lable at ScienceDirect

Neuropharmacology 110 (2016) 519e529

Contents lists avai

Neuropharmacology

journal homepage: www.elsevier .com/locate/neuropharm

Stimulation of brain glucose uptake by cannabinoid CB2 receptors andits therapeutic potential in Alzheimer's disease

Attila K€ofalvi a, b, Cristina Lemos a, 2, Ana M. Martín-Moreno c, d, 1, B�arbara S. Pinheiro a, 2,Luis García-García e, Miguel A. Pozo e, f, Angela Val�erio-Fernandes a, Rui O. Beleza a,Paula Agostinho a, g, Ricardo J. Rodrigues a, b, Susana J. Pasquar�e c, h, Rodrigo A. Cunha a, g,María L. de Ceballos c, d, *

a CNC e Center for Neuroscience and Cell Biology of Coimbra, University of Coimbra, 3004-504 Coimbra, Portugalb Institute for Interdisciplinary Research, University of Coimbra, Coimbra, Portugalc Neurodegeneration Group, Department of Cellular, Molecular and Developmental Neurobiology, Instituto Cajal, CSIC, Doctor Arce, 37, 28002 Madrid, Spaind CIBERNED; Centre for Biomedical Research on Neurodegenerative Diseases, Spaine CAI de Cartografía Cerebral, Instituto Pluridisciplinar, UCM, Madrid, Spainf PET Technology Institute, Madrid, Spaing FMUC, Faculty of Medicine, University of Coimbra, Coimbra, Portugalh Instituto de Investigaciones Bioquímicas de Bahía Blanca (INIBIBB), CONICET-Bahía Blanca and Universidad Nacional del Sur, Edificio E1, Camino LaCarrindanga km 7, 8000 Bahía Blanca, Argentina

a r t i c l e i n f o

Article history:Received 20 June 2015Received in revised form18 February 2016Accepted 7 March 2016Available online 11 March 2016

Chemical compounds studied in this article:AM630 (PubChem CID4302963)Anandamide (PubChem CID5281969)Arachidonyl-2-chloroethylamide (PubChemCID5311006)2-Arachidonoylglycerol (PubChem CID5282280)2-Deoxy-D-glucose (PubChem CID108223)DuP697 (PubChem CID3177)GP1a (PubChem CID10252734)JWH133 (PubChem CID

Abbreviations: 2-AG, 2-arachidonoylglycerol; 3Rs,idonyl-20-chloroethylamide; AEA, anandamide (N-arDulbecco's modified eagle medium; 3HDG, 3H-2-deoxhydroxyethyl)piperazine-N0-(2-ethanesulfonic acid);transgenic amyloid precursor protein; WT, wild-type.* Corresponding author. Neurodegeneration Group

Spain.E-mail address: mceballos@cajal.csic.es (M.L. de C

1 Present address: MD Anderson Cancer Center, Ar2 Present address: Experimental Psychiatry Unit, Ce

http://dx.doi.org/10.1016/j.neuropharm.2016.03.0150028-3908/© 2016 The Authors. Published by Elsevier

a b s t r a c t

Cannabinoid CB2 receptors (CB2Rs) are emerging as important therapeutic targets in brain disorders thattypically involve neurometabolic alterations. We here addressed the possible role of CB2Rs in theregulation of glucose uptake in the mouse brain. To that aim, we have undertaken 1) measurement of 3H-deoxyglucose uptake in cultured cortical astrocytes and neurons and in acute hippocampal slices; 2) real-time visualization of fluorescently labeled deoxyglucose uptake in superfused hippocampal slices; and 3)in vivo PET imaging of cerebral 18F-fluorodeoxyglucose uptake. We now show that both selective(JWH133 and GP1a) as well as non-selective (WIN55212-2) CB2R agonists, but not the CB1R-selectiveagonist, ACEA, stimulate glucose uptake, in a manner that is sensitive to the CB2R-selective antagonist,AM630. Glucose uptake is stimulated in astrocytes and neurons in culture, in acute hippocampal slices, indifferent brain areas of young adult male C57Bl/6j and CD-1 mice, as well as in middle-aged C57Bl/6jmice. Among the endocannabinoid metabolizing enzymes, the selective inhibition of COX-2, rather thanthat of FAAH, MAGL or a,bDH6/12, also stimulates the uptake of glucose in hippocampal slices of middle-aged mice, an effect that was again prevented by AM630. However, we found the levels of the endo-cannabinoid, anandamide reduced in the hippocampus of TgAPP-2576 mice (a model of b-amyloidosis),and likely as a consequence, COX-2 inhibition failed to stimulate glucose uptake in these mice. Together,these results reveal a novel general glucoregulatory role for CB2Rs in the brain, raising therapeutic in-terest in CB2R agonists as nootropic agents.© 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

replacement, refinement and reduction of animals in research; ABDH6/12, a/b-hydrolase domain 6/12; ACEA, arach-achidonoylethanolamine); CB1Rs and CB2Rs, cannabinoid CB1 and CB2 receptors; COX-2, cycloxygenase-2; DMEM,yglucose; DMSO, dimethyl-sulfoxide; FAAH, fatty acid amide hydrolase; 18FDG, 18F-fluoro-deoxyglucose; HEPES, N-(2-MAGL, monoacylglycerol lipase; 2-NBDG, 2-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)-2-deoxyglucose; TgAPP,

, Dept. of Cellular, Molecular and Developmental Neurobiology Cajal Institute, CSIC, Doctor Arce, 37, 28002 Madrid,

eballos).turo Soria 270, 28033 Madrid, Spain.nter for Psychiatry and Psychotherapy, Innsbruck Medical University, Austria.

Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

A. K€ofalvi et al. / Neuropharmacology 110 (2016) 519e529520

6918505)JZL184 (PubChem CID

25021165)LY2183240 (PubChem CID11507802)Ouabain (PubChem CID439501)WIN55212-2 (PubChem CID5311501)WWL70 (PubChem CID:17759121)

Keywords:Anandamide (AEA)b-amyloidCerebral glucose uptakeCannabinoid CB2 receptorCyclooxygenase-2 (COX-2)Positron emission tomography (PET)

1. Introduction

Brain disorders, including Alzheimer's disease (AD), ofteninvolve specific early alterations in themetabolism of glucose in thebrain (Mosconi, 2005; Teune et al., 2010). The idea of alleviatingsymptoms of dementia by boosting cerebral energy metabolismhas been toyed with for decades (Branconnier, 1983), yet safepharmacological agents with well characterized mechanism ofaction are still lacking. In this sense, we have investigated here thelocal cerebral glucoregulatory potential of the endocannabinoidsystem in rodents.

The endocannabinoid systemmainly consists of two established(CB1 and CB2) receptors that are activated by endocannabinoidsignalling molecules, such as N-arachidonoylethanolamine (anan-damide, AEA) and 2-arachidonoylglycerol (2-AG) (Di Marzo and DePetrocellis, 2012; Pertwee, 2012). At the cellular level, endocanna-binoid signalling via CB1 receptors (CB1Rs) regulates peripheralenergy metabolism and glucose uptake (Matias et al., 2008). In thebrain, acute activation of the psychoactive CB1R inhibits mito-chondrial respiration and glucose metabolism in neurons and as-trocytes, without robust effect on the uptake of glucose (B�enardet al., 2012; Duarte et al., 2012; Lemos et al., 2012). Similarly, theperipheral glucoregulatory role of the non-psychoactive CB2 re-ceptors (CB2Rs) has also been recognized (Agudo et al., 2010;Romero-Zerbo et al., 2012), although this remains to be docu-mented in the brain.

CB2Rs were first described in peripheral organs and cells(Gali�egue et al., 1995), and subsequently in activated microglialcells (Carlisle et al., 2002) - the resident immune cell of the brain.However, during the last decade, the presence and function ofCB2Rs have been described in other cell types in the brain. An earlyevidence was provided by Van Sickle et al. (2005) who identifiedand functionally characterized CB2Rs in brainstem neurons. Otherstudies have reported CB2R involvement in neurogenesis, axonguidance, and neuronal excitability (Den Boon et al., 2012; Duffet al., 2013; Sierra et al., 2014; Zhang et al., 2014). Furthermore,astrocytes (in particular, activated astrocytes) are also endowedwith CB2R (Sheng et al., 2005; Benito et al., 2008; Bari et al., 2011;Rodríguez-Cueto et al., 2014). However, a recent report added thatthe vast majority of specific CB2R immunoreactivity in the healthymouse neocortex is neuronal (Savonenko et al., 2015). Nonetheless,conventional CB2R immunstaining techniques may have possiblelimitations (Ashton, 2011). Therefore, it is important that the highlysensitive and specific RNAscope technique also proved the presenceof CB2R mRNA in excitatory and inhibitory neurons (but hardly inresting microglia) throughout the hippocampus of adult mice (Li

and Kim, 2015). Furthermore, these hippocampal (and medial en-torhinal) CB2Rs appear to control the synaptic release of GABA(Morgan et al., 2009; And�o et al., 2012), and the deletion of CB2Rs isassociated with impaired synaptic plasticity at excitatory synapses,too, in the mouse hippocampus (Li and Kim, 2016a). Not surpris-ingly, the CB2R KO mice exhibit memory impairments as well (Liand Kim, 2016b).

Besides these indefinite physiological roles, the therapeuticpotential of central CB2Rs have also been recognized. There is anincreasing clinical interest in endorsing novel selective CB2R ago-nists to treat different illnesses (Atwood and Mackie, 2010;Pertwee, 2012; Han et al., 2013), including AD. Indeed, prolongedoral treatment with a CB2R selective agonist conferred anti-inflammatory effects and reduced microglial activation, whichhave been demonstrated to mitigate the cognitive deficits of TgAPPmice (Martín-Moreno et al., 2012). Similarly, subchronic treatmentwith another CB2R agonist promoted b-amyloid (Ab) clearance, andrestored synaptic plasticity, cognition, and memory in Ab injectedrats (Wu et al., 2013). These results were extended in other reportsusing TgAPP/PS1 mice (Aso and Ferrer, 2014; Cheng et al., 2014).Additionally, the severity of Ab pathology correlates positively withthe density of the CB2R rather than the CB1R (Esposito et al., 2007;Solas et al., 2013; Savonenko et al., 2015). Therefore, to map theputative glucoregulatory potential of CB2Rs in healthy mice and in amodel of b-amyloidosis, we have taken advantage of recentlyoptimized in vitro and in vivo protocols (Lemos et al., 2012; Martín-Moreno et al., 2012). Here we report that acute CB2R activationincreases glucose uptake in the brain of young and middle-agedmice both in vitro and in vivo, which highlights a novel cerebralrole for the CB2R with therapeutic potential.

2. Materials and methods

2.1. Chemicals

3H-2-deoxy-D-glucose (3HDG; specific activity: 60 Ci/mmol),arachidonylethanolamide [ethanolamide 1e3H] (50 Ci/mmol) andunlabelled AEA were obtained from American RadiolabeledChemicals, Inc. (Saint Louis, MO, USA). 2-arachidonoyl-[1,2,3-3H]glycerol (3H-2-AG; 40 Ci/mmol) was bought from PerkinElmer,Madrid, Spain). 2-NBDG was purchased from Invitrogen (Carslbad,California, USA). Non-organic reagents, HEPES and DMSO werepurchased from Sigma-Aldrich Portugal (Sintra, Portugal) or MerckBiosciences (Darmstadt, Germany). ACEA, AM630, DuP697, GP1a,JWH133, JZL184, LY2183240, WIN55212-2 and WWL70 were ob-tained from Tocris Bioscience (Bristol, U.K.). All non-water soluble

A. K€ofalvi et al. / Neuropharmacology 110 (2016) 519e529 521

substances and 2-NBDG were reconstituted in DMSO at 1000� thefinal concentration, and stored in aliquots at �20 �C.

2.2. Animals

All studies were carried out in accordancewith the EU (2010/63/EU) and FELASA guidelines, and theywere approved by the Spanishand Portuguese Ministries of Agriculture, as well as the localinstitutional Animal Care Committees (license No. 280279-31-Aand 025781). All studies involving animals are reported in accor-dance with the ARRIVE guidelines for reporting experimentsinvolving animals (Kilkenny et al., 2010; McGrath et al., 2010). Atotal of 73 animals were used in the experiments described here.

For the experiments which are summarized in Figs. 3AB, 4 and 5,we used 12-month-old (middle-aged) C57Bl/6j mice (i.e. wild type[WT] littermates as controls), and 12-month-old C57Bl/6j trans-genic 2576 mice, containing the human APP695 transgene withmutations at KM670/671NL (hereafter, TgAPP mice). TgAPP micehave b-amyloid-burden but no evident cell loss (Hsiao et al., 1996).At 12 months, TgAPP mice did not differ in weight from the WTC57Bl/6j mice (WT: 39.6 ± 2.4 g; TgAPP: 38.6 ± 2.3 g), yet theydisplayed compromised new object recognition compared to WTmice (data not shown). Data shown in Fig. 3C were obtained from10-week-old male WT CD-1 mice and their CB1R global KO litter-mates (Ledent et al., 1999). All these transgenic and WT mice weregenotyped from tail tip biopsies.

Data shown in Fig. 1 were obtained from young adult (8-10week-old) male C57bl/6 mice (Charles-River, Barcelona, Spain),while to obtain embryos for cell cultures, we mated young adultC57Bl/6 mice (Charles-River).

The animals were group-housed under controlled temperature(23 ± 2 �C) and subjected to a fixed 12 h light/dark cycle, with freeaccess to food and water. We applied the principles of the ARRIVEguideline in designing and performing the in vivo and in vitropharmacological experiments (see below), as well as for datamanagement and interpretation. All efforts were made to reducethe number of animals used and to minimize their stress anddiscomfort. The animals used to perform the in vitro studies weredeeply anesthetized with halothane before sacrifice (no reaction tohandling or tail pinching, while still breathing).

2.3. In vitro real-time fluorescent measurement of deoxyglucoseuptake in hippocampal slices

Experiments were performed as before (Lemos et al., 2015). Fiveyoung C57bl/6 mice were sacrificed, and their brain was quicklyremoved and placed in ice-cold Krebs-HEPES solution (in mM: NaCl113, KCl 3, KH2PO41.2, MgSO41.2, CaCl2 2.5, NaHCO3 25, glucose 5.5,HEPES 1.5, pH 7.2). Coronal vibratome slices (300 mm) were left torecover for 1 h at room temperature in Krebs-HEPES solutiongassed with 5% CO2 and 95% O2, and then gently mounted oncoverslips with the hippocampus in the center, and placed in a RC-20 superfusion chamber on a PH3 platform (Warner Instruments,Harvard, UK). The slices were superfused with gassed Krebs-HEPESsolution at a rate of 0.5 mL min-1 in a closed circuit, and they werethen photographed with a CoolSNAP digital camera (Roper Scien-tific, Trenton, NJ, USA) every 30 s over the following 30min (using a5� PlanNeofluar-objective [NA 0.25] on an inverted Axiovert 200Mfluorescence microscope [Carl Zeiss, Germany], coupled to aLambda DG-4 integrated 175 Watt light source and wavelengthswitching excitation system [Sutter Instrument Company, Novato,CA, USA] to allow real-time video imaging). The datawas band-passfiltered for excitation (BP470/40) and emission (BP525/50).

After recording 4 images for auto-fluorescence (to establish thebaseline), 2-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)-2-

deoxyglucose (2-NBDG; 30 mM) was applied through the reser-voir and 15 min later, JWH133 (1 mM), GP1a (100 nM) or the DMSOvehicle alone (0.1% v/v) were administered. All measurements wereobtained in duplicate from each animal (see Supplementary Fig. 1for additional details).

2.4. Cell culture preparation

Because the embryonic hippocampus is too small to provideenough cells for our assays, neuronal and astrocyte primary cul-tures were prepared from the neocortex as before (Lemos et al.,2015). The neocortex of E18 male and female C57Bl/6 rat embryoswas digested for 15 minwith 0.125% trypsin (type II-S, from porcinepancreas, Sigma-Aldrich Portugal) and 50 mg/mL DNAse (Sigma-Aldrich) in Hank's balanced salt solution without calcium andmagnesium (in mM: NaCl 137, KCl 5.36, KH2PO4 0.44, NaHCO3 4.16,Na2HPO4 0.34, D-glucose 5, pH 7.2). After dissociation, astrocyteswere grown in plastic Petri dishes (p100) for 14 days in DMEMwith10% fetal bovine serum, 50 U/mL penicillin and 50 mg/mL strepto-mycin (Sigma-Aldrich), at 37 �C in an atmosphere of 95%/5% air/CO2, replacing half of the medium at DIV 7. The cells were thentrypsinized and plated onto poly-D-lysine (100 mg/mL, Sigma-Aldrich) and laminin (10 mg/mL, Sigma-Aldrich) coated 24-wellculture plates at a cell density of 150,000/cm2. Uptake assayswere performed 24 h later. For neuronal cultures, cells were plateddirectly onto poly-D-lysine and laminin coated 24-well cultureplates, at the same cell density, in DMEM plus 10% fetal bovineserum, 50 U/mL penicillin and 50 mg/mL streptomycin. Two hoursafter seeding, the medium was replaced by Neurobasal mediumwith 2% B27 supplement (GIBCO, Life technologies), 50 U/mLpenicillin, 50 mg/ml streptomycin and 2 mM glutamine (Sigma-Aldrich), and cells were grown at 37 �C in an atmosphere of 95%/5%air/CO2 until they were used 14 days later. The medium waspartially (40%) replaced every 4 days. On average, 15% of cells werefound positive for glial fibrillar acidic protein immunostaining inrandomly chosen wells (Lemos et al., 2015).

2.5. In vitro 3H-deoxyglucose uptake in neocortical neuron andastrocyte cultures

The assays were carried out according to Lemos et al. (2015).For these assays, we used 7 different astrocyte cultures and 5neuron cultures from 7 and 5 independent preparations (pregnantdams). After 15 days in culture, the culture medium was replacedwith 250 mL Krebs-HEPES assay solution (pH 7.4; 37 �C) (seeabove), containing either the vehicle alone (DMSO 0.1%, in half ofthe plate), or the CB2R-selective antagonist, AM630 (1 mM, in theother 12 wells). The cultures were incubated for 60 min, at 37 �C,and the CB2R agonists JWH133 (100 nM) (4 wells/plate) or GP1a(100 nM) (4 wells/plate), or their vehicle (DMSO 4 wells/plate),was gently added to the wells in a volume of 250 mL, together withAM630 (i.e. 3 � 4 wells/plate) or its vehicle (i.e. 3 � 4 wells/plate),and the radioactive glucose analogue 3H-2-deoxyglucose (3HDG;16.7 nM, final concentration). After a further 30 min incubation at37 �C, the plates were transferred onto ice and the 3H content wasmeasured in an aliquot from each well (to assess the small dif-ferences in the individual 3H concentration in each well thatpipetting and mixing solutions may introduce). The cells werethen washed gently 3 times with 1 mL ice-cold Krebs-HEPES,300 mL NaOH (0.5 M) was added to each well, and the plates wereleft on a shaker for an hour to recover the proteins and 3HDGtaken up by the cells. Tritium was measured in a Tricarb b-counter(PerkinElmer, USA), and these values were adjusted for proteinquantity and the 3H concentrations in each individual wells, andmultiplied by 3.3 � 105 (the ratio between 3HDG and D-glucose in

Fig. 1. Time-course and sub-regional variation of the effect of CB2R agonists on the uptake of the fluorescent glucose analogue, 2-NBDG in hippocampal slices of young C57Bl/6jmale mice. For technical details and additional information, see Supplementary Methods, Supplementary Fig. 1 and Supplementary Video. (A,A0) Representative images illustratingthe distribution of the fluorescence signal, using the false color scale of blue (low signal), green (moderate signal) and yellow (strong signal). As indicated in the time-course graph(B), image (A) was taken just before the treatment (minute 15 of baseline), and image (A0) was photographed close to the end of the exposure to GP1a (100 nM: see also Sup-plementary Video). (B) JWH133 (1 mM) and GP1a (100 nM) both increased the uptake in astrocyte-rich zones of the hippocampus. The displayed curves represent the mean þ S.E.M.of individual net effects (in duplicate) in each animal, which was obtained after the subtraction of the respective DMSO controls (in duplicates per animal) from the raw treatmentdata. Therefore, the DMSO control is represented with the dashed line crossing the zero value. (C) The individual points represent the average of 2 different slices from the samemouse in duplicate (n ¼ 5): *P < 0.05 and **P < 0.01.

Fig. 2. CB2R activation stimulates hippocampal glucose uptake in astrocytes andneurons in vitro. The CB2R-selective agonists JWH133 (30 nM) and GP1a (100 nM)stimulated glucose uptake in (A) primary cortical astrocyte and (B) neuronal culturesfrom C57Bl/6 mice, as assessed using the 3HDG uptake assay (30 min). The CB2R se-lective antagonist, AM630 (1 mM) prevented the action of the CB2R agonists, while ithad no effect on the uptake per se. D-glucose uptake is expressed as mean þ SEM innmol/mg protein (n ¼ 7 astrocyte and n ¼ 5 neuron cultures, performed in quadru-plicate, at a density of 15 � 104 cells/cm in 24-well plates). *P < 0.05 and **P < 0.01 vs.DMSO control; n.s., not significant.

A. K€ofalvi et al. / Neuropharmacology 110 (2016) 519e529522

the medium).In this experimental model, cytochalasin B (10 mM) largely in-

hibits glucose uptake in both cell cultures, thus confirming the

specificity of the transport process (Lemos et al., 2015). Addition-ally, the plasma membrane Naþ/Kþ-ATPase inhibitor, ouabain (10mM), also inhibited neuronal glucose uptake in this assay by42.8 ± 2.1% (n ¼ 3, P < 0.01; figure not shown).

2.6. In vitro 3H-deoxyglucose uptake in hippocampal slices

To better characterize the glucoregulator role of CB2R signallingpharmacologically, wemoved to an in vitro glucose uptake protocolpreviously optimized in acute hippocampal slices which allows theeffect of various treatments to be compared simultaneously indifferent strains in a pairwise arrangement (Lemos et al., 2012,2015). Animals were anesthetized with halothane, sacrificed(around 14:00 o'clock each experimental day to reduce potentialcircadian hormonal effects), and their brain was immediatelyplaced in ice-cold Krebs-HEPES assay. The hippocampus was cutinto 450 mm-thick transverse slices with the help of a McIlwaintissue chopper, and the slices were gently separated in ice-coldcarboxygenated (95% O2 and 5% CO2) assay solution, and main-tained at 37 �C in a multichamber slice incubator with 50 ml ofcarboxygenated assay solution until the end of the experiment.Four slices were used from each animal and in each experimentalcondition. The CB2R-selective antagonist, AM630 or its vehicle(DMSO 0.1%) were bath-applied from min 55 of the preincubationperiod. WIN55212-2, JWH133, ACEA or DuP697, LY2183240, as wellas JZL184 andWWL70 combined or their vehicle (DMSO 0.1%) wereadded frommin 60 of the preincubation period, and the radioactiveglucose analogue, 3H-2-deoxy-D-glucose (3HDG; final cc. 2 nM) wasapplied to the bath 5min later. After 30min, the slices werewashedtwice in ice-cold assay solution for 5 min and collected in 1 mLNaOH (0.5 M) to dissolve the slices. An aliquot of 800 mL was thenassayed for 3H in a Tricarb b-counter, while the remaining solutionwas used to quantify the protein in a bicinchoninic acid assay(Merck Biosciences, Germany). Tritium uptake was multiplied by afactor of 2.75 � 106 to estimate the total glucose uptake (corre-sponding to the concentration difference between D-glucose and3HDG in the uptake medium; Lemos et al., 2012).

2.7. Blood glucose measurements

Blood glucose from non-fasted WT and TgAPP mice wasassessed before and one hour following the acute administration ofJWH133 0.2 mg/kg i.p. Glucose was measured in one drop of blood

Fig. 3. CB2R activation stimulates glucose uptake in hippocampal slices of both middle-aged and young mice ex vivo. A) The CB2R-selective agonist, JWH133 (1 mM), the mixed CB1R/CB2R agonist, WIN55212-2 (1 mM), as well as the COX-2 inhibitor, DuP697 (500 nM) all increased glucose uptake in 12-month-old wild-type (WT) C57Bl/6j mice. All these effectswere prevented by a 5 min preincubation with the CB2R-selective antagonist, AM630 (1 mM). Inhibition of other endocannabinoid metabolizing enzymes, namely, FAAH, MAGL ora,bDH6, with LY2183240, JZL184 and WWL70 (all at 1 mM) respectively, did not affect glucose uptake. B) By contrast, only JWH133 and WIN55212-2 but not DuP697 stimulatedglucose uptake in the TgAPP 2567 mice. Data represent the mean þ SEM of individual measurements from 6 to 12 animals, with the dashed line indicating the DMSO control values:*P < 0.05 vs. DMSO-treated slices; n.s., not significant. C) JWH133 (1 mM), stimulated the uptake of glucose in acute hippocampal slices of 10 week-old CB1R KO mice and their WTlittermates. Each of the connected pairs of squares and circles represent one animal. *P < 0.05 vs. DMSO-treated slices (n ¼ 6).

Fig. 4. AEA levels were reduced in the hippocampus of TgAPP mice. Of the twoprincipal endocannabinoids, the levels of AEA (B) but not of 2-AG (A) were significantlylower (by 38%) in the TgAPP mice (dark bars) than in the WT (light bars), as assessed bycombined liquid chromatography-mass spectrometry. Data represent the mean þ SEMof individual measurements from 7 animals. 2-AG hydrolysis (C) was also unaltered inhippocampal membranes of TgAPP mice in vitro (n ¼ 6), however, AEA metabolism (D)by FAAH into ethanolamine was significantly smaller in the transgenic hippocampiin vitro (n ¼ 6). *P < 0.05, ***P < 0.001; n.s., not significant.

A. K€ofalvi et al. / Neuropharmacology 110 (2016) 519e529 523

from the tail using Glucocard Memory meter strips (Menarini,Spain), in a range of 40e500 mg/mL.

2.8. Endocannabinoid level measurement in mouse hippocampus

Endocannabinoids were measured externally (Division ofApplied Medicine, School of Medicine and Dentistry, University ofAberdeen, UK). Lipid extraction was performed by homogenizingthe frozen hippocampi on ice in 600 mL of acetonitrile/methanol(50/50) using an Ultra-Turrax. Upon addition of 6 pmol of N-(2-hydroxyethyl-1,1,2,2-d4)-5Z,8Z-,11Z,14Z-eicosatetraenamide[deuterated (d4)-AEA; Tocris Bioscience, UK] as an internal

standard, the samples were diluted to 70% water and centrifuged.The pellets were kept for total protein determination. Solid-phaseextraction (Strata-X 33 mm columns, Phenomenex) was per-formed before the LC/MS analysis on a Thermo Surveyor liquidchromatographic system coupled to an electrospray ionizationtriple quadrupole TSQ Quantum mass spectrometer operated inpositive ionizationmode. Chromatographic analysis was performedon an ACE 5 mm C8 (150 � 2.1 mm, Hichrom) column and theanalytes were eluted using an isocratic mobile phase of water/methanol/formic acid (15/85/0.5 v/v/v) at a flow rate of 200 mL/min.The column was maintained at 30 �C and the sample tray at 4 �Cduring the analysis. The extraction efficiency was >95% and thelimit of quantification was 0.01 pmol for AEA and <25 pmol for 2-AG. The sample concentrations were determined from calibrationcurves and the endocannabinoid levels were expressed per mgprotein (Bradford assay).

2.9. Endocannabinoid hydrolysis assay

Experiments were carried out as before (Mulder et al., 2011),and adapted to mice (Pascual et al., 2014). Cortices were mechan-ically homogenized in TriseHCl (20 mM, pH 7.2) containing 0.32 Msucrose and protease inhibitors, and centrifuged to eliminate nucleiand cellular debris (1000g, 10 min, 4 �C). Protein concentrationswere determined by Bradford's colorimetric method. The 2-AG-degrading activity was assessed by incubating sample fractions(50 mg protein) in a solution containing TriseHCl (10 mM, pH 7.2),1 mM EDTA, fatty acid-free bovine serum albumin (1.25 mg/mL),and 3H-2-AG in a final volume of 200 ml for 15min at 37 �C. Identicalconditions were used to assess AEA degrading activity using 3H-AEA ([1-H]ethanolamine, 25 mM, 40 Ci/mmol specific activity; ARCInc.). Reactions were stopped by the addition of 400 mL chlor-oform:methanol (2:1, v/v) and vigorous vortexing, after centrifu-gation (2000g, 10 min, 4 �C), ethanolamine or glycerol, the AEA and2-AG hydrolysis products respectively, were obtained in the upperphase, and they were transferred into scintillation vials andradioactivity measured by liquid scintillation spectroscopy.

The experimental conditions were selected from pilot experi-ments establishing the relationship of substrate and protein con-centrations to identify optimal AEA-degrading enzymatic activity(data not shown), allowing 50% of AEA converted intoethanolamine.

Blanks were prepared identically except that the membraneswere omitted or inactivated by boiling for 10 min or by the additionof chloroform:methanol (1:1, v/v) before use. No differences wereobserved among the different blanks. These values were subtracted

Fig. 5. Evidence for the stimulation of brain glucose uptake by a single low-dose injection of JWH133 in middle-aged mice in vivo. Basal (A1,B1) and JWH133-stimulated A2,B2) 18F-fluoro-deoxy-glucose (18FDG) uptake in the frontal and temporal cortices, and the hippocampus. A1) Representative color-coded PET of basal 18FDG uptake in 12 month old wild-type (WT) and B1) TgAPP 2567 mice. Panels A10,B10) illustrate the same images merged with the respective CT images, seen from three directions. The effect of acute JWH133injection (0.2 mg/kg, i.p.) is shown in panels A2,A20 ,B1,B20). Color code for FDG activity can be seen vertically on the left of the figure. C) Mean basal uptake values in the threeregions of interest (n ¼ 9 mice). D) Acute JWH133 injection does not alter the blood glucose levels in the two strains and hence, JWH133 does not appear to influence glucose uptakein the brain through systemic mechanisms. E-G0) Before-after graphs summarizing the basal uptake, and 4 days later, the JWH133-stimulated (JWH) uptake in the frontal E, E0) andtemporal F, F0) cortices, as well as in the hippocampus G, G0), of 4 wild-type (WT, E-G) and 4 TgAPP 2567 (E0-G0) mice. Uptake values were corrected for the dose injected and forbody weight, *P < 0.05.

A. K€ofalvi et al. / Neuropharmacology 110 (2016) 519e529524

from each enzymatic activity.Enzyme activity was expressed as picomole [1,2,3-3H]glycerol

formed � mg protein � 15 min and picomole [3H]ethanolamineformed � mg protein � 15 min respectively.

2.10. 18F-fluorodeoxy-D-glucose positron emission tomography(18FDG-PET)

These experiments were carried out as described previously(Martín-Moreno et al., 2012) with only slight modifications. In thePET studies the images of basal (no treatment) uptake were ac-quired, and 4 days later the animals were injected with JWH1330.2 mg/kg ip, and after 30 min, the radiotracer was injected andnew PET images were acquired (see below).

PET images were acquired in fasted mice injected (i.p.) with18FDG (approx. 11.1 MBq or 300 mCi/200 ml saline, PET TechnologyInstitute, Madrid). After a 30 min period for radiotracer uptake, theanimals were anesthetized by isoflurane (2%) inhalation, to assureimmobilization during the scanning process, and they were placedon the bed of the tomograph. Images of 18FDG uptake were ac-quired over 30 min, immediately followed by a CT (computed to-mography) acquisition (Albira ARS small animal PET/CT dual

scanner, Oncovision, Spain). PET images were reconstructed usingthe ordered subset expectation maximization algorithm (OSEM, 3iterations), applying corrections for dead time, radioactive decay,random coincidences and scattering. For the CT images a highresolution filtered back projection (FBP) algorithm was used.

Metabolic activity was quantified by matching the CT image ofthe skull of each animal to a common magnetic resonance (MR)mouse brain template, in which regions of interest were previouslydelineated. After saving the spatial transformation, it was appliedto the corresponding fused PET image, to allow the correct co-registration of the PET image to the MR brain template. Resultswere normalized to the actual radioactivity dose injected and thebody weight of each animal (Kuntner et al., 2009). All these pro-cesses were carried out with PMOD software version 3.0 (PMODTechnologies, Switzerland).

2.11. Data presentation and statistical analysis

All data are expressed as themeans ± orþ SEM of the number ofindependent observations (n) indicated. Normalized data weretested for normality with the Kolmogorov-Smirnov normality testsand statistical significance was calculated by one sample t-test

A. K€ofalvi et al. / Neuropharmacology 110 (2016) 519e529 525

against the hypothetical value of zero (Figs. 1C and 2). Alternatively,a paired t-test (Fig. 3C) or repeated measures of ANOVA followed byBonferroni's post-hoc test were used, and a value of P < 0.05 wasaccepted as significant.

3. Results

3.1. CB2R activation rapidly enhances glucose uptake inhippocampal slices

First we assessed whether CB2R activation affects glucose up-take in superfused hippocampal slices, and the rate of fluorescence-labeled deoxyglucose (2-NBDG) accumulation was tested at 30 sresolution in such slices. The basal accumulation of 2-NBDG wasmost evident in astrocyte-rich areas of transverse hippocampalslices, i.e. in the in the stratum lacunosum, followed by the strataradiatum and moleculare, and the smallest signal was found in thestrata pyramidale and granulare (Fig. 1AeA0), consistent with pre-vious findings (Jakoby et al., 2014). Importantly, both of the twoselective CB2R agonists, GP1a (100 nM, P < 0.01) and JWH133 (1 mM,P < 0.05), rapidly increased the velocity of 2-NBDG accumulation inthese hippocampal slices (Fig. 1 B,C, and Supplementary Video).

Supplementary data related to this article can be found online athttp://dx.doi.org/10.1016/j.neuropharm.2016.03.015.

3.2. CB2R activation stimulates glucose transport in culturedneurons and astrocytes

Since the use of 2-NBDG underrepresents neuronal glucosetransport (Jakoby et al., 2014), we measured the effect of acutein vitro CB2R activation on glucose uptake by primary mousecortical astrocytes and neurons using a 3HDG transport assay. In thepresence of JWH133 (30 nM) over 30 min, glucose uptakeaugmented by 35.4 ± 7.5% in astrocytes (n ¼ 7, P < 0.01 vs. DMSOcontrol) and by 13.5 ± 4.0% in neurons (n ¼ 5, P < 0.05: 2 A,B).Notably, JWH133 exhibits an EC50 value close to 1 mM at the CB1R(Huffman et al., 1999). To avoid the possible non-specific co-acti-vation of CB1R in these cell cultures, we opted for testing JWH133here at 30 nM, i.e. 9-times above its EC50 at the CB2R but more thanan order of magnitude below its EC50 at the CB1R (Huffman et al.,1999).

Similarly, GP1a at a concentration ~2700-times above its EC50 atthe CB2R but ~4-fold below its EC50 at the CB1R, i.e. at 100 nM(Murineddu et al., 2006) also increased glucose uptake in astrocytesby 22.4 ± 10.7% (n ¼ 7, P < 0.05) and by 13.7 ± 2.4% in neurons(n ¼ 5, P < 0.01: Fig. 2 A,B). Therefore, with the chosen ligandconcentrations we achieved good selectivity toward the CB2R whilestill observing maximal effect amplitudes. The CB2R antagonistAM630 (1 mM) did not alter glucose uptake per se in cultures ofastrocytes (n ¼ 7) or neurons (n ¼ 5). However, in the presence ofAM630, neither GP1a nor JWH133 significantly altered glucoseuptake in either of these two cell types (P > 0.05 vs. AM630 control;astrocytes n ¼ 5, neurons n ¼ 4: Fig. 2 A,B). In summary, CB2Ractivation was more effective at increasing glucose uptake in as-trocytes compared with neurons. As stated previously, our neuroncultures contained 15% of astrocytes, which likely contributed tothe effect of the CB2R agonists. However, this low astrocyticcontamination cannot solely account for the effect amplitudesobserved neuronal cultures.

3.3. b-amyloidosis hinders endogenous CB2R activation

A similar pharmacological analysis was then performed in hip-pocampal slices of 12-month-old TgAPP mice and their age-matched controls. The normalized control uptake values of the

TgAPP hippocampus amounted to 105.4 ± 8.9% of the WT controluptake (i.e. þ5.4 ± 8.9% over DMSO control) when assessed in apairwise arrangement (n ¼ 19 pairs, P > 0.05: Fig. 3 A,B). Glucoseuptake was significantly enhanced by the CB2R-selective agonistJWH133 (1 mM) in both the WT (þ26.3 ± 10.0%, n ¼ 12, P < 0.05:Fig. 3A) and the TgAPP slices (þ16.7± 6.7%, n¼ 12, P< 0.05: Fig. 3B).Due to the limited availability of 12-month old TgAPP mice, wemade a risk-free decision when testing JWH133 at 1 mM, i.e. at themaximal concentration where good CB2R-selectivity is still ach-ieved (Huffman et al., 1999).

Importantly, glucose uptake was also stimulated by the non-selective (“mixed”) CB1R/CB2R agonist, WIN55212-2 (1 mM), inboth WT (þ22.4 ± 9.2%, n ¼ 12, P < 0.05: Fig. 3A) and TgAPP slices(þ17.5 ± 7.1%, n ¼ 12, P < 0.05: Fig. 3B). While the selective CB2Rantagonist, AM630 (1 mM) had no effect on glucose uptake per se ineither strain (n¼ 12), it did prevent JWH133 andWIN55212-2 fromstimulating glucose uptake in both WT and TgAPP slices (n ¼ 6:Fig. 3 A,B).

Cannabinoid agonists exert their actions predominantly viaCB1R activation in themammalian brain (Katona and Freund, 2012).To check if CB1Rs are also involved in the stimulation of glucoseuptake, first we tested the CB1R-selective agonist ACEA at a con-centration half way between its Kd values at the CB1R and the CB2R,i.e. at 1 mM (Hillard et al., 1999). ACEA did not affect glucose uptakein the middle-aged C57Bl/6 mice (þ5.3 ± 7.9%, n ¼ 7, P > 0.05:Fig. 3A). Subsequently, we asked if CB1Rs were involved in the effectof JWH133, taken that this ligand has some affinity toward the CB1Rat 1 mM (Huffman et al., 1999). Thus, we tested JWH133 (1 mM) inacute hippocampal slices of young adult CB1R KO mice and theirWT littermates on the CD-1 strain. JWH133 (1 mM) significantlystimulated glucose uptake þ13.4 ± 3.7% (n ¼ 6, P < 0.05) in the WTmice and þ26.7 ± 9.0% (n ¼ 6, P < 0.05) in the CB1R KO, thusvirtually, JWH133 was more twice as more effective in the absenceof the CB1R. These data together with the results with ACEA alsoconfirm the absence of acute modulation of hippocampal glucoseuptake by CB1Rs.

Next, we assessed whether endogenous activation of CB2R bythe major endocannabinoids, AEA and 2-AG also bolstered glucoseuptake. To that end, we inhibited either fatty acid amide hydrolase(FAAH) or cyclooxygenase-2 (COX-2), two enzymes known to beinvolved in AEA degradation (Egertov�a et al., 2003; Fowler et al.,2013). The treatment of slices with DuP697 (500 nM), a selectiveCOX-2 inhibitor (IC50 at COX-2, 10 nM; IC50 at COX-1, 800 nM;Gierse et al., 1995), augmented glucose uptake by þ16.6 ± 7.7% intheWTmice (n¼ 6, P < 0.05: Fig. 3A) - an effect that was preventedby CB2R blockade (n ¼ 6: Fig. 3A). However, unlike in the WT,DuP697 failed to modify glucose uptake in the TgAPP mice (n ¼ 6,Fig. 3B). In contrast, LY2183240 (100 nM), a potent dual inhibitor ofFAAH and AEA reuptake (Nicolussi et al., 2014), did not affectglucose uptake in WT mice, and therefore, it was not tested inTgAPP mice slices. The major enzymes that degrade 2-AG in themammalian brain are monoacylglycerol lipase (MAGL) and a/b-hydrolase domain 6/12 (ABDH6/12) (Savinainen et al., 2012;Murataeva et al., 2014). The inhibition of 2-AG hydrolysis by theconcurrent application of JZL184 (1 mM), a MAGL inhibitor (Longet al., 2009), and WWL70 (1 mM), an a,bDH6/12 inhibitor (Li et al.,2007), also failed to modify glucose uptake in WT mice (Fig. 3 A).

3.4. Lower AEA levels in the hippocampus of TgAPP mice

The inability of COX-2 inhibition to stimulate glucose uptake in aCB2R-dependent fashion in the TgAPP hippocampus suggests a lossof endocannabinoid function, which prompted us to measureendocannabinoid levels in these two strains. The hippocampallevels of AEA and 2-AG in theWTmice (Fig. 4) were similar to those

A. K€ofalvi et al. / Neuropharmacology 110 (2016) 519e529526

reported in the rat brain (Stella et al., 1997), and while no differencein the 2-AG levels were detected between WT and TgAPP mice(n ¼ 7, P > 0.05), AEA levels in the latter were 37.8 ± 11.4% lowerthan in WT (n ¼ 7, P < 0.05) (Fig. 4A,B). Accordingly, there was a2.4 ± 0.3 fold increase in the 2-AG/AEA ratio in TgAPP mice (n ¼ 7,P < 0.001). The degradation of 2-AG was also not statisticallydifferent between the WT and TgAPP (n ¼ 6, P > 0.05) (Fig. 4D).Interestingly, the degradation of AEA into ethanolamine by FAAHwas significantly reduced in TgAPP by 33.5 ± 8.9% (n¼ 6, P < 0.001)(Fig. 4D).

3.5. Effect of CB2R activation on 18FDG uptake e a PET analysisin vivo

Finally, we tested the feasibility of targeting CB2R to stimulatebrain glucose uptake in middle-aged mice in vivo. We first evalu-ated the basal uptake of 18FDG in the brain regions typically bearingthe hallmarks of the diminished 18FDG signal in the Alzheimer'sdisease-affected human brain, i.e. the frontal and temporal cortices,and the hippocampus (Mosconi, 2005). Like the ex vivo data, weobserved no difference in the 18FDG signal in the aforementionedareas of WT and TgAPP mice (n ¼ 9: Fig. 5 A,A0 ,B,B0,C). Four dayslater, the same animals were injected with a small dose of JWH133(0.2 mg/kg, i.p.) and they were re-evaluated for 18FDG uptake.While this CB2R agonist did not alter blood glucose levels (Fig. 5 D),it did stimulate glucose uptake by an average of 31.5% in each of thethree regions in theWTmice (n¼ 4, P < 0.05: Fig. 5EeG), as well asby an average of 21.3% in the TgAPP mice (n ¼ 4, P < 0.05:Fig. 5E0eG0).

4. Discussion

4.1. A novel role for cannabinoid CB2 receptors: stimulating glucoseuptake in the brain

We show here that CB2R activation in vitro, and more impor-tantly, in vivo rapidly enhances cerebral glucose uptake by ~30%over the basal levels. These data are in concert with a recent findingthat the intraperitoneal injection of the mixed CB1R/CB2R agonist,HU210 increased cerebral 18FDG PET signal by 21% in the rat(Nguyen et al., 2012). Since CB1R activation decreases mitochon-drial glucose metabolism (Duarte et al., 2012) CB2Rs are likelyinvolved in the stimulation of brain metabolism in the above study.These data support the view of the recently described antagonisticinteraction between CB1Rs and CB2R in neurons (Call�en et al., 2012).

Given that brain glucose availability controls cognition andmemory in humans (Messier, 2004) and that central metabolicboosting alleviates the cognitive symptoms of dementia, these datasuggest that CB2R agonists might represent a novel class of noot-ropic drugs (Branconnier, 1983). This raises the interesting hy-pothesis that by promoting AEA release, neural activity can activatecentral CB2Rs and thereby refuel circuits under heavy load. Indeed,it was recently found that JWH133 enhanced aversive memoryconsolidation in mice, while the genetic ablation or pharmacolog-ical inhibition of CB2R impaired it (García-Guti�errez et al., 2013).

2-NBDG uptake in slices suggested a contribution of astrocytesin the effects of CB2R activation. Consistent with this, our data incultures show that CB2R influences both astrocytic and neuronalglucoregulation. CB2R are found at low density onmost cell types inthe brain, including activated microglia and astrocytes controllingneuroinflammation (Ramírez et al., 2005; Martín-Moreno et al.,2012), and inhibiting astrocytoma growth (S�anchez et al., 2001).Moreover, CB2R mRNA or protein has also been found in severaltypes of neurons, including cortical, hippocampal and midbrainneurons (Atwood and Mackie, 2010; Call�en et al., 2012; Den Boon

et al., 2012; Zhang et al., 2014). Interestingly, CB2R expression anddensity in neurons apparently surpasses that in microglia or as-trocytes in the mouse neocortex (Savonenko et al., 2015) and hip-pocampus (Li and Kim, 2015).

The therapeutic activation of CB2R is safe and it does not triggerpsychoactivity (Atwood and Mackie, 2010; Pertwee, 2012).Furthermore, CB2R density is known to increase in Alzheimer'sdisease and Down's syndrome, as well as in b-amyloidosis in vitro(Esposito et al., 2007; Ruiz-Valdepe~nas et al., 2010; Solas et al.,2013). This probably reflects an adaptive self-defense processsince CB2R activation confers neuroprotection in several experi-mental models of Alzheimer's disease (Ramírez et al., 2005;Esposito et al., 2007; Ruiz-Valdepe~nas et al., 2010; Martín-Moreno et al., 2012).

4.2. AEA and b-amyloidosis

The first step in common pathomechanisms toward AD pa-thology is excessive b-amyloid accumulation (Sperling et al., 2013;Zahs and Ashe, 2013). Remarkably, the antagonistic relationshipbetween AEA and b-amyloid appears to be bi-directional: b-amy-loid toxicity is prevented by AEA in human cell lines in vitro (Milton,2002) and conversely, b-amyloid treatment decreases AEA levels inC6 rat astroglioma cells (Esposito et al., 2007). Accordingly, AEAlevels decrease in the AD brain and they are inversely correlatedwith b-amyloid levels (Jung et al., 2012), consistent with the lowerhippocampal AEA levels in TgAPP mice. b-amyloidosis may lowerAEA levels by either bolstering catabolism and/or by impairing itssynthesis. COX-2 is an important enzyme in AEA metabolism(Glaser and Kaczocha, 2010; Pamplona et al., 2010), furthermore, b-amyloidosis induces COX-2 expression in astrocytes (Giovanniniet al., 2002) and in AD-affected neurons. Hence, COX-2 inhibitionmay help to decrease the incidence or slow the progress of AD (Berket al., 2013). In fact, in a previous work we treated TgAPP 2576 miceorally for four months with the CB2R-selective agonist JWH133 andwe observed an attenuation of the AD phenotype, including re-covery from memory impairment and neuroinflammation, as wellas a normalization of the increase in COX-2 protein levels (Martín-Moreno et al., 2012).

Strikingly, COX-2 inhibition by the selective DuP697 antagonistfacilitated glucose uptake via CB2R activation in WT but not in theTgAPP mice, probably due to the rescue of AEA from COX-2-driventonic metabolism in WT mice (Glaser and Kaczocha, 2010;Pamplona et al., 2010). In the TgAPP mice, we found a 38%decrease of AEA, suggesting that these levels were already too lowto stimulate the CB2Rs, even following COX-2 blockade. In fact, thismay illustrate a deficit in synthesis, since AEA production requiresthe concurrent stimulation of both NMDA and acetylcholine re-ceptors (Stella and Piomelli, 2001), both of which become hypo-functional in AD (Pavía et al., 1998; Giovannini et al., 2002). In thepresent work FAAH activity was reduced in TgAPP mice, as judgedby the decrease in AEA hydrolysis, again in agreement with itsreduction in AD brain, which we recently reported (Mulder et al.,2011; Pascual et al., 2014). This reduction is mimicked by patho-logically relevant Ab1-40 peptide concentrations (Pascual et al.,2014) and may serve as to preserve AEA levels under lower AEAproduction, as a compensatory mechanism. Even though, FAAHinhibition did not stimulate glucose uptake, suggesting that AEAmetabolism by FAAH is not a rate limiting factor for CB2R activationin the WT hippocampus.

4.3. Cannabinoid receptors and glucose uptake

Although the role of cannabinoid receptors in peripheral glu-coregulation is extensively studied (Matias et al., 2008), the

A. K€ofalvi et al. / Neuropharmacology 110 (2016) 519e529 527

mechanism through which CB2R activation facilitates glucosetransport in the brain awaits further interrogation. Several mech-anisms could be invoked on that respect. For instance, cannabi-noids stimulate AMP-activated protein kinase (Dagon et al., 2007), amajor intracellular energy sensor, which leads to enhanced glucoseuptake by stimulating glucose transporter (GLUT) membraneexpression and transport efficiency (Rutter et al., 2003; Weisov�aet al., 2009). More importantly, the effect was absent in CB2Rknockout mice (Dagon et al., 2007). A possible direct effect of CB2Ragonists on glucose transporters remains unexplored. New syn-thesis of transporters would not parallel the acute pharmacologicaleffects reported here. Therefore, we suggest that CB2R agonists mayfavor transporter localization at the plasma membrane. On theother hand, it is less likely that CB2Rs increased glucose uptakedriven by the stimulation of its metabolism, because we foundpreviously that WIN55212-2 slows rather than facilitates themitochondrial oxidative metabolism of glucose via CB1R activationboth in neurons and astrocytes (Duarte et al., 2012).

4.4. Conclusions

In summary, the present results provide the first direct phar-macological evidence in vitro and in vivo of a role of CB2R in centralglucoregulation. Additionally, we found that glucoregulation byendogenous CB2R signalling is negatively affected by b-amyloidosis,thought to be the first pathological step in AD. Therefore, it wouldbe interesting to perform further studies to define how CB2Rmediated glucoregulation contributes to the recently discoveredtherapeutic potential of CB2R agonists in animal models of AD(Martín-Moreno et al., 2012; Wu et al., 2013; Aso and Ferrer, 2014;Cheng et al., 2014).

Conflicts of interest

The authors have no conflict of interests to report, and receivedno financial support or compensation from any individual orcorporate entity over the past 3 years for research or professionalservices, and there are no personal holdings that could be perceivedas constituting a potential conflict of interest.

Author contributions

AK and MLC conceived and designed the work, performed dataanalysis, writing and edition of the manuscript; AK, CL, BSP and AV-F performed glucose uptake in slices and in the cell cultures; ROB,PA and RJR made the cell cultures; AMM-M assessed glucose inplasma, genotyped by PCR and performed the respective dataanalysis; LG-G and MAP conducted the PET studies and therespective data analysis; MLC and SJP performed the endocanna-binoid hydrolysis assay; and RAC interpreted results, and contrib-uted to the writing and the edition of the manuscript, as well as tothe purchase of some supplies. All authors have reviewed andcommented on the manuscript.

Acknowledgements

This work was supported by the Council of Madrid (S-BIO/0170/2006 and P2010/BMD-2349 to MLC and MAP). AMM-M received afellowship from the Spanish Ministry of Education and Science.Authors from the University of Coimbra acknowledge funding fromthe following sources: NARSAD, Santa Casa da Miserc�ordia, DARPA(09-68-ESR-FP-010), FEDER (QREN) (CENTRO-07-ST24-FEDER-002006), Programa Mais Centro under project CENTRO-07-ST24-FEDER-002006 and Programa Operacional Factores de Com-petitividade - COMPETE, and national funds (PTDC/SAU-OSM/

105663/2008 and Pest-C/SAU/LA0001/2013e2014) viaFCTeFundaç~ao para a Ciencia e a Tecnologia.

We thank Dr M. Delgado for excellent assistance in the PETstudies, and Drs L. M. García-Segura and M. A. Ar�evalo for com-ments on the manuscript. We are grateful to Drs Catherine Ledentand Catherine Vroonen (IRIBHM, Brussels) for providing the CB1RKO mice and their WT littermates.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.neuropharm.2016.03.015.

References

Agudo, J., Martin, M., Roca, C., Molas, M., Bura, A.S., Zimmer, A., Bosch, F.,Maldonado, R., 2010. Deficiency of CB2 cannabinoid receptor in mice improvesinsulin sensitivity but increases food intake and obesity with age. Diabetologia53, 2629e2640. http://dx.doi.org/10.1007/s00125-010-1894-6.

And�o, R.D., Bír�o, J., Cs€olle, C., Ledent, C., Sperl�agh, B., 2012. The inhibitory action ofexo- and endocannabinoids on [3H]GABA release are mediated by both CB1 andCB2 receptors in the mouse hippocampus. Neurochem. Int. 60, 145e152.

Aso, E., Ferrer, I., 2014. Cannabinoids for treatment of Alzheimer's disease: movingtoward the clinic. Front. Pharmacol. 5, 37. http://dx.doi.org/10.3389/fphar.2014.00037.

Ashton, J.C., 2011. Knockout controls and the specificity of cannabinoid CB2 receptorantibodies. Br. J. Pharmacol. 163, 1113. http://dx.doi.org/10.1111/j.1476-5381.2010.01139.x.

Atwood, B.K., Mackie, K., 2010. CB2: a cannabinoid receptor with an identity crisis.Br. J. Pharmacol. 160, 467e479. http://dx.doi.org/10.1111/j.1476-5381.2010.00729.x.

Bari, M., Bonifacino, T., Milanese, M., Spagnuolo, P., Zappettini, S., Battista, N.,Giribaldi, F., Usai, C., Bonanno, G., Maccarrone, M., 2011. The endocannabinoidsystem in rat gliosomes and its role in the modulation of glutamate release. Cell.Mol. Life Sci. CMLS 68, 833e845. http://dx.doi.org/10.1007/s00018-010-0494-4.

B�enard, G., Massa, F., Puente, N., Lourenço, J., Bellocchio, L., Soria-G�omez, E.,Matias, I., Delamarre, A., Metna-Laurent, M., Cannich, A., Hebert-Chatelain, E.,Mulle, C., Ortega-Guti�errez, S., Martín-Fontecha, M., Klugmann, M.,Guggenhuber, S., Lutz, B., Gertsch, J., Chaouloff, F., L�opez-Rodríguez, M.L.,Grandes, P., Rossignol, R., Marsicano, G., 2012. Mitochondrial CB receptorsregulate neuronal energy metabolism. Nat. Neurosci. 15, 558e564. http://dx.doi.org/10.1038/nn.3053.

Benito, C., Tol�on, R.M., Pazos, M.R., Nú~nez, E., Castillo, A.I., Romero, J., 2008.Cannabinoid CB2 receptors in human brain inflammation. Br. J. Pharmacol. 153,277e285. http://dx.doi.org/10.1038/sj.bjp.0707505.

Berk, M., Dean, O., Drexhage, H., McNeil, J.J., Moylan, S., O'Neil, A., Davey, C.G.,Sanna, L., Maes, M., 2013. Aspirin: a review of its neurobiological properties andtherapeutic potential for mental illness. BMC Med. 11, 74. http://dx.doi.org/10.1186/1741-7015-11-74.

Branconnier, R.J., 1983. The efficacy of the cerebral metabolic enhancers in thetreatment of senile dementia. Psychopharmacol. Bull. 19, 212e219.

Call�en, L., Moreno, E., Barroso-Chinea, P., Moreno-Delgado, D., Cort�es, A., Mallol, J.,Casad�o, V., Lanciego, J.L., Franco, R., Lluis, C., Canela, E.I., McCormick, P.J., 2012.Cannabinoid receptors CB1 and CB2 form functional heteromers in brain. J. Biol.Chem. 287, 20851e20865. http://dx.doi.org/10.1074/jbc.M111.335273.

Carlisle, S.J., Marciano-Cabral, F., Staab, A., Ludwick, C., Cabral, G.A., 2002. Differ-ential expression of the CB2 cannabinoid receptor by rodent macrophages andmacrophage-like cells in relation to cell activation. Int. Immunopharmacol. 2,69e82.

Cheng, Y., Dong, Z., Liu, S., 2014. b-Caryophyllene ameliorates the Alzheimer-likephenotype in APP/PS1 Mice through CB2 receptor activation and the PPARgpathway. Pharmacology 94, 1e12. http://dx.doi.org/10.1159/000362689.

Dagon, Y., Avraham, Y., Ilan, Y., Mechoulam, R., Berry, E.M., 2007. Cannabinoidsameliorate cerebral dysfunction following liver failure via AMP-activated pro-tein kinase. FASEB J. Off. Publ. Fed. Am. Soc. Exp. Biol. 21, 2431e2441. http://dx.doi.org/10.1096/fj.06-7705com.

Den Boon, F.S., Chameau, P., Schaafsma-Zhao, Q., van Aken, W., Bari, M., Oddi, S.,Kruse, C.G., Maccarrone, M., Wadman, W.J., Werkman, T.R., 2012. Excitability ofprefrontal cortical pyramidal neurons is modulated by activation of intracellulartype-2 cannabinoid receptors. Proc. Natl. Acad. Sci. U. S. A. 109, 3534e3539.http://dx.doi.org/10.1073/pnas.1118167109.

Di Marzo, V., De Petrocellis, L., 2012. Why do cannabinoid receptors have more thanone endogenous ligand? Philos. Trans. R. Soc. Lond. B Biol. Sci. 367, 3216e3228.http://dx.doi.org/10.1098/rstb.2011.0382.

Duarte, J.M.N., Ferreira, S.G., Carvalho, R.A., Cunha, R.A., K€ofalvi, A., 2012. CB1 re-ceptor activation inhibits neuronal and astrocytic intermediary metabolism inthe rat hippocampus. Neurochem. Int. 60, 1e8. http://dx.doi.org/10.1016/j.neuint.2011.10.019.

Duff, G., Argaw, A., Cecyre, B., Cherif, H., Tea, N., Zabouri, N., Casanova, C., Ptito, M.,Bouchard, J.-F., 2013. Cannabinoid receptor CB2 modulates axon guidance. PloS

A. K€ofalvi et al. / Neuropharmacology 110 (2016) 519e529528

One 8, e70849. http://dx.doi.org/10.1371/journal.pone.0070849.Egertov�a, M., Cravatt, B.F., Elphick, M.R., 2003. Comparative analysis of fatty acid

amide hydrolase and cb1 cannabinoid receptor expression in the mouse brain:evidence of a widespread role for fatty acid amide hydrolase in regulation ofendocannabinoid signaling. Neuroscience 119, 481e496.

Esposito, G., Iuvone, T., Savani, C., Scuderi, C., De Filippis, D., Papa, M., Di Marzo, V.,Steardo, L., 2007. Opposing control of cannabinoid receptor stimulation onamyloid-beta-induced reactive gliosis: in vitro and in vivo evidence.J. Pharmacol. Exp. Ther. 322, 1144e1152. http://dx.doi.org/10.1124/jpet.107.121566.

Fowler, C.J., Bj€orklund, E., Lichtman, A.H., Naidu, P.S., Congiu, C., Onnis, V., 2013.Inhibitory properties of ibuprofen and its amide analogues towards the hy-drolysis and cyclooxygenation of the endocannabinoid anandamide. J. Enzyme.Inhib. Med. Chem. 28, 172e182. http://dx.doi.org/10.3109/14756366.2011.643304.

Gali�egue, S., Mary, S., Marchand, J., Dussossoy, D., Carriere, D., Carayon, P.,Bouaboula, M., Shire, D., Le Fur, G., Casellas, P., 1995. Expression of central andperipheral cannabinoid receptors in human immune tissues and leukocytesubpopulations. Eur. J. Biochem. 232, 54e61.

García-Guti�errez, M.S., Ortega-�Alvaro, A., Busquets-García, A., P�erez-Ortiz, J.M.,Caltana, L., Ricatti, M.J., Brusco, A., Maldonado, R., Manzanares, J., 2013. Synapticplasticity alterations associated with memory impairment induced by deletionof CB2 cannabinoid receptors. Neuropharmacology 73, 388e396. http://dx.doi.org/10.1016/j.neuropharm.2013.05.034.

Gierse, J.K., Hauser, S.D., Creely, D.P., Koboldt, C., Rangwala, S.H., Isakson, P.C.,Seibert, K., 1995. Expression and selective inhibition of the constitutive andinducible forms of human cyclo-oxygenase. Biochem. J. 305, 479e484.

Giovannini, M.G., Scali, C., Prosperi, C., Bellucci, A., Vannucchi, M.G., Rosi, S.,Pepeu, G., Casamenti, F., 2002. Beta-amyloid-induced inflammation andcholinergic hypofunction in the rat brain in vivo: involvement of the p38MAPKpathway. Neurobiol. Dis. 11, 257e274.

Glaser, S.T., Kaczocha, M., 2010. Cyclooxygenase-2 mediates anandamide meta-bolism in the mouse brain. J. Pharmacol. Exp. Ther. 335, 380e388. http://dx.doi.org/10.1124/jpet.110.168831.

Han, S., Thatte, J., Buzard, D.J., Jones, R.M., 2013. Therapeutic utility of cannabinoidreceptor type 2 (CB2) selective agonists. J. Med. Chem. 56, 8224e8256. http://dx.doi.org/10.1021/jm4005626.

Hillard, C.J., Manna, S., Greenberg, M.J., DiCamelli, R., Ross, R.A., Stevenson, L.A.,Murphy, V., Pertwee, R.G., Campbell, W.B., 1999. Synthesis and characterizationof potent and selective agonists of the neuronal cannabinoid receptor (CB1).J. Pharmacol. Exp. Ther. 289, 1427e1433.

Hsiao, K., Chapman, P., Nilsen, S., Eckman, C., Harigaya, Y., Younkin, S., Yang, F.,Cole, G., 1996. Correlative memory deficits, Abeta elevation, and amyloid pla-ques in transgenic mice. Science 274, 99e102.

Huffman, J.W., Liddle, J., Yu, S., Aung, M.M., Abood, M.E., Wiley, J.L., Martin, B.R.,1999. 3-(10,10-Dimethylbutyl)-1-deoxy-delta8-THC and related compounds:synthesis of selective ligands for the CB2 receptor. Bioorg. Med. Chem. 7,2905e2914.

Jakoby, P., Schmidt, E., Ruminot, I., Guti�errez, R., Barros, L.F., Deitmer, J.W., 2014.Higher transport and metabolism of glucose in astrocytes compared withneurons: a multiphoton study of hippocampal and cerebellar tissue slices.Cereb. Cortex N. Y. N. 1991 (24), 222e231. http://dx.doi.org/10.1093/cercor/bhs309.

Jung, K.-M., Astarita, G., Yasar, S., Vasilevko, V., Cribbs, D.H., Head, E., Cotman, C.W.,Piomelli, D., 2012. An amyloid b42-dependent deficit in anandamide mobili-zation is associated with cognitive dysfunction in Alzheimer's disease. Neuro-biol. Aging 33, 1522e1532. http://dx.doi.org/10.1016/j.neurobiolaging.2011.03.012.

Katona, I., Freund, T.F., 2012. Multiple functions of endocannabinoid signaling in thebrain. Annu. Rev. Neurosci. 35, 529e558.

Kilkenny, C., Browne, W., Cuthill, I.C., Emerson, M., Altman, D.G., NC3Rs ReportingGuidelines Working Group, 2010. Animal research: reporting in vivo experi-ments: the ARRIVE guidelines. Br. J. Pharmacol. 160, 1577e1579. http://dx.doi.org/10.1111/j.1476-5381.2010.00872.x.

Kuntner, C., Kesner, A.L., Bauer, M., Kremslehner, R., Wanek, T., Mandler, M.,Karch, R., Stanek, J., Wolf, T., Müller, M., Langer, O., 2009. Limitations of smallanimal PET imaging with [18F]FDDNP and FDG for quantitative studies in atransgenic mouse model of Alzheimer's disease. Mol. Imaging Biol. MIB Off.Publ. Acad. Mol. Imaging 11, 236e240. http://dx.doi.org/10.1007/s11307-009-0198-z.

Ledent, C., Valverde, O., Cossu, G., Petitet, F., Aubert, J.F., Beslot, F., B€ohme, G.A.,Imperato, A., Pedrazzini, T., Roques, B.P., Vassart, G., Fratta, W., Parmentier, M.,1999. Unresponsiveness to cannabinoids and reduced addictive effects of opi-ates in CB1 receptor knockout mice. Science 283, 401e404.

Lemos, C., Pinheiro, B.S., Beleza, R.O., Marques, J.M., Rodrigues, R.J., Cunha, R.A.,Rial, D., K€ofalvi, A., 2015. Adenosine A2B receptor activation stimulates glucoseuptake in the mouse forebrain. Purinergic Signal 11, 561e569. http://dx.doi.org/10.1007/s11302-015-9474-3.

Lemos, C., Val�erio-Fernandes, A., Ghisleni, G.C., Ferreira, S.G., Ledent, C., deCeballos, M.L., K€ofalvi, A., 2012. Impaired hippocampal glucoregulation in thecannabinoid CB1 receptor knockout mice as revealed by an optimized in vitroexperimental approach. J. Neurosci. Methods 204, 366e373. http://dx.doi.org/10.1016/j.jneumeth.2011.11.028.

Li, Y., Kim, J., 2015. Neuronal expression of CB2 cannabinoid receptor mRNAs in themouse hippocampus. Neuroscience 311, 253e267.

Li, Y., Kim, J., 2016a. Deletion of CB2 cannabinoid receptors reduces synaptictransmission and long-term potentiation in the mouse hippocampus. Hippo-campus 26, 275e281.

Li, Y., Kim, J., 2016b. CB2 cannabinoid receptor knockout in mice impairs contextuallong-term memory and enhances spatial working memory. Neural Plast.9817089 http://dx.doi.org/10.1155/2016/9817089.

Li, W., Blankman, J.L., Cravatt, B.F., 2007. A functional proteomic strategy to discoverinhibitors for uncharacterized hydrolases. J. Am. Chem. Soc. 129, 9594e9595.

Long, J.Z., Li, W., Booker, L., Burston, J.J., Kinsey, S.G., Schlosburg, J.E., Pav�on, F.J.,Serrano, A.M., Selley, D.E., Parsons, L.H., Lichtman, A.H., Cravatt, B.F., 2009. Se-lective blockade of 2-arachidonoylglycerol hydrolysis produces cannabinoidbehavioral effects. Nat. Chem. Biol. 5, 37e44. http://dx.doi.org/10.1038/nchembio.129.

Martín-Moreno, A.M., Brera, B., Spuch, C., Carro, E., García-García, L., Delgado, M.,Pozo, M.A., Innamorato, N.G., Cuadrado, A., de Ceballos, M.L., 2012. Prolongedoral cannabinoid administration prevents neuroinflammation, lowers b-amy-loid levels and improves cognitive performance in Tg APP 2576 mice.J. Neuroinflammation 9, 8. http://dx.doi.org/10.1186/1742-2094-9-8.

Matias, I., Marzo, V.D., K€ofalvi, A., 2008. Endocannabinoids in energy homeostasisand metabolic disorders. In: K€ofalvi, A. (Ed.), Cannabinoids and the Brain.Springer, US, pp. 277e316.

McGrath, J.C., Drummond, G.B., McLachlan, E.M., Kilkenny, C., Wainwright, C.L.,2010. Guidelines for reporting experiments involving animals: the ARRIVEguidelines. Br. J. Pharmacol. 160, 1573e1576. http://dx.doi.org/10.1111/j.1476-5381.2010.00873.x.

Messier, C., 2004. Glucose improvement of memory: a review. Eur. J. Pharmacol.490, 33e57. http://dx.doi.org/10.1016/j.ejphar.2004.02.043.

Milton, N.G.N., 2002. Anandamide and noladin ether prevent neurotoxicity of thehuman amyloid-beta peptide. Neurosci. Lett. 332, 127e130.

Morgan, N.H., Stanford, I.M., Woodhall, G.L., 2009. Functional CB2 type cannabinoidreceptors at CNS synapses. Neuropharmacology 57, 356e368.

Mosconi, L., 2005. Brain glucose metabolism in the early and specific diagnosis ofAlzheimer's disease. FDG-PET studies in MCI and AD. Eur. J. Nucl. Med. Mol.Imaging 32, 486e510. http://dx.doi.org/10.1007/s00259-005-1762-7.

Mulder, J., Zilberter, M., Pasquar�e, S.J., Alp�ar, A., Schulte, G., Ferreira, S.G., K€ofalvi, A.,Martín-Moreno, A.M., Keimpema, E., Tanila, H., Watanabe, M., Mackie, K.,Hortob�agyi, T., de Ceballos, M.L., Harkany, T., 2011. Molecular reorganization ofendocannabinoid signalling in Alzheimer's disease. Brain J. Neurol. 134,1041e1060. http://dx.doi.org/10.1093/brain/awr046.

Murataeva, N., Straiker, A., Mackie, K., 2014. Parsing the players: 2-arachidonoylglycerol synthesis and degradation in the CNS. Br. J. Pharmacol.171 (6), 1379e1391. http://dx.doi.org/10.1111/bph.12411, 2014 Mar.

Murineddu, G., Lazzari, P., Ruiu, S., Sanna, A., Loriga, G., Manca, I., Falzoi, M., Dessì, C.,Curzu, M.M., Chelucci, G., Pani, L., Pinna, G.A., 2006. Tricyclic pyrazoles. 4.Synthesis and biological evaluation of analogues of the robust and selective CB2cannabinoid ligand 1-(20,40-dichlorophenyl)-6-methyl-N-piperidin-1-yl-1,4-dihydroindeno[1,2-c]pyrazole-3-carboxamide. J. Med. Chem. 49, 7502e7512.http://dx.doi.org/10.1021/jm060920d.

Nicolussi, S., Chicca, A., Rau, M., Rihs, S., Soeberdt, M., Abels, C., Gertsch, J., 2014.Correlating FAAH and anandamide cellular uptake inhibition using N-alkyl-carbamate inhibitors: from ultrapotent to hyperpotent. Biochem. Pharmacol.92, 669e689. http://dx.doi.org/10.1016/j.bcp.2014.09.020.

Nguyen, V.H., Verdurand, M., Dedeurwaerdere, S., Wang, H., Zahra, D., Gregoire, M.-C., Zavitsanou, K., 2012. Increased brain metabolism after acute administrationof the synthetic cannabinoid HU210: a small animal PET imaging study with18F-FDG. Brain Res. Bull. 87, 172e179. http://dx.doi.org/10.1016/j.brainresbull.2011.11.011.

Pamplona, F.A., Menezes-de-Lima, O., Takahashi, R.N., 2010. Aspirin-triggered lip-oxin induces CB1-dependent catalepsy in mice. Neurosci. Lett. 470, 33e37.http://dx.doi.org/10.1016/j.neulet.2009.12.050.

Pascual, A.C., Martín-Moreno, A.M., Giusto, N.M., de Ceballos, M.L., Pasquar�e, S.J.,2014. Normal aging in rats and pathological aging in human Alzheimer's dis-ease decrease FAAH activity: modulation by cannabinoid agonists. Exp. Ger-ontol. 60, 92e99. http://dx.doi.org/10.1016/j.exger.2014.10.011.

Pavía, J., de Ceballos, M.L., Sanchez de la Cuesta, F., 1998. Alzheimer's disease:relationship between muscarinic cholinergic receptors, beta-amyloid and tauproteins. Fundam. Clin. Pharmacol. 12, 473e481.

Pertwee, R.G., 2012. Targeting the endocannabinoid system with cannabinoid re-ceptor agonists: pharmacological strategies and therapeutic possibilities. Philos.Trans. R. Soc. Lond. B Biol. Sci. 367, 3353e3363. http://dx.doi.org/10.1098/rstb.2011.0381.

Ramírez, B.G., Bl�azquez, C., G�omez del Pulgar, T., Guzm�an, M., de Ceballos, M.L.,2005. Prevention of Alzheimer's disease pathology by cannabinoids: neuro-protection mediated by blockade of microglial activation. J. Neurosci. Off. J. Soc.Neurosci. 25, 1904e1913. http://dx.doi.org/10.1523/JNEUROSCI.4540-04.2005.

Rodríguez-Cueto, C., Benito, C., Fern�andez-Ruiz, J., Romero, J., Hern�andez-G�alvez, M., G�omez-Ruiz, M., 2014. Changes in CB(1) and CB(2) receptors in thepost-mortem cerebellum of humans affected by spinocerebellar ataxias. Br. J.Pharmacol. 171, 1472e1489. http://dx.doi.org/10.1111/bph.12283.

Romero-Zerbo, S.Y., Garcia-Gutierrez, M.S., Su�arez, J., Rivera, P., Ruz-Maldonado, I.,Vida, M., Rodriguez de Fonseca, F., Manzanares, J., Bermúdez-Silva, F.J., 2012.Overexpression of cannabinoid CB2 receptor in the brain induces hyper-glycaemia and a lean phenotype in adult mice. J. Neuroendocrinol. 24,1106e1119. http://dx.doi.org/10.1111/j.1365-2826.2012.02325.x.

Ruiz-Valdepe~nas, L., Benito, C., Tol�on, R.M., Martínez Orgado, J.A., Romero, J., 2010.

A. K€ofalvi et al. / Neuropharmacology 110 (2016) 519e529 529

The endocannabinoid system and amyloid-related diseases. Exp. Neurol. 224,66e73. http://dx.doi.org/10.1016/j.expneurol.2010.03.024.

Rutter, G.A., Da Silva Xavier, G., Leclerc, I., 2003. Roles of 50-AMP-activated proteinkinase (AMPK) in mammalian glucose homoeostasis. Biochem. J. 375, 1e16.http://dx.doi.org/10.1042/BJ20030048.

S�anchez, C., de Ceballos, M.L., Gomez del Pulgar, T., Rueda, D., Corbacho, C.,Velasco, G., Galve-Roperh, I., Huffman, J.W., Ram�on y Cajal, S., Guzm�an, M., 2001.Inhibition of glioma growth in vivo by selective activation of the CB(2)cannabinoid receptor. Cancer Res. 61, 5784e5789.

Savinainen, J.R., Saario, S.M., Laitinen, J.T., 2012. The serine hydrolases MAGL,ABHD6 and ABHD12 as guardians of 2-arachidonoylglycerol signalling throughcannabinoid receptors. Acta Physiol. (Oxf) 204, 267e276. http://dx.doi.org/10.1111/j.1748-1716.2011.02280.x.

Savonenko, A.V., Melnikova, T., Wang, Y., Ravert, H., Gao, Y., Koppel, J., Lee, D.,Pletnikova, O., Cho, E., Sayyida, N., Hiatt, A., Troncoso, J., Davies, P., Dannals, R.F.,Pomper, M.G., Horti, A.G., 2015. Cannabinoid CB2 receptors in a mouse model ofAb amyloidosis: immunohistochemical analysis and suitability as a PETbiomarker of neuroinflammation. PLoS One 10 (6), e0129618. http://dx.doi.org/10.1371/journal.pone.0129618.

Sheng, W.S., Hu, S., Min, X., Cabral, G.A., Lokensgard, J.R., Peterson, P.K., 2005.Synthetic cannabinoid WIN55,212-2 inhibits generation of inflammatory me-diators by IL-1beta-stimulated human astrocytes. Glia 49, 211e219. http://dx.doi.org/10.1002/glia.20108.

Sierra, S., Luquin, N., Rico, A.J., G�omez-Bautista, V., Roda, E., Dopeso-Reyes, I.G.,V�azquez, A., Martínez-Pinilla, E., Labandeira-García, J.L., Franco, R., Lanciego, J.L.,2014. Detection of cannabinoid receptors CB1 and CB2 within basal gangliaoutput neurons in macaques: changes following experimental parkinsonism.Brain Struct. Funct. 1e18. http://dx.doi.org/10.1007/s00429-014-0823-8.

Solas, M., Francis, P.T., Franco, R., Ramirez, M.J., 2013. CB2 receptor and amyloidpathology in frontal cortex of Alzheimer's disease patients. Neurobiol. Aging 34,805e808. http://dx.doi.org/10.1016/j.neurobiolaging.2012.06.005.

Sperling, R.A., Karlawish, J., Johnson, K.A., 2013. Preclinical Alzheimer disease-thechallenges ahead. Nat. Rev. Neurol. 9, 54e58. http://dx.doi.org/10.1038/nrneurol.2012.241.

Stella, N., Piomelli, D., 2001. Receptor-dependent formation of endogenous canna-binoids in cortical neurons. Eur. J. Pharmacol. 425, 189e196. http://dx.doi.org/10.1523/jneurosci.4540-04.2005.

Stella, N., Schweitzer, P., Piomelli, D., 1997. A second endogenous cannabinoid thatmodulates long-term potentiation. Nature 388, 773e778.

Teune, L.K., Bartels, A.L., de Jong, B.M., Willemsen, A.T.M., Eshuis, S.A., de Vries, J.J.,van Oostrom, J.C.H., Leenders, K.L., 2010. Typical cerebral metabolic patterns inneurodegenerative brain diseases. Mov. Disord. Off. J. Mov. Disord. Soc. 25,2395e2404. http://dx.doi.org/10.1002/mds.23291.

Van Sickle, M.D., Duncan, M., Kingsley, P.J., Mouihate, A., Urbani, P., Mackie, K.,Stella, N., Makriyannis, A., Piomelli, D., Davison, J.S., Marnett, L.J., Di Marzo, V.,Pittman, Q.J., Patel, K.D., Sharkey, K.A., 2005. Identification and functionalcharacterization of brainstem cannabinoid CB2 receptors. Science 310, 329e332.http://dx.doi.org/10.1126/science.1115740.

Weisov�a, P., Concannon, C.G., Devocelle, M., Prehn, J.H.M., Ward, M.W., 2009.Regulation of glucose transporter 3 surface expression by the AMP-activatedprotein kinase mediates tolerance to glutamate excitation in neurons.J. Neurosci. Off. J. Soc. Neurosci. 29, 2997e3008. http://dx.doi.org/10.1523/JNEUROSCI.0354-09.2009.

Wu, J., Bie, B., Yang, H., Xu, J.J., Brown, D.L., Naguib, M., 2013. Activation of the CB2receptor system reverses amyloid-induced memory deficiency. Neurobiol. Ag-ing 34, 791e804. http://dx.doi.org/10.1016/j.neurobiolaging.2012.06.011.

Zahs, K.R., Ashe, K.H., 2013. b-Amyloid oligomers in aging and Alzheimer's disease.Front. Aging Neurosci. 5, 28. http://dx.doi.org/10.3389/fnagi.2013.00028.

Zhang, H.-Y., Gao, M., Liu, Q.-R., Bi, G.-H., Li, X., Yang, H.-J., Gardner, E.L., Wu, J., Xi, Z.-X., 2014. Cannabinoid CB2 receptors modulate midbrain dopamine neuronalactivity and dopamine-related behavior in mice. Proc. Natl. Acad. Sci. U. S. A. 111,E5007eE5015. http://dx.doi.org/10.1073/pnas.1413210111.