logy 565 (2007) 26–36www.elsevier.com/locate/ejphar

European Journal of Pharmaco

Inhibition of fatty acid amide hydrolase, a key endocannabinoid metabolizingenzyme, by analogues of ibuprofen and indomethacin

Sandra Holt a,1, Ben Paylor a, Linda Boldrup a,2, Kirsi Alajakku a, Séverine Vandevoorde a,3,Anna Sundström a, Maria Teresa Cocco b, Valentina Onnis b, Christopher J. Fowler a,⁎

a Department of Pharmacology and Clinical Neuroscience, Umeå University, SE-901 87 Umeå, Swedenb Dipartimento di Tossicologia, Università degli Studi di Cagliari, Cagliari, Italy

Received 20 November 2006; received in revised form 5 February 2007; accepted 26 February 2007Available online 7 March 2007

Abstract

There is evidence in the literature that the nonsteroidal anti-inflammatory drugs indomethacin and ibuprofen can interact with the cannabinoidsystem both in vitro and in vivo. In the present study, a series of analogues of ibuprofen and indomethacin have been investigated with respect totheir ability to inhibit fatty acid amide hydrolase, the enzyme responsible for the hydrolysis of the endogenous cannabinoid anandamide. Of thefourteen compounds tested, the 6-methyl-pyridin-2-yl analogue of ibuprofen (“ibu-am5”) was selected for further study. This compound inhibitedrat brain anandamide hydrolysis in a non-competitive manner, with IC50 values of 4.7 and 2.5 μM being found at pH 6 and 8, respectively. Bycomparison, the IC50 values for ibuprofen were 130 and 750 μM at pH 6 and 8, respectively. There was no measurable N-acylethanolaminehydrolyzing acid amidase activity in rat brain membrane preparations. In intact C6 glioma cells, ibu-am5 inhibited the hydrolysis of anandamidewith an IC50 value of 1.2 μM. There was little difference in the potencies of ibu-am5 and ibuprofen towards cyclooxygenase-1 and -2 enzymes,and neither compound inhibited the activity of monoacylglycerol lipase. Ibu-am5 inhibited the binding of [3H]-CP55,940 to rat brain CB1 andhuman CB2 cannabinoid receptors more potently than ibuprofen, but the increase in potency was less than the corresponding increase in potencyseen for inhibition of FAAH activity. It is concluded that ibu-am5 is an analogue of ibuprofen with a greater potency towards fatty acid amidehydrolase but with a similar cyclooxygenase inhibitory profile, and may be useful for the study of the therapeutic potential of combined fatty acidamide hydrolase–cyclooxygenase inhibitors.© 2007 Elsevier B.V. All rights reserved.

Keywords: Anandamide; Fatty acid amide hydrolase; Cyclooxygenase; Nonsteroidal anti-inflammatory drug; Cannabinoid receptor

1. Introduction

There is evidence indicating that the cannabinoid system cancontribute to the in vivo pharmacological effects of nonsteroidalanti-inflammatory drugs (NSAIDs). Thus, the effects of spinallyadministered indomethacin in the formalin test of inflammatorypain are blocked by the CB1 cannabinoid receptor antagonist

⁎ Corresponding author.E-mail address: cf@pharm.umu.se (C.J. Fowler).

1 Present address: Medical Products Agency, SE-751 03 Uppsala, Sweden.2 Present address: Department of Medical Biosciences, Umeå University, SE-

901 87 Umeå, Sweden.3 Present address: Unité de Chimie pharmaceutique et de Radiopharmacie,

Université catholique de Louvain, Avenue Mounier, 73, UCL-CMFA 73.40, B-1200 Brussels, Belgium.

0014-2999/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.ejphar.2007.02.051

AM251, and are not seen in CB1-receptor knockout mice(Gühring et al, 2002). A similar result was seen for intrathecallyadministered flurbiprofen (Ates et al., 2003). Locally adminis-tered ibuprofen produces synergistic effects with the endoge-nous cannabinoid anandamide (arachidonoylethanolamide,AEA) in the formalin test, the synergy being prevented byAM251 (Guindon et al., 2006a).

The mechanisms involved in these effects are as yet unclear.Guindon et al. (2006b) reported that the combination of locallyadministered ibuprofen and anandamide produced increases inthe paw levels of AEA and the related N-acylethanolaminespalmitoylethanolamide and oleoylethanolamide over andabove those seen with either compound per se. Similar results(and a similar synergistic effect) were seen with the cyclo-oxygenase-2 (COX-2) inhibitor rofecoxib (Guindon et al.,

27S. Holt et al. / European Journal of Pharmacology 565 (2007) 26–36

2006b). Gühring et al. (2002) and Ates et al. (2003) suggestedthat indomethacin and flurbiprofen may allow for an increasedsynthesis of endocannabinoids from arachidonic acid byblocking the cyclooxygenase (COX) catalyzed removal ofarachidonic acid.

A key pathway for removal of anandamide and other N-acyl-ethanolamines such as palmitoylethanolamide (PEA) is theenzyme fatty acid amide hydrolase (FAAH). Loss of activity ofthis enzyme, either by genetic deletion or by the use of selectiveinhibitors such as URB597 leads to increased levels of N-acyl-ethanolamines and potentially beneficial effects in models ofinflammation and inflammatory pain (see e.g. Lichtman et al.,2004; Holt et al., 2005; Jayamanne et. al., 2006). In 1996, Pariaet al. reported that indomethacin reduced the rate of anandamidehydrolysis in the mouse uterus (Paria et al., 1996). Moresystematic studies have since found thatmanyNSAIDs, includingindomethacin and ibuprofen, inhibit the activity of FAAH (Fowleret al., 1997, 1999, 2003), particularly at low pH (Holt et al., 2001;Holt and Fowler, 2003; Fowler et al., 2003), such as is seen ininflamed tissue (Häbler, 1929; Andersson et al., 1999). AEA isalso a substrate for COX-2 (see Kozak andMarnett, 2002), and soit is possible that the effects of NSAIDs upon inflammation andinflammatory pain may not only be due to a decreased productionof prostaglandin, but also a reduced removal of AEA (and otheranti-inflammatory N-acylethanolamines) by COX-2 and/orFAAH (Holt et al., 2005; Guindon et al., 2006b). Certainly,when it is considered that AEA and PEA can also interact withother targets that are involved in the inflammation and pain, suchas TRPV1 receptors (Zygmunt et al., 1999) and the peroxisomeproliferator-activated receptorγ andα isoforms (Bouaboula et al.,2005; Lo Verme et al., 2005), and that the relative roles played byCB- and TRPV1-meditated responses to AEA are different innormal and inflamed tissue (Singh Tahim et al., 2005), thesituation becomes very complicated indeed.

The pharmacological properties of FAAH inhibitors, togetherwith the potential role of FAAH in the action of NSAIDs raise thepossibility that compounds with dual actions upon FAAH andCOX-2 may be useful novel compounds for the treatment ofinflammation and inflammatory pain. An obvious template forthe identification of such compounds is the NSAIDs themselves.However, structure–activity relationships of analogues ofclinically used NSAIDs with respect to FAAH inhibition havenot been examined in the literature, in contrast to the situation forthe relative COX-1 and -2 potencies of such compounds (see e.g.Kalgutar et al., 2000a,b). In consequence, in the present study, aseries of ibuprofen and indomethacin analogues have beeninvestigated with respect to their ability to inhibit FAAH. Theibuprofen analogues have not previously been characterizedbiochemically, but have been shown to produce analgesic effectscomparable to ibuprofen in the acetic acid writhingmodel, and insome cases to possess considerably less ulcerogenic properties(Cocco et al., 2003). Five of the indomethacin analogues testedshow selectivity for COX-2 vs. COX-1 (Kalgutar et al., 2000a,b)but have not been tested vis-à-vis the endocannabinoid system.Finally, indomethacin morpholinamide has been reported to be aCB2 receptor-selective inverse agonist (Gallant et al., 1996; Newand Wong, 2003).

2. Methods

2.1. Materials

Anandamide [ethanolamine-1-3H] (specific activity 60 Ci/mmol), anandamide [arachidonoyl 5,6,8,9,11,12,14,15-3H](specific activity 200 Ci/mmol), palmitoylethanolamide [etha-nolamine-1-3H] ([3H]-PEA; specific activity 20 Ci/mmol) and2-oleoylglycerol [glycerol-1,2,3-3H] (specific activity 20 Ci/mmol) were purchased from American Radiolabeled Chemi-cals, Inc. St. Louis, MO, U.S.A. [3H]-CP55,940 ((−)-cis-3-[2-hydroxy-4-(1,1-dimethylheptyl)phenyl]-trans-4-(3-hydroxy-propyl) cyclohexanol) [side chain 2,3,4 (N)-3H] (specificactivity 160.6 Ci/mmol) was obtained from Perkin Elmer Lifeand Analytical Sciences, Boston, MA, USA). Rat C6 gliomaand mouse P19 embryonic carcinoma cells were obtained fromthe European Collection of Cell Cultures, Porton Down, U.K.Phenyl sepharose was obtained from Amersham BiosciencesAB (Uppsala, Sweden). The ibuprofen analogues were syn-thesized by co-authors Cocco and Onnis (Cocco et al., 2003).Non-radioactive anandamide, indomethacin ester n-heptyl,indomethacin N-octyl amide, N-(2-phenylethyl)-indomethacinamide, N-(4-acetamidophenyl)-indomethacin amide, methylarachidonoyl fluorophosphonate (MAFP), URB597 (3′-carba-moyl-biphenyl-3-yl-cyclohexyl-carbamate), COX activityassay kits and recombinant human monoacylglycerol lipasewere purchased from Cayman Chemical Company, Ann Arbor,MI, U.S.A. Ibuprofen, (R)-ibuprofen, indomethacin, non-radio-active 2-oleoylglycerol and active charcoal were obtained fromSigmaAldrich, St. Louis, MO, U.S.A. CP55,940, WIN55,212-2((R)-(+)-[2,3-dihydro-5-methyl-3-(4-morpholinylmethyl) pyrrolo[1,2,3-de]-1,4-benzoxazin-6-yl]-1-naphthalenylmethanonemesylate) and HU210 ((6aR)-trans-3-(1,1-dimethylheptyl)-6a,7,10,10a-tetrahydro-1-hydroxy-6,6-dimethyl-6H-dibenzo[b,d]pyran-9-methanol) were obtained from Tocris Bioscience(Ellisville, MO, USA). Fatty acid-free bovine serum albuminand indomethacin ester 4-methoxyphenyl were purchased fromCalbiochem, San Diego, CA, U.S.A. Human cannabinoid CB2

receptor membrane preparations were obtained from Bio-Xtal(Mundolsheim, France) via the Axxora platform (SwedishDistributor In vitro Sweden AB, Stockholm, Sweden). Testcompounds were dissolved in ethanol, with the exception ofCP55,940, WIN55,212-2 and HU210, which were dissolved indimethylsulfoxide (DMSO). Solvent concentrations were keptconstant throughout the assays.

2.2. Preparation of brain membrane and cytosolic fractions

Membrane preparations were obtained by homogenizingwhole brains (minus cerebella) from adult Wistar or Sprague–Dawley rats, or whole brains from BALB/cJBom mice in20 mM HEPES, 1 mM MgCl2, pH 7.0. The homogenates werecentrifuged at ∼35,000 ×g for 20 min (4 °C), resuspended inbuffer and recentrifuged. The pellets were resuspended in bufferand incubated at 37 °C for 15 min in order to hydrolyze allendogenous FAAH substrates. After this, they were recentri-fuged and the pellets were resuspended in 50 mM Tris–HCl

28 S. Holt et al. / European Journal of Pharmacology 565 (2007) 26–36

buffer, pH 7.4, containing 1 mM EDTA and 3 mM MgCl2 andfrozen at −80 °C in aliquots until used for assay. Cytosolicpreparations were obtained by homogenizing cerebella fromadult Sprague–Dawley rats in 50 mM sodium phosphate,0.32 M sucrose, pH 8.0, and thereafter centrifuging thehomogenates for 1 h at 100,000 g, 4 °C. Aliquots of thesupernatants (cytosolic fractions) were stored frozen in aliquotsat −80 °C until used for assay. Ethical permission for the animalexperiments was obtained from the local animal research ethicalcommittee.

2.3. Assay of [3H]-AEA hydrolysis in brain membranefractions

[3H]-AEA and [3H]-PEA hydrolysis was assayed asdescribed by Boldrup et al. (2004). Briefly, membranehomogenates were diluted with buffer (10 mM Tris–HCl,1 mM EDTA at the appropriate pH) to give a proteinconcentration/assay that was on the linear phase of the protein:activity curve and a volume of 165 μl/assay. Note that the pHvalues given in the text refer to the pH of the diluting bufferrather than the final pH at assay. The protein concentrationsused are indicated in the figure legends. Test compounds (10 μl,in ethanol) were added and the samples were preincubated whenindicated in the text. [3H]-AEA (25 μl, labelled in theethanolamine part of the molecule, final concentration 2 μMin a buffer consisting of 10 mM Tris–HCl, 1 mM EDTA and 1%w/v fatty acid-free bovine serum albumin) was added and thesamples were incubated for 5–10 min, as appropriate. Reactionswere stopped by putting the tubes on ice, and adding 80 μlcharcoal mixed in 320 μl of 0.5 M HCl. The samples weremixed, left at room temperature and then centrifuged at 2000 ×gfor 10 min at room temperature to sediment the charcoal.Aliquots (200 μl) of the “supernatants” were removed andanalysed for tritium content by liquid scintillation spectroscopywith quench correction. Blanks contained buffer instead of themembrane preparation.

2.4. Uptake of [3H]-AEA by P19 cells

The method of Rakhshan et al. (2000) as adapted bySandberg and Fowler (2005) was used. Briefly, P19 cells(passage number 34–36, cultured in MEM alpha mediumsupplemented with 10% foetal bovine serum, 100 units/mlpenicillin plus 100 μg/ml streptomycin and 1% non-essentialamino acids) were plated into 24 well culture plates andincubated overnight at 37 °C in humidified atmosphere with 5%CO2. After the incubation period, the cells were washed oncewith warm Krebs–Henseleit–Bicarbonate (KRH) buffer solu-tion containing 1% bovine serum albumin and then preincu-bated with URB597 (1 μM, or vehicle) in 340 μl KRH buffercontaining 0.1% fatty acid-free bovine serum albumin at 37 °Cfor 10 min. Test compounds (10 μl, or vehicle) were added andthe cells incubated for a further 10 min at 37 °C prior to additionof [3H]-AEA (50 μl, final concentration 200 nM) andincubation for 2 or 10 min at 37 °C. After the incubation, theplates were placed on ice and the cells were carefully washed

three times with ice-cold buffer containing 1% bovine serumalbumin. After aspiration, NaOH (0.2 M, 500 μl/well) wasadded and the wells were incubated for 15 min at 75 °C.Aliquots (300 μl) were then transferred to scintillation vials andassayed for tritium content by liquid scintillation spectroscopywith quench correction.

2.5. FAAH activity in C6 glioma cells

The assay of Paylor et al. (2006) was used. Briefly, C6glioma cells (passage range 16–25) were seeded into 24 wellculture plates (2×105 cells/well, culture medium F-10 Hamwith 10% foetal bovine serum, 100 units/ml penicillin and100 μg/ml streptomycin). The cells were then incubatedovernight at 37 °C, 5% CO2 in humidified atmosphericpressure. On the day of assay, the wells were washed twicewith 500 μl of a buffer consisting of 116 mM NaCl, 5.4 mMKCl, 1.8 mM CaCl2, 25 mM HEPES, 1 mM NaH2PO4, 0.8 mMMgSO4, pH 6 or 8 containing 1% (w/v) bovine serum albumin.After removal of the buffer by aspiration, 345 μl of the samebuffer but with 0.1% fatty acid-free bovine serum albumin inplace of 1% bovine serum albumin was added followed by 5 μlof the test compound or ethanol vehicle. The wells wereincubated for 10 min at 37 °C. Substrate (50 μl of [3H]-AEAlabelled in the ethanolamine part of the molecule) was added, togive a final assay substrate concentration of 0.25 μM, and thewells were incubated for 5 min at 37 °C. The culture plates wereplaced on ice, and methanol (400 μl) was added, after which thecells were collected by scraping the wells. Aliquots (400 μl)were transferred to glass tubes and chloroform (200 μl) wasadded. The samples were vortex mixed twice, centrifuged toseparate the phases, and aliquots (200 μl) of the aqueous phasewere assayed for tritium content by liquid scintillationspectroscopy with quench correction. Blanks were wells notcontaining cells.

2.6. Measurement of cyclooxygenase (COX) activity

COX activity was undertaken using a commercial kit(Cayman Chemical Co., Cat. No. 760111) by monitoring theproduction of oxidized N,N,N',N'-tetramethyl-p-phenylenedia-mine (TMPD) at 590 nm following incubation of either ovineCOX-1 or ovine COX-2 with arachidonic acid. The enzymeswere preincubated for 5 min at 25 °C with the test compoundsprior to addition of arachidonic acid (final concentration100 μM) and TMPD and incubation for 5 min at 25 °C.Blank values were obtained in the absence of enzyme.

2.7. Binding to CB1 and CB2 cannabinoid receptors

Radioligand binding experiments were undertaken in 96 wellplates with [3H]-CP55,940 (0.5 nM assay concentration) usingeither rat brain (minus cerebellum) membrane fractions (forCB1 receptors, 15 μg protein/assay, membranes prepared asdescribed above) or recombinant human cannabinoid CB2

receptors (BioXTal, Axxora Cat. no. BXT-C103720-4, 0.925 U/assay) and an assay buffer comprising 50 mM Tris–HCl pH 7.4,

29S. Holt et al. / European Journal of Pharmacology 565 (2007) 26–36

1 mM EDTA, 3 mM MgCl2 and 0.5% (w/v) BSA (total assayvolume 250 μl). Samples were incubated for 60 min at 37 °Cafter which time the samples were filtered through WhatmanGF/C filters pretreated with 0.2% (v/v) polyethylenimine (washbuffer 50 mM Tris–HCl pH 7.4, 0.1% (w/v) BSA). Tritiumretained by the filters was measured by liquid scintillationspectroscopy with quench correction. Non-specific binding wasdetermined in the presence of 1 μM HU-210.

2.8. Hydrolysis of [3H]-2-OG

Hydrolysis of 2-OG was undertaken using the method ofBrengdahl and Fowler (2006) using 96 well plates and an assayvolume of 50 μl. Briefly, inhibitor and enzyme source (rat braincytosolic fractions, 2.5 μg protein/assay or human recombinantmonoacylglycerol lipase, Cayman Chemical Co., Cat. No.10008354, 0.03 μl/assay) in 10 mM Tris–HCl, 1 mM EDTA,pH 7.4 were preincubated for 10 min at 25 °C. Substrate ([3H]-2-OG, 2 μM assay concentration in 10 mM Tris–HCl, 1 mMEDTA, pH 7.4 containing 1% w/v fatty acid-free bovine serumalbumin) was added and the samples were incubated for 2 h(cytosolic fractions) or 1 h (recombinant monoacylglycerollipase) at 25 °C prior to addition of 100 μl of a mixture ofphenyl sepharose gel (20 μl) in 1.5 M NaCl+0.5 M HCl. Thegel was allowed to sink and aliquots (30 μl) of the buffer phasewere measured for tritium content by liquid scintillationspectroscopy with quench correction. Initial experimentsindicated that the enzyme concentrations used were on thelinear phase of the substrate utilized: enzyme concentrationcurve.

2.9. Statistical analysis

pI50 values, and hence IC50 values were obtained using thebuilt-in programme “sigmoidal dose response curve, variableslope” of the GraphPad Prism computer programme (GraphPadSoftware Inc., San Diego, CA). “Top” (uninhibited) values wereset to 100% and “bottom” (minimum activity remaining) valueswere set either to zero or allowed to float, as appropriate. Km

and Vmax values were calculated from the mean data using thedirect linear plot (Eisenthal and Cornish-Bowden, 1974)analysis and the Enzyme Kinetics v1.4 software programme(Trinity Software, Campton, NH, USA).

3. Results

3.1. Inhibition of [3H]-AEA hydrolysis in brain membranes byibuprofen and related compounds

Initial assays of [3H]-AEA were undertaken with theibuprofen analogues using mouse brain homogenates. Themouse is rather appropriate in this respect, since it has beenshown that genetic deletion of FAAH results in a N99% loss inthe ability of brain homogenates to hydrolyze AEA at an assaypH of 7.2 (Cravatt et al., 2001). Assuming no dramatic mousestrain differences, an inhibition of the hydrolytic activity of[3H]-AEA by test compounds in the mouse brain homogenates

represents an inhibition of FAAH. The effects of a series ofibuprofen analogues upon the FAAH-catalyzed hydrolysis ofAEA by mouse brain membranes at an assay pH of 7.2 areshown in Fig. 1A. There was a large variation in potency of thecompounds, ranging from an IC50 value of 220 μM for the 4,6-dimethyl-pyridin-2-yl analogue (compound no. 11 in the studyof Cocco et al., 2003) to 8.3 μM for the 6-methyl-pyridin-2-ylanalogue (compound no. 5 in the study of Cocco et al., 2003,here given the name “ibu-am5”). Ibuprofen was assayedconcomitantly, and 30 and 100 μM concentrations of thiscompound gave % of control values of 89±2 and 61±0.7,respectively (means±S.E.M., n=3).

3.2. Effect of ibu-am5 upon the uptake of AEA by P19 cells

In order to establish whether ibu-am5 was capable ofinteracting with intact cells, we elected to study the effects of[3H]-AEA into mouse embryonic carcinoma P19 cells. Thesecells have been well characterized in this respect and shown toaccumulate AEA in a manner driven at least in part by FAAH(Sandberg and Fowler, 2005; Thors et al., in press). The uptakeexperiments are shown in Fig. 1B. In the absence of URB597, aselective FAAH inhibitor, a significant reduction in the rate ofuptake of AEA (labelled in the arachidonoyl part of themolecule and incubated with the cells for 2 min at 37 °C) wasseen with 10 and 30 μM ibu-am5, whereas ibuprofen at theseconcentrations was without effect. The compounds did notaffect the retention of the AEA by wells alone. The inhibitionproduced by ibu-am5 was not additive to that seen followingpretreatment with URB597. This result is consistent with thehypothesis that the reduction of AEA accumulation in the cellsis secondary to an action upon FAAH and indicates that ibu-am5 can affect the removal of AEA by intact cells. Uptakeexperiments were also undertaken using a 10 min AEAincubation time. In these experiments, URB597 still reducedthe uptake, but there was a larger data spread and the significanteffect of ibu-am5 was no longer seen (data not shown).

3.3. pH sensitivity of [3H]-AEA hydrolysis by ibu-am5

pH sensitivity studies were performed using membranes andcells from rats. The switch in species from mouse to rat was fortwo reasons: a) most of our previous work on the pH sensitivityof NSAIDs has been done using rat brain membranes and rat C6glioma cells (Holt et al., 2001; Fowler et al., 2003; Holt andFowler, 2003), thus providing a frame of reference for thepresent data; and b) the only in vivo data available on ibu-am5 isfrom the rat (Cocco et al., 2003).

Ibu-am5 was tested initially at pH 7.2 in membranepreparations from rat brain (minus cerebellum) in order toconfirm its activity towards [3H]-AEA hydrolysis in thisspecies. The compound effectively inhibited AEA metabolismwith a pI50 value of 5.57±0.06, corresponding to an IC50 valueof 2.7 μM. In comparison, (R)-ibuprofen and indomethacinwere less potent, with pI50 values (IC50 values in brackets) of3.87±0.05 (130 μM) and 4.35±0.06 (45 μM), respectively(data not shown).

Fig. 1. Panel A: inhibition by analogues of ibuprofen of the hydrolysis of 2 μM [3H]-AEA by mouse brain membranes. An assay buffer pH of 7.2, a protein content of5 μg/assay and a [3H]-AEA incubation time of 10 min without a preincubation phase were used. Shown are means for 1–3 experiments (n=1 only at the highestconcentrations). The numbers in bold type in the box refer to the compound numbering used in the original paper of Cocco et al. (2003). Ibuprofen itself has an –OHgroup in place of the –NH–R group. Panel B: effect of ibu-am5 and ibuprofen upon the accumulation of AEA by intact cells. P19 cells or wells alone werepreincubated for 10 min with either vehicle (whole columns) or 1 μMURB597 (dark columns enclosed within the whole columns) for 10 min prior to addition of thetest compounds and incubation for a further 10 min. AEA (labelled in the arachidonoyl part of the molecule) was added (assay concentration 200 nM) and the cells (orwells alone) incubated for 2 min at 37 °C. Data are means±S.E.M., n=4. ⁎Significantly different from control, Dunnett's post-hoc test following significant one-wayANOVA for repeated measures. The concentration of ibu-am5 and ibuprofen (“ibu") used for the wells alone was 30 µM.

30 S. Holt et al. / European Journal of Pharmacology 565 (2007) 26–36

Rat brain membranes in assay buffer of pH 6 or pH 8 werepreincubated with either ibu-am5 or ibuprofen prior to additionof [3H]-AEA and incubation for 10 min at 37 °C (Fig. 2A). Asexpected, ibuprofen was more potent at pH 6 than at pH 8, withpI50 values (IC50 values in brackets) of 3.87±0.02 (130 μM)and 3.12±0.04 (750 μM). The data with ibuprofen is consistentwith our previous data using either the racemate or theindividual enantiomers, and using either rat brain homogenatesor COS cells transfected with either wild type or mutant FAAH(Holt et al., 2001; Fowler et al., 2003). In contrast to ibuprofen,ibu-am5 was slightly less potent as an inhibitor of rat brain [3H]-AEA hydrolysis at pH 6 (pI50 value 5.33±0.02, correspondingto an IC50 value of 4.7 μM) than at pH 8 (pI50 value 5.61±0.03,corresponding to an IC50 value of 2.5 μM) (Fig. 2A). The 95%confidence limits of the pI50 values did not overlap.

The effect of ibu-am5 upon the hydrolysis of AEA (labelledin the ethanolamine part of the molecule) by intact C6 gliomacells is shown in Fig. 2B. Under the conditions used here, thehydrolysis is inhibited completely by URB597 (Paylor et al.,2006; see also Thors et al., in press). Ibu-am5 inhibited AEA

hydrolysis with a pI50 value of 5.91 (corresponding to an IC50

value of 1.2 μM) at both pH 6 and pH 8.

3.4. Evaluation of the contribution of N-acylethanolamine-hydrolyzing acid amidase to the hydrolytic activity of rat brainmembranes

One drawback of the use of membranes from rat brain is that,at least in theory, additional hydrolytic enzymes can contributeto [3H]-AEA metabolism. In particular, an N-acylethanolamine-hydrolyzing acid amidase (NAAA) has been shown to beexpressed in the rat brain (Ueda et al., 2001). NAAA is almostinactive at alkaline pH, but has an optimum pH at 5 (Ueda et al.,2001), raising the possibility that the shift in pH seen with ibu-am5 in fact reflects a changed contribution of NAAA to thehydrolysis rather than a changed activity at FAAH.

In order to establish whether or not NAAA activity is foundin rat brain membrane preparations, we used [3H]-palmitoy-lethanolamine ([3H]-PEA) as substrate. We have previouslyused this substrate to assess the hydrolytic activity of different

Fig. 3. Hydrolysis of 2 μM [3H]-PEA by rat brain (minus cerebellum)membranes. In panel A, the membranes were incubated for 60 min at 37 °C atpH 7.4 (●) or pH 5 (○) with protein concentration ranging from 0 to 5 μg/assayor 0 to 30 μg/assay (inset). In panel B, the membranes (1 μg/assay) werepreincubated at pH 7.4 (●) or pH 5 (○) for 10 min with MAFP prior to theaddition of 2 μM [3H]-PEA and incubation for a further 60 min. In panel C, themembranes (1 μg/assay) were preincubated for 10 min with either vehicle or100 nM MAFP prior to the addition of the indicated concentrations ofdithiothreitol (DTT) and/or Triton X-100 (“Tri”) and incubation for a further10 min. [3H]-PEA (2 μM) was then added and the samples were incubated for afurther 2 h. In all cases, data are means±S.E.M., n=3.

Fig. 2. pH sensitivity of [3H]-AEA hydrolysis to ibu-am5. In panel A, rat brain(minus cerebellum) membranes were used at protein contents of 2.5 μg/assay(pH 6) or 0.8 μg/assay (pH 8), a preincubation phase of 10 min and an [3H]-AEAincubation time of 10 min. The difference in protein content reflects the pHprofile of the enzyme (Schmid et al., 1985). Data are means±S.E.M., n=3. Inpanel B, C6 glioma cells were incubated for 10 min at 37 °C with ibu-am5 priorto addition of [3H]-AEA (labelled in the ethanolamine part of the molecule,assay concentration 250 nM) and the wells were incubated for 5 min at 37 °C.Data are means±S.E.M., n=4.

31S. Holt et al. / European Journal of Pharmacology 565 (2007) 26–36

subcellular fractions of rat brain, and demonstrated thatsynaptosomal, mitochondrial, microsomal and myelin fractionsall contain hydrolytic activity sensitive to inhibition byoleoyltrifluoromethylketone, phenylmethylsulfonyl fluorideand the two enantiomers of ibuprofen with the expectedpotencies for FAAH (Tiger et al., 2000). This substrate is moreavidly metabolized by NAAA than [3H]-AEA, whereas thereverse is true for FAAH (Ueda et al., 2001; Bisogno et al.,1997), thereby allowing us to maximize our chances ofidentifying an NAAA activity in the membranes. As expected,[3H]-PEA was metabolized by the membranes in a mannerdependent upon the assay protein concentration, but the rate ofhydrolysis was much greater at pH 7.4 than at pH 5 (Fig. 3A),suggesting at best a moderate relative contribution by NAAA.This was confirmed pharmacologically using the non-selectiveserine hydrolase inhibitor methylarachidonoylfluorophospho-nate (MAFP), which is an effective inhibitor of FAAH but doesnot block NAAA activity (Ueda et al., 2001). The compoundtotally blocked the hydrolysis of [3H]-PEA at both pH 7.4 and 5,with very similar potencies (pI50 values of 10.40±0.03 and10.73±0.06, corresponding to IC50 values of 39 and 19 pM at

pH 7.4 and 5, respectively) (Fig. 3B). Finally, a combination ofdithiothreitol and Triton X-100 was added to the membranes,since these have been shown to stimulate NAAA activity (Uedaet al., 2001). Even in the presence of these agents, the observed

Fig. 4. Mode of inhibition of rat brain (minus cerebellum) membranes by ibu-am5. An assay pH of 6, protein content of 2.5 μg/assay and a [3H]-AEAincubation time of 10 min without a preincubation phase was used. In the mainfigure, data are means±S.E.M., n=3. A secondary replot of the data is shown inthe insert to illustrate the non-competitive nature of the interaction.

32 S. Holt et al. / European Journal of Pharmacology 565 (2007) 26–36

activity in the presence of MAFP was extremely low comparedto the activity of the membranes in the absence of MAFP (Fig.3C). It should be pointed out as caveat that the membranes usedfor these experiments were from different animals to those usedfor the experiments with ibu-am5 and ibuprofen. However, the

Fig. 5. Inhibition of the FAAH-catalyzed hydrolysis of 2 μM [3H]-AEA in the rat br(for comparative purposes). For the experiments undertaken using an assay buffer oan assay buffer of pH 8.4, 3 μg protein/assay was used. In these experiments, there w10 min (pH 6.2 and pH 8.4) or 15 min (pH 7.2). For the experiments undertaken atbold given in parentheses refer to the compound numbers in the original paper oindomethacin ester 4-methoxyphenyl at pH 6.2 is the same in both panels, and ha

membrane preparation protocol was the same, and the datawould suggest that the contribution of NAAA to the hydrolysisof PEA, and by extension AEA, is extremely small.

3.5. Mode of inhibition of rat brain [3H]-AEA hydrolysis byibu-am5

Rat brain membranes were preincubated with 0, 1, 5 and 20 μMibu-am5 at either pH 6 or 8 for 3, 15, 30, 45 and 60 min prior toaddition of [3H]-AEA.No large changes in inhibitory potencywereseen. Thus for example, % of control activities (means±S.E.M.,n=3) at pH 6 and an ibu-am5 concentration 5 μM were 52±5,53±1, 51±2, 51±3 and 55±2 at preincubation times of 3, 15, 30,45 and 60 min, respectively. The corresponding values at pH8were 36±2, 33±1, 32±3, 35±0.7 and 28±2% respectively. Theinhibition at pH 6 was non-competitive in nature, with a Ki valueof ∼8 μM (Fig. 4). The kinetics of inhibition were alsoinvestigated at pH 8. In this case however, the Km value for theuninhibited enzyme was lower than the lowest concentration ofsubstrate tested (0.8 μM), making detailed analysis difficult.However, Vmax(app) values could be measured with accuracy andwere 11.5, 7.7, 7.0 and 4.8 nmol/mg protein/min at concentrationsof 0, 2, 4 and 6 μM ibu-am5 (calculated from pooled data fromthree experiments using the direct linear plot method of Eisenthaland Cornish-Bowden, 1974), giving a Ki(intercept) value of 4.5 μM(data not shown).

ain (minus cerebellum) membranes by analogues of indomethacin and ibu-am5f pH 6.2, 7.5 μg protein/assay was used. For the experiments undertaken usingas no preincubation phase, and the incubation phase with [3H]-AEAwas eitherpH 7.2 (on a different occasion), 2 μg protein/assay was used. The numbers inf Kalgutar et al. (2000b). Data are means of 2–3 experiments. The data fors been included in panel B for comparative purposes.

33S. Holt et al. / European Journal of Pharmacology 565 (2007) 26–36

3.6. Inhibition of rat brain [3H]-AEA hydrolysis by indometha-cin analogues

Six analogues of indomethacin were investigated using therat membranes. In contrast to the ibuprofen analogues andindomethacin itself, none of the analogues produced a completeinhibition of [3H]-AEA hydrolysis over the concentration rangetested (Fig. 5). This most likely reflects the limited solubility ofthese compounds, and a similar phenomenon has been seen inour hands with other lipophilic compounds (see e.g. Vande-voorde et al., 2003). The most potent of the indomethacinanalogues, the 4-methoxyphenyl ester of indomethacin (com-pound 25 of Kalgutar et al., 2000b), inhibited the activity by amaximum of about half at both pH 6.2 and pH 8.4, and with pI50

Fig. 6. Effects of ibu-am5, ibuprofen and indomethacin upon A, ovine COX-1 activitycerebellum) membranes (CB1 receptors); D, [

3H]-CP55,940 specific binding to humhuman monoacylglycerol lipase and rat cerebellar cytosolic fractions. Data are meconcentration of the test compounds, where means and ranges, n=2 are shown.

values for the inhibitable fraction of 5.75±0.06 (IC50 value1.8 μM) and 5.49±0.05 (3.3 μM), respectively. The corre-sponding values for indomethacin at pH 6.2 and pH 8.4, in thiscase with complete inhibition of [3H]-AEA hydrolysis, were4.72±0.02 (19 μM) and 4.02±0.04 (96 μM).

Two other NSAIDs were investigated for their ability toinhibit rat brain [3H]-AEA hydrolysis. Fenbufen was tested atpH 7.2 (11 concentrations in the range 0.2–200 μM, n=3) andwas found to inhibit [3H]-AEA hydrolysis with a pI50 value of3.95±0.06, corresponding to an IC50 value of 110 μM.Meloxicam was tested using 10 concentrations (range from 2to 1000 μM, n=3) and gave pI50 values of 3.68±0.04 (210 μM)and 3.42±0.03 (380 μM) at pH 6 and 8, respectively (data notshown).

; B, ovine COX-2 activity; C, [3H]-CP55,940 specific binding to rat brain (minusan recombinant CB2 receptors; and E, [3H]-2-OG hydrolysis by recombinant

ans±S.E.M., n=3–4 except for the cytosolic 2-OG hydrolysis data at 10 μM

34 S. Holt et al. / European Journal of Pharmacology 565 (2007) 26–36

3.7. Interaction of ibu-am5, indomethacin and ibuprofen withcyclooxygenase-1 and -2, cannabinoid receptors and mono-acylglycerol lipase

In order to establish whether the greater potency of ibu-am5towards FAAH as compared with ibuprofen and indomethacinwas accompanied by corresponding changes in the potencytowards cyclooxygenase enzymes, a commercial kit was used.The data are shown in Fig. 6A and B for ovine COX-1 and -2,respectively. At the outset, it should be pointed out thatpotencies reported in the literature towards COX isoforms varyconsiderably between methods, enzyme sources and laborato-ries (see e.g. Mitchell et al., 1993; Riendeau et al., 1997). It isthus more important that the relative potencies between thethree compounds are considered, rather than the absolutepotencies. In our hands, indomethacin inhibited the activities ofCOX-1 and COX-2 with pI50 values of 6.75±0.15 and 6.51±0.17, respectively, corresponding to IC50 values of 180 and310 nM, respectively. Ibuprofen was, as expected, less potentthan indomethacin. The potency of ibu-am5 was similar to, andcertainly not greater than, that of ibuprofen, indicating that theincreased potency of this compound towards FAAH is notaccompanied by an increased potency towards COX enzymes.

The effects of the three compounds on the specific binding of[3H]-CP55,940 to rat CB1 and human CB2 receptors are shownin Fig. 6C and D. The rank order of potency at both receptorswas ibu-am5N indomethacinN ibuprofen. The pI50 values (withIC50 values in brackets) for ibu-am5 towards CB1 and CB2

receptors, respectively, were 4.38±0.17 (41 μM) and 4.62±0.04 (24 μM), whilst the IC50 values for indomethacin andibuprofen were in the range of 100–300 μM. It is potentiallymisleading to compare potencies of lipophilic compounds indifferent assays undertaken using different methods. However,interpretation of relative potencies can once again be consid-ered. Thus, ibu-am5 is about ten-fold more potent that ibuprofenwith respect to its ability to interact with CB receptors, whilst itis 28-fold (pH 6) and 300-fold (pH 8) more potent with respectto its ability to inhibit [3H]-AEA hydrolysis in rat brainmembranes.

The effects of the three compounds upon the ability ofhuman recombinant monoacylglycerol lipase and rat cerebellarcytosolic fractions to hydrolyze 2-oleoylglycerol are shown inFig. 6E. None of the compounds inhibited the rate of hydrolysisover the concentrations tested. In contrast, MAFP inhibitedhydrolysis of 2-OG by human recombinant monoacylglycerollipase and rat cytosolic fractions with pI50 values of 9.29±0.06and 8.87±0.11, respectively, corresponding to IC50 values of0.51 and 1.3 nM, respectively (B. Jacobsson, A. Lenman & C.J.Fowler, unpublished data).

4. Discussion

In the present study a series of indomethacin and ibuprofenanalogues were tested with the aim of finding a compound witha dual action upon FAAH and COX-2, but with a morepronounced FAAH action than the original parent compounds.With respect to the indomethacin series, only one compound

showed an increased potency towards FAAH, in contrast to thesituation for COX-2 (Kalgutar et al., 2000a,b). Furthermore, theFAAH inhibition produced by this compound was not complete,presumably a reflection of a solubility issue. In consequence,the indomethacin series of compounds was not studied further.

The ibuprofen series of compounds, on the other hand,showed useful properties, in that they inhibited FAAHcompletely, and with a greater potency than ibuprofen, acompound which is known to potentiate the effects of AEA invivo (Guindon et al., 2006a). Thus, substitution of the carboxylgroup of ibuprofen with the amide heterocycle group improvesthe ability of the analogues to interact with FAAH. Among thesecompounds, the pyridin-3-yl substituent was more potent thanthe pyridin-2-yl substituent (compounds 1 vs. 2). Introductionof a methyl group in either the 4, 5 or 6 position of the pyridin-2-yl substituent also increased the potency of the compound(compounds 3, 4 and 5 vs. 1) whereas introduction of a 3-methyl group (compound 6), a 4-chloro group (compound 8) ormethyl groups at both positions 4 and 6 of the pyridin-2-ylsubstituent (compound 11) reduced the potency by a factor of2.7–8. The most potent compound of the series, ibu-am5 (5),inhibited FAAH in a non-competitive manner with an IC50 inthe low micromolar range, was effective in blocking AEAmetabolism in intact cells, was without effect upon MAGL, anddid not show a more pronounced interaction with CB receptorsrelative to its potency as an FAAH inhibitor than ibuprofen. Thenon-competitive inhibition of ibu-am5 differs from the mixed-type inhibition seen with ibuprofen (or its (−)enantiomer)(Fowler et al., 1997; Holt et al., 2001), but the only practicalconsequence of this is that the inhibitory potency of ibu-am5will not be dependent upon the AEA concentration, whereas theinhibitory potency of ibuprofen is likely to decrease slightly asthe local AEA concentration increases.

Previous data has demonstrated that the potency of ibuprofenincreases as the pH is lowered (Holt et al., 2001), as is the casefor other acidic NSAIDs (Fowler et al., 2003). The most simpleexplanation for this behaviour is that the unionized form of themolecules contributes primarily to the inhibition of FAAH (Holtet al., 2001). Replacement of the carboxyl group of ibuprofenwith an amide–pyridinyl group will of course change theionization properties of the molecule, so it is perhaps notsurprising that ibu-am5 did not show the increase in potencytowards FAAH seen with ibuprofen at low pH values— indeed,a slight reduction in potency was seen at the lower pH valuetested. However, even under these conditions, ibu-am5 wasN25-fold more potent than ibuprofen as an inhibitor of FAAH,IC50 values of 4.7 and 130 μM being seen at pH 6 for the twocompounds. Furthermore, the increase in potency towardsFAAH compared with ibuprofen was not mirrored by anincrease in potency towards the COX enzymes. Ibu-am5 was infact a slightly weaker COX inhibitor than ibuprofen in ourhands, although the difference was small, and is unlikely toexplain the large difference in ulcerogenic potency between thetwo compounds (Cocco et al., 2003). A more likely explanationis that the physicochemical properties of the compound willaffect its ability to affect COX isoenzymes in gastric surfacecells, given the pH gradient there (for discussion with respect to

Fig. 7. Comparison of the pI50 values for the inhibition of mouse FAAH, pH 7.2(taken from Fig. 1A) by the ibuprofen analogues and the inhibition of acetic acidwrithing produced by a 20 mg/kg i.p. dose of the compounds to rats (data fromTable 1 of Cocco et al., 2003).

35S. Holt et al. / European Journal of Pharmacology 565 (2007) 26–36

indomethacin and the “ion-trapping” hypothesis, see Kavvadaet al., 2006).

The present study was not designed to evaluate the compoundsin vivo, rather to identify a compound that can be useful for futurestudies. However, in 2003, Cocco et al. reported the ability of thecompounds to inhibit acid inducedwrithing in rats. Inmice, this testis sensitive to cannabinoids (Sofia et al., 1975; Ulugöl et al., 2006)and it is not unreasonable to presume that the same is true for rats.Comparison of the data of Cocco et al. (2003) with the potencies ofthe compounds towards FAAH shows no obvious correlation(Fig. 7). This lack of correlation is perhaps not surprising since itdoes not take into account either species or pharmacokineticdifferences between the compounds, or the contribution to theanalgesic activity made by COX inhibition. In this respect Ulugölet al. (2006) demonstrated that the effects of the NSAID ketorolacand the cannabinoid receptor agonist WIN55,212-2 in the mouseacetic acid writhing model were additive, rather than synergistic,suggesting that the cannabinoid and COX make separatecontributions rather than potentiate each other. However, theobservation that ibu-am5 is more efficacious than ibuprofen in theacetic acid writhing model (Cocco et al., 2003), and that thedifference between ibuprofen and ibu-am5 is not attributable to anincreased efficacy towards COX isoforms alone would support thecontention that compounds combining COX- and FAAH-inhibi-tory properties may be useful analgesics. Ibu-am5 is thus a valuablecompound with which to explore this contention further.

Acknowledgements

The authors are indebted to Britt Jacobsson and IngridPersson for expert technical assistance with some of theexperiments reported here. C.F. would like to thank the SwedishResearch Council (Grant no. 12158, medicine), Gun and BertilStohne's Foundation, Konung Gustaf V's and DrottningVictorias Foundation, Stiftelsen för Gamla Tjänarinnor andthe Research Funds of the Medical Faculty, Umeå Universityfor financial support. The postdoctoral research of S.V. wassupported by a grant from the foundation Wenner-GrenskaSamfundet. S.V. thanks the Belgian National Funds for

Scientific Research (F.N.R.S.) for the current grant ofpostdoctoral researcher (Chargée de recherches du F.N.R.S.).

References

Andersson, S.E., Lexmüller, K., Johansson, A., Ekström, G.M., 1999. Tissueand intracellular pH in normal periarticular soft tissue and during differentphases of antigen induced arthritis in the rat. J. Rheumatol. 26, 2018–2024.

Ates, M., Hamza, M., Seidel, K., Kotalla, C.E., Ledent, C., Gühring, H., 2003.Intrathecally applied flurbiprofen produces an endocannabinoid-dependentantinociception in the rat formalin test. Eur. J. Neurosci. 17, 597–604.

Bisogno, T., Maurelli, S., Melck, D., De Petrocellis, L., Di Marzo, V., 1997.Biosynthesis, uptake, and degradation of anandamide and palmitoylethano-lamide in leukocytes. J. Biol. Chem. 272, 3315–3323.

Boldrup, L., Wilson, S.J., Barbier, A.J., Fowler, C.J., 2004. A simple stoppedassay for fatty acid amide hydrolase avoiding the use of a chloroformextraction phase. J. Biochem. Biophys. Methods 60, 171–177.

Bouaboula, M., Hilairet, S., Marchand, F., Fajas, L., Le Fur, G., Casellas, P.,2005. Anandamide induced PPARγ transcriptional activation and 3T3-L1preadipocyte differentiation. Eur. J. Pharmacol. 517, 174–181.

Brengdahl, J., Fowler, C.J., 2006. A novel assay for monoacylglycerolhydrolysis suitable for high-throughput screening. Anal. Biochem. 359,40–44.

Cocco, M.T., Congiu, C., Onnis, V., Morelli, M., Cauli, O., 2003. Synthesis ofibuprofen heterocyclic amides and investigation of their analgesic andtoxicological properties. Eur. J. Med. Chem. 38, 513–518.

Cravatt, B.F., Demarest, K., Patricelli, M.P., Bracey, M.H., Giang, D.K., Martin,B.R., Lichtman, A.H., 2001. Supersensitivity to anandamide and enhancedendogenous cannabinoid signaling in mice lacking fatty acid amidehydrolase. Proc. Natl. Acad. Sci. U.S.A. 98, 9371–9376.

Eisenthal, R., Cornish-Bowden, A., 1974. The direct linear plot. A newgraphical procedure for estimating enzyme kinetic parameters. Biochem. J.139, 715–720.

Fowler, C.J., Tiger, G., Stenstrom, A., 1997. Ibuprofen inhibits rat braindeamidation of anandamide at pharmacologically relevant concentrations.Mode of inhibition and structure–activity relationship. J. Pharmacol. Exp.Ther. 283, 729–734.

Fowler, C.J., Janson, U., Johnson, R.M., Wahlström, G., Stenström, A.,Norström, K., Tiger, G., 1999. Inhibition of anandamide hydrolysis by theenantiomers of ibuprofen, ketorolac and flurbiprofen. Arch. Biochem.Biophys. 362, 191–196.

Fowler, C.J., Holt, S., Tiger, G., 2003. Acidic nonsteroidal anti-inflammatorydrugs inhibit rat brain fatty acid amide hydrolase in a pH-dependent manner.J. Enz. Inhib. Med. Chem. 18, 55–58.

Gallant, M., Dufresne, C., Gareau, Y., Guay, D., Leblanc, Y., Prasit, P., Rochette,C., Sawyer, N., Slipetz, D.M., Tremblay, N., Metters, K.M., Labelle, M.,1996. New class of potent ligands for the human peripheral cannabinoidreceptor. Bioorganic Med. Chem. Letters 6, 2263–2268.

Gühring, H., Hamza, M., Sergejeva, M., Ates, M., Kotalla, C.E., Ledent, C.,Brune, K., 2002. A role for endocannabinoids in indomethacin-inducedspinal antinociception. Eur. J. Pharmacol. 454, 153–163.

Guindon, J., De Léan, A., Beaulieu, P., 2006a. Local interactions betweenanandamide, an endocannabinoid, and ibuprofen, a nonsteroidal anti-inflammatory drug, in acute and inflammatory pain. Pain 121, 85–93.

Guindon, J., LoVerme, J., De Léan, A., Piomelli, D., Beaulieu, P., 2006b.Synergistic antinociceptive effects of anandamide, an endocannabinoid, andnonsteroidal anti-inflammatory drugs in peripheral tissue: a role forendogenous fatty-acid ethanolamides? Eur. J. Pharmacol. 550, 68–77.

Häbler, C., 1929. Über den K- und Ca-gehalt von eiter und exsudaten und seinebeziehungen zum entzündungsschmerz. Klin. Wochenschr. 8, 1569–1572.

Holt, S., Fowler, C.J., 2003. Anandamide metabolism by fatty acid amidehydrolase in intact C6 glioma cells. Increased sensitivity to inhibition byibuprofen and flurbiprofen upon reduction of extra- but not intracellular pH.Naunyn-Schmiedeberg's Arch. Pharmacol. 367, 237–244.

Holt, S., Nilsson, J., Omeir, R., Tiger, G., Fowler, C.J., 2001. Effects of pH onthe inhibition of fatty acid amidohydrolase by ibuprofen. Br. J. Pharmacol.133, 513–520.

36 S. Holt et al. / European Journal of Pharmacology 565 (2007) 26–36

Holt, S., Comelli, F., Costa, B., Fowler, C.J., 2005. Inhibitors of fatty acid amidehydrolase reduce carrageenan-induced hind paw inflammation in pentobarbital-treated mice: comparison with indomethacin and possible involvement ofcannabinoid receptors. Br. J. Pharmacol. 146, 467–476.

Jayamanne, A., Greenwood, R., Mitchell, V.A., Aslan, S., Piomelli, D.,Vaughan, C.W., 2006. Actions of the FAAH inhibitor URB597 inneuropathic and inflammatory chronic pain models. Br. J. Pharmacol. 147,281–288.

Kalgutar, A.S., Crews, B.C., Rowlinson, S.W., Marnett, A.B., Kozak, K.R.,Remmel, R.P., Marnett, L.J., 2000a. Biochemically based design ofcyclooxygenase-2 (COX-2) inhibitors: facile conversion of nonsteroidalantiinflammatory drugs to potent and highly selective COX-2 inhibitors.Proc. Natl. Acad. Sci. U. S. A. 97, 925–930.

Kalgutar, A.S., Marnett, A.B., Crews, B.C., Remmel, R.P., Marnett, L.J., 2000b.Ester and amide derivatives of the nonsteroidal anti-inflammatory drug,indomethacin, as selective cyclooxygenase-2 inhibitors. J. Med. Chem. 43,2860–2870.

Kavvada, K.M., Murray, J.G., Moore, V.A., Coombes, A.G.A., Hanson, P.J.,2006. High permeability of the anionic form restrics accumulation ofindomethacin by cultured gastric surface epithelial cells exposed to lowapical pH. Eur. J. Pharmacol. 549, 41–49.

Kozak, K.R., Marnett, L.J., 2002. Oxidative metabolism of endocannabinoids.Prostaglandins Leukot. Essent. Fatty Acids 66, 211–220.

Lichtman, A.H., Shelton, C.C., Advani, T., Cravatt, B.F., 2004. Mice lackingfatty acid amide hydrolase exhibit a cannabinoid receptor-mediatedphenotypic hypoalgesia. Pain 109, 319–327.

Lo Verme, J., Fu, J., Astarita, G., La Rana, G., Russo, R., Calignano, A.,Piomelli, D., 2005. The nuclear receptor PPAR-α mediates the antiin-flammatory actions of palmitoylethanolamide. Mol. Pharmacol. 67, 15–19.

Mitchell, J.A., Akarasereenont, P., Thiemermann, C., Flower, R.J., Vane, J.R.,1993. Selectivity of nonsteroidal antiinflammatory drugs as inhibitors ofconstitutive and inducible cyclooxygenase. Proc. Natl. Acad. Sci. U.S.A. 90,11693–11697.

New, D.C., Wong, Y.H., 2003. BML-190 and AM251 act as inverse agonists atthe human cannabinoid CB2 receptor: signalling via cAMP and inositolphosphates. FEBS Lett. 536, 157–160.

Paria, B.C., Deutsch, D.D., Dey, S.K., 1996. The uterus is a potential site foranandamide synthesis and hydrolysis: differential profiles of anandamidesynthase and hydrolase activities in the mouse uterus during theperiimplantation period. Mol. Reprod. Dev. 45, 183–192.

Paylor, B., Holt, S., Fowler, C.J., 2006. The potency of the fatty acid amidehydrolase inhibitor URB597 is dependent upon the assay pH. Pharmacol.Res. 54, 481–485 Erratum published in Pharmacol. Res. 55, 80 (2007).

Rakhshan, F., Day, T.A., Blakely, R.D., Barker, E.L., 2000. Carrier-mediateduptake of the endogenous cannabinoid anandamide in RBL-2H3 cells.J. Pharmacol. Exp. Ther. 292, 960–967.

Riendeau, D., Charleson, S., Cromlish, W., Mancini, J.A., Wong, E., Guay, J.,1997. Comparison of the cyclooxygenase-1 inhibitory properties ofnonsteroidal anti-inflammatory drugs (NSAIDs) and selective COX-2inhibitors, using sensitive microsomal and platelet assays. Can. J. Physiol.Pharm. 75, 1088–1095.

Sandberg, A., Fowler, C.J., 2005. Measurement of saturable and non-saturablecomponents of anandamide uptake into P19 embryonic carcinoma cells inthe presence of fatty acid-free bovine serum albumin. Chem. Phys. Lipids134, 131–139.

Schmid, P.C., Zuzarte-Augustin, M.L., Schmid, H.H.O., 1985. Properties of ratliver N-acylethanolamine amidohydrolase. J. Biol. Chem. 260, 14145–14149.

Singh Tahim, A., Sántha, P., Nagy, I., 2005. Inflammatory mediators convertanandamide into a potent activator of the vanilloid type 1 transient receptorpotential receptor in nociceptive primary sensory neurons. Neuroscience136, 539–548.

Sofia, R.D., Vassar, H.B., Knobloch, L.C., 1975. Comparative analgesic activityof various naturally occurring cannabinoids in mice and rats. Psychophar-macologia 40, 285–295.

Thors, L., Alajakku, K., Fowler, C.J., in press. The “specific” tyrosine kinaseinhibitor genistein inhibits the enzymic hydrolysis of anandamide.Implications for anandamide uptake. Br. J. Pharmacol.

Tiger, G., Stenström, A., Fowler, C.J., 2000. Pharmacological properties of ratbrain fatty acid amidohydrolase in different subcellular fractions usingpalmitoylethanolamide as substrate. Biochem. Pharmacol. 59, 647–653.

Ueda, N., Yamanaka, K., Yamamoto, S., 2001. Purification and characterization ofan acid amidase selective forN-palmitoylethanolamine, a putative endogenousanti-inflammatory substance. J. Biol. Chem. 276, 35552–35557.

Ulugöl, A., Özyigit, F., Yeşilyurt, Ö., Dogrul, A., 2006. The additiveantinociceptive interaction between WIN55,212-2, a cannabinoid agonist,and ketorolac. Anesth. Analg. 102, 443–447.

Vandevoorde, S., Jonsson, K.-O., Fowler, C.J., Lambert, D.M., 2003.Modifications of the ethanolamine head in N-palmitoylethanolamine:synthesis and evaluation of new agents interfering with the metabolism ofanandamide. J. Med. Chem. 46, 1440–1448.

Zygmunt, P.M., Petersson, J., Andersson, D.A., Chuang, H.-h., Sørgård, M., DiMarzo, V., Julius, D., Högestätt, E.D., 1999. Vanilloid receptors on sensorynerves mediate the vasodilator action of anandamide. Nature 400, 452–457.