An anorexic lipid mediator regulated by feeding

Oleylethanolamide (OEA) is a natural analogue of the endogenous cannabinoid anandamide. Like anandamide, OEA is produced in cells in a stimulus-dependent manner and is rapidly eliminated by enzymatic hydrolysis, suggesting a function in cellular signalling 1 . However, OEA does not activate cannabinoid receptors and its biological functions are still unknown 2 . Here we show that,

Oleylethanolamide (OEA) is a natural analogue of the endogenous cannabinoid anandamide. Like anandamide, OEA is produced in cells in a stimulus-dependent manner and is rapidly eliminated by enzymatic hydrolysis, suggesting a function in cellular signalling 1 . However, OEA does not activate cannabinoid receptors and its biological functions are still unknown 2 . Here we show that, in rats, food deprivation markedly reduces OEA biosynthesis in the small intestine. Administration of OEA causes a potent and persistent decrease in food intake and gain in body mass. This anorexic effect is behaviourally selective and is associated with the discrete activation of brain regions (the paraventricular hypothalamic nucleus and the nucleus of the solitary tract) involved in the control of satiety. OEA does not affect food intake when injected into the brain ventricles, and its anorexic actions are prevented when peripheral sensory ®bres are removed by treatment with capsaicin. These results indicate that OEA is a lipid mediator involved in the peripheral regulation of feeding.
Fatty acid ethanolamides (FAEs) are unusual components of animal and plant lipids 3±5 that are synthesized in response to a variety of physiological and pathological stimuli, including activation of neurotransmitter receptors in rat brain neurons 1,6 and exposure to metabolic stressors in mouse epidermal cells 7 . The primary mechanism underlying FAE generation in mammalian tissues involves two concerted biochemical reactions: cleavage of the membrane phospholipid N-acyl phosphatidylethanolamine (NAPE), catalysed by an unknown phospholipase D; and NAPE re-synthesis, mediated by an N-acyltransferase (NAT) that is regulated by calcium ions and cyclic AMP 8,9 . After release, FAEs are transported back into cells 10 and eventually broken down to fatty acid and ethanolamine by an intracellular fatty acid amide hydrolase (FAAH) 11,12 . That animal cells release FAEs in a stimulus-dependent manner suggests that these compounds may participate in cell-tocell communication. Further support for this idea comes from the discovery that the polyunsaturated FAE anandamide (arachidonylethanolamide) serves as an endogenous ligand for cannabinoid receptors 13 . However, the pharmacological effects of saturated or mono-unsaturated FAEs such as OEA cannot be accounted for by activation of any of the known cannabinoid receptor subtypes 2 , and the biological roles of these compounds remain elusive.
Because anandamide may regulate feeding 14 , we investigated the effects of OEA on food intake in rats. Systemic administration of OEA caused a dose-and time-dependent suppression of food consumption (Fig. 1). Under the same conditions anandamide and oleic acid had no effect, palmitylethanolamide was signi®cantly less potent than OEA and elaidylethanolamide was similar in potency to OEA (Fig. 1a). These results indicate that OEA reduces eating in a structurally selective manner, and suggest that the molecular requisites for this effect are distinct from those involved in the interaction of anandamide with recognized cannabinoid targets 15 . In further support of this idea, CB1 and CB2 cannabinoid antagonists (SR141716A and SR144528, respectively) did not affect OEA hypophagia (data not shown).
To test whether tolerance develops to the hypophagic actions of OEA, we administered it subchronically in rats. Daily injections of OEA (5 mg kg -1 i.p.) resulted in a small but signi®cant decrease in cumulative food intake (Fig. 2a), which was accompanied by a marked inhibition of body mass gain (Fig. 2b). By contrast, subchronic OEA administration had no effect on water intake (Fig. 2c) or on plasma levels of various metabolites and liver enzymes (see Supplementary Information). Pair-feeding experi-ments suggest that decreased food consumption is suf®cient to account for the mass-reducing actions of OEA (Table 1). We cannot exclude, however, the possible contribution of other factors to this effect, such as stimulation of energy expenditure or inhibition of energy accumulation, as suggested by the signi®cantly lower triglyceride levels in OEA-treated rats than in controls (see Supplementary Information).
Although potent when administered peripherally, OEA is ineffective after intracerebroventricular injection (see Supplementary  Information), leading us to hypothesize that its primary sites of action are located outside the central nervous system (CNS). To test this idea, we destroyed peripheral sensory ®bres by treating adult rats with capsaicin 17 . Capsaicin-treated rats failed to respond to systemically administered cholecystokinin-8 (CCK-8) (Fig. 3a, d) and drank more water than controls (Fig. 3c, f), two indications that the neurotoxin had removed sensory afferents 18 . Capsaicin-treated animals also failed to become hypophagic in response to OEA (5 mg kg -1 i.p.), but responded normally to the compound CP-93129, which targets brain 5-HT 1B receptors (Fig. 3a, d) 19 . These ®ndings support the hypothesis that OEA reduces food intake by acting at a peripheral site and that sensory ®bres are required for this effect. Interestingly, the ability of a high OEA dose (20 mg kg -1 i.p.) to reduce motor activity was not affected by capsaicin treatment (Fig. 3b, e)   brainstem and the paraventricular (PVN) nucleus in the hypothalamus 20 . To identify brain pathways engaged during OEAevoked hypophagia, we mapped messenger RNA levels for the activity regulated gene c-fos 21 by in situ hybridization after systemic administration of OEA, oleic acid or vehicle. When compared with controls, OEA (10 mg kg -1 , i.p.) evoked a highly localized increase in c-fos mRNA levels in the PVN, supraoptic nucleus ( Fig. 4a) and NST (Fig. 4c). This enhancement was speci®c, insofar as c-fos expression in other brain regions was not signi®cantly affected by OEA (Fig. 4b, d). The stimulation of c-fos expression in the NST (which processes vagal sensory inputs to the CNS) and the PVN (a primary site for the coordination of central catabolic signals) 20 , is consistent with a role for OEA as a peripheral regulator of feeding behaviour.
The anorexic effects of OEA are reminiscent of those produced by gut peptides such as CCK. Because the release of CCK from duodenum is regulated by nutrients 22 , we studied the impact of feeding on intestinal OEA biosynthesis. Analyses with highperformance liquid chromatography and mass spectrometry (HPLC/MS) revealed that small-intestinal tissue from free-feeding rats contains substantial amounts of OEA (354 6 86 pmol g -1 , n = 3). Intestinal OEA levels were markedly decreased after food deprivation, but returned to baseline after refeeding (Fig. 5a). By contrast, no such changes were observed in stomach (control, 210 6 20 pmol g -1 ; starvation, 238 6 84 pmol g -1 ; starvation± refeeding, 239 6 60 pmol g -1 ; n = 3). Variations in intestinal OEA levels were accompanied by parallel alterations in NAT activity, which participates in OEA formation 4 , but not in FAAH activity, which catalyses OEA hydrolysis (Fig. 5b, c) 11,12 . These ®ndings suggest that starvation and feeding reciprocally regulate OEA biosynthesis in small intestine. In agreement with an intraabdominal source of OEA, we found that plasma OEA levels in starved rats are higher in portal than in caval blood (porta, 14.6 6 1.8 pmol ml -1 ; cava, 10.3 6 2.8 pmol ml -1 ; n = 5). The contribution of other intra-abdominal tissues to OEA formation cannot be excluded at present.
Our results suggest a hypothetical model for the role of OEA in feeding behaviour. According to this model, food intake may stimulate NAT activity, enhancing OEA biosynthesis in the small intestine and possibly other intra-abdominal tissues. Newly produced OEA may activate local sensory ®bres, which may in turn inhibit feeding by engaging brain structures such as the NST and PVN. A number of questions remain, such as what physiological stimuli initiate and terminate OEA biosynthesis, whether there is a functional relationship between OEA and other nutritional signals, and what the molecular targets of OEA are. Irrespective of the answers, our results reveal an unexpected role for OEA in the peripheral regulation of feeding, and provide a framework to develop medicines for the treatment of eating disorders.

Animals
We used male Wistar rats (200±350 g). All procedures met the National Institutes of Health guidelines for the care and use of laboratory animals, and the European Communities directive 86/609/EEC regulating animal research.

Enzyme assays
In all biochemical experiments, rats were killed and tissues collected between 14:00 and 16:00. Microsome fractions were prepared as described 24 . NAT assays were performed using 1,2-di[ 14 C]palmityl-sn-glycerophosphocholine as a substrate (108 mCi mmol -1 , Amersham) under conditions that were linear with time and protein concentrations 9

HPLC/MS analyses
FAEs were extracted from tissues with a methanol±chloroform mixture and fractionated by column chromatography 23 . FAEs were quanti®ed by HPLC/MS, with an isotope dilution method 25 .

Blood chemistry
We used commercial kits to measure glucose, insulin, leptin and other plasma metabolites and enzymes (Sigma and Linco Research). Plasma prolactin, corticosterone and luteinizing hormone were quanti®ed by radioimmunoassay 26 .

Feeding experiments
For acute experiments, we measured food intake in rats deprived of food for 24 h (ref. 27) that were habituated to the experimental setting. We administered drugs 15 min before food presentation. For subchronic experiments, freely fed rats received vehicle injections for two days. On day 3, we divided the animals into two equal groups and gave them daily injections of vehicle or OEA (5 mg kg -1 at 19:00) for seven consecutive days, while measuring body mass, food intake and water intake. A third group of rats (pair fed) received an amount of food identical to that consumed by the OEA group. On day 9 of the study, the animals were killed and plasma was collected for biochemical analyses.

Conditioned taste aversion
Rats were deprived of water for 24 h and then accustomed to drinking from a graded bottle during a 30-min test period for 4 days. On day 5, water was substituted with a 0.1% saccharin solution and, 30 min later, the animals received injections of vehicle, OEA (20 mg kg -1 ) or lithium chloride (0.4 M, 7.5 ml kg -1 ). During the following 2 days, water consumption was recorded over 30-min test periods. The animals were then presented with water or saccharin, and drinking was measured.

Operant responses for food
Rats were trained to press a lever for food on a ®xed ratio (FR) 1 schedule of reinforcement, while restricted to 20 g of food per rat per day. Once a stable response was achieved, the animals were trained to acquire an FR5, time out 2-min schedule of food reinforcement and kept with limited access to food. When a stable baseline was obtained, the animals were used to test the effects of vehicle or OEA (5 or 20 mg kg -1 ) administered 15 min before lever presentation. Tests lasted 60 min.

Other behavioural assays
The elevated plus maze test was conducted as described 26 after the administration of vehicle or OEA (20 mg kg -1 i.p.). Horizontal activity in an open ®eld and pain threshold in the hot plate test (55 8C) were measured 15 min after injection of vehicle or OEA (20 mg kg -1 ). Rectal temperature was measured with a digital thermometer.

In situ hybridization
We accustomed rats to the handling and injection procedure for 5 days. On day 6, we administered vehicle, OEA (10 mg kg -1 i.p.), or oleic acid (10 mg kg -1 ) and killed the rats 60 min later by decapitation under anaesthesia. In situ hybridization analyses were conducted using [ 35 S]-labelled antisense RNA probes for c-fos 28 and choline acetyl transferase 29 . Average hybridization densities were determined from at least three tissue sections per rat. Statistical signi®cance was evaluated using a one-way analysis of variance (ANOVA) followed by the Tukey±Kramer post-hoc test for paired comparisons.  Figure 5 Feeding regulates OEA biosynthesis in small intestine. Effects of free feeding, starvation and starvation±refeeding on OEA levels (a), NAT activity levels (b) and FAAH activity levels (c). Asterisk, P , 0.05; n = 3±5 per group.