Deletion of Monoglyceride Lipase in Astrocytes Attenuates Lipopolysaccharide-induced Neuroinflammation*

Monoglyceride lipase (MGL) is required for efficient hydrolysis of the endocannabinoid 2-arachidonoylglyerol (2-AG) in the brain generating arachidonic acid (AA) and glycerol. This metabolic function makes MGL an interesting target for the treatment of neuroinflammation, since 2-AG exhibits anti-inflammatory properties and AA is a precursor for pro-inflammatory prostaglandins. Astrocytes are an important source of AA and 2-AG, and highly express MGL. In the present study, we dissected the distinct contribution of MGL in astrocytes on brain 2-AG and AA metabolism by generating a mouse model with genetic deletion of MGL specifically in astrocytes (MKOGFAP). MKOGFAP mice exhibit moderately increased 2-AG and reduced AA levels in brain. Minor accumulation of 2-AG in the brain of MKOGFAP mice does not cause cannabinoid receptor desensitization as previously observed in mice globally lacking MGL. Importantly, MKOGFAP mice exhibit reduced brain prostaglandin E2 and pro-inflammatory cytokine levels upon peripheral lipopolysaccharide (LPS) administration. These observations indicate that MGL-mediated degradation of 2-AG in astrocytes provides AA for prostaglandin synthesis promoting LPS-induced neuroinflammation. The beneficial effect of astrocyte-specific MGL-deficiency is not fully abrogated by the inverse cannabinoid receptor 1 agonist SR141716 (Rimonabant) suggesting that the anti-inflammatory effects are rather caused by reduced prostaglandin synthesis than by activation of cannabinoid receptors. In conclusion, our data demonstrate that MGL in astrocytes is an important regulator of 2-AG levels, AA availability, and neuroinflammation.

Monoglyceride lipase (MGL) is required for efficient hydrolysis of the endocannabinoid 2-arachidonoylglyerol (2-AG) in the brain generating arachidonic acid (AA) and glycerol. This metabolic function makes MGL an interesting target for the treatment of neuroinflammation, since 2-AG exhibits anti-inflammatory properties and AA is a precursor for pro-inflammatory prostaglandins. Astrocytes are an important source of AA and 2-AG, and highly express MGL. In the present study, we dissected the distinct contribution of MGL in astrocytes on brain 2-AG and AA metabolism by generating a mouse model with genetic deletion of MGL specifically in astrocytes (MKO GFAP ). MKO GFAP mice exhibit moderately increased 2-AG and reduced AA levels in brain. Minor accumulation of 2-AG in the brain of MKO GFAP mice does not cause cannabinoid receptor desensitization as previously observed in mice globally lacking MGL. Importantly, MKO GFAP mice exhibit reduced brain prostaglandin E2 and pro-inflammatory cytokine levels upon peripheral lipopolysaccharide (LPS) administration. These observations indicate that MGL-mediated degradation of 2-AG in astrocytes provides AA for prostaglandin synthesis promoting LPS-induced neuroinflammation. The beneficial effect of astrocytespecific MGL-deficiency is not fully abrogated by the inverse cannabinoid receptor 1 agonist SR141716 (Rimonabant) suggesting that the anti-inflammatory effects are rather caused by reduced prostaglandin synthesis than by activation of cannabinoid receptors. In conclusion, our data demonstrate that MGL in astrocytes is an important regulator of 2-AG levels, AA availability, and neuroinflammation.
Published data suggest that MGL has a dual function. First, it controls the levels of 2-AG, which exhibits neuroprotective effects by preventing excitotoxicity in neurons (8,9), but also by directly acting on microglia (10). Second, by degrading 2-AG, MGL provides AA for eicosanoid synthesis, a process which was originally thought to be exclusively dependent on phospholipase A2 (11). Accordingly, mice globally lacking MGL exhibit reduced brain AA and prostaglandin levels and are protected against LPS-induced neuroinflammation (12).
Astrocytes actively contribute to brain endocannabinoid signaling by secreting 2-AG (13) and by releasing gliotransmitters upon activation of CBRs (14). Furthermore, they are considered as key players in the synthesis and provision of polyunsaturated fatty acids in brain (15) and, together with microglia, they critically determine the inflammatory response of the CNS (16). Since MGL is highly expressed in astrocytes (13,17), we dissected the distinct contribution of MGL in astrocytes on brain 2-AG levels and AA metabolism. For this purpose, we generated a mouse model with specific genetic deletion of MGL in astrocytes and examined brain 2-AG and AA metabolism, behavior, and neuroinflammation. Our data demonstrate that astrocyte-specific deletion of MGL increases brain 2-AG levels without causing CBR desensitization or cannabimimetic behavioral effects. Yet, MGL-deficiency in astrocytes reduces the availability of AA for prostaglandin synthesis and thereby attenuates LPS-induced neuroinflammation.

Experimental Procedures
Animals-Mice were maintained on a regular dark light cycle (12 h light, 12 h dark) at 22 Ϯ 1°C in a specific pathogen free (SPF) environment and were and kept ad libitum on a standard laboratory chow diet (M-Z extrudate, V1126, ssniff). MKO GFAP mice were generated by targeted homologous recombination, similar as described for MKO global mice (7). In brief, a targeting vector harboring Mgll exons 3 and 4 flanked by a floxed neomycin resistance gene cassette and an additional loxP site was used for homologous recombination in murine embryonic stem cells. Embryonic stem cell clones with a manipulated Mgll locus were then transfected with a plasmid encoding Cre-recombinase to obtain embryonic stem cell clones with a floxed Mgll allele. Two independent clones were selected for microinjection into 3.5-day-old C57BL/6J blastocysts followed by implantation into recipient mice. Chimeric animals were bred with C57BL/6J mice to obtain Mgll floxed mice. Astrocyte specific deletion was achieved by breeding Mgll floxed mice with mice transgenic for Cre-recombinase under control of the human GFAP-promotor (FVB-Tg(GFAP-cre)25Mes/J, The Jackson Laboratory). Subsequently, female mice homozygous for the floxed Mgll allele and heterozygous for the Cre-recombinase were bred with male mice homozygous for the floxed Mgll allele. Mgll floxed littermates were used as wild-type controls.
To exclude effects of Cre-recombination per se (18) C57BL/6J mice carrying the Cre-recombinase under control of the human GFAP-promotor (WT GFAP ) (19,20) were used as controls. No differences between wild-type and WT GFAP mice were observed (data not shown). The study was approved by the Ethics committee of the University of Graz, the Austrian Federal Ministry of Science and Research, and is in accordance with the council of Europe Convention (ETS 123).
Primary Astrocyte Cultures-Brains of 1 day old pups were excised, meninges were removed, and brain tissue was washed in MEM alpha medium (Life Technologies, Thermo Scientific). Subsequently, brain tissue was mechanically triturated, filtered through a 70-m cell strainer and centrifuged at 1,000 ϫ g for 5 min. The cell pellet was resuspended in MEM␣ containing 10% FCS, 2 mM glutamine (Life Technologies, Thermo Scientific), 100 IU/ml penicillin/streptomycin (Gibco, Thermo Scientific), and 100 g/ml Primocin (InvivoGen, San Diego, CA). Cells were seeded in flasks coated with 15 g/ml poly-D-lysine (Merck KGaA) and cultivated under standard conditions at 37°C, 5% CO 2 , and 95% humidified atmosphere. Microglial cells were mechanically detached and removed via changing the medium every 3 days.
Primary Microglia Cultures-Microglia were obtained from mixed astrocyte/microglia cultures. Therefore, astrocytes were isolated and cultivated as described, but no detachment of microglia was performed. 10 days after astrocyte cultures reached confluency, microglia were harvested by mechanical detachment and centrifugation at 1,000 ϫ g for 5 min. Microglia were seeded in 15 g/ml poly-D-lysine (Merck KGaA)coated dishes and maintained at 37°C, 5% CO 2 , and 95% humidified atmosphere.
Primary Neuron Cultures-Cortical and hippocampal neurons were isolated from day 15.5 embryonic brains after removal of meninges via dissection in papain solution (Worthington). Neurons were seeded on cultivation dishes coated with 50 g/ml poly-D-lysine (Merck KGaA) at a density of 700 cells per mm 2 . Neurons were cultivated in Neurobasal Medium (Gibco, Thermo Scientific) containing 2% B-27 supplement (Gibco, Thermo Scientific), 100 IU/ml penicillin/streptomycin (Gibco, Thermo Scientific), 100 g/ml Primocin (Invivo-Gen), and 0.5 mM glutamine (Life Technologies, Thermo Scientific) and maintained at 37°C, 5.8% CO 2 , and 95% humidified atmosphere. Half of the nutrition medium was exchanged every 2 days and experiments were performed 7 days after neuron dissection.
Determination of in Vitro Monoglyceride Hydrolase (MGH) Activities-MGH activities of tissues and cells were determined as previously described (7). For MGL inhibition studies, samples were pretreated with either DMSO or 1 M JZL184 (Cayman Chemical, Ann Arbor, MI) for 10 min at 37°C.
Metabolic Phenotyping-For determination of food intake, locomotor activity, oxygen consumption, and respiratory quotient of mice, the Lab Master phenotyping platform (TSE Systems, Bad Homburg, Germany) was used. Therefore, 3-monthold mice were accustomed to Lab Master drinking bottles for at least 3 days in their home cages. Subsequently, mice were single housed and acclimatized to phenotyping cages for 3 days before measurement.
Behavioral Testing-The open field test was performed as described (21). Mice were individually placed in the center of the open field, and their behavior during a 5 min test period was tracked by a video camera and recorded with the VideoMot2 software (TSE Systems). For the tail suspension test, mice were suspended by their tail with strapping tape (Leukotape classic; BSN Medical S.A.S., Le Mans, France) to a lever for 6 min, and their behavior was recorded and analyzed via VideoMot2 software (TSE Systems). A trained blinded observer analyzed the video recordings with the event monitoring module for 3 types of behavior: swinging, curling, and immobility (22,23).
The social memory test was performed in a three-chambered apparatus (24,25). Mouse behavior was tracked with the Vid-eoMot2 software (TSE Systems) during a 5-min test period evaluating the time spent in the immediate vicinity (5 cm) of the grid enclosures and number of entries into each.
Acute thermal nociception was assessed with a Plantar Test apparatus (model 7370, Ugo Basile, Comerio, Italy) as described (26,27). The mean paw withdrawal latencies for both hind paws were calculated from the average of 3 separate trials, taken at 2-min intervals.
Spatial memory was analyzed using the Barnes maze (28). All sessions were recorded and analyzed with the VideoMot2 software (TSE Systems).

Immunohistochemistry (IHC) and Immunofluorescence
(IF)-IHC and IF were performed as previously described (31). In brief, mice were deeply anesthetized and transcardially perfused with PBS and 4% paraformaldehyde (PFA). Brains were excised, postfixed in 4% PFA overnight, saturated in 30% sucrose solution, and sectioned with a cryotome at 30 m. IHC staining of microglia was performed using anti-IBA1 antibody (Wako Chemicals, Neuss, Germany) and imaged using a Zeiss AXIO Scope A1 equipped with a Plan-APOCHROMAT 20ϫ/ 0.8 objective. Images were acquired at room temperature with the AxioRel Vision 4.8 software and quantified in ImageJ. For double IF goat anti-GFAP antibody (Sigma-Aldrich) and rabbit anti-MGL antibody (kind gift from Ken Mackie, Indiana University Bloomington) and anti-goat (Alexa Fluor 488-conjugated) and anti-rabbit (Alexa Fluor 594-conjugated) secondary antibodies were used. Sections were imaged by confocal laser scanning microscopy using a Leica TCS SP5 II equipped with a HCX PL APO 63 ϫ 1.3 NA objective. Images were obtained at room temperature with Leica LAS AF software.
Cytokine Determination-Tissues were excised, briefly washed in PBS, snap frozen in liquid N 2 within 30 s, and stored at Ϫ80°C. For cytokine measurements tissues were homogenized in 1 ml of PBS ϩ protease inhibitor (cOmplete Protease Inhibitor Mixture Tablets, Roche) and centrifuged at 1,000 ϫ g for 10 min at 4°C. The supernatant was transferred into a new tube and centrifuged at 100,000 ϫ g for 40 min at 4°C. For determination of cytokine levels the Custom Mix 'N Match Multi-Analyte ELISArray (Qiagen, Hilden, Germany) was used according to the manufacturer's protocol. In brief, ELISA wells were incubated with antigen standards or 100 g sample for 2 h at room temperature, washed, and incubated with respective Biotin labeled detection antibodies for 1 h at room temperature. After washing wells were incubated for 30 min with Avidin-HRP at room temperature. Subsequently, wells were incubated with development solution for 15 min at room temperature followed by addition of stop solution and measurement of absorbance at 450 nm. For quantification readings at 567 nm were subtracted.
Statistical Analysis-Values are represented as means Ϯ S.D. Statistical significance was determined by Student's unpaired t test (two-tailed) for the comparison of genotypes. For analysis of multiple measurements, analysis of variance followed by Bonferroni post hoc test was used. Group differences were considered statistically significant for genotypes: *, p Ͻ 0.05; **, p Ͻ 0.01; and ***, p Ͻ 0.001; for treatments: #, p Ͻ 0.05; ##, p Ͻ 0.01; and ###, p Ͻ 0.001.

Results
Generation and Gross Characterization of Mice with Astrocyte-specific Deletion of MGL-Astrocyte-specific MGL knockout mice (MKO GFAP ) were generated by crossing MGL-floxed animals to mice expressing Cre-recombinase under control of the hGFAP-promotor. Primary astrocytes from MKO GFAP mice exhibited markedly reduced MGL protein expression, whereas a minor reduction in MGL expression was observed in whole brain lysates (Fig. 1A). Brain lysates of MKO global mice showed no MGL expression confirming the specificity of the antibody (Fig. 1A). In contrast to astrocytes, MGL protein expression was preserved in primary neurons from cortex and hippocampus. In accordance with published data, primary microglia exhibited low MGL expression (32), which was barely detectable in cultures obtained from control and MKO GFAP mice (Fig. 1B). To confirm astrocyte-specificity of MGL deletion, we performed double immunofluorescence analysis of MGL and the astrocyte marker GFAP. In cortical brain sections of control mice, we observed a co-localization of MGL and GFAP-positive cells (Fig. 1C). Although MGL expression was detected in sections of MKO GFAP mice, we could not observe any co-localization of MGL and GFAP (Fig. 1C). Together, these observations implicate that MKO GFAP mice lack MGL specifically in astrocytes. Gross phenotypic characterization revealed that MKO GFAP mice exhibit no alterations in bodyweight, body composition, food intake, locomotor activity, oxygen consumption, and respiratory quotient (RQ) (Fig. 2).

Astrocyte-specific Deletion of MGL Reduces Total Brain Monoglyceride Hydrolase Activity and Increases 2-AG Levels-
Next, we determined the impact of MGL deletion on in vitro MGH activity in isolated primary astrocytes. In lysates of MKO GFAP astrocytes, we found 65% reduction in MGH activity as compared with controls. Addition of the MGL inhibitor JZL184 (1 M) reduced MGH activity of control and MKO GFAP astrocytes to 20% of total activity observed in controls (Fig. 3A). The reduced MGH activity led to a 3-fold increase in 2-AG levels in the medium of MKO GFAP astrocytes (Fig. 3B). To assess the contribution of astrocyte MGL to total brain MGH activity, we compared activities detected in brain lysates of control and MKO GFAP mice. As shown in Fig. 3C, MKO GFAP mice exhibited 35% reduction in total brain MGH activity. Addition of the MGL inhibitor JZL184 (1 M) reduced MGH activity in MKO GFAP and control lysates to 15% of total activity observed in controls (Fig. 3C). Reduced MGH activity led to a 1.8-fold increase in total brain 2-AG levels as compared with controls (Fig. 3D). In accordance with the tissue specificity of the hGFAP promotor (19), no differences in MGH activities were observed in lysates from liver, gonadal white adipose tissue, brown adipose tissue, small intestine, and cardiac muscle (Fig. 3E).

Global but Not Astrocyte-specific MGL Deletion Leads to CBR
Desensitization-As shown in previous studies, global genetic deletion or chronic pharmacological inhibition of MGL results in CBR desensitization (5,6). We and others previously demonstrated that MKO global mice are resistant against the hypometabolic effects of the CBR agonist CP55940 due to 2-AG dependent CBR desensitization (6,7). Thus, we investigated whether moderately elevated 2-AG levels in the brain of MKO GFAP mice affect CBR sensitivity. Administration of CP55940 (0.15 mg/kg, intraperitoneal) led to a similar reduction of food intake (Fig. 4A), locomotor activity (Fig. 4B), and RQ (Fig. 4C) in control and MKO GFAP mice, indicating that CBR sensitivity in MKO GFAP mice is not affected. As a control, we equally treated MKO global mice, and found that these mice exhibit increased food intake (Fig. 4D), locomotor activity (Fig.  4E), and RQ (Fig. 4F) in comparison to wild-type controls. These observations confirm that global MGL deficiency leads to functional antagonism of the endocannabinoid system and indicate that the moderate increase in 2-AG levels in the brain of MKO G -FAP mice is not sufficient to cause CBR desensitization.
Astrocyte-specific MGL Deletion Does Not Cause Cannabimimetic Behavioral Effects-To investigate whether increased 2-AG levels cause cannabimimetic effects, MKO GFAP mice were subjected to several behavioral tests. Yet, we could not observe any differences in anxiety-like behavior in the open field test (Fig. 5A), depression-like behavior in the tail suspension test (Fig. 5B), social interaction (Fig. 5C), spatial memory in the Barns maze (Fig. 5D), or thermal nociception (Fig. 5E). These data implicate that deletion of MGL in astrocytes is not sufficient to cause major behavioral changes as observed in response to cannabinoid treatment (33).
Astrocyte-specific MGL Deletion Attenuates LPS Induced Neuroinflammation-Published data suggest that global deletion of MGL in mice has beneficial effects on neuroinflammation by reducing the amount of AA available for prostaglandin synthesis (12). Likewise, 2-AG has been shown to have neuroprotective properties per se (9). Since astrocytes play a major role in neuroinflammation, we hypothesized that astrocyte-specific deletion of MGL has similar effects on neuroinflammation as observed in MKO global mice. To induce systemic inflammation, mice were treated with LPS (5 mg/kg, intraperitoneal). Subsequently, we monitored LPS induced sickness behavior using the RQ as sickness index, which is reduced in response to LPS treatment (34). As shown in Fig. 6A, control mice showed a pronounced reduction in RQ upon LPS administration and this effect was attenuated in MKO GFAP mice. We confirmed the beneficial effect of global MGL deficiency in LPS induced sickness in MKO global mice (Fig. 6B). Next, we investigated brain 2-AG, AA, PGE2, and cytokine levels. LPS treated MKO GFAP mice exhibited a ϳ2-fold increase in brain 2-AG (Fig. 6C) similar as observed under basal conditions (Fig. 3D). Brain AA levels in vehicle treated MKO GFAP mice were reduced by ϳ20% in comparison to wild-type controls and further decreased slightly in response to LPS treatment in both control and MKO GFAP mice (Fig. 6D). Notably, LPS substantially increased brain PGE2 levels in control but not in MKO GFAP mice (Fig. 6E). This observation implicates that MGL activity in astrocytes strongly determines the availability of AA for prostaglandin synthesis upon inflammatory stimuli. LPS treatment leads to activation of brain microglia, which are important in the inflammatory response of the CNS (35). To test the role of astrocyte MGL in microglia activation, we treated mice with 0.5 mg/kg LPS on 4 consecutive days to induce microglia activation as previously described for MKO global mice (12). Microglia morphology and density was determined via IHC of cortical brain sections using the microglia marker IBA1. As shown in representative images, microglia morphology and density was similar in vehicle- treated control (Fig. 7A) and MKO GFAP (Fig. 7C) mice. Microglia were similarly activated in both genotypes after LPS treatment (Fig. 7, B and D). Accordingly, quantification of IBA1 positive area revealed unchanged microglia density in both genotypes (Fig. 7E).
Next, we assessed whether reduced PGE2 is associated with altered cytokine production. LPS treatment induced a time-dependent increase in cytokines TNF␣ (Fig. 8A), IL-1␣ (Fig. 8B), and IL-6 ( Fig. 8C) in the brain of control mice and this effect was blunted in MKO GFAP mice. Finally, we investigated whether the attenuated neuroinflammatory response in MKO GFAP mice is dependent on CBR1 activation. We therefore treated mice with the inverse CBR1 agonist SR141716 (Rimonabant; 1 mg/kg, intraperitoneal) (36) 1 h before LPS administration. As shown in Fig. 8, A-C, TNF␣, IL-1␣, and IL-6 levels remained reduced in MKO GFAP mice treated with SR141716 indicating that the reduction in neuroinflammation is rather independent of CBR1 activation. Taken together, our observations demonstrate that MGL in astrocytes is crucial for providing AA for prostaglandin synthesis upon LPS-induced neuroinflammation.

Discussion
MGL is required for efficient degradation of 2-AG in brain and thus inversely regulates 2-AG levels and AA availability (1).  This distinct metabolic function makes MGL an interesting target for the treatment of neuroinflammation, since 2-AG exhibits anti-inflammatory properties and AA is a precursor of proinflammatory prostaglandins (37). Accordingly, it has been shown that pharmacological or global genetic inactivation of MGL has beneficial effects in animal models of neuroinflammatory and neurodegenerative diseases (12,38).
In the present study, we generated mice with genetic deletion of MGL specifically in astrocytes and investigated the impact on brain 2-AG and AA metabolism, behavior, and inflammation. MKO GFAP mice exhibit only minor reduction in total brain MGH activities and a moderate increase of brain 2-AG levels as compared with MKO global mice showing more than 10-fold increased brain 2-AG levels (5-7). The strongly elevated 2-AG levels in MKO global mice cause tolerance to CBR agonists, which is also observed upon chronic pharmacological MGL inhibition or prolonged exposure to CBR agonists (6). The observed tolerance is caused by CBR desensitization in the brain and also in peripheral tissues such as the intestine (39). We previously demonstrated that measurement of physiological parameters in metabolic cages upon CBR agonist treatment can be used for determination of CBR desensitization in vivo (7). In contrast to MKO global mice, MKO GFAP mice respond to CBR agonist treatment similar as wild-type controls suggesting that the moderate increase in 2-AG in MKO GFAP mice is not sufficient to cause CBR desensitization. Thus, 2-AG accumulation could affect animal behavior, since the endocannabinoid system plays a key role in learning, memory, anxiety, depression, and pain (33, 40 -42). Additionally, CBR dependent signaling between astrocytes and neurons is involved in the regulation of synaptic plasticity and memory formation (43,44). Therefore, we performed a variety of standard tests for social and spatial memory, anxiety and depression like behavior, and thermal nociception. However, we could not find any differences between control and MKO GFAP mice. We conclude that the moderate increase of 2-AG in MKO GFAP mice is not sufficient to elicit cannabimimetic behavioral effects.
Astrocytes are considered as major source of AA in brain (15,45) and published data implicate that rather MGL than cytosolic phospholipase A2 controls the availability of AA for prostaglandin synthesis in the brain (12). Accordingly, we hypothesized that MGL in astrocytes may be crucial in providing AA for prostaglandin production and that MGL deletion has beneficial effects on neuroinflammation. Our observations demonstrate that deletion of MGL in astrocytes is sufficient to improve sickness behavior of mice in response to LPS treatment similar as observed for MKO global mice. To reveal the underlying mechanisms, we measured 2-AG and AA levels. We found that 2-AG is moderately increased in the brain of MKO GFAP mice under basal and inflammatory conditions. In accordance with previous studies, peripheral LPS administration altered brain 2-AG levels neither in wild-type (12) nor in MGL GFAP mice. Free AA levels were moderately decreased in MKO GFAP mice in comparison to wild-type controls and further decreased slightly in both genotypes upon LPS treatment. Importantly, MGL GFAP mice completely lacked the LPS-induced increase in PGE2 synthesis and exhibited reduced levels of pro-inflammatory cytokines. Dysregulated PGE2 synthesis is associated with the onset and progression of neuroinflammatory and neurodegenerative diseases (46). Our data indicate that MGL in astrocytes is crucial for providing AA as substrate for PGE2 synthesis. These observations support the recently discovered role of MGL as interesting pharmacological target for treatment of neuroinflammation, providing a novel strategy for reduction of prostaglandin synthesis independent of COX inhibition (12). Several studies demonstrated the beneficial effects of COX1 inhibition in inflammatory processes of the CNS (47)(48)(49). Both COX1 and MGL inhibitors exhibit anti-inflammatory and analgesic properties (37,50). Use of MGL inhibitors, however, can avoid gastrointestinal side effect of non steroidal anti-inflammatory drugs (NSAIDs) and even improve GI bleeding induced by NSAIDs (51).
Both astrocytes and microglia are thought to contribute to the onset and progression of neuroinflammation by producing pro-inflammatory mediators, whereby prostaglandins predominantly derive from activated microglia (52). Notably, microglia activation is unchanged in control and MKO GFAP mice after LPS treatment. Very recent data support our observations by demonstrating that the endocannabinoid dependent metabolic interplay between neurons and astrocytes is involved in the regulation of inflammatory processes of the CNS. Using an inducible knock-out system, Viader et al. demonstrate that rather MGL in astrocytes than in microglia or neurons affects neuroinflammation in response to single and repeated LPS treatment (53).
To investigate whether increased 2-AG levels in MKO GFAP mice contribute to the anti-inflammatory effects, we treated mice with the inverse CBR1 agonist SR141716. Interestingly, pro-inflammatory cytokine levels remained reduced in LPStreated MKO GFAP mice, after SR141716 treatment. This suggests that the reduced mobilization of AA for prostaglandin synthesis rather than the accumulation of 2-AG is responsible for the beneficial effects on neuroinflammation. Accordingly, studies in CBR1/CBR2 double knock-out mice implicate that neuroprotective effects of MGL inhibition are independent of CBR (12). Nevertheless, another study demonstrates that CBR1 mediates the hypothermic effects of LPS in rats (54). Thus, the activation of CBRs and the reduced availability of AA might regulate different aspects of LPS-induced inflammation.
In summary, we demonstrate that MGL in astrocytes affects brain 2-AG hydrolysis, AA availability, and prostaglandin synthesis. Astrocyte-specific deletion of MGL is not sufficient to cause cannabimimetic behavioral effects or CBR desensitization, but reproduces the neuroprotective effects of global MGL deficiency.