The endocannabinoid system (ECS) and, in particular, cannabinoid CB1 receptors, their endogenous agonists (the endocannabinoids anandamide and 2-arachidonoylglycerol) and enzymes for the biosynthesis and degradation of the latter mediators are emerging as key players in the control of all aspects of food intake and energy balance. The ECS is involved in stimulating both the homoeostatic (that is, the sensing of deficient energy balance and gastrointestinal load) and the hedonic (that is, the sensing of the salience and the incentive/motivational value of nutrients) aspects of food intake. The orexigenic effects of endocannabinoids are exerted in the brain by CB1-mediated stimulatory and inhibitory effects on hypothalamic orexigenic and anorectic neuropeptides, respectively; by facilitatory actions on dopamine release in the nucleus accumbens shell; and by regulating the activity of sensory and vagal fibres in brainstem–duodenum neural connections. In turn, the levels of anandamide and 2-arachidonoylglycerol and/or CB1 receptors in the brain are under the control of leptin, ghrelin and glucocorticoids in the hypothalamus, under that of dopamine in the limbic forebrain and under that of cholecystokinin and ghrelin in the brainstem. These bi-directional communications between the ECS and other key players in energy balance ensure local mediators such as the endocannabinoids to act in a way coordinated in both ‘space’ and ‘time’ to enhance food intake, particularly after a few hours of food deprivation. Alterations of such communications are, however, also among the underlying causes of overactivity of the ECS in hyperphagia and obesity, a phenomenon that provided the rationale for the development of anti-obesity drugs from CB1 receptor antagonists.
Brief introduction to the endocannabinoid system
It has been known for a long time that smoking preparations from Cannabis sativa, such as marijuana and hashish, and self-administering the major psychoactive principle of this plant, Δ9-tetrahydrocannabinol (THC), stimulate an appetite, especially for very palatable foods (the ‘munchies’).1 Therefore, it is not surprising to find THC or its synthetic analogue, nabilone, among the drugs approved in the US and other countries to increase body weight in cancer and AIDS patients, and to inhibit chemotherapy-induced emesis. The molecular mechanisms through which marijuana stimulates an appetite and inhibits vomiting have been characterized only relatively recently, when a specific binding site for THC was identified in the CNS, including in brain nuclei that control reward, motivation and satiety. In fact, we now know that THC binds to and activates not only the cannabinoid CB1 receptor but also another specific G-protein-coupled receptor, the cannabinoid CB2 receptor, which is expressed abundantly in several immune cells and tissues. Brain CB1 receptors are coupled, among other things, to neurotransmitter release inhibition, whereas CB2 receptors participate in the regulation of cytokine release and immune cell function.2 The discovery of CB1 and CB2 receptors led to the finding of endogenous compounds capable of binding to and activating them: the endocannabinoids (ECs). The two best-studied ECs are N-arachidonoylethanolamine (anandamide) and 2-arachidonoylglycerol (2-AG) (Figure 1). It is now well established that these two ECs are not stored in secretory vesicles but are biosynthesized de novo after an increase in intracellular concentrations of Ca2+ within a framework of phospholipid-originated metabolic reactions. In fact, both the formation of the two direct and distinct biosynthetic precursors for anandamide and 2-AG and their conversion into the two ECs are catalysed by Ca2+-sensitive enzymes.3 This means that the whole cascade of EC production is triggered ‘on demand’, usually in response to an acute or chronic perturbation of cell homoeostasis and with the function of returning the cell to its steady state before this perturbation. The cannabinoid receptors, the ECs and the proteins catalysing EC biosynthesis and inactivation constitute the EC system (ECS).3
It is now clear that the activation or blockade of CB1 receptors can significantly influence eating behaviours in both animals and men. The increased food intake caused by marijuana smoking in humans was shown to be because of an augmentation of food consumed as ‘between-meal’ snacks rather than by an increase in meal size per se.1 In satiated rats, doses of 0.5, 1.0 and 2.0 mg kg–1 of oral THC produce substantial hyperphagia during the first hour of testing. Subsequently, rats compensate for this effect so that 24-h intakes are similar to those in vehicle-treated animals. Like THC, anandamide (0, 1.0, 5.0 or 10.0 mg kg–1, subcutaneously) stimulates feeding, with a marked reduction in the ‘between-meal’ latency. Apart from its more frequent and/or rapid onset, cannabinoid-induced eating retains the normal, species-typical pattern that is characteristic of normal free-feeding rats.4 Studies carried out in guinea pigs show that the orexigenic effect of a CB1 agonist is manifested by increases not only in meal frequency but also in the amount of food eaten per meal. On the other hand, CB1 antagonist administration to this species produces an anorexic effect associated with decreases in both meal frequency and duration.5 In the next sections, we review the mechanisms proposed so far through which CB1 receptors and ECs control all aspects of food intake, thus providing an ideal link between homoeostatic and hedonic pathways involved in energy-balance regulation.
Homoeostatic pathways: the hypothalamic endocannabinoid system and food intake
Cannabinoid CB1 receptors are widely expressed in all hypothalamic nuclei at both the somatic and, predominantly, the axonal level. CB1-immunoreactive axons innervate densely the majority of hypothalamic nuclei in the mouse, except for the suprachiasmatic and lateral mammillary nuclei in which only scattered CB1-labelled fibres occur.6 CB1-positive axonal innervation of the arcuate, innervation of the arcuate nuclei (ARC), paraventricular nuclei (PVN), ventromedial hypothalamus (VMH), dorsomedial hypothalamus (DMH), and of the external zone of the median eminence, is consistent with the important role of this receptor in the regulation of energy homoeostasis and neuroendocrine functions. Indeed, injection of either anandamide into the VMH, or that of THC into the PVN, causes a significant increase in food intake in satiated animals.7, 8 Ultrastructural studies have established that CB1 immunoreactivity preferentially appears in the preterminal and terminal portions of axons.6 This distribution, together with the finding of complementary immunoreactivity for diacylglycerol lipase-α, the enzyme mostly responsible for 2-AG biosynthesis in the somatodendritic compartments in the LH (Figure 2), is in agreement with the proposed role of ECs as retrograde signals. Thus, ECs would be released from the depolarized post-synaptic neuron to act on pre-synaptic CB1 receptors to inhibit either excitatory or inhibitory neurotransmitter release. Indeed, such a mechanism has been shown to be responsible for (1) the disinhibition of the release of the orexigenic neuropeptide, a melanin-concentrating hormone (MCH), from LH neurons through a retrograde inhibition of γ-aminobutyric acid (GABA) release;9 and (2) the inhibition of anorectic neurons in the PVN after a sequential activation of either ghrelin or ‘fast’ glucocorticoid receptors in these neurons (which likely produce the corticotropin-releasing hormone (CRH)), the biosynthesis and release of ECs, and subsequent retrograde CB1-mediated inhibition of glutamate release from parvocellular neurons innervating these neurons.10, 11 In some cases,9, 10 these retrograde orexigenic actions of ECs are under the negative control of post-synaptic leptin receptors that reduce EC biosynthesis by decreasing intracellular calcium. In fact, it had been shown earlier that hypothalamic EC levels are inhibited by an intravenous injection of leptin in rats, an effect that is likely responsible for the permanently elevated EC levels in the hypothalamus of obese rodents that lack leptin (ob/ob mice) or fully functional leptin receptors (db/db mice and fa/fa Zucker rats).12 Unlike ghrelin, glucocorticoids and leptin, other hormones like insulin and melanocortin α-melanocyte-stimulating hormone (α-MSH), do not affect hypothalamic EC levels under conditions shown to inhibit food intake, despite the presence of their receptors in the hypothalamus.13
CB1 receptors are also postsynaptic in hypothalamic nuclei, for example, in anorectic CRH-expressing neurons of the PVN, and seem to negatively control the expression of CRH and cocaine- and amphetamine-regulated transcript (CART) in the ARC. In fact, CRH mRNA levels in the PVN are higher in CB1-deficient mice,14 whereas an elevation of endogenous anandamide levels, obtained by ‘knocking-out’ fatty acid amide hydrolase (FAAH: the enzyme responsible for anandamide degradation), is accompanied by reduced CART release in the ARC, DMH and PVN, an effect antagonized by a CB1 antagonist.15 Another action of post-synaptic CB1 receptors may be to enhance orexin A signalling. In fact, it has been shown that orexin A receptors are sensitized by CB1 receptors co-expressed in the same cells,16 and that CB1 and orexin A receptors are often co-localized in mouse PVN.17
The above mechanisms might intervene in both acute and chronic modulation of eating behaviours. In fact, it has been shown that hypothalamic 2-AG levels are elevated after food deprivation and reduced after food consumption.18 This may be a consequence of increased ghrelin and glucocorticoid and decreased leptin plasma levels that follow food deprivation, with subsequent stimulation and disinhibition, respectively, of EC biosynthesis. However, hypothalamic CB1 receptor and/or 2-AG levels are also elevated in rodents that develop obesity because of leptin deficiency12 or a high-fat diet (V Di Marzo and L Cristino, unpublished observations19), and this might contribute to hyperphagia and increased body weight. By contrast, prolonged semi-starvation reduces hypothalamic 2-AG levels,20 possibly as an adaptive response aimed at reducing appetite during food shortage.
Hedonic pathways: the mesolimbic endocannabinoid system and food intake
The components of the ECS, including CB1 receptors, ECs and the diacylglycerol lipase-α, are all present in the mesolimbic system and, in particular, in the nucleus accumbens shell (NAcS) and ventral tegmental area (VTA), wherein they play a role in the circuits involved in motivation and reward. The activity of CB1 receptors was shown to play a facilitatory role in the self-administration of substances of abuse, particularly alcohol, nicotine and morphine, although not of psychostimulants, such as cocaine and metamphetamine. In fact, either genetic or pharmacological blockade of CB1 receptors counteracts both the chronic self-administration of alcohol, nicotine and morphine and their effect to stimulate dopamine release in the NAcS,21, 22 which is considered to be one of the key neurochemical correlates for the rewarding and addictive properties of these substances,23 as well as of palatable foods in, for example, binge eating.24 It is important that the CB1 receptor antagonist, rimonabant, was found to inhibit novel palatable food-induced dopamine release in NAcS25 and to reduce the intake of sweet food in non-food-deprived animals.26 These data suggest that exposure to foods with high salience and incentive properties might stimulate an EC tone to induce dopamine release in this limbic area. This latter event might, in turn, lead to both increased motivation to consume palatable foods and heightened rewarding effects after the consumption of such foods. Accordingly, although the effect of exposure to palatable foods on NAcS EC levels has not been evaluated, food deprivation increases and food consumption decreases both anandamide and 2-AG levels in the limbic forebrain (which includes the NAcS).18 On the other hand, leptin receptor-deficient Zucker rats, which are hyperphagic and obese, have an enhanced CB1 receptor expression in NAcS.27 A prolonged high-fat diet leading to obesity causes either no change19 or a decreased expression28 of CB1 receptors in NAcS, which the authors of the latter study interpreted as a possible adaptive response to permanently elevated EC levels in this nucleus. Finally, elevation of anandamide levels in the limbic forebrain was observed in rats chronically treated with alcohol and nicotine,29 whereas CB1 blockade reduced self-administration of these substances.21, 22 Thus, as seen with some substances of abuse, chronic consumption of a high-energy diet, leading to obesity, might be accompanied by a high limbic forebrain EC tone, and hence, its counteraction can be observed after administration of CB1 antagonists.
On the basis of data showing that dopamine inhibits EC biosynthesis in the limbic forebrain,30 and that feeding after food deprivation is accompanied by increased dopamine release in the nucleus accumbens,31 it can be hypothesized that, in lean animals, ECs increase in this brain area during food deprivation and then decrease after food consumption18 because of correspondingly lower and higher dopamine levels, respectively. The putative increased EC tone in NacS, which may occur in food-deprived animals after exposure to palatable food, would enhance the incentive value of food (that is, the ‘wanting’) before food intake, and heighten its rewarding effects after consumption (and hence, its ‘liking’). Both these effects might occur through, among other things, CB1-mediated reinforcement of dopamine release in this brain area, which, eventually, should feedback negatively on EC levels. After prolonged consumption of high-fat diets and the subsequent obesity, this negative feedback might become impaired and this, as observed in the hypothalamus, might lead to a chronically elevated EC tone in NAcS, which, in this case, would contribute to hyperphagia by reinforcing the aforementioned ‘hedonic’ aspects of food intake.
It is interesting that the effects of ECs on dopamine release in NAcS are unlikely to be direct as there is no evidence of the co-expression of CB1 with dopamine in this nucleus. It has been suggested that activation of pre-synaptic CB1 receptors on glutamatergic afferents from the prefrontal cortex instead reduces glutamate release in NAcS, thereby inhibiting the GABAergic accumbens neurons projecting to the VTA and hence disinhibiting the dopaminergic VTA neurons that go back to the NAcS, thus causing enhanced dopamine release in this latter area.22
Two questions that ‘behaviouralists’ have been attempting to answer are (1) does CB1 activation cause a preferential stimulation of the ‘hedonic’ aspects of food intake, thus affecting preferentially the ‘wanting’ and ‘liking’ of foods? And, as a consequence, (2) does CB1 blockade preferentially affect the consumption of palatable foods? Although experiments aimed at answering the former question have given consistent results, data obtained using CB1 receptor antagonists have been more controversial. Higgs et al.32 observed that THC and anandamide systemically administered to rats significantly increased the total number of licks of a sucrose solution, a behaviour that the authors considered as a measure of the ‘liking’ of food. Administration of the CB1 antagonist, SR141716 (1 and 3 mg kg–1), instead, significantly decreased total licks, consistent with EC and CB1 involvement in the mediation of food palatability.32 In a subsequent study, Ward and Dykstra33 suggested that CB1 receptors are preferentially involved in the reinforcing effects of sweet food as compared with those of pure-fat food. More recently, the effect of anandamide microinjections into the medial NAcS on the affective reactions to sweet and bitter tastes in rats was tested. Anandamide doubled the number of positive ‘liking’ reactions elicited by intraoral sucrose, without altering negative ‘disliking’ reactions to bitter quinine. It is important that, on the basis of the localization of anandamide microinjection-induced cFos activation, integrated with behavioural data, the authors identified in the dorsal medial shell of accumbens an anatomical ‘hotspot’ of roughly 1.6 mm3, which is responsible for the hedonic amplification and the subsequent stimulation of eating behaviour by ECs.34 In agreement with these studies, some authors have reported that the CB1 antagonist, rimonabant, preferentially counteracts the intake of palatable food,26 and that, as opposed to the effect on regular food, the inhibitory action of this compound on the consumption of chocolates does not undergo tolerance after repeated administration.35 By contrast, other groups, using several types of experimental paradigms, have shown that food palatability is not a necessary determinant of the anorectic effects of this or other CB1 antagonists in rats36 and in non-human primates.37 One way to reconcile these findings is by proposing the intermediacy of different populations of CB1 receptors in the effects of exogenous and endogenous CB1 agonists. Thus, systemic CB1 agonists might preferentially activate CB1 populations involved in enhancing the incentive and salient properties of food. By contrast, those CB1 receptors that are tonically engaged by ECs, and the activity of which is more strongly shown by the use of systemic CB1 antagonists, may be involved in both the homoeostatic and hedonic aspects of food intake, and possibly also in modulating the connections and signal integration among brain nuclei that are more or less specialized in these aspects.
Homoeostatic and hedonic pathways: the brainstem endocannabinoid system and food intake
Nuclei in the brainstem receive sensory and vagal fibres from the stomach and duodenum, thereby sensing distension and contraction of these tissues and, hence, gastrointestinal load. Furthermore, important gastric peptides are released after food consumption (for example, cholecystokinin (CCK)) or deprivation (ghrelin) to convey anorectic or orexigenic signals, respectively, to brain areas controlling food intake. Although CCK is particularly important in inhibiting gastric emptying and mediating satiety signalling in the brainstem, ghrelin also acts in the NAcS and brainstem, and is one of the signals that coordinate the activity of the three major brain areas involved in food intake. It has been shown that CB1 receptor expression in the rat nodose ganglion, in which vagal fibres terminate in the brainstem, is under the negative control of CCK (the receptors of which are co-expressed with CB1). Food deprivation elevates the number of neurons in this ganglion that are immunoreactive for CB1, whereas CCK or food consumption reduces it.38 Importantly, ghrelin reverses the negative control on CB1 expression by CCK in the nodose ganglion.39 These data indicate that, as seen in the hypothalamus and NAcS, EC tone changes in the brainstem during the different phases of eating, the highest being after food deprivation and the lowest during food consumption, an effect that is possibly because of the opposing regulatory effects of CCK and ghrelin.
The ECS might control the activity of brainstem neurons involved in the sensing of a gastric load by acting in the duodenum. In fact, duodenal anandamide and 2-AG levels are highest after food deprivation and lowest during food consumption,40, 41 and the anorectic effect of intra-peritoneal rimonabant is attenuated by the destruction of sensory fibres ascending to the brainstem.40 EC levels are also dramatically increased in the duodenum of Zucker rats, again suggesting that the dysregulation of EC tone might contribute to hyperphagia and obesity in these animals.41
Recent data implicate another nucleus of the brainstem in the effects of ECS: the pontine parabrachial nucleus (PBN). Although considered the second central gustatory relay involved in taste and hence, indirectly, in reward, the PBN also responds to gastric sensory stimulation. Immunofluorescence and in vitro [35S]GTP-γ-S autoradiography of rat tissue sections containing the PBN showed the presence of CB1 receptors and their functional coupling to G-proteins after incubation with 2-AG. Microinfusions of 2-AG into PBN in rats stimulated the feeding of pellets containing either fat and sucrose, pure sucrose or pure fat during the first 30 min after infusion. In contrast, 2-AG failed to increase the consumption of standard chow. The orexigenic responses to 2-AG were attenuated by the CB1 antagonist, AM251. Furthermore, the responses were regionally specific, because 2-AG failed to alter intake when infused into sites ∼500 μm caudal to infusions that successfully stimulated feeding. On the basis of these data, the authors suggested that the hedonically positive sensory properties of food enable ECs to act on CB1 receptors in the PBN to stimulate further eating.42 Accordingly, the authors also reported that elevation of endogenous EC levels by inhibition of FAAH causes a similar selective increase of palatable food consumption, again by activating CB1 receptors.43 These findings indicate that the integration of neural substrates modulating feeding, energy balance and behavioural responses for natural reward are regulated by the ECS.
Role of the endocannabinoid system in the interactions between brain nuclei involved in food intake
As mentioned above, the control of the three major aspects of food intake, that is, the sensing of energy deficiency/abundance, gastric load and salience of nutrients, is not uniquely regulated by the hypothalamus, brainstem and the mesolimbic system, respectively. Instead, these three brain areas communicate with each other to integrate these three key components regulating food consumption. The ECs also targets these communications. Apart from the aforementioned effects of CB1 activation in the PBN, an important role is likely to be played by ECS in MCH-expressing neurons of the LH that also regulate the activity of the nucleus accumbens while receiving inhibitory control from this area through GABAergic neurons that are also controlled by CB1 receptors.9 Further evidence of the role of ECS in the neuronal circuits integrating the activity of the hypothalamus and NAcS came from a recent study by Soria-Gomez et al.44 The orexigenic effects of (1) anandamide, (2) a FAAH inhibitor and (3) an inhibitor of anandamide cellular uptake, injected into NAcS, were evaluated parallel to their activation of cFos in various hypothalamic nuclei, including the LH. The results indicated that all the treatments stimulate food intake during 4 h post-injection, whereas also increasing cFos immunoreactivity in all hypothalamic nuclei analysed, in a way sensitive to CB1 antagonism. These data support the involvement of the ECS in the mechanisms integrating the activity of the NAcS and the hypothalamus and leading to food intake.
Here we have reviewed how the ECS stimulates, through the activation of CB1 receptors, the consumption of food by affecting both the sensing of energy deficiency/abundance and gastric load (homoeostatic mechanisms), and the salience as well as the incentive/motivational value of food (hedonic aspects). Taken together with the several emerging mechanisms, ECS also seems to be involved in the peripheral control of lipid and glucose metabolism and of energy expenditure.45 It can be concluded that this signalling system is perhaps one of the few that can coordinate, by acting at several locations in a way regulated by hormones, peptides and neuropeptides, all the determinants of energy balance, including intake, processing, transformation and expenditure. This concept should foster more studies aimed at exploring the possible exploitation of the ECS in the treatment of obesity and eating disorders, on the basis of the ever increasing evidence for a dysregulation of this system during imbalanced metabolic states.45 It is now clear that CB1 antagonists that have undergone clinical development as anti-obesity drugs can directly target metabolic disorders in a way partly independent from their anorectic and body weight-reducing actions.45 More efforts need to be made, instead, to establish if CB1 agonists and inhibitors of EC degradation can be used in the clinic against types of anorexia (that is, anorexia nervosa) other than cachexia in cancer patients.
Conflict of interest
V Di Marzo has received consulting fees, lecture fees and grant support from Sanofi-Aventis. V Di Marzo has also received grant support from Allergan and GW Pharmaceuticals. The remaining authors declare no conflict of interest.
The work of the authors on the role of ECS in energy balance is partly supported by a research grant from Sanofi-Aventis and by the Programma Neuroscienze, Compagnia San Paolo, Grant no. 2008.1262.
About this article
Free-choice high-fat diet alters circadian oscillation of energy intake in adolescent mice: role of prefrontal cortex
European Journal of Nutrition (2017)