-cannabinoid plant natural products have been reported to act as cannabinoid receptor ligands. This prompts us to define ‘phytocannabinoids’ as any plant-derived natural product capable of either directly interacting with cannabinoid receptors or sharing chemical similarity with cannabinoids or both. Direct cannabinoid receptor ligands are compounds that show high binding affinities (in the lower nM range) for cannabinoid receptors and exert discrete functional effects (i.e. agonism, neutral antagonism or inverse agonism). By contrast, indirect ligands target either key proteins within the ECS that regulate tissue levels of endocannabinoids or allosteric sites on the CB₁ receptor. Certain plant natural products, including some cannabinoids, possess at least some of these properties. Given the often high variability of molecular pharmacological data obtained in different laboratories and the distinct degrees of scrutiny of the experimental setup, molecular pharmacological data on natural products should always be interpreted with care (Gertsch, 2009). For example, the availability of CB receptor KO mice provides a powerful means of investigating the actual cannabimimetic nature of a particular compound in vivo. This commentary focuses on natural products from medicinal and dietary plants which have been reported to interact with the ECS.

Fatty acid derivatives

Despite the fact that N-acylethanolamines (NAEs) (Table 1) from plants do not interact with CB receptors (plants do not generally produce arachidonic acid, which is the acyl scaffold favoured for CB interaction) they have been shown to inhibit FAAH, thus leading to an increase in endocannabinoid tone. N-linoleoylethanolamide and N-oleoylethanolamide, which are found not only in chocolate (Theobroma cacao L.) but also other plants (Di Marzo et al., 1998), and the widespread NAE palmitoylethanolamide, inhibit anandamide breakdown (Maurelli et al., 1995; Di Tomaso et al., 1996). Certain N-alkylamides (alkamides) from Echinacea spp. (Table 2) have been shown to interact functionally with the human CB₂ receptor with low nM to µM K_i values (Gertsch et al., 2006). These N-isobutylamides selectively act at the CB₂ receptor over the CB₁ receptor, leading to an increase in intracellular calcium which could be blocked by the selective CB₂ receptor inverse agonist SR144528, but they do not modulate the G_αi signalling pathway. Intriguingly, CB₂ receptor-binding N-alkylamides show similar anti-inflammatory effects as anandamide (e.g. inhibition of TNF-α) at low nM concentrations (Raduner et al., 2006). Certain Echinacea N-alkylamides inhibit anandamide reuptake in vitro (Chicca et al., 2009). Like anandamide, N-alkylamides also target PPAR-gamma (Spelman et al., 2009). Different Echinacea N-isobutylamides are orally bioavailable resulting in nM plasma levels in humans (Woelkart et al., 2008). The polyacetylenic polyyne falcarinol, which is found in different plants of the Apiaceae family (e.g. in carrots) shows significant binding interactions with both cannabinoid receptors, but appears to selectively undergo an alkylation reaction with the CB₁ receptor (K_i value <1 µM), leading to relatively potent inverse agonistic and pro-inflammatory effects in human skin (Leonti et al., 2010). Finally, it has been proposed that certain dietary fatty acids, which can also be found in plants, can modulate the ECS by influencing the availability of phospholipid biosynthetic precursors of endocannabinoids (Banni and Di Marzo, 2009).

Table 1. Plant natural products that have been suggested to exert cannabimimetic effects but do not interact directly with cannabinoid (CB) receptors

Structure	Name	Origin	CB receptor affinity	Function	In vivo efficacy	Other targets (ECS)	References
	N-acylethanolamines	Widespread in plants	No affinity	FAAH inhibitors	Validated in CB1 and CB2 KO mice	GPR55	Maurelli et al., 1995; Di Tomaso et al., 1996; Di Marzo, 2008
				Indirect cannabimimetics
	Salvinorin A	Salvia divinorum	Insignificant affinity to CB receptors	Indirect cannabimimetic effects at CB1 (mechanism unknown)	No data	KOP agonist	Capasso et al., 2008; Fichna et al., 2009;
		Epling & Jativa-M
	Pristimerin	Relatively widespread in the Celastraceae	No data	Potent reversible	No data	No data	King et al., 2009
				MAGL inhibitor (IC50 value <100 nM)
	Kaempferol	Widespread in plants	No affinity	FAAH inhibitor (IC50 value <1 µM)	No data	No data	Thors et al., 2007; 2008
	Trans-resveratrol	Relatively widespread in plants (e.g. Vitis vinifera L.)	Insignificant affinity	Insignificant effects	No data	No data	Prather et al., 2009
	Curcumin	Curcuma spp.	Insignificant affinity	Insignificant effects	No data	No data	Prather et al., 2009
	Epigallocatechin-3-O-gallate	Relatively widespread in plants (e.g. Camellia sinensis L.)	Insignificant affinity	No data	No data	No data	Korte et al., 2010

• ECS, endocannabinoid system; FAAH, fatty acid amide hydrolase; MAGL, monoacylglycerol lipase.

Table 2. Plant natural products that have been shown to interact directly with cannabinoid (CB) receptors

Structure	Name	Origin	CB receptor affinity	Function	In vivo efficacy	Other targets (ECS)	References
	Δ9-THC	Cannabis sativa L.	Non-selective CB1 and CB2 affinity (Ki values <50 nM) (human)	Parial agonist Gi/o	Validated in CB1 and CB2 KO mice	GPR55	Mechoulam, 1986; Pertwee, 2006
				Inhibition by SR141716 and SR144528		PPARs
						Different ion channels
	N-alkylamide	Echinacea spp.	Selective CB2 affinity (Ki value <100 nM) (human)	Parial agonist	No data	PPAR-γ	Raduner et al., 2006; Chicca et al., 2009
				[Ca2+]i Inhibition by SR144528		Inhibition of AEA transport
						Partial FAAH inhibtion
	N-alkylamide	Echinacea spp.	Selective CB2 affinity (Ki value <100 nM) (human)	Parial agonist [Ca2+] i	No data	PPAR-γ	Raduner et al., 2006; Chicca et al., 2009
				Inhibition by SR144528		Inhibition of AEA transport
						Parial FAAH inhibition
	β-caryophyllene	Widespread in plants	Selective CB2 affinity (Ki value <200 nM) (human)	Full agonist	Validated in CB2 KO mice	No data	Gertsch et al., 2008
				Gi/o
				[Ca2+]i
	Falcarinol	Relatively widespread in Apiaceae (e.g. Daucus carota L.)	Non-selective CB1 affinity (Ki value <1 µM) (human)	CB1 receptor-selective inverse (covalent) agonist	No data	No data	Leonti et al., 2010
				Inhibition of AEA/WIN55212-2
	Rutamarin	Ruta graveolens L.	Selective CB2 affinity (Ki value <10 µM) (human)	No data	No data	No data	Rollinger et al., 2009
	DIM 3,3-diindolylmethane metabolite from indole-3-carbinol	Relatively widespread in the Brassica genus	Selective CB2 affinity (Ki value ≅1 µM) (human)	Partial agonist at CB2 receptor	No data	No data	Yin et al., 2009

• Δ⁹-THC is shown as the major phytocannabinoid from Cannabis sativa L. but there are several other structurally related cannabinoids that interact with CB receptors.
• Δ⁹-THC, Δ⁹-tetrahydrocannabinol; DIM, 3,3′-diindolylmethane; ECS, endocannabinoid system; FAAH, fatty acid amide hydrolase; PPAR, peroxisome proliferator-activated protein.

Terpenes

The bicyclic sesquiterpene, β-caryophyllene (trans-isomer) (Table 2), which is a plant volatile very frequently found in plants, has been shown to selectively target the CB₂ receptor at nM concentrations (K_i= 155 nM) and to act as a full agonist (Gertsch et al., 2008). Remarkably, β-caryophyllene is also a major compound in Cannabis sativa L. essential oil. Thus, Cannabis produces two entirely different chemical scaffolds able to differentially target CB receptors. While studies on the pharmacokinetics of β-caryophyllene are still ongoing, it is already clear that this cyclobutane-ring containing terpene is readily bioavailable, and, unlike many polyphenolic natural products, is not metabolized immediately but shows a Tmax >1 h after one single oral administration (J.G., unpublished data). Orally administed β-caryophyllene (<5 mg·kg⁻¹) produces strong anti-inflammatory and analgesic effects in wild-type mice but not in CB₂ receptor knockout mice, which is a clear indication that it may be a functional CB₂ ligand. Ongoing studies show that β-caryophyllene is effective at reducing neuropathic pain in a CB₂ receptor-dependent manner (Zimmer et al., 2009). Therefore, the FDA approved food additive β-caryophyllene has the potential to become an attractive candidate for clinical trials targeting the CB₂ receptor (Gertsch, 2008). Interestingly, the diterpene salvinorin A from Salvia divinorum Epling & Jativa-M (Table 1) has been reported to be a selective high-affinity kappa-opioid receptor (KOP) agonist, but recent data also suggest that it may interact with a putative CB receptor/KOP heterodimer which may be formed during inflammatory conditions (Fichna et al., 2009). To date, binding experiments have shown that salvinorin A has very low affinity for homomeric cannabinoid receptors and does not inhibit endocannabinoid degradation (Capasso et al., 2008). Consequently, further research is needed to establish whether salvinorin A interacts with a putative cannabinoid/KOP heterodimeric receptor or whether the cannabimimetic effects reported are indirectly mediated via KOP. More recently, two naturally occurring quinonoid triterpenoids, pristimerin (Table 1) and euphol, were found to inhibit MAGL with high potency (IC₅₀= 93 nM and 315 nM respectively) through a reversible mechanism (King et al., 2009). As this class of triterpenes is relatively frequent in nature, it may not be unusual to find ‘indirect’ rather than ‘direct’ agonists of cannabinoid receptors among plant secondary metabolites. Several distinct triterpenes are known to modulate immune functions through yet unknown mechanisms (Rios, 2010) and it will thus be interesting to see in a more systematic study whether other similar triterpenoids are also able to inhibit MAGL.

Polyphenols

The dietary polyphenols trans-resveratrol and curcumin (Table 1) were reported to bind selectively to the human CB₁ cannabinoid receptor with low nM K_i values (5.9 nM and 45 nM respectively) and to exert potent pharmacological effects in mice similar to those induced by the CB₁ receptor inverse agonist rimonabant (Seely et al., 2009). Intrigued by this unexpected finding, our research groups independently measured the binding affinities of these compounds for CB₁ and CB₂ receptors in our laboratories. In our experiments, trans-resveratrol and curcumin only displaced [³H]CP55 940 from cannabinoid receptors at high µM concentrations, suggesting that they lack significant affinity for these receptors. Also polydatin, a glycosilated form of resveratrol, was inactive in these binding assays. Recently, the senior author of the original report retracted the paper (Prather et al., 2009). Hence, neither trans-resveratrol nor curcumin interact functionally with the CB₁ receptor, despite the fact that these compounds appear to share the ability of the CB₁ receptor inverse agonist, rimonabant, to induce weight loss in mice.

More recently, catechin-derivatives were shown to bind to human cannabinoid receptors rather non-selectively at high µM concentrations (Korte et al., 2010). Among these, epigallocatechin 3-gallate and (-)-epigallocatechin (Table 1) were reported to bind to the CB₁ receptor with K_i values of 33.6 and 35.7 µM respectively. However, these K_i values may not be correct. For the calculation of the K_i values the Cheng-Prussof equation (K_i= IC₅₀/1+([S]/K_d) was not applied correctly. The EC₅₀ values used to calculate the K_i values were approximations as neither compound produced more than 60% radioligand displacement even at the highest concentration used. Catechins are very widespread plant secondary metabolites which may provide nutritional health benefits. The same group has recently reported similar CB₁ and CB₂ receptor K_i values for delphinidin and cyanidin, two hydrophilic anthocyanidins (Korte et al., 2009). In both reports, no functional data were shown. In our hands, flavonoid-type compounds (catechins, anthocyanidins, flavones) lead to negligible or very high K_i values, which likely reflect a nonspecific molecular denaturation of the protein surface rather than a functional binding interaction. Similar potentially artefactual effects would most likely be observed with other GPCRs.

Plant polyphenols, such as phenylpropanoids (e.g. epigallocatechin 3-O-gallate, curcumin, resveratrol) possess chemical scaffolds which at µM concentrations bind to protein targets in vitro with limited specificity. This is clearly reflected by numerous reports on protein binding interactions that such compounds undergo in the µM range (Anand et al., 2008; Bisht et al., 2009). At the macroscopic level, polyphenols (i.e. tannins) have been used to tan leather by denaturing of proteins, and at the microscopic level µM concentrations of polyphenols interact with multiple protein binding sites (via their hydroxyl groups) non-specifically and therefore such compounds score as frequent hitters in vitro. The great majority of established cannabinoid receptor ligands are highly lipophilic, which reflects the nature of the active site within cannabinoid receptors. Thus, hydrophilic polyphenols like catechins and anthocyanidins would clearly be atypical cannabinoid receptor ligands.

More interesting are findings that certain flavonoids inhibit fatty acid amide hydrolase (FAAH), which is the enzyme responsible for the breakdown of the endogenous CB receptor ligand anandamide (Thors et al., 2007; 2008). Both the isoflavonoid genistein and the flavonoids kaempferol (Table 1), 7-hydroxyflavone and 3,7-dihydroxyflavone have been shown to concentration-dependently inhibit anandamide hydrolysis in rat brain homogenates, albeit at relatively high concentrations (IC₅₀ values between 2 and 10 µM). Nevertheless, the authors of these studies showed a preliminary structure-activity relationship with 7-hydroxyflavone being the most potent inhibitor (IC₅₀ value <1 µM).

An abundant literature is devoted to mechanisms underlying the biological activity of plant polyphenols (Landis-Piwowar and Dou, 2008; Bisht et al., 2009). However, although most beneficial and potentially therapeutic effects of trans-resveratrol, curcumin, catechins and kaempferol-type flavonoids are typically detected in the low µM range in vitro, all such compounds show limited bioavailability and poor pharmacokinetics in vivo with reported plasma concentrations in the low nM range (DuPont et al., 2004; Garcea et al., 2004; Boocock et al., 2007).

Other plant natural products with binding affinity to the CB₂ receptor

Other plant natural products have been shown to bind weakly to the CB₂ receptor. These include the coumarin derivative rutamarin from the medicinal plant Ruta graveolens L. (Rollinger et al., 2009) and 3,3′-diindolylmethane (DIM) (Table 2), which is an anticarcinogenic metabolite generated by ingestion of indole-3-carbinol. Indole-3-carbinol is commonly found in Brassica vegetables. DIM has been shown to be a weak CB₂ receptor partial agonist (Yin et al., 2009).

Conclusions

There is no doubt that phytocannabinoids from Cannabis have greatly influenced research on the ECS and without the milestone discovery that Δ⁹-THC is the main psychoactive principle (reviewed in Mechoulam, 1986) many of the subsequent discoveries in the field of cannabinoid research would probably not have been made. Furthermore, with the development of therapeutic Cannabis extracts, as with Sativex™, this plant is also likely to provide new pharmaceutical applications in the future. The question remains as to why Cannabis sativa L. appears to be the only plant that produces a metabolite (Δ⁹-THC acid) that readily leads to its decarboxylation product Δ⁹-THC, which is the most potent phytocannabinoid activator of the CB₁ receptor. Interestingly enough, while nature may have been rather parsimonious in its provision of botanical secondary metabolites that activate the CB₁ receptor, there is an increasing number of plant-derived natural products reported to target the CB₂ receptor to varying degrees. Flavonoids, which belong to natural polyphenols that readily interact with proteins, may target some of the proteins within the ECS, such as FAAH. However, no convincing evidence has been provided that polyphenols modulate cannabinoid receptors with significant potency. The finding that certain triterpenes potently inhibit MAGL further adds to the repertoire of plant-produced ‘indirect’ cannabinoid receptor agonists. Although higher plants do not contain endocannabinoids and lack the classical G-protein-coupled cannabinoid receptors, they do express enzyme isoforms that resemble some of the enzymes known to be important in the processing of endocannabinoids (Shrestha et al., 2006). Plants produce fatty acid amides, some of which are able to inhibit the degradation of anandamide but do not generally bind with significant affinity to CB receptors (Gertsch et al., 2006; Di Marzo et al., 2007). At present, the only phytocannabinoid that has been discovered to also exist in plants other than Cannabis is β-caryophyllene, which is among the most abundant plant essential oil components. Although Δ⁹-THC is a partial agonist at both CB₁ and CB₂ receptors, it has significant lower efficacy at the CB₂ receptor. Another phytocannabinoid, Δ⁹-tetrahydrocannabivarin, has also recently been shown to be a CB₂ receptor partial agonist, but is also a CB₁ receptor antagonist (Bolognini et al., 2010). Therefore, β-caryophyllene, which is also one of the most abundant secondary metabolites in Cannabis essential oil, could be considered as a true CB₂ receptor-selective Cannabis constituent. During mammalian evolution contacts with ‘direct’ CB₂ receptor active plant metabolites like β-caryophyllene or ‘indirect’ cannabinoid receptor agonists (FAAH and MAGL inhibitors) in diet may have led to hitherto unrecognized physiological effects. Although it is tempting to believe that these compounds exert beneficial effects in humans, clinical evidence is lacking. Future studies will have to determine whether there are additional apparently nontoxic CB₂ receptor-selective ligands in plants other than Cannabis and whether they could in fact be exploited therapeutically.

Conflict of interest

The authors state no conflicts of interest.