Inhibition of monoacylglycerol lipase by troglitazone, N-arachidonoyl dopamine and the irreversible inhibitor JZL184: comparison of two different assays

Abstract

BACKGROUND AND PURPOSE

Drugs used clinically usually have a primary mechanism of action, but additional effects on other biological targets can contribute to their effects. A potentially useful additional target is the endocannabinoid metabolizing enzyme monoacylglycerol lipase (MGL). We have screened a range of drugs for inhibition of MGL and compared the observed potencies using different MGL enzyme assays.

EXPERIMENTAL APPROACH

MGL activity was screened using recombinant human MGL (cell lysates and purified enzyme) with 4-nitrophenyl acetate (NPA) as substrate. 2-Oleolyglycerol metabolism by rat cerebellar cytosolic MGL and by recombinant MGL was also investigated.

KEY RESULTS

Among the 96 compounds screened in the NPA assay, troglitazone, CP55,940, N-arachidonoyl dopamine and AM404 inhibited NPA hydrolysis by the lysates with IC₅₀ values of 1.1, 4.9, 0.78 and 3.1 µM, respectively. The potency for troglitazone is in the same range as its primary pharmacological activity, activation of peroxisome proliferator-activated receptor (PPAR) γ. Among PPARγ ligands, the potency order towards human MGL was troglitazone > ciglitazone > rosiglitazone > 15-deoxy-Δ^12,14-prostaglandin J₂≈ CAY 10415 > CAY 10514. In contrast to the time-dependent inhibitor JZL184, the potency of troglitazone was dependent upon the enzyme assay system used. Thus, troglitazone inhibited rat cytosolic 2-oleoylglycerol hydrolysis less potently (IC₅₀ 41 µM) than hydrolysis of NPA by the human MGL lysates.

CONCLUSIONS AND IMPLICATIONS

‘Hits’ in screening programmes for MGL inhibitors should be assessed in different MGL assays. Troglitazone may be a useful lead for the design of novel, dual action MGL inhibitors/PPARγ activators.

Introduction

The endocannabinoid system, comprising the endogenous cannabinoids (CB) anandamide (AEA, arachidonoyl ethanolamide) and 2-arachidonoylglycerol (2-AG), their target CB receptors, and their synthetic and degradative enzymes, are involved in a wide number of physiological processes in the body, ranging from cognition to bone metabolism (review, see Pacher et al., 2006). AEA is primarily metabolized by the enzyme fatty acid amide hydrolase (FAAH, also known as FAAH1, Alexander et al., 2009), and selective inhibitors of this enzyme such as URB597 (3′-(aminocarbonyl)[1,1′-biphenyl]-3-3-yl)-cyclohexylcarbamate, Kathuria et al. 2003) have shown potentially useful activity in models of pain, inflammation, anxiety and depression (see, e.g. Kathuria et al., 2003; Gobbi et al., 2005; Holt et al., 2005; Jayamanne et al., 2006). 2-AG is also a substrate for FAAH, cyclooxygenase-2, and the hydrolytic enzymes α/β-hydrolase domain (ABHD)12 and ABHD6, but in the mouse brain, the most important hydrolytic enzyme is monoacylglycerol lipase (MGL), which accounts for ∼85% of the hydrolysis of this substrate (Goparaju et al., 1998; Dinh et al., 2002; Blankman et al., 2007). JZL184 (4-nitrophenyl-4-(dibenzo[d][1,3]dioxol-5-yl(hydroxy)methyl)piperidine-1-carboxylate, Long et al., 2009a), a selective irreversible inhibitor of MGL, has been shown to produce anti-nociceptive effects in the chronic constriction injury model of neuropathic pain (Kinsey et al., 2009).

The wide range of effects produced by the endocannabinoid system in the body raises the possibility that many drugs that are, or have been, used clinically could interact with this system and that this interaction may contribute to their biological effects in humans over and above their primary pharmacological effect. Compounds directly activating CB₁ receptors would presumably have been discovered by recreational drug abusers, but compounds with inhibitory effects upon FAAH and MGL are not likely to be of interest in this respect since they do not produce cannabis-like behaviours, unless both enzymes are inhibited by the compound in question (Kathuria et al., 2003; Long et al., 2009a,b;). There is evidence that non-steroidal anti-inflammatory drugs inhibit FAAH in vitro and that these drugs can influence endocannabinoid-mediated responses in vivo (Fowler et al., 1997; Gühring et al., 2002; Guindon et al., 2006; Naidu et al., 2009; Bishay et al., 2010). The intravenous anaesthetic propofol is also a FAAH inhibitor, a property that may contribute to its pharmacological effects in vivo (Patel et al., 2003).

It is not known, however, whether MGL is also inhibited by drugs that are either currently or previously in our therapeutic arsenal or used experimentally, for pharmacological properties other than MGL inhibition. One way of identifying such compounds is the use of an assay permitting the characterization of a large number of compounds. One such assay, whereby the hydrolysis of 4-nitrophenyl acetate (NPA) by recombinant MGL can be followed spectrophotometrically, has recently been reported (Muccioli et al., 2008). In the present study, we investigated a series of drugs with primary effects in therapeutic areas known to be sensitive to CB actions, such as non-steroidal anti-inflammatory drugs, antidepressants and drugs for the treatment of type 2 diabetes. Active compounds have been characterized further both in the NPA assay and in a standard radiochemical assay using 2-oleoylglycerol (2-OG) as substrate. We report that the peroxisome proliferator-activated receptor (PPAR) γ ligand troglitazone and the endogenous vanilloid receptor agonist, N-arachidonoyl dopamine, are relatively potent MGL inhibitors using the NPA assay, but that the potency of troglitazone is dependent upon the MGL assay used.

Methods

Spectrophotometric assay for MGL activity using recombinant human MGL

Assays were carried out in a 96-well microtiter plate (100 µL total volume) using a method based on the assay of Muccioli et al. (2008). Human recombinant MGL (either clear lysates or purified enzyme, as indicated) in 10 mM Tris–HCl, 1 mM EDTA, pH 7.4 (added volume 70 µL) was added to each well which also contained test compounds [3 µL in vehicle, DMSO except for N-arachidonoyl dopamine (ethanol), 15-deoxy-Δ^12,14-prostaglandin J₂ and CAY 10541 (methyl acetate)] or vehicle alone. Buffer (7 µL) was added to each well. To start the hydrolysis, NPA (20 µL, final concentration of 0.25 mM, unless otherwise shown) was added rapidly. Blanks contained buffer alone. The concentration of MGL was chosen to ensure that initial velocities were measured. The samples were incubated at room temperature. The absorbance was measured at 405 nm after 0 min (to rule out effects of the compounds per se on the absorbance) and at least two 20 min intervals, thereafter using a Thermomax Microplate Reader (ThermoMax Kinetic Microplate Reader, Molecular Devices, Sunnyvale, CA, USA). For the experiments reported here, the 20 min time point was used to ensure that initial activities were measured but at the same time allow for sufficient product formation. In a typical experiment, the OD values for the blanks were ∼0.045 (with very little variation between assays), while controls were two- to threefold higher than blanks (with a small intra-assay variation). Measurements at 40 min were also made to confirm the inhibition at a higher control : blank ratio.

Radiochemical assay for 2-OG hydrolysis by rat brain extracts

All animal care and experimental procedures complied with national guidelines and laws and were approved by the local animal ethical committee. Assays were carried out essentially as described in Ghafouri et al. (2004), but with a charcoal separation (Boldrup et al., 2004) in place of a chloroform : methanol extraction. Briefly, cerebella from adult Wistar or Sprague-Dawley rats that had been obtained previously and stored frozen at −80°C were thawed and homogenized in 0.32 M sucrose containing 50 mM sodium phosphate, pH 8. Homogenates were centrifuged at 100 000 g for 60 min at 4°C and the supernatants (‘cytosolic fractions’) were collected and stored at −80°C in aliquots until used for assay. Protein concentrations for the assays were 3 µg/assay, the fractions being diluted with 10 mM Tris–HCl, 1 mM EDTA, pH 7.4.

The radiochemical assays contained enzyme source (cytosol or recombinant MGL), 10 mM Tris–HCl, 1 mM EDTA and test compound (10 µL, in vehicle [DMSO for troglitazone and JZL184; ethanol for N-arachidonoyl dopamine) in a volume of 175 µL. After pre-incubation, when appropriate, 25 µL of a solution of [³H]2-OG (4 µM, to give an assay concentration of 0.5 µM) in 10 mM Tris buffer, pH 7.4 containing 1% (w v⁻¹) fatty acid-free bovine serum albumin (BSA, assay concentration 0.125% w v⁻¹) was added and the samples were incubated for 10 min at 37°C. Thereafter, reactions were stopped by the addition of 400 µL of active charcoal mixture (80 µL charcoal + 320 µL 0.5 M HCl). After vortex mixing and phase separation, aliquots of the aqueous phase were counted for tritium content by liquid scintillation spectroscopy with quench correction. Blanks were prepared in the same manner but without enzyme source. FAAH assays with JZL184 were carried out using homogenates of whole brain (minus cerebella, 0.6 µg protein per assay) from adult Wistar or Sprague-Dawley rats using the method of Boldrup et al. (2004) and [³H]AEA (labelled in the ethanolamine part of the molecule, final concentration 0.5 µM) as substrate.

Analysis of data

K_m^app, V_max^app, pI₅₀ and IC₅₀, values, linear regressions and confidence limits were determined using the GraphPad Prism computer programme (GraphPad Software Inc., San Diego, CA, USA). The pI₅₀, and thereby IC₅₀ values, were calculated using the built-in programme ‘sigmoidal dose–response (variable slope)’ from the data expressed as % of vehicle controls using top (i.e. uninhibited) values of 100% and bottom (residual activity) values that were either set to zero or allowed to float. The two curves were compared using Akaike's informative criteria and the preferred model thereafter used.

Compounds

2-Oleoylglyceerol [glycerol – 1,2,3-³H] (1.48 TBq.mmol⁻¹) and AEA [ethanolamine-1-³H] (specific activity 2.22 TBq.mmol⁻¹) were purchased from American Radiolabeled Chemicals, Inc (St. Louis, MO, USA). Human recombinant His-tagged MGL expressed in Escherichia coli (hMGL), either as clear lysates, catalogue nos. #10008354 and #705194; (according to the manufacturers, the two lysate preparations are the same enzyme preparation, but are used as part of different enzyme assay kits. The #10008354 is now discontinued. We have indicated the catalogue number of the lysate used in the Figure legends) or as the enzyme that had been further purified using a nickel column, catalogue no. #10007812; non-radiolabelled AEA, troglitazone, rosiglitazone, ciglitazone, 15-deoxy-Δ^12,14-prostaglandin J₂, CAY10415 (5-[[4-[2-(5-ethyl-2-pyridinyl)-2-oxoethoxy]phenyl]methyl]-2,4-thiazolidinedione, compound 10 of Tanis et al. 1996), CAY10514 (8-hydroxy-8-[2-(pentyloxy)phenyl]-5-octynoic acid, methyl ester, Compound 14a of Caijo et al. 2005), JZL184 (4-nitrophenyl-4-(dibenzo[d][1,3]dioxol-5-yl(hydroxy)methyl)piperidine-1-carboxylate) and N-arachidonoyl dopamine were obtained from the (Cayman Chemical Co., Ann Arbor, MI, USA). AM404 (N-(4-hydroxyphenyl)-5Z,8Z,11Z,14Z-eicosatetraenamide), CP 55 940 ((-)-cis-3-[2-hydroxy-4-(1,1-dimethylheptyl)phenyl]-trans-4-(3-hydroxypropyl)cyclohexanol), JWH015 ((2-methyl-1-propyl-1H-indol-3-yl)-1-naphthalenylmethanone) and WIN55,212-2 ((R)-(+)-[2,3-dihydro-5-methyl-3-(4-morpholinylmethyl)pyrrolo[1,2,3-de]-1,4-benzoxazin-6-yl]-1-naphthalenylmethanone mesylate) were obtained from Tocris Bioscience (Ellisville, MO, USA). NPA, fatty acid-free BSA, 4-nitrophenol and non-radioactive 2-OG were obtained from the Sigma Chemical Co., St. Louis, MO, USA.

Results

Preliminary screen of selected compounds for inhibition of MGL using the NPA assay

The NPA assay was used with the clear lysates of recombinant human His-tagged MGL expressed in E. coli (‘hMGL lysate’) as enzyme source. Initial experiments indicated that the OD₄₀₅ for the NPA hydrolysis product, 4-nitrophenol, was linear over the range 0.01–0.15 mM (r² = 0.998, seven concentrations tested) and that the OD₄₀₅ measured when the hMGL lysates were incubated with NPA showed the appropriate dependency upon the incubation time and enzyme concentration (data not shown). Over a range of NPA concentrations (0.08–0.64 mM), the hydrolysis produced by the hMGL lysates was saturable, with a K_m value of 0.23 mM (data not shown), consistent with the study of Muccioli et al. (2008) who used a different hMGL preparation. Finally, the rate of NPA hydrolysis showed the expected pH optimum (∼7), consistent with the literature (Di Marzo et al., 1999; Goparaju et al., 1999), and was reduced by the alternative substrates, 2-AG and 2-OG, but not by AEA, which is not a substrate for this enzyme (Dinh et al., 2002) (data not shown).

A series of 96 compounds available at the department were tested at two concentrations, 3 and 10 µM. The compounds tested included a number of agents known to interact with the CB system, non-steroidal anti-inflammatory agents, compounds with different actions upon targets in the brain (such as, for example, antipsychotic and antidepressant drugs) and some naturally occurring compounds. The initial screen was carried out with single assays, albeit with separate controls for the 3 µM and 10 µM concentrations, to allow the identification of ‘hits’, and the values should be considered in this light. Four compounds were found to produce more than 60% inhibition at concentrations of both 3 and 10 µM: the CB receptor agonist CP55,940, the endogenous transient receptor potential vanilloid 1 (TRPV1) agonist N-arachidonoyl dopamine, the FAAH inhibitor and TRPV1 antagonist, N-arachidonoyl serotonin, and the PPARγ ligand, troglitazone (see Table 1 for a selection of the compounds). Three compounds produced >60% inhibition at 10 but not 3 µM concentrations: the AEA uptake inhibitor AM404, the lipase (including diacylglycerol lipase) inhibitor tetrahydrolipstatin, and the CB₂ receptor-selective agonist JWH015. Among other compounds tested, it was noted that in some cases, the compounds increased the observed rate of NPA hydrolysis, a case in point being JWH133 (see Table 1). The high value for JWH133 was not due to assay interference, as this compound, at the concentrations tested, did not affect the OD at the 0 min time point. In a repeat experiment, activation was also seen when the compound was pre-incubated with the enzyme for 30 or 60 min prior to addition of substrate, but the degree of activation was very variable, and it was not seen in the absence of a pre-incubation phase (see legend to Table 1). This property of the compound was not studied further, given its limited robustness and given that the concentrations used are much higher than needed to interact with CB₂ receptors (Huffman et al., 1999), but it may reflect some form of allosteric action on the enzyme. The raw data for the compounds corresponding to the major groups described above are given for reference purposes in Table 1.

Table 1.

Effects of 76 compounds with different biological actions upon the activity of human recombinant MGL (lysate assay, 0.25 mM NPA as substrate)

		MGL activity (% of control)
Compound	Primary pharmacological action	3 µM	10 µM
CB/TRPV1 receptor ligands and related compounds
ACEA	CB1 receptor agonist	126	81
AM251	CB1 receptor inverse agonist	69	62
AM404**	Blocks endocannabinoid reuptake	63	18
AM630	CB2 receptor inverse agonist	90	55
N-arachidonoyl serotonin	FAAH and TRPV1 inhibitor	37	32
N-arachidonoyl dopamine**	Endogenous TRPV1 ligand	15	−6
Capsazepine	TRPV1 antagonist	90	63
CP 55,940**	Non-selective CB receptor agonist	28	26
JWH015**	CB2 receptor agonist	53	37
JWH133a	CB2 receptor agonist	172	170
(±)-2-Methylarachidonyl-2′-fluoroethylamide	CB1 receptor agonist	89	99
Resveratrol	Antioxidant, CB1 receptor ligand	101	86
RHC-80267	DAG lipase inhibitor	102	71
(±)SLV 319	CB1 receptor inverse agonist	99	79
Tetrahydrolipstatin	Lipase inhibitor (incl. DAG lipase)	95	37
Win 55,212-2 mesylate**	Non-selective CB receptor agonist	67	25
Non-steroidal anti-inflammatory drugs
Acetylsalicylic acid*, diclofenac, dipyrone, fenbufen*, fenoprofen, flufenamic acid, ketoprofen*, meloxicam*, nabumetone, niflumic acid, nimesulide, sulindac, suprofen	Inhibition of cyclooxygenase	range 76–147	range 95–161
Antidepressants
Amitriptyline, clomipramine, desipramine, imipramine, nortriptyline, maprotiline	Monoamine reuptake inhibitors	range 82–120	range 78–122
Phenelzine	Monoamine oxidase inhibitor
Other CNS-active drugs and related compounds
Haloperidol, cis-(Z)-flupenthixol	Dopamine receptor antagonists	range 86–116	range 72–116
Spiperone*	DA/5-HT1A receptor antagonists
Ketanserin, pirenperone*	5-HT2 receptor antagonists
Promethazine	Histamine H1 receptor antagonist
Diazepam, flurazepam, temazepam	Benzodiazepines
Sodium barbital	Barbiturate
Nipecotic acid	GABA reuptake inhibitor
(+)MK801	NMDA receptor antagonist
Naloxone	Opioid receptor antagonist
Physostigmine sulphate	Acetylcholinesterase inhibitor
Other drugs and related compounds
Lidocaine, tetracaine	Na+-channel blockers	range 87–170	range 90–114
Verapamil	L-type Ca2+ channel blocker
Lovastatin	HMG-CoA reductase inhibitor
S(−)Atenolol, R(+)atenolol	ß-Adrenoceptor antagonists (S>R)
Captopril	ACE inhibitor
Theophylline, 8-phenyltheophylline	Adenosine receptor antagonists
Cromolyn sodium	‘Mast cell stabiliser’
Progesterone	Sex steroid hormone
Troglitazone**	PPARγ ligand	21	2
Naturally occurring, biologically active compounds
Coumestrol	Non-steroidal phytoestrogen	range 74–99	range 45–143
Baicalein, fisetin, luteolin, morin, phloretin, quercetin	Flavonoids; antioxidant and other properties
Other compounds (a selection)
3-Hydroxytyramine		Range 92–126	range 76–103
Benzamide
Hydrocinnamic acid
N-Acetyl-D-sphingosine
8-Methyl-N-vanillyl-6-noneamide
Thimerosal		61	45

Data are for single experiments, albeit run on different plates for the 3 µM and the 10 µM test concentrations, with ‘hits’ (>60% inhibition at the test concentration shown) given in bold text. Selected compounds (marked with asterisks) were retested at a later date at a concentration of either 10 µM (inactive compounds, *) or over a concentration range (**, compounds shown in Figure 1A) to confirm the reliability, but the values shown here are from the initial experiments.

^aJWH133 was also retested at a later stage, and the % of control activities seen after 0, 30 and 60 min pre-incubation, respectively, were: 3 µM, 100 ± 17, 129 ± 17 and 133 ± 25; 10 µM, 98 ± 20, 146 ± 8 and 159 ± 21 (means ± SEM, n = 3).

Inhibition of MGL by troglitazone and the other ‘hits’ in the primary screen

Six of the ‘hits’ described above were characterized further using hMGL lysates and NPA as substrate. Concentration–response curves for troglitazone, CP55,940, N-arachidonoyl dopamine and AM404, over the range 0.1–10 µM were undertaken and pI₅₀ values (IC₅₀ values in brackets) of 5.95 ± 0.04 (1.1 µM), 5.31 ± 0.07 (4.9 µM), 6.11 ± 0.08 (0.78 µM) and 5.51 ± 0.09 (3.1 µM), respectively, were found (Figure 1A). For WIN55,212-2 and JWH015, there was a large variability at the highest concentration tested, presumably related to the solubility of the compounds, precluding accurate determinations of their IC₅₀ values. The inhibition of MGL by troglitazone was compared with that for other PPARγ ligands (Figure 1B; note that the datasets for troglitazone in the two panels of Figure 1 are different, but show very similar potencies), and the potency order was found to be troglitazone > ciglitazone > rosiglitazone > 15-deoxy-Δ^12,14-prostaglandin J₂≈ CAY 10415 > CAY 10514 (Table 2). This differs considerably from their potency order as inhibitors of FAAH or as activators of PPARγ (see Table 2).

Figure 1.

Inhibition of the hydrolysis of 4-nitrophenyl acetate by human recombinant monoacylglycerol lipase lysates by (A) ‘hits’ from the initial screen of a small compound library and (B) peroxisome proliferator-activated receptor (PPAR) γ agonists. Shown are means and SEM (when not enclosed by the symbols), n = 3–4. The catalogue numbers of the lysates were for Panel A, #705194 and for Panel B 10008354.

Table 2.

Inhibition of NPA hydrolysis in hMGL lysates by PPARγ ligands

Compound	Structure	hMGL pI50[IC50 (µM)]	rat brain FAAH IC50 (µM)	PPARγ EC50 (µM) (h/m)
Troglitazone		5.94 ± 0.03 (1.2)	∼200 µMa	0.55/0.78b
Ciglitazone		5.60 ± 0.05 (2.5)	84 µMa	−/3c
Rosiglitazone		4.53 ± 0.04 (29)	∼200 µMa	0.043/0.076b
CAY 10415		4.31 ± 0.04 (49)
15-deoxy-Δ12,14-prostaglandin J2		4.32 ± 0.06 (48)	87 µMa	−/2d
CAY 10514		<4 (>100)		0.64/−e

The pI₅₀ values and hence IC₅₀ values were calculated from the raw data shown in Figure 1B, using hMGL lysates (catalogue no. #10008354) and 0.25 mM NPA.

^aThe FAAH data are for rat brain homogenates, 0.5 µM [³H]AEA as substrate, and an assay pH of 7.4, and are taken from Lenman et al. (2007). The PPARγ values for troglitazone, ciglitazone, rosiglitazone, 5-deoxy-Δ^12,14-prostaglandin J₂ and CAY10514 are for transactivation assays using human (h) or mouse (m) chimeric receptors and are from

^bWillson et al. (2000)

^cWillson et al. (1996)

^dKliewer et al. (1995))

^eCaijo et al. (2005).

Note that CAY10514 showed a low efficacy compared with rosiglitazone, which in the same study reported to have an EC₅₀ value of 0.004 µM. Comparable data are not (to our knowledge) reported for CAY10415. However, this compound increases insulin-stimulated inccorporation of [¹⁴C]acetate into lipid in differentiated mouse 3T3-L1 cells at 3–10 µM concentrations (Tanis et al., 1996).

The mode of inhibition by troglitazone of NPA hydrolysis by hMGL lysates was investigated. The inhibition showed no time dependency (Figure 2A), but dilution experiments suggested that it was not rapidly reversible. In the latter experiments, troglitazone was incubated with the human MGL lysates (catalogue no. #705194) for 60 min at room temperature prior to addition of substrate and either a fivefold dilution or no dilution prior assay for activity. For a fully reversible compound, the inhibitory potency at, say, 10 µM prior to dilution should be the same as that seen for an undiluted sample using 2 µM inhibitor, whereas for an irreversible or tight-binding compound, the inhibitory potency should resemble that seen for an undiluted sample using 10 µM inhibitor. For the irreversible compound JZL184, used as a positive control, the % of vehicle control activity (means ± SEM, n = 3) seen after pre-incubation with 50 nM compound followed by fivefold dilution was 7 ± 0.8, which was similar to that seen with the undiluted samples incubated with 50 nM compound (5 ± 1) and much lower than seen with the undiluted samples incubated with 10 nM JZL184 (43 ± 3). A similar pattern was seen with troglitazone: the % of control activity (means ± SEM, n = 3) seen after pre-incubation with 10 µM compound followed by fivefold dilution was 34 ± 3, which was similar to that seen with the undiluted samples incubated with 10 µM compound (33 ± 2) and much lower than seen with the undiluted samples incubated with 2 µM troglitazone (77 ± 4).

Figure 2.

Mode of inhibition by troglitazone of 4-nitrophenyl acetate (NPA) hydrolysis by human monoacylglycerol lipase (MGL) lysates (catalogue number #10008354). In Panel A, the lysates were pre-incubated with troglitazone for the times shown at room temperature prior to addition of NPA (0.25 mM) and assay for activity. In Panel B, no pre-incubation phase was used. Shown are means and SEM (when not enclosed by the symbols), n = 3, except for the data in Panel B with 0.32 mM NPA, where n = 2. The inset in Panel B shows a double reciprocal plot of the mean data with 0 and 0.75 µM troglitazone, to illustrate the competitive nature of the inhibition.

The saturation curves for NPA incubated with troglitazone (no pre-incubation phase) are shown in Figure 2B. The K_m^app values calculated from the mean data with 0 and 0.75 µM troglitazone were 0.17 and 0.70 mM, respectively. The corresponding V_max^app values (calculated using a post hoc calibration curve for 4-nitrophenol) were 0.44 and 0.57 nmol substrate hydrolysed per minute and assay tube, respectively, each assay tube containing 0.01 µL of the lysate preparation. At higher troglitazone concentrations, the K_m^app values were greater than the highest substrate concentration tested, and so could not be determined with accuracy. A slope re-plot with 0 and 0.75 µM troglitazone gave an approximate K_i value of 0.34 µM. This is in good agreement with the value of K_i 0.45 calculated from the IC₅₀ value of 1.1 µM (Table 2) assuming a competitive interaction and a K_m value of 0.17 mM.

Assay dependency of the inhibition of MGL by troglitazone, N-arachidonoyl dopamine and JZL184

In order to investigate whether the inhibition of MGL by the compounds was assay, substrate and/or species dependent, two assay systems were compared: (i) NPA as substrate, hMGL lysates or purified hMGL as enzyme sources, assays run at room temperature; (ii) 2-OG as substrate, rat cytosol, hMGL lysates or purified hMGL as enzyme sources; 0.125% w·v⁻¹ fatty acid-free BSA present, assays run at 37°C.

In order to establish the degree of assay variation for a well-characterized compound, we used the selective irreversible MGL inhibitor JZL184 (Figure 3). In all assays, the compound inhibited the substrate hydrolysis in a time-dependent manner, consistent with its irreversible mode of action (Long et al., 2009a). JZL184 inhibited NPA hydrolysis by the hMGL lysates with pI₅₀ values of 7.17 ± 0.04 (IC₅₀ value 68 nM) and 8.33 ± 0.04 (IC₅₀ value 4.7 nM) following 0 and 60 min of pre-incubation, respectively. When 2-OG was used as substrate, the potency was about fourfold lower (Figure 3A). In their original study, Long et al. (2009a) used recombinant mouse MGL with 100 µM 2-AG as substrate and reported IC₅₀ values of 6 nM, following a 30 min pre-incubation period. For the 2-OG assays, the control activity was affected by the pre-incubation phase; thus, the rate of hydrolysis of 2-OG by the pre-incubated lysates was 36 ± 0.8% of the corresponding activity seen for the samples that were not pre-incubated. This was even more apparent when the purified hMGL was used: the rate of hydrolysis of the pre-incubated control lysates was only 14 ± 3% of that seen in the absence of pre-incubation. A reduction in activity was also seen with pre-incubation in the NPA assays, and the reduction was more variable. However, the remaining activity after the pre-incubation phase was sensitive to inhibition by JZL184, with pI₅₀ values of 7.78 ± 0.12 (IC₅₀ value 17 nM) and 8.02 ± 0.08 (IC₅₀ value 9.6 nM) being found for NPA and 2-OG as substrates, respectively (Figure 3B). Finally, JZL184 inhibited rat cytosol 2-OG hydrolysis with pI₅₀ values of 6.46 ± 0.04 (IC₅₀ value 350 nM), 7.91 ± 0.10 (IC₅₀ value 12 nM, residual activity 16 ± 6%) and 8.24 ± 0.02 (IC₅₀ value 5.8 nM, residual activity 9 ± 1%) following 0, 30 and 60 min pre-incubation times, respectively (Figure 3C). In a single experiment, the effect of JZL184 upon 2-OG hydrolysis by the hMGL lysates was investigated at room temperature, rather than at 37°C. Approximate IC₅₀ values of 850 and 14 nM were found following pre-incubation times of 0 and 60 min, respectively, in line with the data at 37°C (data not shown). Taken together, the results show there is a time dependency in the sensitivity of MGL to JZL184, but the potencies for a given pre-incubation time are reasonably consistent in the different assays.

Figure 3.

Inhibition by JZL184 of the hydrolysis of 4-nitrophenyl acetate (NPA) or 2-oleoylglycerol (2-OG) by A, human monoacylglycerol lipase (MGL) lysates (catalogue no. #705194); B, purified human MGL; C, rat cerebellar cytosol fractions. Shown are means and SEM (when not enclosed by the symbols), n = 3–9. For comparative purposes, the effect of the compound upon the hydrolysis of anandamide (AEA) (0.5 µM) by rat brain membrane preparations is shown to illustrate the MGL-selectivity of the compound.

In the absence of a pre-incubation phase, N-arachidonoyl dopamine inhibited the hydrolysis by the hMGL lysates of NPA and 2-OG with pI₅₀ values of 6.11 ± 0.08 (IC₅₀ value 0.78) and 5.66 ± 0.03 (IC₅₀ value 2.2 µM), respectively), while the compound was less potent as an inhibitor of the hydrolysis of 2-OG by rat cytosol (pI₅₀ value 4.70 ± 0.04, IC₅₀ value 20 µM) (Figure 4A).

Figure 4.

The potencies of N-arachidonoyl dopamine (A) and troglitazone (B) towards monoacylglycerol lipase (MGL) in different assays. The compounds were not pre-incubated with the enzyme preparations prior to addition of substrate. Shown are means and SEM (when not enclosed by the symbols), n = 3–6. The catalogue number of the lysates was #705194. Note that the curves for the human MGL lysate with NPA as substrate are the same as shown in Figure 1A for the two compounds.

For troglitazone (no pre-incubation), there was a large variation in the inter-assay potencies. With NPA as substrate, the compound inhibited the purified hMGL with a threefold lower potency than seen for the lysate, and a pI₅₀ value of 4.61 ± 0.07 (IC₅₀ value 25 µM, residual enzyme activity 21 ± 7%) was seen for rat cytosols with 2-OG as substrate (Figure 4B). Rosiglitazone was also less potent as an inhibitor of 2-OG hydrolysis by the cytosols than of NPA hydrolysis by hMGL lysates, producing 31 ± 1% inhibition at the highest concentration tested (100 µM) for the cytosols (data not shown) as compared to an IC₅₀ value of 29 µM for the hMGL lysates (Table 2). Most surprisingly, troglitazone was a very poor inhibitor of the hMGL lysate when assayed radiochemically with 2-OG as substrate (Figure 4B). This was not due to the assay temperature used, as a single experiment with 2-OG carried out at room temperature gave an IC₅₀ value of approximately 62 µM for the inhibition by troglitazone of the hMGL lysate-catalysed hydrolysis of 2-OG (data not shown).

Three possible reasons for the discrepancy in potencies seen for troglitazone in the different assays using the same enzyme source were investigated:

1. The effect of troglitazone upon NPA hydrolysis is an artefact of the assay. For all compounds, the initial reading at t = 0 min was made to ensure that the compounds did not produce a signal at the OD used, but it is possible that troglitazone can quench the signal produced by NPA hydrolysis in the samples. However, this would have been expected to affect equally the signal seen in assays with the lysates and the purified enzyme, and would not have been likely to produce the effects it did in the dilution experiments or to give a competitive type of inhibition. Nonetheless, the possibility was checked by adding troglitazone to assay mixtures containing the lysate and NPA after a 20 min incubation period and reading the plate immediately and 2 min later. Over the concentration range tested (0.1–30 µM), troglitazone was without any effect on the observed absorbance at 405 nM (data not shown), indicating that effect of troglitazone is not due to quenching of the 4-nitrophenol signal.

2. The manner in which the substrate is presented to the enzyme may be of importance, given the difference in lipophilicity between 2-OG and NPA. In order to investigate this possibility, the inhibitory potency of troglitazone was investigated when 2-OG was emulsified using Triton X-100 and compared to that in the standard BSA-containing assay. However, preliminary experiments indicated that the enzyme activity towards 2-OG in the presence of Triton X-100 was very much lower than seen in the standard BSA-containing assay (∼12% for both MGL lysates and for rat cytosolic preparations using a Triton X-100 concentration of 5 mM to emulsify the substrate, final assay concentration 0.625 mM, that is, 0.04% w v⁻¹), and in the cytosolic preparations, there was no obvious difference in the sensitivity to troglitazone (data not shown).

3. Differences in fatty acid-free BSA concentration in the two assays affect the observed inhibitory potency of troglitazone. The presence of BSA is likely to reduce the free troglitazone concentration given its propensity to bind to albumin (>99.9% binding to human serum albumin, Shibukawa et al., 1995), and this may result in a reduced observed potency. In order to investigate this possibility, NPA assays were undertaken in the presence of different concentrations of fatty acid-free BSA (0–0.25% w v⁻¹) and with N-arachidonoyl dopamine and troglitazone as test compounds. The experiments were complicated by the fact that serum albumins have esterase activity (see Salvi et al., 1997), and this can be clearly seen in the blank values obtained at each fatty acid-free BSA concentration (Figure 5A and B). At the high BSA concentrations, the blank values are so high relative to the MGL activity that accurate determination of inhibitory potencies is not possible. However, the difference between the blank and enzyme-containing controls is approximately the same at 0 and 0.125% w·v⁻¹ BSA, and in these cases, the inhibition produced by 8 µM N-arachidonoyl dopamine and 10 µM troglitazone, appeared to be greater in the absence than in the presence of 0.125% w·v⁻¹ fatty acid-free BSA. Experiments were also undertaken using the hMGL lysates and 2-OG in the radiochemical assay, but lacking fatty acid-free BSA. In these experiments, N-arachidonoyl dopamine was a very potent inhibitor of the 2-OG hydrolysis (pI₅₀ value 6.58 ± 0.02, IC₅₀ 0.26 µM), that is, greater inhibition than seen in the presence of BSA (Figure 4A), although as a caveat it was noted that the vehicle used (ethanol) reduced the enzyme activity, compared to a DMSO vehicle. Troglitazone was also a better inhibitor of 2-OG hydrolysis by the lysates in the absence of BSA (pI₅₀ 4.43 ± 0.04, IC₅₀ 37 µM), although this value is still lower than seen with the lysates in the NPA assay (Figure 4B).

Figure 5.

Effect of fatty acid-free bovine serum albumin (BSA) upon the rate of hydrolysis of 4-nitrophenyl acetate (NPA) and its inhibition by (A) N-arachidonoyl dopamine, (B) troglitazone. The compounds were not pre-incubated with the enzyme preparations prior to addition of substrate. Note that in the figure, the blank values (containing the same concentration of fatty acid-free bovine serum albumin, but no hMGL lysate) are not subtracted. Shown are means and SEM., n = 3. The catalogue number of the lysates was #705194.

Discussion

In the present study, we used a high-capacity assay to investigate a series of compounds with respect to their MGL inhibitory properties, and characterized the two most potent in the screen with respect to their inter-assay variability. The screen identified several compounds which inhibited NPA hydrolysis by the hMGL lysates in the low micromolar range. With the exception of troglitazone (see below), the compounds were either arachidonoyl analogues (AM404, N-arachidonoyl dopamine) or compounds sharing the ability of 2-AG to interact with CB receptors (CP 55 940; JWH015; WIN 55,212-2). With respect to CP 55,940, JWH015 and WIN 55,212-2, the potencies towards the hMGL in the lysates are lower than the nanomolar concentrations required to interact with CB receptors (see McPartland et al., 2007), and so can best be described as ‘off-target’ actions of these compounds unlikely to contribute to their pharmacological effects. The same is probably true for N-arachidonoyl dopamine, which is a potent TRPV1 receptor agonist, although the IC₅₀ value seen here for human MGL is similar to that seen for the interaction of the compound with CB₁ receptors (Bisogno et al., 2000; Chu et al., 2003). The potency of AM404 towards hMGL in the lysates is similar to its potency as an inhibitor of endocannabinoid uptake (Beltramo et al., 1997), and so its effects upon MGL may contribute to the pharmacological actions of this compound in vivo. AM404 also inhibits the hydrolysis of 2-OG by rat cerebellar cytosolic fractions (IC₅₀ value of ∼20 µM, Vandevoorde and Fowler, 2005).

Two other compounds, N-arachidonoyl serotonin and tetrahydrolipstatin, were found to be ‘hits’ in the original screen, but were not investigated further. Tetrahydrolipstatin was also evaluated by Muccioli et al. (2008) using the NPA assay and a different hMGL preparation. They found it to be quite a potent inhibitor of hMGL, but not to produce more than ∼50% inhibition of the enzyme, even at high (100 µM) concentrations. However, it had an IC₅₀ value >10 µM in cytosolic fractions from monkey COS-7 cells when MGL was assayed with 2-AG as substrate (Ortar et al., 2008). N-arachidonoyl serotonin was reported not to inhibit the hydrolysis of 2-AG by rat cerebellar membranes (Saario et al., 2004), so it is possible that the effect seen here may be dependent upon the species and/or assay used. Muccioli et al. (2008) also tested a series of non-steroidal anti-inflammatory agents (including ibuprofen, naproxen, flubiprofen and diclofenac sodium), and found <50% (100 µM) and ≤33% (10 µM) inhibition in all cases. This lack of effect of this class of drugs upon hMGL is consistent with our data (see Table 1). Finally, the organic mercurial compound thimerosal produced 55% inhibition at 10 µM in our screen, a finding consistent with the ability of other organic mercurials to inhibit MGL activity (see Tornqvist and Belfrage, 1976; Chau and Tai, 1988; Tarzia et al., 2007).

The finding that the PPARγ agonist troglitazone inhibits NPA hydrolysis by the hMGL lysates extends previous work showing that PPARγ agonists can inhibit FAAH (Lenman et al., 2007), and that endocannabinoids and related biological lipids can interact with PPAR subtypes (see O'Sullivan, 2007). PPARγ is a ligand-activated nuclear receptor, and the PPARγ ligands rosiglitazone and pioglitazone are currently used for the treatment of type 2 diabetes. For the thiazolidinedione compounds tested here, the rank order of potencies for inhibition of MGL (troglitazone > ciglitazone > rosiglitazone) was different from that seen for either inhibition of FAAH (ciglitazone > rosiglitazone ≈ troglitazone, Lenman et al., 2007) or activation of PPARγ (rosiglitazone >> troglitazone > ciglitazone; Willson et al., 1996; 2000;). The initial inhibition of NPA hydrolysis in the hMGL lysates by troglitazone was competitive in nature (K_i 0.34 µM). Although the inhibition was not time dependent, dilution experiments indicated that the inhibition was not rapidly reversible.

A degree of variation in potency is expected for a compound when tested in different assay systems, and in the case of JZL184, the primary variable influencing the observed potency is the pre-incubation time, as expected of an irreversible inhibitor. In the case of N-arachidonoyl dopamine, the variation in potency seems to be species-related, since the potencies were reasonably similar for the hMGL lysates when assayed with the two different substrates. For troglitazone, however, the pattern was at first sight more surprising, given that the compound inhibited the hydrolysis of NPA much more potently than the hydrolysis of 2-OG when the hMGL lysates were used. However, troglitazone has a very high degree of albumin binding (Shibukawa et al., 1995), and the difference in potency appears to be related in part to the concentration of fatty acid-free BSA in the assay medium, since its addition reduced the inhibition by both troglitazone and N-arachidonoyl dopamine of NPA hydrolysis by the hMGL lysates. The reverse assay (the use of 2-OG in the absence of fatty acid-free BSA) is more difficult to interpret, since the albumin concentration will affect the level of free substrate, and hence the observed inhibitory potency of a competitive inhibitor like troglitazone. We found that the sensitivity of hMGL to inhibition by N-arachidonoyl dopamine was indeed increased in the 2-OG assay when BSA was excluded. We have previously reported that the potency of VDM11 (an arachidonoyl-based uptake inhibitor with a structure close to AM404) and arachidonoyl serinol (serinol is 2-amino-1,3-propanediol) are roughly twice as potent as inhibitors of the hydrolysis of 2-OG by rat cerebellar membranes when assayed in the absence of fatty acid-free BSA compared to when assayed at 0.125% w·v⁻¹ albumin (Vandevoorde and Fowler, 2005), which is entirely consistent with the present data with N-arachidonoyl dopamine and with the known ability of its analogues arachidonate and AEA to bind to albumin (Bojesen and Bojesen, 1994; Bojesen and Hansen, 2003; Oddi et al., 2009). Why the sensitivity of the 2-OG hydrolysis to inhibition by troglitazone is not similarly increased in the absence of BSA is unclear, but our current hypothesis is that 2-OG presents itself to the enzyme in a different manner to NPA and by doing so is less amenable to inhibition by troglitazone (but not by N-arachidonoyl dopamine or JZL184).

An obvious question raised by the present study is whether the ability of troglitazone to inhibit MGL is seen in man after normal dosing of the drug. Troglitazone was the first thiazolidinedione used clinically for the treatment of diabetes type 2. It was withdrawn from the US market at the request of the FDA in 2000 due to serious hepatotoxicity issues, the mechanism of which is controversial, but may involve the production of reactive metabolites and/or the ability of the compound to impair mitochondrial function (see Masubuchi, 2006). However, pharmacokinetic data were obtained prior to its withdrawal. In healthy subjects, 7 days of treatment with troglitazone at doses of 200, 400 and 600 mg (doses in the range that was used to reduce plasma glucose in patients with type 2 diabetes, Johnson et al., 1998) produced maximum serum concentrations of 0.90, 1.61 and 2.82 µg/mL, respectively, corresponding to concentrations of 2.0, 3.6 and 6.4 µM, respectively. The corresponding trough plasma concentrations were 0.3, 0.5 and 0.8 µM, respectively (Loi et al., 1999). Although these numbers do not consider the influence of the plasma protein binding of troglitazone (>99.8%, Shibukawa et al., 1995), they are at least in the same range as seen for the inhibition of NPA hydrolysis by the hMGL lysates, which in turn are similar to the potency of the compound for activation of human PPARγ in transactivation assays (EC₅₀ value 0.55 µM, Willson et al., 2000). Given the variation in potency of troglitazone for the different MGL assays, a ‘best case’ speculation is that MGL may have been inhibited upon normal troglitazone dosing. On the other hand, it is highly unlikely that the enzyme is inhibited following normal rosiglitazone dosing to patients with type 2 diabetes, given that the compound is used in lower doses than troglitazone (the maximum plasma concentration obtained after an 8 mg tablet is ∼600 ng/mL, corresponding to 1.7 µM; Cox et al., 2000); is an order of magnitude more potent than troglitazone towards PPARγ; and is an order of magnitude less potent than troglitazone as an inhibitor of NPA hydrolysis by the hMGL lysates and of 2-OG hydrolysis by the rat cerebellar cytosol fractions.

In conclusion, the present study has investigated a series of compounds with primary biological actions in areas where the endocannabinoid system is known to operate, and identified the PPARγ ligand troglitazone as a reasonably potent inhibitor of NPA (but not 2-OG) hydrolysis by hMGL lysates. PPARγ ligands may be useful in the treatment of diseases such as arthritis and neuropathic pain (Fehrenbacher et al., 2009; Giaginis et al., 2009), diseases where modulators of the endocannabinoid system are also being explored as treatment strategies (see Yao et al., 2008). It may thus be possible to use troglitazone as a template for the design of novel compounds without its hepatotoxicity but which have potent effects upon both MGL and PPARγ, as potential novel analgesics and anti-arthritis drugs. The study also underlines the importance of investigating MGL inhibitory potency in several different assays, to ensure that the inhibitory properties of the compound are not assay dependent.

Glossary

Abbreviations

NPA

4-nitrophenyl acetate

AEA

anandamide

2-AG

2-arachidonoylglycerol

AM404

N-(4-hydroxyphenyl)-5Z,8Z,11Z,14Z-eicosatetraenamide

BSA

bovine serum albumin

CAY10415

(5-[[4-[2-(5-ethyl-2-pyridinyl)-2-oxoethoxy]phenyl]methyl]-2,4-thiazolidinedione

CAY10514

8-hydroxy-8-[2-(pentyloxy)phenyl]-5-octynoic acid, methyl ester

CB

cannabinoid

CP55 940

(-)-cis-3-[2-hydroxy-4-(1,1-dimethylheptyl)phenyl]-trans-4-(3-hydroxypropyl)cyclohexanol

FAAH

fatty acid amide hydrolase

JWH015

(2-methyl-1-propyl-1H-indol-3-yl)-1-naphthalenylmethanone

JZL184

4-nitrophenyl-4-(dibenzo[d][1,3] dioxol-5-yl(hydroxy)methyl)piperidine-1-carboxylate

MGL

monoacylglycerol lipase

2-OG

2-oleoylglycerol

PPAR

peroxisome proliferator-activated receptor

URB597

3′-(aminocarbonyl)[1,1′-biphenyl]-3-3-yl)-cyclohexylcarbamate

WIN 55,212-2

(R)-(+)-[2,3-dihydro-5-methyl-3-(4-morpholinylmethyl)pyrrolo[1,2,3-de]-1,4-benzoxazin-6-yl]-1-naphthalenylmethanone mesylate

Conflicts of interest

None declared with respect to the present work.