Monoacylglycerol Lipase Inhibition in Human and Rodent Systems Supports Clinical Evaluation of Endocannabinoid Modulators

Monoacylglycerol lipase (MGLL) is the primary degradative enzyme for the endocannabinoid 2-arachidonoylglycerol (2-AG). The first MGLL inhibitors have recently entered clinical development for the treatment of neurologic disorders. To support this clinical path, we report the pharmacological characterization of the highly potent and selective MGLL inhibitor ABD-1970 [1,1,1,3,3,3-hexafluoropropan-2-yl 4-(2-(8-oxa-3-azabicyclo[3.2.1]octan-3-yl)-4-chlorobenzyl)piperazine-1-carboxylate]. We used ABD-1970 to confirm the role of MGLL in human systems and to define the relationship between MGLL target engagement, brain 2-AG concentrations, and efficacy. Because MGLL contributes to arachidonic acid metabolism in a subset of rodent tissues, we further used ABD-1970 to evaluate whether selective MGLL inhibition would affect prostanoid production in several human assays known to be sensitive to cyclooxygenase inhibitors. ABD-1970 robustly elevated brain 2-AG content and displayed antinociceptive and antipruritic activity in a battery of rodent models (ED50 values of 1–2 mg/kg). The antinociceptive effects of ABD-1970 were potentiated when combined with analgesic standards of care and occurred without overt cannabimimetic effects. ABD-1970 also blocked 2-AG hydrolysis in human brain tissue and elevated 2-AG content in human blood without affecting stimulated prostanoid production. These findings support the clinical development of MGLL inhibitors as a differentiated mechanism to treat pain and other neurologic disorders.

Direct pharmacological activation of the CB receptors by Cannabis preparations, tetrahydrocannabinol, and synthetic cannabinoids can elicit therapeutically beneficial effects on pain, spasticity, appetite, and nausea (Pertwee, 2012); however, dose-limiting central nervous system (CNS) effects of such exocannabinoids limit their broad use as pharmaceuticals. An alternative strategy to exocannabinoids is amplification of endocannabinoid signaling through inhibition of endocannabinoid degradation. MGLL is the major 2-AG hydrolase in the CNS and most peripheral tissues, and genetic or pharmacological inactivation of this enzyme reduces 2-AG hydrolysis and elevates tissue 2-AG concentrations in rodents (Long et al., 2009b;Chanda et al., 2010;Schlosburg et al., 2010). MGLL inhibitors produce an array of CB1/2-mediated effects in animal models, including antinociceptive (Grabner et al., 2017), anxiolytic (Patel et  s This article has supplemental material available at jpet.aspetjournals.org. and antiepileptogenic effects (Griebel et al., 2015;von Rüden et al., 2015;Sugaya et al., 2016), without inducing the full spectrum of stereotypical behaviors observed after administration of direct CB1 agonists (Long et al., 2009b;Ignatowska-Jankowska et al., 2014. These observations may reflect that MGLL inhibitors potentiate ongoing 2-AG signaling while preserving the spatial and temporal specificity of the endocannabinoid system, unlike exocannabinoids, which indiscriminately activate CB1 throughout the brain. The product of MGLL-mediated 2-AG hydrolysis, arachidonic acid (AA), is the precursor for prostanoid signaling molecules (Funk, 2001). MGLL-mediated 2-AG hydrolysis serves as one of the sources of AA in the rodent nervous system (Long et al., 2009b) and contributes to brain levels of proinflammatory prostanoids . MGLL inactivation in mice reduces molecular and cellular signs of neuroinflammation and is protective in models of neurodegeneration and status epilepticus through CB-independent mechanisms, suggesting involvement of prostanoid suppression Chen et al., 2012;Piro et al., 2012;Terrone et al., 2018;Zhang and Chen, 2018). The contribution of MGLL to AA metabolism in the mouse is tissue dependent . This anatomic segregation suggests that inhibition of MGLL may afford some of the antiinflammatory effects associated with a reduction in prostanoid signaling without the side effects caused by nonsteroidal antiinflammatory drugs (NSAIDs) or coxibs, which reduce prostanoid signaling through the inhibition of cyclooxygenase COX1 and/or COX2 enzymes. In rodents, MGLL inhibition is devoid of gastrointestinal effects and instead protects from NSAIDinduced gastric hemorrhages via CB1 . The effects of MGLL blockade on prostaglandin production and signaling in human cellular systems have, to our knowledge, not been reported but could potentially further differentiate MGLL inhibitors from COX1/2 inhibitors.
To support clinical investigation of endocannabinoid modulation, we report herein the pharmacological characterization of an advanced compound ABD-1970 [1,1,1,3,3,3hexafluoropropan-2-yl 4-(2-(8-oxa-3-azabicyclo[3.2.1]octan-3yl)-4-chlorobenzyl)piperazine-1-carboxylate] ( Fig.1) (Cisar et al., 2018) that serves as a highly potent and selective MGLL inhibitor. We show that oral administration of ABD-1970 to rodents leads to potent and sustained inhibition of MGLL and accumulation of 2-AG in the brain, which is accompanied by dose-dependent antinociceptive and antipruritic activity. The antinociceptive effects of ABD-1970 combine beneficially with analgesic standards of care and occur without overt cannabimimetic effects. Studies with ABD-1970 confirm that MGLL is a principal regulator of 2-AG in the human brain and demonstrate that in stimulated human blood and endothelial cells, blockade of this enzyme does not affect prostanoid production and thus clearly differentiates from COX inhibition. Together, these results support the pursuit of MGLL inhibitors for the safe and differentiated treatment of neurologic disorders in humans.
Gel-based ABPP analysis was performed as described previously (Leung et al., 2003) using three ABPP probes: FP-Rh probe (Patricelli et al., 2001), HT-01 (Hsu et al., 2012), and JW912 (Chang et al., 2013). Brain membrane homogenates (50 mg), PC3 membrane lysates (100 mg), PC3 cells (100-mm dishes), human tissue homogenates (50 mg), or human whole blood (4 ml) were treated with ABD-1970 or DMSO for 30 minutes at 37°C. Lysates were prepared from compoundtreated PC3 cells and PBMCs were isolated from compound-treated whole blood, as described in the Supplemental Methods. ABPP analysis was performed by treating the tissue or cell homogenates with 1 mM FP-Rh (all tissue homogenates and PBMCs), HT-01 (rodent and dog brain homogenates), or JW912 (PC3 membranes) for 30 minutes at 25°C. Reactions were quenched with 4Â SDS-PAGE loading buffer and separated by SDS-PAGE (10% acrylamide), and fluorescence was visualized in-gel with a ChemiDox XRS fluorescence imager (Bio-Rad, Hercules, CA). Fluorescence is shown in grayscale. Quantitation of fluorescent band intensities was performed by densitometric analysis using ImageJ software (version 1.5; National Institutes of Health, Bethesda, MD). Integrated peak intensities were generated for bands corresponding to MGLL activity (labeled with FP-Rh or JW912) or a/b hydrolase domain containing 6 (ABHD6) activity (labeled with HT-01 or JW912). IC 50 values were calculated by curve fitting semi-log-transformed data (x-axis) by nonlinear regression with a four-parameter, sigmoidal dose-response function (variable slope) in Prism software (version 6; GraphPad, La Jolla, CA).
Quantitative mass spectrometry (MS)-based ABPP was used to profile the selectivity of ABD-1970 in human brain homogenates using the FP-biotin probe. Human prefrontal cortex membrane homogenates were treated with 0.01-10 mM ABD-1970 or DMSO for 30 minutes at 37°C. Samples were treated with FP-biotin, enriched by avidin chromatography, and digested by trypsin, essentially as previously described (Jessani et al., 2005). Tryptic digests of the inhibitor-and DMSO-treated samples were isotopically labeled by reductive dimethylation of primary amines with natural or isotopically heavy formaldehydes (Boersema et al., 2009) and subsequently analyzed by liquid chromatography (LC)-tandem mass spectrometry (MS/MS) on a Velos Elite Orbitrap mass spectrometer (Thermo Fisher, Waltham, MA). Peptide spectral matching was performed with the complete human UniProt database using the ProLuCID algorithm (version 1.3.3, Xu et al. 2015), and the resulting matches were filtered using DTASelect (version 2.0.47, Cociorva et al. 2007). Quantification of light/heavy ratios was performed using the Cimage algorithm Weerapana et al. 2010. Data were plotted as the percent inhibition relative to DMSO-treated samples. Full details can be found in the Supplemental Methods.
Determination of Inhibitor Potency by Substrate Assays. 2-AG hydrolysis activity in human brain cortex membrane homogenates (20 mg in 200 ml PBS) treated with ABD-1970 or DMSO vehicle (30 minutes at 37°C) was assessed using an MS-based substrate assay, essentially as previously described (Blankman et al., 2007). Full experimental details can be found in the Supplemental Methods.
The potency of ABD-1970, independent of preincubation time, was determined as the k inact /K i using purified recombinant human MGLL protein and a fluorogenic substrate assay. Full-length human MGLL was expressed in Escherichia coli bacteria with an N-terminal hexahistidine tag (Evotec, Watertown, MA), purified by nickle nitrilotriacetic acid affinity chromatography followed by size exclusion chromatography (Evotec), and stored in 20 mM Tris, pH 8.0, 100 mM NaCl, and 10% glycerol. To initiate the assay, MGLL (4 nM) was added to the assay system consisting of ABD-1970 (62.5-3000 nM), 10 mM butyl resorufin substrate (Sigma-Aldrich), 200 mM KCl, 1 mM EDTA, and 50 mM Hepes (pH 7.0) in a total volume of 100 ml. Reactions without enzyme or inhibitor were included as controls. Immediately after the addition of MGLL protein, fluorescence was measured in a Neo2 multimode meter (BioTek, Winooski, VT) set for excitation at 530 nm and emission at 587 nm. Data were collected at room temperature over 10 minutes, with individual reads every 12-13 seconds. The MGLL-dependent hydrolysis of the butyl resorufin substrate was monitored continuously to provide a progress curve for the rate of enzyme inactivation at varying inhibitor concentrations. The progress curve data were corrected by subtracting the average of four reactions run without enzyme. Nonlinear fitting (GraphPad Prism) of these corrected progress curves to a single exponential equation provided the first-order rate constant, k obs , at each concentration of inhibitor tested. The k obs value was plotted against ABD-1970 concentration and the curve was fit using the following equation to obtain k inact and K i app : k obs 5 k inact [I]/([I] 1 K i app ). The true K i was subsequently calculated using the following equation: K i app 5 K i (1 1 [S]/K m ) (Copeland, 2000). Each experiment was performed with quadruplicate assay points and the entire experiment was repeated three times.
In Vitro Reversibility of Inhibition and Recovery of Activity in Cultured Cells. The persistence of inhibition of human MGLL by ABD-1970 in vitro was evaluated after removal of free compound by multiple rounds of centrifugation. Human embryonic kidney 293 cells containing the SV40 T-antigen were grown in Dulbecco's modified Eagle's medium (Gibco, Grand Island, NY) containing 10% fetal calf serum (FCS; Omega Scientific, Tarzana, CA) at 37°C under 5% CO 2 to approximately 60% confluence and were transfected with full-length human MGLL with an N-terminal FLAG epitope tag or an empty vector (pFLAG-CMV-6b) using the Fugene 6 reagent (Promega, Madison, WI). After 48 hours, cell membrane lysates were prepared as described in the Supplemental Methods, and an aliquot (490 mg in 490 ml PBS) was treated with 1 mM ABD-1970 (10 ml of 50 mM) or DMSO for 30 minutes at 37°C. Cell lysates were subjected to two rounds of centrifugation (100,000g for 30 minutes at 4°C) and the membrane lysates were collected immediately (time 5 0 hours), after 8 hours, and after 24 hours at room temperature and rapidly frozen on dry ice. Enzymatic activity of human MGLL was assessed in these samples after resuspension in PBS to 1 mg/ml using gel-based ABPP with the FP-Rh probe.
The recovery of MGLL activity in cultured human PC3 cells was evaluated after treatment with ABD-1970 and removal of free compound. PC3 cells were grown to 70%-80% confluence in F-12K medium (Gibco) supplemented with 10% FCS at 37°C with 5% CO 2 , washed twice with PBS, and then incubated for 30 minutes at 37°C with serum-free cell culture medium containing ABD-1970 (10 nM) or DMSO (0.01% final concentration). Cells were washed with PBS to remove unbound ABD-1970, fresh media containing 10% FCS were added, and cells were incubated at 37°C with 5% CO 2 for up to 72 hours. Cells were harvested at 0.5, 6, 24, 48, and 72 hours after the removal of ABD-1970 and cell lysates were prepared. MGLL activity in ABD-1970-or DMSO-treated PC3 lysates was then assessed using gel-based ABPP analysis with the JW912 probe. The average recovery rate of MGLL activity was calculated by linear regression analysis using Prism software. The slope of the fit gave the average MGLL activity recovery rate in percent MGLL activity per hour.

In Vivo Studies
All animal experiments at Abide Therapeutics were performed in accordance with the guidelines outlined by the Guide for the Care and Use of Laboratory Animals and the Animal Welfare Act and were approved by the Explora Biolabs Institutional Animal Care and Use Committee. Animal experiments carried out at external institutions were performed in accordance with the individual institutions and their respective country's policies governing the ethical and humane treatment of laboratory animals. In all studies, animals were maintained under a 12-hour/12-hour light/dark cycle and allowed free access to food and water throughout the duration of the experiments.
Dose and Time-Course Effects of ABD-1970 in Rodents. Male ICR mice (Envigo, Livermore, CA) and male Sprague-Dawley rats (Charles River Laboratories, Wilmington, MA) aged 6-10 weeks at the time of dosing were administered ABD-1970 by oral gavage (5 ml/kg volume). ABD-1970 was prepared fresh on the day of dosing in 0.5% methylcellulose (400cP; Sigma-Aldrich) vehicle for the rat doseresponse study or in 7:2:1 polyethylene glycol 400/ethanol/PBS vehicle for the rat time-course and mouse studies. Maximal dispersal of the compound was achieved by bath and probe sonication until a fine white suspension was formed. Animals were administered single oral doses of ABD-1970 (0.1-30 mg/kg for mice and 0.1-10 mg/kg for rats). At specified time points after ABD-1970 administration, animals were anesthetized with isoflurane, and blood samples were collected by cardiac puncture into syringes precoated with EDTA (500 mM disodium; Teknova, Hollister, CA). After blood collection, animals were killed by decapitation, and the brains were removed and rinsed in PBS before freezing in liquid nitrogen. One brain hemisphere was submitted for ABD-1970 concentration analysis by LC-MS/MS, and the other hemisphere was further dissected to remove two adjacent forebrain sections (approximately 100-200 mg) and was submitted for analysis of MGLL inhibition by ABPP with the FP-Rh probe, as described above, and for brain lipid concentrations by LC-MS/MS, as described below.
Because of the instability of ABD-1970 in rodent blood and plasma that is thought to be a consequence of carboxylesterase (CES) enzymes, which are known to metabolize ester and carbamate xenobiotics and are more abundant in rodent plasma than other higher mammal species, including dogs, primates, and humans (Berry et al., 2009), blood samples were immediately placed into an equal volume of acetonitrile to stabilize ABD-1970. After mixing, the samples were centrifuged (17,000g for 2 minutes), and the resulting whole blood supernatants were transferred and submitted to LC-MS/MS analysis for whole blood ABD-1970 measurements.
Brain Lipid Analysis. Quantitation of endocannabinoid and eicosanoid lipids in brain tissue after organic extraction was performed by LC-MS/MS analysis on an Agilent 1260 LC system coupled to an Agilent 6460 Triple Quadrupole mass spectrometer (Agilent Technologies, Santa Clara, CA). Full experimental details are presented in the Supplemental Methods.
Formalin Pain Model. Male Sprague-Dawley rats (171-235 g at the time of dosing; Harlan, Huntingdon, UK) were administered an intraplantar injection of formalin (50 ml, 2.5%) into the dorsal surface of the right hindpaw to induce acute pain (Saretius Ltd., Reading, UK). Immediately after, paw licking behavior, a correlate of spontaneous pain (Abbott et al., 1995), was tracked using the Laboras (Laboratorium Animal Behavior Observation, Registration and Analysis System; Metris B.V., Hoofddorp, The Netherlands) automated rodent behavioral tracking system, which was validated to differentiate and automatically recognize behaviors and locomotor activity of individually housed rats. Animals injected with formalin were immediately placed into the Laboras cages [22.5 cm (width) Â 42.5 cm (length) Â 19.0 cm (height)] and data on the duration of paw licking behavior (in seconds) were collected from two distinctive phases: 1) the early nociceptive phase occurring 0-5 minutes after formalin injection and 2) the late phase thought to be associated with central sensitization, which peaks 10-30 minutes after formalin injection.
To establish a pharmacokinetic (PK)/pharmacodynamic (PD) relationship, animals were administered vehicle (0.5% methylcellulose), ABD-1970 (1, 3, and 10 mg/kg), or the positive control pregabalin (60 mg/kg) 240 minutes before formalin injection. Forty-five minutes after formalin injection and behavioral monitoring and approximately 285 minutes after treatment, animals were anesthetized with isoflurane and blood samples were collected by cardiac puncture. After blood collection, animals were killed by decapitation and the brains were removed. Samples were dissected and processed as described above for analysis of MGLL inhibition by ABPP and ABD-1970 and 2-AG concentrations by LC-MS/MS.
For the ABD-1970 and pregabalin combination study, male Sprague-Dawley rats (214-235 g; Harlan) were administered ABD-1970 (2 mg/kg) by oral gavage (5 ml/kg volume) immediately followed by a separate oral administration (5 ml/kg volume) of pregabalin (10 mg/kg) 240 minutes before formalin administration. For the monotherapy control groups, ABD-1970 or pregabalin was administered along with separate oral gavage administrations of vehicle (0.5% methylcellulose). Paw licking behavior was monitored for 30 minutes using the Laboras behavior tracking system. For the ABD-1970 and morphine combination study, male Sprague-Dawley rats (225-269 g at the time of dosing; Harlan) were administered vehicle (0.5% methylcellulose) or ABD-1970 (1 mg/kg) by oral gavage. After 210 minutes, vehicle (saline) or morphine (2.49 mg/kg) was administered by subcutaneous injection. After an additional 30 minutes (240 minutes after ABD-1970 and 30 minutes after morphine), formalin was administered into the hindpaw and paw licking behavior was monitored using the Laboras behavior tracking system.
Mechanical allodynia in the CFA-injected paw was measured prior to CFA injection and 0 (24 hours after CFA and before vehicle or ABD-1970), 2, 4, and 6 hours after compound administration by determining withdrawal thresholds to an electronic von Frey instrument (Bioseb, Vitrolles, France) applied perpendicularly with increasing force to the plantar surface of the paw. The threshold for paw withdrawal was calculated by taking the average of two to three repeated von Frey applications. Observers were blinded to the identity of the treatment groups.
Plantar Incision Model of Postoperative Pain. Postoperative pain was induced in male Sprague-Dawley rats (220-250 g; Charles River Laboratories) as described previously (Brennan et al., 1996) (Pharmaron). Briefly, rats were anesthetized with isoflurane and placed in a dorsal recumbent position, and the plantar surface of the left hind foot was draped and prepared aseptically. On the plantar surface of the foot, a 1-cm incision was made at 0.5 cm from the edge of the heel and extending distally through the skin and fascia. The plantaris muscle was exposed by blunt dissection, elevated and incised longitudinally, and the skin was closed with two interrupted mattress sutures (3-0, polydioxanone).
Animals were administered vehicle (0.5% methylcellulose) or ABD-1970 (20 mg/kg) by oral gavage 60 minutes before anesthesia induction and plantar incision. Buprenorphine (0.03 mg/kg) was administered Pharmacological Characterization of MGLL Inhibitor ABD-1970 by intraperitoneal injection immediately before the procedure. Mechanical allodynia in the operated paw was measured before compound administration (t 5 0) to establish presurgical baseline withdrawal thresholds, and at 1, 3, and 5 hours after incision using an electronic von Frey instrument (Bioseb). Observers were blinded to the identity of the treatment groups.
Chronic Constriction Injury Model of Peripheral Neuropathic Pain. Chronic constriction injury (CCI) on the left common sciatic nerve was performed in male Sprague-Dawley rats (Harlan) to induce neuropathic pain as described previously (Bennett and Xie, 1988) (Aquila BioMedical Ltd., Edinburgh, UK). After recovery from anesthesia, animals were recovered for 7 days before behavioral testing was initiated.
Mechanical allodynia was assessed in rats by measuring the withdrawal threshold to von Frey filaments of increasing diameter and force. Animals were placed in a cage with a wire mesh bottom, and a series of calibrated von Frey filaments [3.84 (0.6 g), 4.17 (1.4 g), 4.56 (3.8 g), 4.93 (7.1 g), 5.18 (12.4 g), and 5.46 (20.5 g)] was applied to the plantar surface of the left hind paw. Each paw was tested six times with von Frey fibers presented consecutively. A minimum of three of six withdrawals was considered a positive response. The middle filament (4.56) was applied first; if a positive response was observed, the next weaker filament was then tested. Conversely, if the rat did not withdraw its paw at least three times, a larger filament would be tested until a 50% (three of six) withdrawal response was observed. Testing continued in this manner until a withdrawal threshold was reached.
von Frey thresholds were measured prior to CCI on all rats to establish a baseline withdrawal threshold and then after the CCI procedure on days 7 and 14 to monitor the development of mechanical allodynia. Animals with left paw withdrawal thresholds to von Frey fibers #8 g were deemed to have developed mechanical allodynia and were included in the study. Allocation to treatment groups was based on each animal's pretreatment withdrawal threshold measured on day 14.
On day 14 or 15, animals received a single oral gavage dose of vehicle (0.5% methylcellulose), ABD-1970, or pregabalin (10 ml/kg volume); 240 minutes later, mechanical allodynia was measured to determine treatment effects. All observers were blinded to the identity of the treatment groups.
Serotonin-Induced Scratching Model of Pruritus. The day before the experiment, female Sprague-Dawley rats (202-278 g; Janvier Laboratories, Le Genest-Saint-Isle, France) were shaved on the rostral part of the back and then placed individually into clear cylindrical observation chambers (height 5 35 cm, diameter 5 19 cm) for at least 60 minutes to habituate them to the testing environment (Porsolt, Le Genest-Saint-Isle, France). On the day of the experiment, scratching was induced by intradermal injection of serotonin (50 ml, 50 mg/rat) into the rostral part of the rat's back.
To test the effects of ABD-1970 on serotonin-induced scratching, animals were administered vehicle (0.5% methylcellulose) or ABD-1970 (1, 2, or 10 mg/kg) by oral gavage (5 ml/kg volume) 240 minutes before serotonin injection. The positive control naltrexone (1 mg/kg) or its respective vehicle (0.9% saline) was administered subcutaneously 15 minutes before serotonin. Scratching bouts were monitored for 30 minutes after serotonin injection by an observer blinded to the experimental groups. A single scratching bout was defined as one or more rapid movements of the hindpaws to the area around the site of serotonin injection ending with licking or biting of the toes and/or placement of the hindpaw on the floor of the observation chamber.
Locomotor Activity. Open field behavior was monitored in male Sprague-Dawley rats 240 minutes after oral administration of ABD-1970 (1 and 10 mg/kg, 0.5% methylcellulose vehicle) or 15 minutes after subcutaneous administration of the CB agonist WIN 55,212-2 (3 mg/kg, 1% Tween 80/99% saline vehicle) at Saretius Ltd. To control for the different administration times and routes, all animals received counterbalanced vehicle administrations 240 minutes (0.5% methylcellulose, p.o.) or 15 minutes (1% Tween 80/99% saline, s.c.) prior to behavior assessment. Locomotor activity, rearing, and grooming were monitored for 40 minutes using the Laboras automated behavioral tracking system.

Prostanoid Production in Stimulated Human Blood and Cells
Lipopolysaccharide-Stimulated Whole Blood. Normal human blood was obtained from The Scripps Research Institute Normal Blood Donor Service (La Jolla, CA) and was approved by The Scripps Research Institute institutional review board. Blood samples from two donors per experiment (one male donor and one female donor) were prepared by diluting equal parts heparinized whole blood and serumfree RPMI medium (v/v). Diluted samples (4 ml) were treated with ABD-1970 (0.03-30 mM), indomethacin (10 mM), rofecoxib (10 mM), or vehicle (DMSO) for 30 minutes at 37°C in 24-well plates and were then stimulated with 30 ng/ml lipopolysaccharide (LPS; E. coli serotype 0111:B4, no. 581-012-L002; Enzo Life Sciences, New York, NY) or vehicle for approximately 24 hours at 37°C. To collect plasma, well contents were transferred to a 15-ml conical tube and centrifuged at 1500 rpm for 5 minutes to pellet erythrocytes. The clear supernatant was isolated, transferred to a glass vial, and frozen at 280°C prior to LC-MS/MS analysis as described in the Supplemental Methods.
Ionomycin-Stimulated Whole Blood. Blood from one female donor was prepared by diluting equal parts heparinized whole blood and serum-free RPMI (v/v). Diluted samples (1 ml) were treated with ABD-1970 (0.3 and 3 mM), indomethacin (10 mM), rofecoxib (10 mM), or vehicle (DMSO) for 30 minutes at 37°C in 24-well plates and were then stimulated with 30 mM of the calcium ionophore ionomycin (Calbiochem, San Diego, CA) or vehicle for 30 minutes at 37°C. To collect plasma, plates were centrifuged at 1500 rpm for 5 minutes to pellet erythrocytes. The clear supernatant was isolated, transferred to a glass vial, and frozen at 280°C prior to LC-MS/MS analysis as described in the Supplemental Methods.

Statistical Analyses
Results are presented as the mean 6 S.E.M. Treatment effects on formalin-induced paw licking were analyzed by one-way analysis of variance followed by the Dunnett post-test or Newman-Keuls multiple-comparison tests, as warranted. The CCI study in which the effect of treatment on von Frey thresholds versus pretreatment 498 thresholds was compared was analyzed by a paired Wilcoxon test. Treatment effects on all additional data sets were analyzed by analysis of variance and the Dunnett post-test. Differences were considered statistically significant at P , 0.05. All statistical analyses were carried out using Prism (version 6), SPSS (version 17.0; IBM, Armonk, NY), or Statistica (version 10.1; Statsoft, Palo Alto, CA) software.

ABD-1970 Is a Potent and Selective Inhibitor of MGLL across Species
ABD-1970 (Fig. 1) is a member of the carbamate class of MGLL inhibitors, which have been shown to act via covalent carbamylation of the active-site serine residue (Long et al., 2009a;Chang et al., 2012;Niphakis et al., 2012;Griebel et al., 2015;Butler et al., 2017). Characterization of ABD-1970 potency and selectivity leveraged ABPP, a chemical proteomics platform that utilizes active site-directed chemical probes to evaluate the activity of entire enzyme families in parallel in native biologic matrices . In vitro treatment of tissue or cell homogenates with ABD-1970, followed by incubation with ABPP probes, SDS-PAGE analysis, and in-gel fluorescence imaging, allows for the direct visualization of the targets of ABD-1970. In this gel-based ABPP platform, quantification of serine hydrolase activity is based on the intensity of probe labeling, and inhibition is measured as a reduction of probe labeling compared with control samples.
ABD-1970 (0.001-10 mM, 30 minutes, 37°C) was analyzed by ABPP in brain tissue homogenates from mice, rats, rabbits, dogs, monkeys, and humans, as well as in lysates prepared from PC3 cells. Brain tissue was chosen for profiling because it represents an important therapeutic target organ and exhibits high activity of many serine hydrolases, including MGLL. PC3 cells are a rich source of human MGLL, as well as of ABHD6, an off-target common to published MGLL inhibitors (Long et al., 2009a;Chang et al., 2012;Niphakis et al., 2012;Griebel et al., 2015;Butler et al., 2017). Inhibition of MGLL activity was observed after pretreatment with ABD-1970 in brain tissue homogenates of all species and PC3 cell lysates, with average IC 50 values of ,20 nM (average values presented in Table 1 and representative ABPP gel images and MGLL activity quantitation in Fig. 2). MGLL migrates as multiple bands in the ABPP gels, and each band was equally sensitive to ABD-1970. ABD-1970 showed excellent selectivity in all species tested, crossreacting with only a single off-target (ABHD6) in mouse, rat, rabbit, dog, and monkey brain homogenates, with IC 50 values of 3.2, 0.8, 0.6, 1.4, and 3 mM, respectively. ABHD6 was inhibited in human PC3 cells with an IC 50 of 2.5 mM, whereas ABHD6 activity was not detected by gel-based ABPP in human brain tissue. Notably, at the concentrations tested, ABD-1970 did not inhibit fatty acid amide hydrolase. Since the carbamate functional group of ABD-1970 could be subject to hydrolysis, we evaluated the activity of the amine analog ABD-0038 (Supplemental Fig. 1A) using ABPP. Demonstrating that the hexafluoroisopropanol carbamate functional group is required for the activity of ABD-1970, ABD-0038 did not inhibit MGLL or other serine hydrolases in mouse, rat, and human brain tissue (Supplemental Fig. 1, B-D).
The cellular activity of ABD-1970 was evaluated in two human systems: cultured PC3 cells and PBMCs isolated from human whole blood. When added to the cell culture media, ABD-1970 (0.001-10 mM, 30 minutes, 37°C) was a potent inhibitor of MGLL in intact PC3 cells, displaying an IC 50 value of 3.2 nM. MGLL was also inhibited in human PBMCs after treatment of whole blood with ABD-1970 (0.03-30 mM, 30 minutes, 37°C). In this human blood assay, which provides a basis for a clinical peripheral biomarker assay, ABD-1970 treatment inhibited MGLL with an average IC 50 value of 62 nM (Supplemental Fig. 2A).
To amplify endocannabinoid signaling and modulate neurotransmission clinically, an MGLL inhibitor must effectively prevent the breakdown of 2-AG in the human nervous system. MGLL was previously shown to be the major 2-AG hydrolase in the rodent brain both in vitro and in vivo (Blankman et al., 2007;Long et al., 2009a). Here, we used ABD-1970 to evaluate the contribution of MGLL to 2-AG catabolism in human brain tissue. Inhibition of 2-AG hydrolysis by ABD-1970 (0.0003-1 mM, 30 minutes, 37°C) was assessed using an MS-based 2-AG substrate assay, which revealed that approximately 95% of the 2-AG hydrolytic activity in the particulate fraction of the human brain cortex was blocked by ABD-1970 with an IC 50 of 3 nM (Supplemental Fig. 2B).
Some carbamate/urea inhibitors of serine hydrolases can display slow reversibility, presumably due to hydrolytic turnover of the carbamylated active-site serine residue ( Bar-On et al., 2002;Keith et al., 2015). We therefore evaluated the persistence of the ABD-1970-MGLL interaction in vitro by assessing enzyme activity over a 24-hour period after removal of excess compound. After treatment with ABD-1970 (1 mM, 30 minutes, 37°C), human MGLL in transfected lysates of human embryonic kidney 293 cells containing the SV40 T-antigen remained inhibited by greater than 95% for up to 24 hours after removal of free compound (Supplemental Fig. 2C). Based on these measurements, we conclude that the presumed carbamylated adduct formed between human MGLL and ABD-1970 is highly stable over a 24-hour time period, with very little hydrolysis and reactivation of the enzyme.
Considering that ABD-1970 is an irreversible inhibitor of MGLL, we determined a second-order rate constant (or k inact / K i ) value for the compound, which provides a measurement of potency that is independent of substrate concentration and preincubation time. Using a fluorescent substrate assay, ABD-1970 displayed time-dependent inhibition of purified recombinant human MGLL (Supplemental Fig. 2D). The progress curves were fit to a first-order exponential to determine the observed first-order rate constant for enzyme inactivation (k obs ) at each concentration of ABD-1970 (Supplemental Fig. 2E). Nonlinear curve fitting of the k obs versus ABD-1970 concentration plot allowed determination of k inact , K i , and the second-order rate constant k inact /K i . ABD-1970 displayed an average k inact value of 0.006 s 21 , an average K i value of 0.3 mM, and an average k inact /K i value of approximately 19,000 M 21 s 21 against human MGLL. For irreversible enzyme inhibitors, the recovery of enzymatic activity in vivo and in intact cells is largely mediated by protein biosynthesis. To approximate the rate of recovery of MGLL activity after irreversible inhibition in human cells, we measured MGLL activity in cultured PC3 cells for up to 72 hours after incubation with ABD-1970 (10 nM, 30 minutes, 37°C) and removal of excess compound. ABD-1970 treatment inhibited MGLL activity by approximately 90% at 30 minutes post-treatment, and near-complete recovery of MGLL activity was observed 72 hours after free ABD-1970 was removed from the culture media (Supplemental Fig. 2F). The average recovery rate for MGLL activity over the course of the experiment (0.5-72 hours) was approximately 1% activity per hour.

ABD-1970 Is Highly Selective for MGLL in Human Tissues and Cells
A comprehensive characterization of the selectivity of ABD-1970 across the serine hydrolase class was performed in human tissues and cells using both gel-based and MS-based ABPP methods. In vitro treatment of human brain (prefrontal cortex), liver, kidney, skin, and lung homogenates or PBMC lysates with ABD-1970 (10 mM, 30 minutes, 37°C) revealed selective inhibition of MGLL by gel-based ABPP with the FP-Rh probe (Fig. 3A). The only off-target of ABD-1970 observed in these tissues was the xenobiotic metabolizing enzyme CES1.
The same principles of ABPP can be applied to high-content MS-based proteomic platforms for identification and quantification of protein targets of inhibitors. Here, human brain homogenates were treated with ABD-1970 (0.01-10 mM) or DMSO vehicle for 30 minutes at 37°C and were subjected to MS-based ABPP using the FP-biotin probe to enrich active but not inhibited serine hydrolases via streptavidin-affinity chromatography (Jessani et al., 2005). Isotopic labels were incorporated into peptides from DMSO-and inhibitor-treated samples, permitting accurate quantitation of relative enzyme activities in a single MS experiment (Boersema et al., 2009).  Table 1. ABHD6, a/b hydrolase domain containing 6 FAAH, fatty acid amide hydrolase MGLL, monoacylglycerol lipase.

500
This analysis resulted in the identification and quantification of 78 cerebral human serine hydrolases. Consistent with the gel-based analyses, MGLL was completely inhibited by ABD-1970 at concentrations as low as 0.1 mM, and no offtargets were observed until 10 mM ABD-1970, at which concentration, ABHD6, CES1, and phospholipase A2 group VII were also inhibited (Fig. 3B). ABD-1970 displays a mean IC 50 of 0.013 mM toward MGLL in human brain homogenates (Table 1) and therefore exhibits at least 77-fold selectivity for MGLL against ABHD6, CES1, and phospholipase A2 group VII and at least 770-fold selectivity for MGLL versus the other 74 enzymes profiled.
The time-course effects of ABD-1970 were determined in rats after a single oral dose of 12 mg/kg. Exposure to ABD-1970 in the brain and blood (Fig. 4F) was associated with MGLL inhibition in the brain (Fig. 4G) and increases in brain 2-AG concentrations (Fig. 4H) for up to 12 hours postdose. Partial recovery of MGLL activity and normalization of brain 2-AG concentrations were observed by 24 hours postdose.
Similar dose-and time-dependent inhibition of MGLL and accumulation of 2-AG were observed after single oral doses of ABD-1970 in mice (Supplemental Fig. 3).
MGLL inhibitors have been shown to enhance the antinociceptive effects of analgesic drugs in preclinical models of pain (Crowe et al., 2015Wilkerson et al., 2016). We evaluated the effects of ABD-1970 in combination with the a-2, d-1 calcium channel blocker pregabalin or the m opioid receptor agonist morphine in the rat formalin model. Coadministration of ABD-1970 and pregabalin significantly reduced paw licking pain behavior in formalin-treated rats to a greater extent than either agent individually (Fig. 5E). Treatment of rats with ABD-1970 and morphine similarly reduced formalin responses, whereas administration of either compound alone lacked efficacy in this experiment (Fig. 5F). These results demonstrate that MGLL inhibition potentiates the antinociceptive effects of mechanistically distinct and clinically effective analgesics.

ABD-1970 Demonstrates Broad-Spectrum Antinociceptive and Antipruritic Effects without Cannabimimetic Effects on Locomotor Activity
ABD-1970 was tested in additional preclinical models of pain of distinct pathologic origins and in a model of pruritus.
In the inflammatory pain model caused by plantar injection of CFA, ABD-1970 administration produced dose-dependent antiallodynic effects for up to 6 hours postdose, with a calculated ED 50 value of 2 mg/kg at 4 hours postdose (Fig.  6A). A single dose of 20 mg/kg ABD-1970 reduced mechanical allodynia in rats caused by plantar incision to a similar extent as a nonsedative dose of the opioid buprenorphine (Fig. 6B). In the CCI model of neuropathic pain, a single dose of 30 mg/kg ABD-1970 or pregabalin significantly reversed allodynic responses compared with vehicle-treated rats (Fig. 6C). In a rat model of pruritus induced by intradermal injection of serotonin, ABD-1970 demonstrated significant antipruritic effects at doses as low at 1 mg/kg that were similar to that of the opioid antagonist naltrexone (Fig. 6D).

MGLL and COX Inhibition Differentially Affect Prostanoid Production in Human Cells
The anatomically contextualized contribution of MGLL to prostanoid metabolism in mice  motivated us to evaluate the effects of ABD-1970 in human cellular systems alongside the nonselective COX inhibitor, indomethacin, and selective COX2 inhibitor, rofecoxib. Human whole blood can be stimulated with LPS endotoxin or calcium ionophores to produce inflammatory prostanoids, which are strongly suppressed by COX inhibition (Brideau et al., 1996;Warner et al., 1999). ABD-1970 treatment of whole blood effectively inhibits MGLL activity in PBMCs (IC 50 5 62 nM). Therefore, human blood represents an appropriate system to evaluate the effects of MGLL inhibition on stimulated prostanoid production and compare them with the effects of COX inhibition.
The impact of MGLL inhibition on prostanoid production in human endothelial cells was investigated in primary HUVECs stimulated with the inflammatory cytokine IL-1b. HUVECs have previously been used to study the effect of COX inhibition on prostanoid production, as they generate important prostanoid lipids when stimulated (Niederberger et al., 2004;Salvado et al., 2009). These lipids include thromboxane A 2 (measured as its stable breakdown product TXB 2 ), which stimulates platelet aggregation and vasoconstriction, and prostacyclin (measured as its stable breakdown product 6-ketoprostaglandin F1a), which inhibits platelet aggregation and causes vasodilation (Salvado et al., 2009).
To investigate the functional consequences of MGLL inhibition on prostanoid signaling, ABD-1970 was tested for its effect on collagen-induced platelet aggregation, an assay in which NSAIDs display robust antiaggregatory activity (Armstrong et al., 2008). In human platelet-rich plasma, ABD-1970 (0.3-30 mM) did not promote platelet aggregation or suppress aggregation induced by collagen (0% agonist or antagonist activity at 0.3, 3, and 30 mM ABD-1970). In contrast, indomethacin (3 mM) suppressed collagen-induced aggregation by 98%.

Discussion
Preparations of Cannabis and cannabinoid receptor agonists produce therapeutic effects in humans, which are accompanied by adverse CNS effects that limit the clinical dose. An alternative strategy to harness the clinical utility of cannabinoid receptor activation is to augment endocannabinoid signaling by limiting enzymatic degradation. MGLL is the major 2-AG hydrolase in rodents, and MGLL inhibitors produce a myriad of beneficial effects in animal models of pain, anxiety, epilepsy, traumatic brain injury, neuroinflammation, and neurodegeneration Chen et al., 2012;Piro et al., 2012;Griebel et al., 2015;Katz et al., 2015;von ABD-1970, indomethacin, and rofecoxib on concentrations of 2-AG, AA, and prostanoid lipids were evaluated in human whole blood treated with LPS (A-C) or ionomycin (D and E) or primary HUVECs stimulated with IL-1b (F). In (A) through (C), data are from whole blood from two human donors. In (D) through (F), data are presented as the mean 6 S.E.M. (n = 3 per group; *P , 0.05; **P , 0.01; ***P , 0.001; ****P , 0.0001 vs. stimulated DMSO control). PGE 2 , prostaglandin E2; n.s., not significant; TXB 2 , thromboxane B2.
Pharmacological Characterization of MGLL Inhibitor ABD-1970 Grabner et al., 2017;Patel et al., 2017). To prepare for an MGLL inhibitor clinical program, we developed the selective MGLL inhibitor tool compound ABD-1970 to evaluate the consequences of MGLL inhibition in human tissues and cells and in rodent models of pain and pruritus.
ABD-1970 is a potent inhibitor of MGLL across species that acts via a time-dependent and irreversible mechanism, which is consistent with a covalent interaction, as has been observed for other mechanism-based carbamate-containing MGLL inhibitors (Long et al., 2009a;Chang et al., 2012;Niphakis et al., 2012;Griebel et al., 2015;Butler et al., 2017). Covalent inhibition can be an attractive mechanism for enzyme inactivation because high biochemical efficiency and an extended duration of target engagement, compared with the PK profile, can often be achieved (Johnson et al., 2010;Singh et al., 2011). The time-dependent and irreversible nature of the ABD-1970-MGLL interaction likely contributes to the high potency and sustained activity that ABD-1970 displays in vivo. Once MGLL is inhibited, ABD-1970 does not need to compete with rising 2-AG concentrations, which ultimately reach .100 nmol/g (approximately 100 mM) in the brain, and the maintenance of inhibition is not solely determined by sustained compound exposure but instead is largely dependent on protein biosynthesis.
The exceptional potency and selectivity of ABD-1970 was achieved through a medicinal chemistry campaign that leveraged the chemical proteomics platform ABPP (Niphakis and Cravatt, 2014) to simultaneously evaluate and optimize these parameters (Cisar et al., 2018). Extensive ABPP profiling in human tissues and cells revealed that ABD-1970 has at least 77-fold selectivity toward MGLL over other serine hydrolases and completely inhibits MGLL at concentrations that are devoid of any serine hydrolase off-target interactions. Of the .70 serine hydrolase enzymes screened, many are wholly uncharacterized with regard to their substrates and functions; as such, substrate-agnostic techniques like ABPP represent the only way to functionally evaluate these enzymes and confirm selectivity.
The favorable pharmacological profile demonstrated by ABD-1970 in vitro was maintained in vivo after oral administration to mice and rats, as ABD-1970 selectively and sustainably inhibited MGLL in the brain with concomitant increases in whole brain 2-AG concentrations. Single doses of ABD-1970 effectively inhibited brain MGLL activity in both species, with ED 50 values of approximately 1 mg/kg. The doseresponse relationship revealed marked sensitivity of brain 2-AG to MGLL activity because partial inhibition of MGLL in the brain was sufficient to elevate 2-AG levels, which ultimately increased by .10-fold once MGLL was completely inhibited. In contrast, in the rat, approximately 2-fold reductions in levels of AA and PGE 2 /prostaglandin D 2 in the brain were associated with near-complete inhibition of MGLL.
To understand the relationship between ABD-1970 PK, MGLL target engagement, 2-AG, and antinociceptive efficacy, we used the rat formalin pain model. This model provided the opportunity to evaluate the effects of ABD-1970 administration on two distinct nociceptive modalities: 1) an early phase corresponding to direct activation of peripheral sensory nerves and 2) a late phase corresponding largely to central sensitization (Tjølsen et al., 1992). Single doses of ABD-1970 numerically reduced pain behavior in the early phase and significantly reduced pain behaviors in the late phase of the formalin response, with an ED 50 of 1 mg/kg. The antinociceptive effects observed at the ED 50 were associated with ABD-1970 concentrations less than 10 nM in the brain and blood, 40% inhibition of brain MGLL activity, and 3-fold elevations in brain 2-AG concentrations. Multiple previous studies have demonstrated sustained antinociceptive, antiallodynic, and antihyperalgesic effects after repeated administration of MGLL inhibitors at low doses that partially inhibit MGLL in the brain (Busquets-Garcia et al., 2011;Kinsey et al., 2013;Burston et al., 2016;Curry et al., 2018), whereas tolerance to these effects has been observed after prolonged, complete MGLL inhibition or genetic ablation (Chanda et al., 2010;Schlosburg et al., 2010;Ignatowska-Jankowska et al., 2014). The functional tolerance observed with complete MGLL inactivation and maximal elevation of brain 2-AG concentrations is associated with downregulation and/or desensitization of brain CB1 receptors (Chanda et al., 2010;Schlosburg et al., 2010;Navia-Paldanius et al., 2015). The ABD-1970 PK/PD relationship in the rat formalin model revealed that painmodifying activity was observed at low doses and exposures of ABD-1970 that only partially inhibited brain MGLL and caused submaximal elevations of 2-AG in the brain. Therefore, low doses of ABD-1970 near the ED 50 of 1 mg/kg would be predicted to maintain efficacy after repeated dosing.
In addition to demonstrating significant antinociceptive activity as a single agent in the rat formalin model, ABD-1970 also potentiated the activity of two analgesic standards of care: pregabalin and morphine; this is in agreement with the synergistic activity that has been shown previously for MGLL inhibitors and similar agents in other pain models (Wilkerson et al., 2016;. That MGLL inhibition enhanced the efficacy of these clinically effective drugs suggests that in addition to providing pain relief as a single agent, an MGLL inhibitor may find clinical utility in enhancing the effectiveness of non-narcotic pain killers like pregabalin or limiting the dose needed of opioid drugs to provide adequate pain relief. The effects of ABD-1970 on pain were not limited to acute pain caused by formalin; antinociceptive or antiallodynic effects were observed in rat models of postoperative pain, neuropathic pain, and inflammatory pain. ABD-1970 also significantly reduced itch behavior in the rat serotonin model of pruritus, consistent with previously published reports with MGLL inhibitors in mice (Tosun et al., 2015;Yesilyurt et al., 2016). These effects occurred without obvious cannabimimetic effects or sedation, as ABD-1970 did not reproduce the behavioral effects of the direct cannabinoid receptor agonist WIN 55,212-2 in a novel open field. ABD-1970 did dose-dependently reduce grooming behavior in the novel environment, which may be related to the antipruritic and anxiolytic effects of this mechanism.
Selective MGLL inhibition by ABD-1970 in the rodent brain effectively increased total brain 2-AG concentrations, as has been previously demonstrated after pharmacological or genetic inactivation of MGLL (Long et al., 2009a;Schlosburg et al., 2010). Experiments with ABD-1970 in human tissues and cells suggest that MGLL is likely to similarly regulate 2-AG in humans. ABD-1970 treatment of human brain tissue reduced 2-AG hydrolysis by .95%, which is comparable to the approximately 85% of the total 2-AG hydrolase activity mediated by MGLL in mouse brain preparations (Blankman et al., 2007), which therefore increases confidence that MGLL inhibition in the human brain will amplify 2-AG signaling and 506 produce the therapeutically beneficial effects demonstrated in preclinical studies. Likewise, treatment of human whole blood with ABD-1970 inhibited MGLL in blood cells and elevated plasma 2-AG concentrations. The concentration-dependent inhibition of MGLL observed in PBMCs after treatment of whole blood provides a viable clinical biomarker strategy to measure peripheral MGLL target engagement in human subjects or patients administered an MGLL inhibitor.
In rodents, MGLL serves as a source of AA for COXmediated prostanoid generation in a subset of tissues . Here, we used ABD-1970 to evaluate the contribution of MGLL to the generation of AA-derived prostanoids in human whole blood and endothelial cells. The nonselective COX inhibitor indomethacin and the selective COX2 inhibitor rofecoxib were evaluated in parallel to compare the effects of MGLL and COX inhibition in assays known to be sensitive to COX inhibitors. MGLL inhibition did not suppress stimulated prostanoid production or platelet aggregation and was therefore differentiated from COX inhibition. Our results show that despite their involvement in sequential steps of the AA metabolic pathway, the effects of inhibiting MGLL or COX enzymes in human cells are distinct, suggesting that MGLL inhibitors are unlikely to replicate the toxicity profiles displayed by COX inhibitors.
In conclusion, ABD-1970 is a highly potent, selective, and orally bioavailable MGLL inhibitor, which we have used to evaluate the effects of MGLL inhibition in human tissues and cells and in rodent models of pain and itch. These results support use of ABD-1970 as a tool MGLL inhibitor in preclinical models and further clinical exploration of the mechanism. Toward this end, we have developed a mechanistically similar MGLL inhibitor that is currently in clinical development for neurologic indications, including Tourette syndrome, spasticity in multiple sclerosis, and pain (Cisar et al., 2018).