The search for the palmitoylethanolamide receptor

Palmitoylethanolamide (PEA), the naturally occurring amide of ethanolamine and palmitic acid, is an endogenous lipid that modulates pain and inflammation. Although the anti-inflammatory effects of PEA were first characterized nearly 50 years ago, the identity of the receptor mediating these actions has long remained elusive. We recently identified the ligand-activated transcription factor, peroxisome proliferator-activated receptor-alpha (PPAR- a ), as the receptor mediating the anti-inflammatory actions of this lipid amide. Here we outline the history of PEA, starting with its initial discovery in the 1950s, and discuss the pharmacological properties of this compound, particularly in regards to its ability to activate PPAR- a . D 2005

underprivileged children with dried chicken egg yolk prevented recurrences of rheumatic fever, despite continued streptococcal infections (Coburn and Moore, 1943). In a continuation of this research, it was demonstrated that lipid fractions purified from egg yolk (Coburn et al., 1954;Long and Martin, 1956), as well as peanut oil and soybean lecithin (Long and Miles, 1950), exerted anti-allergic effects in the guinea pig (see Fig. 2 for a timeline of events). Soon thereafter, PEA (N-(2-hydroxyethyl)hexadecanamide, Npalmitoylethanolamine, LG 2110/1) was isolated as the agent responsible for these anti-inflammatory properties (Kuehl et al., 1957). This work ultimately led to its identification in mammalian tissues in 1964 (Bachur et al., 1965).

PEA in the clinic
The years following the discovery of PEA's anti-inflammatory effects (Benvenuti et al., 1968;Perlik et al., 1971) witnessed a series of interesting clinical studies on this compound, which was tested under the trade name of Impulsin by the Czech pharmaceutical company, SPOFA. It is not clear what prompted these clinical trials; nonetheless, PEA was found to reduce the severity and duration of symptoms caused by the influenza virus in school children and soldiers (Kahlich et al., 1979;Masek et al., 1974). The apparent success of these trials led to the clinical use of PEA in the former Czechoslovakia for acute respiratory diseases during the late 1970s. After several years on the market, PEA was withdrawn from clinical use for unknown reasons, which do not appear to be related to toxicity. PEA is currently used to restore skin reactivity in animals in a veterinary composition, marketed under the trade name RedonylR by the Italian company, Innovet. The termination of PEA in the clinic and the failure to identify its molecular target caused a period of research stasis that lasted more than 20 years. It was the discovery that another fatty acid amide, anandamide (arachidonoy- lethanolamide), is an endogenous agonist for cannabinoid receptors (Devane et al., 1992) that caused a renewal of interest in PEA.

Cannabinoid receptors
Cannabinoid receptors, the molecular targets of the psychoactive component in marijuana, delta-9tetrahydrocannabinol (THC), belong to the G-protein-coupled receptor super family (Matsuda et al., 1990;Munro et al., 1993). Activation of either CB 1 receptors (which are primarily localized to the brain but are also present in the periphery) or CB 2 receptors (which are predominantly found in cells of the immune system) results in G i -mediated reduction of adenylyl cyclase activity and subsequent inhibition of cAMP-mediated responses. In the brain, CB 1 receptors are mostly found on presynaptic endings of GABAergic and glutamatergic neurons (for a comprehensive review, see Piomelli, 2003). Behavioral effects produced by activation of central CB 1 receptors include reduced locomotion and body temperature, catalepsy, alterations in memory, sedation, and euphoria-effects that are also commonly observed with recreational marijuana use (Adams and Martin, 1996). In the periphery, activation of CB 1 receptors on nerve endings produces analgesia (Calignano et al., 1998;Jaggar et al., 1998), reduces cough (Calignano et al., 2000), and inhibits intestinal transit (Capasso et al., 2001). The functions served by CB 2 receptors are less clear, but recent evidence suggests that they may modulate immune cell function (Klein et al., 2001) to produce peripheral analgesia and anti-inflammation Malan et al., 2003;Quartilho et al., 2003)-although the mechanism by which these phenomena occur remains enigmatic.
Following the early experiments in the 1950s, a series of more recent studies has shown that PEA inhibits mast cell degranulation (Aloe et al., 1993;Mazzari et al., 1996) and pulmonary inflammation in mice (Berdyshev et al., 1998). These effects have been tentatively linked to reduced nitric oxide production by macrophages (Ross et al., 2000) and/or neutrophil influx (Farquhar-Smith and Rice, 2003). For example, in one study, animals treated with PEA for 3 days (10 mg kg À 1 , orally, once daily) following a carrageenan-induced inflammatory insult displayed significantly lower levels of inducible nitric oxide synthase (iNOS) and cyclooxygenase (COX-2) )-a genomic action reminiscent of steroidal anti-inflammatory drugs. Other studies have demonstrated that PEA reduces inflammation in a matter of hours LoVerme et al., 2005), implying that a nongenomic mechanism of action may also exist.
Less understood are the functions of PEA in the central nervous system (CNS), where PEA is present at high levels (Cadas et al., 1997). Increasing evidence points to an antiepileptic and neuroprotective action of PEA (Lambert et al., 2001;Sheerin et al., 2004). For example, in one study, intraperitoneal administration of the compound was shown to inhibit electroshock-induced and chemically induced seizures with a half-maximally effective dose (ED 50 ) of 9 mg kg À 1 , i.p. (Lambert et al., 2001). Other neuroprotective actions have also been reported; for example, in a separate study, PEA dose-dependently protected cultured mouse cerebellar granule cells from glutamate toxicity (Skaper et al., 1996). In yet Analgesia Acute pain Inflammatory pain Neuropathic pain SR144528 inhibitable

Anti-Inflammation
Mast cell activation iNOS expression COX-2 expression Neutrophil influx

Convulsions Excitotoxicity
Anti-viral Incidences of acute respiratory diseases another study, PEA reduced histamine-induced cell death in hippocampal cultures (Skaper et al., 1996). Finally, PEA was more recently shown to enhance microglial cell motility (Franklin et al., 2003).

PEA biosynthesis and inactivation
Unlike classical neurotransmitters and hormones which are stored in and released from intracellular secretory vesicles, the production of FAEs occurs through on-demand synthesis within the lipid bilayer (Cadas et al., 1996;Schmid et al., 1990). In mammalian tissues, two concerted and independent biochemical reactions are responsible (Fig. 4). The first is the transfer of a fatty acid from membranebound phospholipids to phosphatidylethanolamine (PE), catalyzed by a calcium ion and cyclic-AMP regulated N-acyltransferase (NAT), to form the FAE precursor N-acyl phosphatidylethanolamine (NAPE). Different FAE precursors are generated according to which fatty acid is initially transferred to PE (i.e., the initial transfer of palmitic acid will yield a PEA precursor, while that of arachidonic acid will yield an anandamide precursor). The second step in FAE synthesis is the cleavage of membrane-bound NAPE to release free PEA, which is mediated by a NAPE-specific phospholipase D (PLD). This lipid hydrolase shares little sequence homology to other members of the PLD family and recognizes multiple NAPE species, producing PEA along with other FAEs .
Alternatively, a separate mechanism of synthesis has been proposed involving a similar two-step reaction: (1) the hydrolysis of NAPE to N-palmitoyl-lysoPE (lyso-NAPE) by soluble phospholipase A 2 (sPLA 2 ) and (2) the subsequent cleavage of lyso-NAPE by a lysophospholipase D (lyso-PLD) (Natarajan et al., 1984). The activities of these two enzymes are highest in the stomach, brain, and testis (Sun et al., 2004). The relative contribution of each of these synthetic pathways is unknown at present.
FAAH, a membrane-bound intracellular serine hydrolase, for which PEA is an excellent substrate (Désarnaud et al., 1995;Hillard et al., 1995;Ueda et al., 1995), is present in all mammalian tissues, but is particularly abundant in brain and liver . In fact, mice lacking the faah gene have dramatically reduced PEA hydrolysis and increased PEA levels in brain and liver tissues Lichtman et al., 2002;Patel et al., 2004). In contrast to FAAH, PAA activity is most abundant in the rodent intestine, spleen, and lung (Ueda et al., 2001). PAA recognizes all FAEs, suggesting that it may play a broad role in the deactivation of these compounds by intact cells; however, in the presence of detergent, this activity displays a marked preference for PEA as a substrate (Ueda et al., 2001).

Physiological regulation of PEA levels
The physiological stimuli that regulate PEA levels in mammalian tissues are largely unknown; however, multiple studies indicate that this lipid accumulates during cellular stress, particularly following tissue injury. For example, PEA and other FAEs increase postmortem in the pig and mouse brain (Patel et al., 2004;Schmid et al., 1995). In rat testis, PEA levels were elevated nearly 40-fold following CdCl 2induced tissue degeneration (Kondo et al., 1998). Similar elevations of PEA have been observed in the ischemic heart (Epps et al., 1979) and brain (Franklin et al., 2003;Schabitz et al., 2002). Isolated cells, including brain neurons, macrophages (J774), basophils (RBL-2H3), and neuroblastoma (N 18 TG 2 ) cells, have also been reported to generate PEA in response to ionomycin or digestion with exogenous PLD (Bisogno et al., 1997;Cadas et al., 1997;Di Marzo et al., 1994). PEA levels also increase in response to ultraviolet-B irradiation in mouse epidermal (JB6 P + ) cells (Berdyshev et al., 2000). As discussed, most studies have focused on PEA production in vitro, or when performed in vivo, aggressive inflammatory agents were used, which in some cases induce extensive tissue degeneration (Kondo et al., 1998). In contrast, the regulation of PEA during inflammatory states is less clear. Capasso et al. (2001) demonstrated that during croton oil-induced intestine inflammation, levels of PEA significantly decreased in the mouse small intestine. While this manuscript was under review, Darmani et al. (2005) reported contrasting results demonstrating that PEA levels increased during neuroinflammatory conditions in both humans and rats. We found that phorbol esters/protein kinase C activators that exert profound proinflammatory effects decrease PEA levels in mouse abdominal skin. Using an established protocol (LoVerme et al., 2005;Sheu et al., 2002), we applied the phorbol ester 12-Otetradecanoylphorbol-13-acetate (TPA) (0.06% wt vol À 1 , 50 Al) in acetone to the abdomens of male CD1 mice and quantified PEA levels by high-performance liquid chromatography coupled to mass spectrometry (HPLC/MS) in inflamed skin 4 and 18 h following TPA administration. PEA levels in vehicle (acetone)-treated skin were comparable to those measured in untreated skin, whereas PEA levels in TPA-treated animals were twofold lower (Fig. 5). These results suggest that the reduction of PEA levels may contribute to the normal inflammatory response, albeit in an unknown capacity.

The PEA receptor
Despite its potential clinical significance, the cellular receptor responsible for the actions of PEA has long remained unidentified, and a great deal of controversy has surrounded its identity. The structural and functional similarities between anandamide and PEA first suggested that these two lipid mediators might share the same receptor. In support of this idea, PEA was initially reported to displace the binding of the high-affinity cannabinoid agonist [ 3 H]WIN55,212-2 with a half-maximal inhibitory concentration (IC 50 ) of 1.0 nM from RBL-2H3 cell membranes, which are known to express CB 2 mRNA (Facci et al., 1995). However, these results have not been subsequently replicated (Jacobsson and Fowler, 2001;Lambert et al., 2002;Lambert and Di Marzo, 1999;Showalter et al., 1996;Sugiura et al., 2000) and it is now generally accepted that PEA does not bind to CB 2 receptors. Adding to the PEA mystery, the CB 2 antagonist/inverse agonist SR144528 inhibits many, but not all, of the pharmacological actions of PEA (Calignano et al., 1998(Calignano et al., , 2001Conti et al., 2002;Jaggar et al., 1998). In particular, the analgesic actions of PEA are blocked by SR144528 (Calignano et al., 1998;Rice, 2001, 2003;Helyes et al., 2003), whereas its anti-peristaltic effects are not (Capasso et al., 2001). Further confounding the issue, it appears that the anti-inflammatory actions of PEA are only sensitive to SR144528 in acute models of inflammation Costa et al., 2002;LoVerme et al., 2005).
In light of these apparent discrepancies, several possible scenarios have been proposed to describe the mechanism of action of PEA (Fig. 6). The first possibility is that PEA binds to an unidentified receptor for which SR144528 is a functional antagonist (i.e., a compound that either directly interacts with the receptor, or indirectly blocks a downstream effector of this receptor (CB 2 receptors)). A second possibility is that PEA acts through a so-called bentourage effectQ (Ben-Shabat et al., 1998;Lambert and Di Marzo, 1999). According to this theory, PEA may increase anandamide levels in tissues by competing with this compound for FAAH-mediated hydrolysis, resulting in a potentiation of anandamide's actions. However, bentourage effectsQ have only been observed in vitro. In contrast, administration of PEA in vivo has the opposite result of decreasing endogenous anandamide levels (LoVerme et al., 2005). Finally, Di  suggested that in human breast cancer cells, PEA may work by inhibiting FAAH expression, which should lead to an increase in anandamide levels. This hypothesis does not account, however, for the very rapid actions of PEA, which occur within minutes of its administration.

PPAR-A is the anti-inflammatory target of PEA
Prior work in our laboratory has shown that OEA, a lipid amide structurally related to PEA (Fig. 1), regulates feeding behavior and lipid metabolism in rodents by activating the nuclear receptor peroxisome proliferator-activated receptor-a (PPAR-a) (Fu et al., 2003;Gaetani et al., 2003;Guzmán et al., 2004 Fig. 6. Proposed mechanisms of action for PEA. PEA may exert its effects through an unknown receptor, which may either be sensitive to SR144528 (1) or engage CB 2 receptors as downstream effectors (2). (3) PEA may act as an bentourage compound,Q effectively inhibiting FAAH activity, and thus increasing local anandamide levels (see text for further details). Rodríguez de Fonseca et al., 2001). A growing body of evidence has implicated PPAR-a in the control of inflammatory responses and these receptors are expressed in various cells of the immune system (for review, see Daynes and Jones, 2002). Mice lacking the gene encoding for PPAR-a display prolonged inflammatory responses (Devchand et al., 1996) and synthetic agonists of this receptor cause profound anti-inflammatory effects (Chinetti et al., 2000;LoVerme et al., 2005;Sheu et al., 2002). These effects are accompanied by reduced expression of inducible nitric oxide synthase (iNOS), cyclooxygenase 2 (COX-2), and various inflammatory cytokines, including interleukin-1h (IL-1h), prostaglandin E 2 (PGE 2 ), and tumor necrosis factor alpha (TNF-a)-effects reminiscent of those executed by PEA . In addition, PPAR-a activators have been shown to inhibit monocyte differentiation and neutrophil function, including endothelial extravasation. Although the precise mechanism of PPAR-a anti-inflammation is unclear, the receptor has been linked to a non-genomic inhibition of the proinflammatory signaling pathways mediated by nuclear factor nh (NF-nh) and activated protein-1 (AP-1). Thus, multiple lines of evidence suggest that PPAR-a receptors and their ligands are important modulators of the inflammatory process. Recent evidence has shown that, like its structural analogue OEA, PEA directly activates PPAR-a with a half-maximal effective concentration (EC 50 ) of approximately 3 AM (LoVerme et al., 2005)-a potency comparable to that of the synthetic PPAR-a agonist Wy-14643, which produces robust antiinflammatory actions (Sheu et al., 2002). However, unlike its structural analogue OEA, which activates both PPAR-a and PPAR-h/y (Fu et al., 2003), PEA does not engage either PPAR-h/y or PPAR-g (LoVerme et al., 2005). Moreover, when administered in a topical formulation to inflamed mouse skin, PEA inhibits inflammation and induces the expression of PPAR-a mRNA-an effect characteristic of synthetic high-affinity PPAR-a ligands (Fu et al., 2003;LoVerme et al., 2005).
Most importantly, PEA does not elicit anti-inflammatory effects in mutant PPAR-a-null mice (PPARa À/À mice). Indeed, when assessed in either the carrageen hindpaw or phorbol ester ear pinna tests, PEA reduced inflammation in wild-type, but not in PPAR-a À/À mice (LoVerme et al., 2005). Collectively, these findings indicate that PEA produces its anti-inflammatory actions by acting as a PPAR-a agonist (Fig. 7).

Macrophage, mast cell or keratinocyte?
Anti-inflammation PEA RA RXR Fig. 7. PEA activates PPAR-a to produce its anti-inflammatory effects. According to this hypothetical model, ligation of PPAR-a by PEA induces a heterodimerization event with ligated retinoic acid receptors (RXR) forming the activated receptor complex, which translocates to the nucleus to bind a PPAR-a response element (PPRE) and reduce inflammation.

Concluding remarks
The discovery that PPAR-a mediates the anti-inflammatory effects of PEA (LoVerme et al., 2005) raises several important questions. Is PPAR-a responsible for all other effects of PEA, including analgesia and neuroprotection? And if so, what is the relationship between PPAR-a and CB 2 receptors? How does endogenous PEA interact with PPAR-a to regulate physiological inflammatory processes? Our data suggest that part of this response may involve a reduction in PEA levels in inflamed tissues, which might in turn serve to remove an endogenous anti-inflammatory tone. Alternatively, it has been reported that PEA levels may increase during inflammation to produce local anti-inflammatory and analgesic actions (Darmani et al., 2005). These possibilities remain, however, to be explored. A final question is whether non-cannabinoid FAEs, such as stearoylethanolamide (Maccarrone et al., 2002;Terrazzino et al., 2004), also activate PPAR-a or related transcription factors. As these questions are progressively answered, we may not only gain insights on the PEA signaling system, but also identify new potential targets for analgesic and anti-inflammatory medicines.