Pharmacological Characterization of Hydrolysis-Resistant Analogs of Oleoylethanolamide with Potent Anorexiant Properties

Oleoylethanolamide (OEA) is an endogenous lipid mediator that reduces food intake, promotes lipolysis, and decreases body weight gain in rodents by activating peroxisome proliferator-activated receptor-α (PPAR-α). The biological effects of OEA are terminated by two intracellular lipid hydrolase enzymes, fatty-acid amide hydrolase and N-acylethanolamine-hydrolyzing acid amidase. In the present study, we describe OEA analogs that resist enzymatic hydrolysis, activate PPAR-α with high potency in vitro, and persistently reduce feeding when administered in vivo either parenterally or orally. The most potent of these compounds, (Z)-(R)-9-octadecenamide,N-(2-hydroxyethyl,1-methyl) (KDS-5104), stimulates transcriptional activity of PPAR-α with a half-maximal effective concentration (EC50) of 100 ± 21 nM (n = 11). Parenteral administration of KDS-5104 in rats produces persistent dose-dependent prolongation of feeding latency and postmeal interval (half-maximal effective dose, ED50 = 2.4 ± 1.8 mg kg-1 i.p.; n = 18), as well as increased and protracted tissue exposure compared with OEA. Oral administration of the compound also results in a significant tissue exposure and reduction of food intake in free-feeding rats. These results suggest that the endogenous high-affinity PPAR-α agonist OEA may provide a scaffold for the discovery of novel orally active PPAR-α ligands.

The nuclear receptor peroxisome proliferator-activated receptor-␣ (PPAR-␣) is a key regulator of lipid metabolism and energy balance in mammals (Desvergne and Wahli, 1999). Although OEA may bind to multiple receptors (Wang et al., 2005;Overton et al., 2006), three distinct lines of evidence indicate that PPAR-␣ mediates the satiety-inducing effects of this compound. First, OEA binds with high affinity to the purified ligand-binding domain (LBD) of mouse and human PPAR-␣ (K D 37 and 40 nM, respectively) and activates with high-potency PPAR-␣-driven transactivation in a heterologous expression system (EC 50 ϭ 120 nM) (Fu et al., 2003). Second, two structurally different synthetic PPAR-␣ agonists, the compounds GW7647 and Wy-14643, exert anorexiant effects that are due to a selective increase in feeding latency and thus are behaviorally identical to those produced by OEA (Fu et al., 2003). Third, mutant mice in which the PPAR-␣ gene has been deleted by homologous recombination (PPAR-␣ Ϫ/Ϫ mice) do not respond to OEA or synthetic PPAR-␣ agonists, although they retain normal responses to two mechanistically different anorexiants, cholecystokininoctapeptide and fenfluramine (Fu et al., 2003). Together, the findings outlined above suggest that endogenous OEA, produced in the proximal small intestine during feeding, regulates between-meal satiety by activating PPAR-␣.
The biological deactivation of OEA is not fully understood but is likely to involve the enzymatic hydrolysis of this lipid amide to oleic acid and ethanolamine (Lo Verme et al., 2005b). Two structurally distinct OEA-hydrolyzing enzymes have been characterized: fatty-acid amide hydrolase (FAAH), an intracellular membrane-bound serine hydrolase (Désarnaud et al., 1995;Hillard et al., 1995;Ueda et al., 1995;Cravatt et al., 1996), and N-acylethanolamine-hydrolyzing acid amidase, a lysosomal cysteine hydrolase Ueda et al., 2005). In the present study, we set out to develop analogs of OEA that resist enzymatic hydrolysis while retaining agonist activity at PPAR-␣. Such agents may not only be useful to characterize the pharmacological properties of OEA in live animals but might also provide a starting point for the development of novel antiobesity, antihyperlipidemic, and anti-inflammatory drugs (Willson et al., 2000).

Materials and Methods
Animals. Adult male Wistar rats (250 -300 g, 7 weeks old) and C57/BL6 mice (20 -25 g, 8 weeks old) were housed in standard Plexiglas cages at room temperature. A 12-h light/dark cycle was set with the lights on at 4:45 AM. Water and standard chow pellets (Prolab RMH 2500; PMI Nutrition International, Brentwood, MO) were available ad libitum. All procedures met the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of the University of California, Irvine.
Drugs and Solvents. URB597 was synthesized as described previously (Mor et al., 2004). Solvents were obtained from Burdick and Jackson (Muskegon, MI), and all other chemicals were obtained from Sigma-Aldrich Co. (St. Louis, MO).
Chemical Analyses. Melting points were determined on a Buchi SMP-510 capillary melting point apparatus (Flawil, Switzerland) and were uncorrected. 1 H NMR spectra were recorded on a Bruker Avance 200 MHz spectrometer (Bruker, Newark, DE); chemical shifts are reported in parts per million relative to the central peak of the solvent. Coupling constants (J values) are given in Hertz. EI-MS spectra (70 eV) were taken on a Fisons Trio 1000 spectrometer (Fisons Instruments, Manchester, UK). Infrared spectra were obtained on a Nicolet Avatar 360 Fourier-Transfor-IR spectrometer (ThermoNicolet, Madison, WI); absorbance is reported in centimeter Ϫ1 . Optical rotatory powers were obtained on a PerkinElmer 241 polarimeter (PerkinElmer Life and Analytical Sciences, Boston, MA). Elemental analyses for C, H, and N were performed on a Carlo Erba analyzer and were within Ϯ0. In Vitro Enzymatic Hydrolysis. We incubated test compounds (10 nmol) in 0.1 ml of Tris buffer, pH 8.2, containing mouse liver homogenate (0.7 mg of protein) and 0.05% bovine serum albumin (fatty acid-free; Sigma-Aldrich Co.) for 0 to 60 min at 37°C. Samples were subjected to protein precipitation with 0.25 ml of acetonitrile containing [ 2 H 4 ]OEA as an internal standard. We prepared [ 2 H 4 ]OEA by the reaction of oleoyl chloride with a 10-fold molar excess of [ 2 H 4 ]ethanolamine (Cambridge Isotope Laboratories, Andover, MA). The reaction was conducted in dichloromethane at 0 -4°C for 15 min, with stirring. The product was washed with water, dehydrated over sodium sulfate, filtered, and dried under N 2 . The samples were mixed and centrifuged, and the supernatants were transferred to a 96-well plate for LC tandem MS analyses. LC was carried out on a Waters 2790 Alliance system (Waters, Milford, MA). Separation was performed on a Synergi Polar-RP column (2 ϫ 150 mm, 4 ; Phenomenex, Torrance, CA) using an isocratic mobile phase of acetonitrile/water/formic acid (80:20:0.1 v/v) at a flow rate of 0.3 ml min Ϫ1 and a column temperature of 45°C. The LC system was interfaced with a Micromass Quattro Ultima tandem MS (Beverly, MA). The samples were analyzed using an electrospray probe in the positive ionization mode with cone voltage set at 40 V and capillary voltage at 3.2 kV. The source and desolvation temperatures were 130 and 500°C, respectively. The voltage of the collision-induced dissociation chamber was Ϫ20 eV. Multiple reaction monitoring was used for the detection of 1 (OEA) and 2 (m/z 326 Ͼ 62), 3 (KDS-5104), 4 and 5 (m/z 340 Ͼ 76), 6 (m/z 326 Ͼ 254), and [ 2 H 4 ]OEA (internal standard) (m/z 330 Ͼ 66).
In Vitro PPAR Activation. We used HeLa cells stably transfected with a luciferase reporter plasmid containing a hygromycin resistance gene. These cells were cotransfected with a plasmid containing the LBD of PPAR-␣ or PPAR-␥ fused to the DNA-binding domain of the yeast regulatory protein GAL4 and a neomycin resistant gene under the control of human cytomegalovirus promoter (Fu et al., 2003). The cells were cultured in Dulbecco's modified Eagle's medium and supplemented with fetal bovine serum (10%), hygromycin, and G418. For transactivation assays, we seeded cells in six-well plates and incubated them for 7 h in Dulbecco's modified Eagle's medium containing an appropriate concentration of test compounds. , n ϭ 5-12) were assessed in gene activation assays. b ED 50 values (ϮS.E.M., n ϭ 6 -18) were assessed in feeding assay (using latency of feeding onset as variable).

Hydrolysis-Resistant Analogs of Oleoylethanolamide
We used a luciferase reporter assay system (Promega, Madison, WI) and a MIX Microtiter plate luminometer (Dynex, Chantilly, VA) to determine luciferase activity in the cell lysates. Pharmacokinetic Studies. For parenteral administration, we dissolved OEA and KDS-5104 in dimethyl sulfoxide/saline (70:30, v/v) and administered them (10 mg kg Ϫ1 i.p.) to free-feeding rats. After administrations, we anesthetized rats with halothane and collected tissues at various time points. Tissue samples were subjected to lipid extraction as reported previously (Giuffrida et al., 2000), and [ 2 H 4 ]OEA was added as an internal standard. OEA and KDS-5104 were separated on an 1100-LC system coupled to a 1946A-MS detector (Agilent Technologies, Palo Alto, CA) equipped with an electrospray ionization interface. We used a XDB Eclipse C 18 column (50 ϫ 4.6 mm i.d., 1.8 m; Zorbax; Agilent Technologies) eluted with a rapid gradient of methanol in water (from 85 to 90% methanol in 2.5 min) at a flow rate of 1.5 ml/min. Column temperature was kept at 40°C. MS detection was in the positive ionization mode, capillary voltage was set at 3 kV, and fragmentor voltage was varied from 120 to 140 V. N 2 was used as drying gas at a flow rate of 13 l/min and a temperature of 350°C. Nebulizer pressure was set at 60 p.s.i. We quantified the test compounds monitoring sodium adducts of the molecular ions [MϩNa] ϩ in the selected ion-monitoring mode. For oral administration, we suspended the drugs in water/ carboxymethyl cellulose/Tween 80 (99.1:0.5:0.4, v/v) and administered them (100 mg kg Ϫ1 , gavage) to free-feeding rats. Rats were then anesthetized with halothane at various time points to collect tissues. Lipids were extracted and analyzed by LC/MS/MS as described under In Vitro Enzymatic Hydrolysis.
Behavioral Studies. Food intake and motor activity were recorded using a fully automated system (Technical and Scientific Equipment, Bad Homburg, Germany). The feeding station consists of a food container fixed to a sensor. The amount of food consumed was monitored every second, and the threshold for an eating episode was set at 0.5 g and Ͼ1 min. The motility system consists of 2 ϫ 6 infrared light-barriers per cage disposed at right angles on the x-y axes to determine the animal's center of gravity and its displacement over time. Rats were habituated to test cages for 3 days prior to trials. We administered drugs parenterally (5, 10, and 20 mg kg Ϫ1 i.p.) in a vehicle of dimethyl sulfoxide/sterile saline (70:30, v/v) 15 min before food presentation or orally (100 mg kg Ϫ1 ) in a vehicle of water/carboxymethyl cellulose/Tween 80 (99.1:0.5:0.4, v/v), 60 min before food presentation. Monitoring began at the onset of the dark phase (4:45 PM) and lasted 24 h. Recorded data were analyzed as food ingested per kilogram of body weight and motility counts on XY sensors. A detailed analysis of the meal patterns was performed adopting a minimal PMI of 10 min (Burton et al., 1981). The following meal parameters were analyzed (Reidelberger et al., 2001): latency of feeding onset (minutes), the time interval from trial inception to the first meal episode; first meal size (gram kilogram Ϫ1 ), amount of food consumed during the first meal; first PMI (minute), the time interval between the end of the first meal and the beginning of the second meal; and second PMI, the time interval between the end of the second meal and the beginning of the third meal. ED 50 values for feeding latency data were measured graphically.
Docking and Molecular Dynamics Simulations. Molecular modeling studies were performed with Sybyl 6.9 (Tripos Inc., St. Louis, MO) and with Autodock 3.0.5 (The Scripps Research Institute, La Jolla, CA). The three-dimensional structure of PPAR-␣ LBD was obtained from the Protein Data Bank [1K7L entry, chain C (Xu et al., 2001)], and was prepared for the docking experiments by deleting the cocrystallized agonist GW409544 and water molecules, adding polar hydrogens, checking tautomeric and protonation states, and calculating Kollman-united charges. The structure of KDS-5104 was built by the standard sketch options of Sybyl, and Gasteiger-Hü ckel charges were calculated. Docking experiments were performed within a region of space comprising the whole binding site starting from different conformations of KDS-5104; all of the rotatable bonds of the ligand were free to move, with the exception of the amide bond, which was fixed in the antiposition. A genetic algorithm-local search procedure (Morris et al., 1998) was applied, setting the maximal number of energy evaluations to 1,500,000 and the maximal number of generations to 300,000. For each docked conformation, the estimated free energy of binding was calculated by the standard Autodock scoring function. The conformations with binding energy lower than Ϫ7.00 kcal/mol were inspected to select those in which the three polar groups of KDS-5104 interacted with the protein. These conformations were inserted into the PPAR-␣ LBD, and the complexes were subjected to geometry optimization and molecular dynamics (MD) simulations. Energy minimization was performed with the MMFF94s force-field (Halgren, 1999) to an energy gradient of 0.05 kcal (mol Å) Ϫ1 . MD simulations were performed with the MMFF94 force-field (Halgren, 1996) for 300 ps at 310 Kelvins applying a time step of 1 fs. The average structure obtained from the last 200 ps of MD simulation was retrieved and submitted to geometry optimization. During energy minimization and MD simulations, only KDS-5104 and the amino acid side chains within 12 Å from the ligand were allowed to move.
Statistical Analyses. Biochemical data were analyzed by twoway ANOVA followed by Bonferroni's test as post hoc. Behavioral data were analyzed by one-way ANOVA followed by Dunnett's test. All statistical analyses were conducted using Prism software version 4.0 (GraphPad Software Inc., San Diego, CA), and differences were considered significant if P Ͻ 0.05.

Synthesis of Hydrolysis-Resistant OEA Analogs.
To develop hydrolysis-resistant analogs of OEA, we modified the chemical scaffold of this fatty-acid ethanolamide in three points (Table 1). Firstly, we replaced the oleic acid chain, which contains a Z double bond in the ⌬ 9 position, with an elaidic acid chain, in which the ⌬ 9 olefin has an E configuration (compound 2). Secondly, we increased the steric bulk around the amide bond by introducing a methyl group either on the ethanolamine ␣-carbon (stereoisomers 3-4) (Lin et al.,  1998) or the amide nitrogen (5). Finally, we inverted the natural orientation of the carboxamide moiety to produce the reverse amide 6. Chemical syntheses and analyses are described under Materials and Methods. We evaluated the abil-ity of compounds 1 to 6 to resist degradation by mouse liver homogenates, which contain significant amounts of the OEAhydrolyzing enzymes FAAH and N-acylethanolamine-hydrolyzing acid amidase (Désarnaud et al., 1995;Cravatt et al., 1996;Ueda et al., 1999;Tsuboi et al., 2005). As shown in Fig.  1a, incubation with liver homogenates resulted in the rapid degradation of OEA and its geometric isomer 2 but not the methyl-substituted derivatives 3 (KDS-5104), 4, and 5, or the reverse amide 6. Median times to reach maximal hydrolysis (t 1/2 ) were 14 Ϯ 3 min for OEA and 35 Ϯ 2 min for compound 2 (n ϭ 6). Incubation of liver homogenates with the highly selective FAAH inhibitor URB597 (Mor et al., 2004) (1 M) decreased OEA hydrolysis by 85.9 Ϯ 0.7% (n ϭ 6), indicating that this reaction is predominantly catalyzed by FAAH.
Molecular modeling studies suggest the existence of multiple potential binding modes of KDS-5104 onto the structure of the PPAR-␣ LBD. One such mode assigned the acyl chain of KDS-5104 to the same hydrophobic region occupied by synthetic agonists in crystals of the PPAR-␣ LBD (Xu et al., 2001;Cronet et al., 2001) (Fig. 2). For this solution, the hydroxyl group of KDS-5104 is close to the carboxyl group of Glu282, the amide CO interacts with the backbone NH of Tyr334, and the amide NH is hydrogen-bonded to the hydroxyl group of Thr279. Thr279 also donates a hydrogen bond to the hydroxyl group of KDS-5104 (Fig. 2). This binding mode is consistent with the small set of structure-activity relationship data obtained from OEA derivatives. In fact, it requires that both the amide NH and CO fragments and the ending OH group participate in the interaction in a concerted arrangement, which can explain the loss of activity observed for the N-methyl derivative 5 and the retroamide 6. Moreover, whereas the position and interactions of compound 3 were stable during MD simulation, the less potent (S) enantiomer 4 could not maintain the same interactions and moved

Hydrolysis-Resistant Analogs of Oleoylethanolamide
away from the initial position (data not shown). However, in the absence of crystallization data, this binding mode must be considered tentative. Initial Pharmacokinetic Assessment. The pharmacological effects of OEA are curtailed by a rapid clearance Oveisi et al., 2004). To begin to characterize the pharmacokinetic properties of KDS-5104, we examined the distribution of this compound in plasma and liver following parenteral administration in rats. KDS-5104 (10 mg kg Ϫ1 i.p.) reached maximal plasma levels (C max ) 15 min after injection and was slowly cleared from the circulation. Its C max in plasma was ϳ2-fold higher than that of OEA (10 mg kg Ϫ1 i.p.), and the area under the curve (AUC) was approximately 5 times greater ( Fig. 3a; Table 2). An even more striking difference between the two compounds was observed in liver tissue, where the C max of KDS-5104 was 13 times higher than that of OEA and its AUC was 23 times greater ( Fig. 3b; Table 2). We next determined the distribution of KDS-5104 (100 mg kg Ϫ1 ) after oral administration in free-feeding rats. The compound exhibited significantly higher C max and AUC values in plasma and liver than did OEA ( Fig. 4; Table 3).
Behavioral Studies. Because OEA inhibits food intake in rodents, we examined the effects of analogs 2 to 6 on feeding behavior after i.p. administration in free-feeding rats. Analysis of meal parameters revealed that OEA and KDS-5104 produced a dose-dependent increase in the latency of feeding onset, whereas the reverse amide 6 and the tertiary amide 5 were inactive (Fig. 5 and data not shown). The ED 50 values for all compounds, assessed on feeding latency, are reported in Table 1. Two-way ANOVA analysis revealed a significant difference between OEA and KDS-5104 treatment (P Ͻ 0.001, F 1/68 ϭ 15.58) and a significant difference between 10 mg kg Ϫ1 doses of the two compounds (P Ͻ 0.01, Bonferroni's post hoc test). Moreover, KDS-5104, but not OEA, caused a significant prolongation of both first and second PMI, which is suggestive of a persistent satiety-inducing action (Fig. 5). Oral administration of KDS-5104 (100 mg kg Ϫ1 ) resulted in a significant reduction in food intake in free-feeding rats, which lasted for 4 h following food presentation (Fig. 6, a and  b). By contrast, no such effect was observed with the same oral dose of OEA (Fig. 6, a and b). Orally administered KDS-5104 or OEA exerted no significant effect on motor activity (Fig. 6, c and d).

Discussion
In the present study, we describe a new series of hydrolysis-resistant analogs of the endogenous PPAR-␣ ligand OEA, some of which display potent agonist activity at PPAR-␣. Our synthetic approach focused on structural modifications of OEA that were expected to result in increased resistance to enzymatic hydrolysis based on previous studies with other lipid ethanolamides (Désarnaud et al., 1995;Khanolkar et al., 1996;Vandevoorde et al., 2003). We specifically modified three substructures of OEA-the fatty-acid chain, carboxamide group, and ethanolamine moiety-while leaving the general template of this natural ligand unaltered. As anticipated from our design, OEA analogs with increased steric bulk around the amide bond (compounds 3-5) or inverted orientation of the carboxamide moiety (6), resisted hydrolysis by mouse liver homogenates. Among these compounds, the two C-␣ methyl stereoisomers 3 (KDS-5104) and 4 engaged PPAR-␣ with high potency in a reporter gene assay, whereas the tertiary amide 5 and the reverse amide 6 were inactive. It is noteworthy that the (R)-methyl isomer KDS-5104 was three times more potent at activating PPAR-␣ than was enantiomer 4 (EC 50 100 Ϯ 21 and 333 Ϯ 27 nM, respectively). We interpret these results to indicate that first the insertion of a methyl group on the C-␣ of ethanolamine protects OEA analogs from catalytic hydrolysis and second the presence of a free amide group in a precise orientation is essential for PPAR-␣ activation by this class of molecules.
Administration of OEA to live animals results in a variety of PPAR-␣-mediated effects, which include prolongation of feeding latency (after i.p. injection) (Rodríguez de Fonseca et al., 2001) and PMI (after oral administration of gastroprotected capsules) , increased fatty-acid mobilization (Guzman et al., 2004), and reduced inflammation (after topical application) (Lo Verme et al., 2005a). However, many of these effects are rather short-lived, presumably on account of the liability of OEA to catalytic hydrolysis. The prolonged tissue exposure and duration of action of KDS-5104, which is resistant to catalytic hydrolysis in vitro, strongly supports this possibility. Furthermore, the in vivo anorexiant potency of KDS-5104 and indeed of all of the compounds tested in the present study is generally well correlated with their in vitro agonist potency at PPAR-␣, providing further evidence for a role of this receptor in the biological actions of OEA.

568
tions in plasma and liver and was pharmacologically active following oral administration in free-feeding rats. These results confirm that resistance to enzymatic hydrolysis is an important requisite for the design of orally active OEA analogs, which mimic the anorexiant actions of this endogenous PPAR-␣ agonist. Future studies will need to provide a more complete evaluation of the pharmacological properties of KDS-5104. It will be particularly important first to determine whether chronic administration of KDS-5104 exerts weight-and lipid-lowering effects similar to those of OEA (Rodríguez de Fonseca et al., 2001;Fu et al., 2003;Guzman et al., 2004;Lo Verme et al., 2005b) and second to characterize the impact of this compound on peroxisome proliferation, a prominent side effect of synthetic PPAR-␣ agonists in rodents (but not humans) (Klaunig et al., 2003).
In conclusion, our findings define a new class of potent orally active PPAR-␣ agonists designed on the scaffold of the endogenous ligand OEA. These agents provide a new set of molecular tools to investigate the pleiotropic functions of PPAR-␣ in feeding behavior, energy balance, and inflammation. Fig. 6. a and b, effects of OEA (analog 1) and KDS-5104 (analog 3) on cumulative food intake 1 (a) and 4 h (b) after food presentation. Effects of OEA (analog 1) and KDS-5104 (analog 3) on locomotor activity 1 (c) and 4 h (d) after food presentation. Compounds were administered by gavage in free-feeding rats at a dose of 100 mg kg Ϫ1 (n ϭ 8). ‫,ء‬ P Ͻ 0.05 versus vehicle. Fig. 5. a to c, effects of OEA (analog 1) (a), KDS-5104 (analog 3) (b), and compound 6 (c) on latency of feeding onset (left), PMI after the first meal (PMI-1, center), and PMI after the second meal (PMI-2, right). Compounds were administered by i.p. injection in free-feeding rats at doses of 5 to 20 mg kg Ϫ1 (n ϭ 6 -18). ‫,ء‬ P Ͻ 0.05; and ‫,ءء‬ P Ͻ 0.01 versus vehicle.