Astroglial CB₁ cannabinoid receptors regulate leptin signaling in mouse brain astrocytes

Abstract

Type-1 cannabinoid (CB₁) and leptin (ObR) receptors regulate metabolic and astroglial functions, but the potential links between the two systems in astrocytes were not investigated so far. Genetic and pharmacological manipulations of CB₁ receptor expression and activity in cultured cortical and hypothalamic astrocytes demonstrated that cannabinoid signaling controls the levels of ObR expression. Lack of CB₁ receptors also markedly impaired leptin-mediated activation of signal transducers and activators of transcription 3 and 5 (STAT3 and STAT5) in astrocytes. In particular, CB₁ deletion determined a basal overactivation of STAT5, thereby leading to the downregulation of ObR expression, and leptin failed to regulate STAT5-dependent glycogen storage in the absence of CB₁ receptors. These results show that CB₁ receptors directly interfere with leptin signaling and its ability to regulate glycogen storage, thereby representing a novel mechanism linking endocannabinoid and leptin signaling in the regulation of brain energy storage and neuronal functions.

1. Introduction

The type-1 cannabinoid (CB₁) receptor is highly expressed throughout the brain where it mediates the effects of (endo)cannabinoids on neuronal transmission, plasticity and functions [1,2]. Beside the predominant neuronal location, low but functionally relevant levels of CB₁ receptors are present in astrocytes [3]. Early investigations suggested a role for CB₁ receptors located on astrocytes in the release of inflammatory mediators [4,5] and in the control of cellular metabolism [6]. However, more recent studies have demonstrated that, by acting at astroglial CB₁ receptors, endocannabinoids can modulate neuron–astrocytes communication and astrocytic glutamate signaling [7,8]. Furthermore, astroglial CB₁ receptor has been recently found to modulate synaptic spike timing-dependent long-term depression of transmission (t-LTD) in the neocortex [9], and to mediate cannabinoid-induced long-term depression (CB-LTD) and impairment of working memory in the hippocampus [10].

Due to their pivotal location between blood vessel and neurons, astrocytes are primarily involved in supplying energy to neurons [11], but also in sensing peripheral energy state [12]. Indeed, astrocytes express different metabolism-related receptors and mediators [13–16]. Consistent with a metabolic sensor role, diet- or genetic-induced obesity is accompanied with an increase in glial fibrillary acidic protein (GFAP) in the hypothalamus [17,18]. Interestingly, the enhanced astrocytic activity induced by high-fat diet was recently shown to be under the control of the endocannabinoid system [19].

Endocannabinoids are known to affect energy balance and metabolism at multiple levels by interacting with the actions of neurotransmitter systems, neuropeptides and hormones [20,21]. Among these, leptin is a major anorectic adipokine, which plays a key role in the regulation of energy balance in the central nervous system (CNS) [22,23]. Whereas most of the investigations studying leptin CNS functions have focused on its actions at neuronal level, astrocytes from different brain structures express the active isoform of the leptin receptor (ObRb) [17,24–26]. Strong relationships between leptin and the endocannabinoid signaling have been evidenced in the control of metabolic processes, which have been so far studied in the context of the functions of CB₁ receptors in neurons [21,27,28]. However, the possible direct role of astroglial CB₁ receptors on leptin-dependent signaling has not been investigated so far.

In the present study we addressed the interactions between CB₁ receptor and leptin signaling systems at astroglial level. In particular, we investigated whether astroglial CB₁ receptors could directly modulate leptin signaling in primary astrocyte cultures and found that astroglial CB₁ receptors participate in the control of astroglial energy storage functions through the maintenance of functional leptin signaling.

2. Methods

2.1. Drugs

The CB₁ receptor antagonist SR141716A was obtained from NIMH Chemical Synthesis and Drug Supply Program. The dual MGL and FAAH inhibitor JZL195 was synthesized as previously described [29] and generously provided by Dr. B. Cravatt, The Scripps Research Institute (La Jolla, California, USA). Mouse recombinant leptin was obtained from Dr. A. Parlow, National Hormone and Peptides Program (Torrance, California, USA). The STAT5 inhibitor sc-355979 was purchased from Santa Cruz Biotechnology (Heidelberg, Germany).

2.2. Animals

All experiments were conducted in strict compliance with the European Union recommendations (2010/63/EU) and were approved by the French Ministry of Agriculture and Fisheries (authorization number 3306369) and the local ethical committee. Constitutive and conditional CB₁-mutant mice and their wild-type littermates were in a mixed genetic background, with a predominant C57BL/6NCrl contribution (6–7 backcrossing generations). Previous extensive anatomical characterizations showed that the mutant mice used in the present study carry deletions of CB₁ receptors from all the cells of the body (CB₁-KO mice; [30]). GFAP-CB₁-KO mice were generated using the Cre/loxP system as previously described [10]. Mice carrying the “floxed” CB₁R gene (CB₁^f/f) [31] were crossed with GFAP-CreERT2 mice [32], using a three-step backcrossing procedure to obtain CB₁R^{f/f;GFAP-CreERT2} and CB₁R^f/f littermates, called GFAP-CB₁-KO and GFAP-CB₁-WT, respectively. As CreERT2 protein is inactive in the absence of tamoxifen treatment [32], deletion of the CB₁R gene was obtained in adult mice (8 weeks-old) by daily i.p. injections of tamoxifen (1 mg dissolved at 10 mg/ml in 90% sesame oil, 10% ethanol, Sigma-Aldrich, St Quentin, France) for 8 days. Mice were used 4 weeks after tamoxifen treatment. Given the influence of the female hormonal cycle on cannabinoid system, only male WT, CB₁-KO and GFAP-CB₁-KO were used for the electron microscopy and the immunohistochemistry experiments.

2.3. CB₁-GFAP double immunocytochemistry for electron microscopy

GFAP-CB₁-KO, GFAP-CB₁-WT and CB₁-KO mice were deeply anesthetized (ketamine/xilazine 80/10 mg/kg, i.p.). and transcardially perfused with phosphate buffer saline (0.1 M PBS, pH 7.4) and then with the fixative solution [either 4% paraformaldehyde (PFA), 0.1% glutaraldehyde and 0.2% saturated picric acid in phosphate buffer (0.1 M PB, pH 7.4) or Zamboni's fixative solution made up of 2% PFA and 15% saturated picric acid [33] in 0.1 M PB]. Coronal sections (50 µm) collected in 0.1 M PB were prepared for double pre-embedding staining of CB₁ receptor and GFAP with silver-intensified immunogold method and immunoperoxidase method as previously described [10]. Figure compositions were scanned at 500 dots per inch (dpi). Labeling and minor adjustments in contrast and brightness were made using Adobe Photoshop (CS, Adobe Systems, San Jose, CA, USA).

2.4. Primary astrocyte cultures

Primary cultures of astrocytes were prepared at postnatal day 2 or 3, from neonatal WT or full CB₁-KO mice [30] of both sexes obtained from first generation homozygote breeding. The genotype of the parent mice was checked before and after several rounds of breeding. Cortex and hypothalamus were carefully dissected in PBS supplemented with 0.6% glucose, pooled (3 animals/structure), and rinsed in HBSS solution containing 0.01 M Hepes, 100 U/ml penicillin and 100 µg/ml streptomycin. After enzymatic and mechanic dissociations, cells were grown in Dulbecco's modified Eagle's medium containing 10% heat inactivated fetal bovine serum, 0.6% glucose, 100 U/ml penicillin, 100 µg/ml streptomycin, 2.5 µg/ml fungizone, 2 mM Glutamine and 1 mM sodium pyruvate. After 10–12 days, the cultures were shaken on an orbital shaker for 3 h at 37 °C to remove debris. Three days later, cells were subcultured and grown for another 5–7 days before use. All cell culture media and reagents were obtained from Invitrogen (Saint Aubin, France). Cells were cultured at 37 °C in an atmosphere of humidified air and 5% CO₂. All drug treatments were conducted in the culture medium, in similar conditions. SR141716A was used for 24 h (1 µM or ranging concentrations from 1 nM to 10 µM for dose response curves) or for 6 h when used in combination with JZL195 (0.1 µM). Leptin was used at 0.3 or 1 µg/ml, and sc-355979 was used at 500 µM. A preliminary characterization of our cultures indicated a purity of about 97% as 3.04±1.44% of the cells were labeled with CD11b, a widely used microglial marker. Neither neuronal nor olygodendrocyte labeling was found.

2.5. Transfection of astrocytes

Primary cortical astrocytes grown in 24-well plates on poly-l-lysine-coated 12 mm round glass coverslips were transfected with a vector expressing either the fluorescent protein mCherry alone (mCherry) or the CB₁ receptor and the mCherry protein (mCherry-CB₁) using a lipofection method. Cells were transfected by adding 0.1 ml of the plasmid mixture [1.5 μl of Lipofectamin LTX (Invitrogen), 0.5 μg of plasmid, 0.5 μg of PLUS reagent (Invitrogen) in Opti-MEM medium, incubated for 30 min at room temperature (RT)] to 0.5 ml of culture medium exempted of antibiotics and containing 5% FBS. After 18 h, the incubation medium was replaced by normal growth medium. Seventy two hours after the transfection, immunocytochemistry was performed.

2.6. Immunocytochemistry

Cells grown on poly-l-lysine-coated 12 mm round glass coverslips were rinsed 3× with PBS, fixed for 30 min with 4% PFA, permeabilised with 0.05% saponin in PBS for 30 min, and incubated for an additional 30 min (0.05% saponin, 3% BSA in PBS). Cells were then incubated for 18 h at 4 °C with primary chicken anti long-isoform leptin receptor (ObRb) antibody (CD295 antibody; Gentaur, Paris, France; 1:300, 0.05% saponin, 0.3% BSA in PBS). For phosphorylated STAT3 (P-STAT3) or phosphorylated STAT5 (P-STAT5) immunostainings, fixed cells were permeabilised with methanol (−20 °C, 20 min) and incubated 1 h in 3% BSA and 0.3% Triton X-100 PBS. Rabbit anti mouse P-STAT3 (Tyr705) and anti mouse P-STAT5 (Tyr694) (both at 1/500, Cell Signalling, Beverly, MA, USA) were diluted in 0.3% BSA and 0.03% Triton X-100 PBS, and incubated 18 h at 4 °C. After 3 rinses (0.01% Tween-20 PBS), cells were incubated 1 h at RT with the secondary antibody Alexa 488-conjugated goat anti rabbit (1:1000, Invitrogen) and anti GFAP antibody (Cy3^™ conjugated, Sigma-Aldrich). After rinsing, cells were stained for 10 min with DAPI (1:10,000) rinsed again in PBS, then in deionised water and mounted with Fluoromount-G. Acquisitions were collected using the same exposition settings. When the primary antibody was omitted, no signal was detected. After correcting for background intensity, images were processed with Metamorph software (Version 717) by measuring the fluorescence intensities of the cells (ObRb) or the nuclei (P-STAT3 and 5) as determined by the DAPI staining.

2.7. Immunohistochemistry

GFAP-CB₁-WT and GFAP-CB1-KO mice were deeply anesthetized (Nembutal 50 mg/kg, i.p.) and transcardially perfused with PBS solution. Dissected brains were stored at −80 °C and coronal cryosections (30 µm) were collected, fixed for 1 h in 4% PFA and permeabilized with methanol 10 min at −20 °C. After 1 h blocking (3% donkey serum and 0.3% Triton X-100 in 0.1 M PB), sections were incubated with rabbit anti P-STAT5 antibody (1/300 in, 1% BSA and 0.3% Triton X-100 in 0.1 M PB, 48 h, 4 °C). After washes (0.1 M PB), sections were incubated for 2 h at RT with Alexa 488-conjugated goat anti rabbit antibody and Cy3^™ conjugated-GFAP antibody (both 1:1000, 1% BSA and 0.3% Triton X-100 in 0.1 M PB). After 3 rinses in PB, sections were stained with DAPI (10 min, 1:10,000 in PBS) rinsed 3 times more in PBS, then in deionised water and mounted with Fluoprep (BioMerieux Benelux, Bruxelles, Belgium). Acquisitions were collected using the same exposition settings. When the primary antibody was omitted, no signal was detected. After correction for background, total number of nuclei was counted using the DAPI staining. The number of P-STAT5 positive nuclei was counted amongst all the cells using thresholded Alexa-488 fluorescence pictures.

2.8. Quantitative real time PCR (Q-PCR)

Total RNA extractions were performed using the Trizol reagent (Invitrogen) according to the manufacturer's instruction. Genomic DNA contaminations were removed using the turboDNA free kit (Ambion, Saint Aubin, France). Only samples displaying RIN above 7 as determined using the RNA 6000 Nano Labchip kit and the Bioanalyser 2100 (Agilent Technologies, Massy, France) were used for following analysis. cDNA was synthesized from 2 µg of total RNA with Revert Aid^™ Premium Reverse Transcriptase (Fermentas, St. Leon-Rot, Germany) and random primers (Fermentas). PCR amplification was performed in a 10 µl reaction volume containing 4 ng cDNA, 600 pM primers and the DyNAmo^TM SYBR Green qPCR kit (Finnzymes, Fisher Scientific, Illkirch, France) using a DNA Engine Opticon2 fluorescence detection system (MJResearch/Bio-Rad, Marnes La Coquette, France) with following cycles (95 °C for 15 min followed by 40 cycles with 95 °C, 20 s and 61 °C for 35 s). Primer sequences are reported in Supplementary Table 1. Absence of genomic DNA contamination was confirmed by using RNA samples that were not reverse transcribed. Ct values for the gene of interest were normalized against that of ubiquitin C. The relative levels of expression were calculated using the comparative (2^−ΔΔCT) method and controls were arbitrarily set at 1.

2.9. Western blotting analysis

Cell lysates [in Tris-Base solution (50 mM, pH 6.8), with complete mini protease inhibitor (Roche Diagnostics, Meylan, France) inhibitor cocktail I and II (Sigma-Aldrich), 0.5% 2-mercaptoethanol and 1% Triton X-100] were resuspended in 1× Roti-Load 1 (Carl Roth, Lauterbourg, France), boiled for 5 min, separated on 10% SDS-polyacrylamide gels and electro-transferred to PVDF membranes. Membranes were blocked for 2 h (5% BSA,0.05% Tween-20 in TBS) and probed overnight at 4 °C with rabbit primary antibodies anti P-STAT3 (1:1500) or anti P-STAT5 (1:2500) diluted in blocking buffer. After several washes (0.05% Tween-20 in TBS) membranes were incubated with rabbit anti-goat HRP-conjugated secondary antibodies (Invitrogen) 1 h at RT. Immunoreactivity was detected by ECL Plus detection kit (GE Healthcare Life Sciences, Orsay, France). Membranes were stripped (100 mM 2-mercaptoethanol, 2% SDS, 62.5 mM Tris–HCl, pH 6.7, 60 °C, 30 min) and reprobed with anti-total STAT3 or anti-total STAT5 antibodies (1:3000 and 1:2000 respectively, Cell Signalling) that recognized both phosphorylated and unphosphorylated forms of STAT3 and STAT5, respectively. Densitometry analysis was done using GS-800 densitometer (Bio-Rad). Results were expressed as phospho-protein density normalized against the density of the corresponding total protein.

2.10. Glycogen content

Culture medium was replaced by medium exempted of phenol red 24 h before the experiment. Cells were then exposed for 2 h (cortical astrocytes) or 30 min (hypothalamic astrocytes) to indicated treatments; rinsed 3× with PBS and scrapped in distilled water. Glycogen content was analyzed using glycogen assay kit (Biovision, Lyon, France) according to manufacturer's instructions. Results are expressed as the glycogen (µg) normalized to the protein content (mg) determined in each well (6 well culture plates).

2.11. Statistical analysis

All the results are expressed as mean±SEM of indicated replicated number of experiments. Statistical analysis and dose–response curves were generated using GraphPad Prism 5.03 (GraphPad Software Inc., San Diego, CA, USA). χ² test was used to see differences in the percentage of CB₁ receptor expressing astrocytes in the different WT and mutant mice. Student's t-tests were used for the comparison between two groups. When comparing more than two groups, one-way ANOVA followed by Dunnet's or Tukey's post-hoc test (as appropriate for comparisons relative to control or between all groups) was used. Two-way ANOVA followed by Bonferroni's post-hoc test was used to compare results from experiments with two independent variables. p values lower than 0.05 denote statistical significance.

3. Results

3.1. Astrocytes of neocortex and hypothalamus contain CB₁ receptors

Functional and anatomical investigations using CB₁^{f/f;GFAP-CreERT2} mice (hereafter called GFAP-CB₁-KO; [10]), carrying specific deletion of CB₁ gene in GFAP positive cells, were recently carried out to demonstrate the presence of CB₁ receptor protein in astrocytes from the CA1 region of the hippocampus [10]. We tested if this holds true also in the cerebral cortex (Cx) and the ventromedial hypothalamus (VMH). Electron microscopic detection of both GFAP and CB₁ receptor protein in Cx and VMH revealed that the proportion of astrocytes expressing CB₁ receptors was as previously described in the hippocampus [10], with 55.1±4.4% and 40.1±5.0% of GFAP-positive elements containing CB₁ immunogold labeling in the Cx and the VMH, respectively (Figure 1A and B). These percentages dropped to 9.2±2.6% and 12.1±4.6% in the Cx and VMH of conditional GFAP-CB₁-KO mutant mice (Figure 1C and D), and to the respective background levels of 2.3±1.0% and 5.7±1.7% in the same regions of constitutive CB₁-KO mice [30] (Figure 1E and F), confirming the specificity of the astroglial CB₁ detection in these two regions. Accordingly, CB₁ receptor mRNA was detected in primary astrocytes extracted from both cortex and hypothalamus of WT mice but not in cultures derived from constitutive CB₁-KO mice (Supplementary Figure 1).

Figure 1.

Electron microscopic detection of CB₁ receptors on GFAP positive cells in the cerebral cortex and the hypothalamus. Electron micrographs showing immunogold labeling of CB₁ receptors (arrows) in the Cx (A, C and E) and the VMH (B, D and F) of GFAP-CB₁-WT (A and B), GFAP-CB₁-KO (C and D), and CB₁-KO (E and F) mice. CB₁ immunogold labeling is visible on both neuronal terminals (*) and GFAP positives cells (O) in the GFAP-CB₁-WT animals while in the GFAP-CB₁-KO mice it is only detected on neuronal terminals. In full CB₁-KO animals no immunogold labeling is observed. Scale bar: 500 nm.

3.2. Expression of leptin receptors is controlled by CB₁ receptor-dependent mechanism in cultured astrocytes

Astrocytes where recently shown to express significant levels of leptin receptors (ObR) [24–26]. Immunocytochemistry using an antiserum specifically targeting the active isoform of leptin receptor (ObRb) revealed that cultured astrocytes from cortex (Figure 2A and B) express ObRb receptors. Interestingly, however, in cortical cultures prepared from constitutive CB₁-KO animals, the level of ObRb were clearly reduced both at protein (Figure 2A and B) and mRNA level (Figure 2C). To verify that the decreased levels of expression of ObR were specifically due to blockade of CB₁ receptor signaling in astrocytes and not to secondary effects of genetic CB₁ deletion, we treated WT and constitutive CB₁-KO cortical astrocyte cultures with the CB₁ receptor antagonist SR141716A for 24 h. In WT astrocytes, the treatment dose-dependently reduced ObR mRNA expression, which reached the same levels as in constitutive CB₁-KO cultures starting from the concentration of 100 nM (Figure 2D). Conversely, SR141716A had no effect on constitutive CB₁-KO primary astrocytes, confirming the specificity of the drug (Figure 2D). These results suggest that CB₁ receptor signaling regulates ObR expression in astrocytes.

Figure 2.

CB₁ receptor-dependent mechanism control leptin receptor expression in astrocytes. (A) Immunolabeling for GFAP (red) and functional isoform of leptin receptor (ObRb) (green) in primary cultures of cortical astrocytes from WT and CB₁-KO mice. (B) Quantification of the A-488 fluorescence intensity (ObRb labeling) in WT and CB₁-KO astrocytes. Results are expressed as the relative intensity versus intensity in WT astrocytes. Data shown are means with SEM values of at least 3 different experiments. (C–F) Quantitative PCR of ObR mRNA performed on total RNA extracts from primary astrocyte cultures prepared from the cortex (C–E) and the hypothalamus (F) of WT and CB₁-KO mice. Cortical (D) and hypothalamic (F) astrocyte cultures were exposed for 24 h to decreasing or 1 µM concentrations of SR141716A, respectively. (D) The expression of ObR mRNA was normalized against ubiquitin expression and results are expressed as the relative expression versus controls (WT, vehicle treated cultures). Cells used for ObR mRNA analysis were treated for 6 h with JZL195 (0.1 µM) in the absence or in the presence of 1 µM SR141716A. Data are means with SEM values of at least 4 experiments performed in duplicate. ***p<0.001, **p<0.01, relative to WT control cells. ### p<0.001, genotype effect.

The cellular levels of the two main endocannabinoids 2-arachidonoylglycerol and anandamide are under the control of the specific degrading enzymes monoacylglycerol lipase (MGL) and fatty acid amide hydrolase (FAAH), respectively [1,34]. To test whether the endocannabinoid signaling is involved in the regulation of astrocyte ObR expression, we applied the dual MGL and FAAH inhibitor JZL195 [29] to primary astrocyte cultures. JZL195 increased ObR mRNA expression in WT, but not in constitutive CB₁-KO primary astrocytes. This effect was also fully blocked by SR141716A (Figure 2E).

Considering the important influence of leptin on hypothalamic functions [22,35], ObR expression was investigated on astrocyte cultures prepared from the hypothalamus as well. Using these cultures, a similar decrease in ObR mRNA expression was found after treatment with SR141716A or in hypothalamic astrocytes coming from constitutive CB₁-KO mice (Figure 2F), indicating that similar interactions between CB₁ receptors and ObR are shared among anatomically distinct astrocyte subpopulations.

Although the data obtained with the CB₁ receptor antagonist SR141716A argue against this possibility, the suppression of ObR expression in astrocytes derived from constitutive CB₁-KO mice might result from an adaptation mechanism due to neuronal CB₁ receptor deletion. Thus, we directly tested if CB₁ receptor expression in astrocytes is sufficient to regulate the levels of ObRb expression, by re-expressing the receptor in primary cortical astrocytes derived from constitutive CB₁-KO mice. This manipulation fully rescued the levels of ObRb protein expression (Figure 3A and B). In addition, the co-expression of the mCherry tracer protein together with the CB₁ receptor indicated that ObR expression is restored specifically in transfected cells, but not in adjacent untransfected astrocytes (Figure 3C). Altogether, these data show that astroglial expression of ObR is under the tight control of astroglial CB₁ receptors, which play a necessary and sufficient role to guarantee physiological levels of leptin receptors.

Figure 3.

ObR expression in CB₁-KO astrocytes is rescued by re-expression of CB₁ receptors. (A) Immunolabeling for the functional isoform of leptin receptor (ObRb) (green) in primary cultures of cortical astrocytes from WT or CB₁-KO mice after expression of the fluorescent protein mCherry (red) alone or together with the CB₁ receptor (mCherry-CB1 vector). (B) Quantification of the A-488 fluorescence intensity (ObRb labeling) in transfected cells. Results are expressed as the relative intensity versus intensity in WT astrocytes transfected with the mCherry vector. ***p<0.001, **p<0.01 as indicated between groups. (C) A-488 fluorescence intensity in transfected and non-transfected cells from the CB₁-KO astrocyte cultures exposed to the mCherry-CB₁ vector. *p<0.05 relative to the intensity in transfected astrocytes. Data shown are means with SEM values of at least 3 different experiments.

3.3. Astroglial CB₁ receptor deletion affects leptin signaling in astrocytes

One of the main intracellular effects of ObR signaling is the activation of the Janus kinase/signal transducers and activators of transcription (JAK/STAT) cascade. In particular, STAT3 and STAT5 are the transcription factors mainly activated by leptin [36–38]. Treatment with leptin moderately and transiently stimulated STAT3 phosphorylation in cultured WT cortical astrocytes (Figure 4A and B). In contrast, leptin failed to induce STAT3 phosphorylation in astrocytes prepared from constitutive CB₁-KO mice (Figure 4A and C). These results were confirmed by immunocytochemical detection and quantification of nuclear expression of phosphorylated STAT3 (P-STAT3) protein (Figure 4D).

Figure 4.

Deletion of astroglial CB₁ receptors impairs leptin-mediated STAT3 phosphorylation. (A) Immunoblot detection of the Tyr705-phosphorylated form of STAT3 (P-STAT3). WT (B) or CB₁-KO (C) cortical astrocytes were treated with vehicle, leptin 0.3 µg/ml or leptin 1 µg/ml, during 5, 10 or 30 min. The densitometric analysis of P-STAT3 expression was normalized against total STAT3 (Tot-STAT3) expression and results are expressed as the relative density versus density in non-treated WT astrocytes. Data shown are means with SEM values of at least 4 experiments. *p<0.05 relative to control. (D) Immunolabeling for P-STAT3 (green) and GFAP (red) in WT and CB₁-KO astrocytes treated during 10 min with vehicle or leptin (1 µg/ml). The nuclear localization of P-STAT3 is confirmed by colocalization with DAPI staining.

In addition, leptin elicited a considerable time-dependent increase in STAT5 phosphorylation in WT cultures as detected by Western immunoblotting (Figure 5A and B) and immunocytochemistry (Figure 5D and E). Similar to STAT3, the genetic deletion of the CB₁ receptors impeded the stimulating effect of leptin on STAT5 phosphorylation (Figure 5A and C). Interestingly, however, cortical astrocytes derived from constitutive CB₁-KO mice displayed a strong hyperphosphorylation of STAT5 in basal conditions, which was independent of leptin treatment. Indeed, in constitutive CB₁-KO cultures the basal expression of phosphorylated STAT5 (P-STAT5) was about 2-fold higher than in WT cultures (non-treated CB₁-KO astrocytes: 215.6±15.72% of non-treated WT astrocytes, ⁎⁎⁎ p<0.001, see also Fig. 5C and E). Very similar results were obtained in astrocyte cultures from hypothalami of CB₁-KO mice (Figure 5F). Thus, deletion of CB₁ receptors in astrocytes leads to a constitutive activation of the STAT5 pathway, which becomes unresponsive to leptin. Therefore, to investigate whether CB₁-dependent modulation of STAT5 activity could be relevant in vivo, we studied P-STAT5 expression in the hypothalamus of WT and GFAP-CB₁-KO mice (Figure 6). Nuclear P-STAT5 staining was identified in approximately 30% of all cells in the arcuate nucleus of WT animals, whereas this percentage was drastically increased up to more than 50% (Figure 6C) in GFAP-CB₁-KO mice.

Figure 5.

Deletion of astroglial CB₁ receptors modulates both basal and leptin-mediated STAT5 phosphorylation. (A) Immunoblot detection of the Tyr694-phosphorylated form of STAT5 (P-STAT5). WT (B) or CB₁-KO (C) cortical astrocytes were treated with vehicle, leptin 0.3 µg/ml or leptin 1 µg/ml, during 5, 10 or 30 min. The densitometric analysis of P-STAT5 expression was normalized against total STAT5 (Tot-STAT5) expression and results are expressed as the relative density versus density in non-treated WT astrocytes. Data shown are means with SEM values of at least 4 experiments. *p<0.05, **p<0.01 relative to control. (D) Immunolabeling for P-STAT5 (green) and GFAP (red) in WT and CB₁-KO cortical astrocytes treated during 10 min with vehicle or leptin (1 µg/ml). The nuclear localization of P-STAT5 is confirmed using DAPI staining. (E) Quantification of the A-488 fluorescence intensity (P-STAT5 labeling) in the nuclei (identified by the DAPI staining) of treated and non-treated WT and CB₁-KO astrocytes. (F) Quantification of a similar immunolabeling performed in hypothalamic astrocyte cultures. Results are expressed as the relative intensity versus intensity in WT control cultures. Data shown are mean values from 3 different experiments. ### p<0.001, general genotype effect as determined in two-way ANOVA, ***p<0.001 relative to control using Bonferroni's post hoc test.

Figure 6.

Increase in P-STAT5 labeled nuclei in arcuate nucleus of GFAP-CB₁-KO mice. Immunolabeling for P-STAT5 (green) and GFAP (red) in the arcuate nucleus or WT and GFAP-CB₁-KO mice (A). DAPI staining colocalizes with the P-STAT5 detection, confirming the nuclear localization of the phosphorylated protein V, third ventricle. (B) Quantification of the number of P-STAT5 labeled nuclei in WT and GFAP-CB₁-KO mice. Results are mean values with SEM of 12 sections series obtained from 3 different animals and are expressed as the percent of labeled nuclei in comparison to the total number of nuclei in WT versus in GFAP-CB₁-KO mice. *p<0.05 relative to WT.

3.4. Hyperactivation of STAT5 signaling alters leptin receptor expression in astrocytes

To analyze whether the reduced expression of ObR in cultured astrocytes from constitutive CB₁-KO mice is linked to STAT5 hyperactivation, we treated WT and mutant cultures with the STAT5 inhibitor sc-355979, and analyzed the mRNA levels of ObR. This treatment had no effect on WT cultures, but it rescued the expression of ObR mRNA in constitutive CB₁-KO astrocytes (Figure 7), suggesting that ObR down-regulation is due to STAT5 over-activation.

Figure 7.

Overactivation of STAT5 signaling pathway in CB₁-KO astrocytes is involved in the down-regulation of ObR expression. Quantitative PCR of ObR mRNA obtained from primary astrocyte cultures prepared from the cortex of WT and CB₁-KO animals. Cultures were exposed to vehicle or to the STAT5 inhibitor sc-355979 (500 µM) for 24 h. The expression of ObR mRNA was normalized against ubiquitin expression and results are expressed as the relative expression versus controls (WT, vehicle treated cultures). Data shown are means with SEM values of at least 3 experiments performed in duplicate. **p<0.01, relative to WT control cells. ## p<0.01 as indicated between groups.

3.5. Deletion of astroglial CB₁ receptors impacts leptin-induced glycogen accumulation in astrocytes

Leptin has been previously associated with alterations of glycogen metabolism in cultures of rat hepatocytes and mouse astrocytes [39,40]. In the brain, glycogen is primarily stored in astrocytes, where its accumulation correlates with neuronal activity [41]. Therefore, we examined the involvement of CB₁ receptors in the control of astrocyte glycogen storage by leptin. Leptin treatment of astrocyte cultures produced a marked increase in glycogen content in astrocytes isolated from both cortex (Figure 8A) and hypothalamus (Figure 8B). Interestingly, the response to leptin was triggered in a shorter interval of time (30 min exposure versus 2 h) and required lower concentration of the hormone in hypothalamic as compared to cortical astrocytes (Figure 8A and B). This suggests that hypothalamic astrocytes have higher sensitivity to leptin than cortical astrocytes. Importantly, leptin treatment did not affect the expression of inflammatory marker and/or enzymes controlling their synthesis (Supplementary Figure 2) suggesting that ObR does not control inflammatory responses in astrocytes. In addition, leptin failed to enhance intracellular glycogen concentration in astrocyte cultures derived from constitutive CB₁-KO mice (Figure 8 A and B), implying that astroglial CB₁ receptors are necessary for leptin's effect on glycogen storage. Finally, the inhibitor of STAT5 phosphorylation sc-355979 fully abolished the effect of leptin on glycogen content in WT cultures (Figure 8C). Altogether, these results strongly suggest that astroglial CB₁ receptors are required in order to maintain STAT5 signaling and ObR expression in cultured astrocytes and have a crucial impact on leptin-dependent metabolic effects.

Figure 8.

Deletion of astroglial CB₁ receptors impairs leptin-mediated glycogen storage in astrocytes. (A and B) Quantification of intracellular glycogen content in WT and CB₁-KO astrocyte cultures. Cortical astrocyte cultures (A) were exposed 2 h to vehicle, leptin 0.3 µg/ml or leptin 1 µg/ml while hypothalamic cultures (B) were treated during 30 min only in the same conditions. The influence of STAT5 signaling was investigated using the STAT5 inhibitor sc-355979. (C) Cortical astrocytes were treated for 2 h with leptin 1 µg/ml in the presence or the absence of 500 µM sc-355979. Data shown are means with SEM values of at least 4 experiments performed in duplicate and are expressed as glycogen content relative to control (non-treated WT astrocytes). **p<0.01, ***p<0.001 relative to WT control cells.

4. Discussion

Astrocytes have been classically considered as “passive” physical and nutritional support to neurons [42]. However, by tightly regulating brain energy production, delivery, storage and utilization [41], these cells have emerged as active players of synaptic activity. Moreover, astrocytes, by taking up and releasing neurotransmitters and “gliotransmitters” are now recognized as one of the key elements of the so-called “tripartite” synapse, formed by pre- and post-synaptic neuronal components and astrocytic processes [42]. To guarantee a fine-tuned regulation of brain energy metabolism, astrocytes must detect central and peripheral metabolic changes that may affect brain homeostasis. Recent studies have shown that leptin, one of the main peripheral adiposity signals, affects astrocyte morphology [43] as well as their functional and metabolic responses [17,40]. Here, we demonstrate the involvement of the astroglial CB₁ receptors in the regulation of crucial metabolic functions through the control of the leptin-dependent signaling.

Particularly, our results indicate that the deletion of CB₁ receptors from astrocytes impairs ObR expression and leptin-mediated functional responses in astrocytes. Previous investigations have shown that astrocytes produce different endocannabinoids and endocannabinoid-related compounds [44,45], therefore modulation of endocannabinoid levels could explain the requirement of CB₁ receptors for leptin-mediated responses. Accordingly, the treatment of astrocyte cultures with the dual FAAH and MGL inhibitor JZL195 [29] significantly induced ObR expression in WT astrocytes, while the CB₁ receptor antagonist SR141716A induced a robust reduction of ObR expression. Altogether, these findings support a direct role for astroglial CB₁ receptors concomitantly with a basal endocannabinoid tone in the maintenance of optimal ObR expression and functions. Importantly, the expression of ObR was completely rescued by re-expression of CB₁ receptor in constitutive CB₁-KO astrocytes, confirming that astroglial CB₁ receptors play a necessary and sufficient role to guarantee physiological levels of ObR.

Amongst the different splice variants of ObR, only the longest isoform (ObRb) can activate intracellular effectors [46]. While earlier investigations have initially associated the expression of the ObRb exclusively with neurons [35,47], more recent studies have reported the expression of this isoform in astrocytes [17,24–26]. Consistent with the latter studies, we herein demonstrated through different approaches that primary astrocyte cultures express functional ObR. Although our quantitative PCR experiments did not discriminate the different ObR isoforms, the specific expression of the active ObRb isoform was confirmed using an antibody specifically rose against this long-isoform (Gentaur, CD295 chicken antibody, [17]). Amongst the different STAT proteins, STAT3 is regarded as the major intracellular mediator of the actions of leptin in the brain [48,49]. However, our primary astrocyte cultures only displayed a modest phosphorylation of STAT3 in response to leptin. Conversely, we found that leptin considerably induced STAT5 phosphorylation. Interestingly, a recent study showed that nuclear STAT5 signal was increased in astrocytes after leptin administration while STAT3 was specifically enhanced in neurons and in endothelial cells [50]. Thus, the increase in STAT5 signaling seems to represent the predominant intracellular cascade triggered by ObRb activation in astrocytes. Constitutive CB₁-KO astrocytes displayed significant impairments in both leptin-mediated STAT3 and STAT5 signaling, likely due to lower levels of ObRb expression. However, quite surprisingly, we also found a constitutive increase in STAT5 phosphorylation in constitutive CB₁-KO cultures. Although we do not yet completely understand how CB₁ receptors and ObR interact, treatment of constitutive CB₁-KO astrocytes with sc-355979, a selective STAT5 inhibitor, completely restored ObR expression, thus implying that a down-regulation of ObR is likely the consequence of an excessive STAT5 constitutive activation. Possibly supporting the physiological implication of such an alteration in the leptin-mediated signaling cascade, the overall nuclear P-STAT5 signal was significantly higher in the hypothalamus of mice specifically deleted for the expression of CB₁ receptors in astrocytes (GFAP-CB₁-KO mice) as compared to their WT littermates.

While the essential role of STAT3 in leptin-mediated regulation of energy balance is well known [36,37], the function of leptin-induced STAT5 phosphorylation in this context remains unclear. Nevertheless, the obese phenotype of animals lacking a functional STAT5 in the brain suggests a role in the regulation of energy balance [51]. Related to this, leptin has been shown to regulate glucose and glycogen metabolism in different ex vivo models [52–55] as well as in astrocytes [40]. While in this latter study it was found that leptin blunts insulin-induced glycogen synthesis, we found that leptin, per se induced a significant increase in astrocyte glycogen content. This major functional response was abrogated in constitutive CB₁-KO cultures. While previous investigations already described a direct regulation of glycogen metabolism by CB₁ receptor activation in astrocytes [56], no differences were observed in the basal level of glycogen content between WT and constitutive CB₁-KO astrocytes. Rather than a CB₁-mediated regulation of basal glycogen content, this suggests a role for astroglial CB₁ receptors in the control of leptin-dependent brain energy metabolism as a consequence of the ObR down-regulation. Additionally, we found that leptin stimulation of glycogen storage was STAT5-dependent, further supporting the importance of this signaling cascade for astroglial leptin-mediated functions.

Beside the control of brain energy storage, leptin is also a cytokine involved in the modulation of immune responses and STAT proteins have a predominant role in the control of inflammation, since they are activated by several cytokines [57]. While STAT3 has been identified as the major signaling molecule engaged by cytokines during inflammatory processes in astrocytes [58,59], inflammatory mediators are also able to induce STAT5 activation [60]. Therefore, the observed regulation of STAT signaling as well as the STAT5-dependent metabolic effect could represent a consequence of an inflammatory response triggered by leptin. However, arguing against this hypothesis, leptin neither regulates pro and/or anti-inflammatory cytokine release, nor does it modify the expression of enzymes controlling their synthesis in our culture model (Supplementary Figure 2). Thus, despite the importance of astrocyte neuroinflammatory processes, this suggests that astroglial ObR receptors are essentially involved in the control of intracellular energy storage independently of the control of inflammatory responses.

Astrocytes provide neurons with anaplerotic metabolites for generation of energy [41] mainly through direct glucose delivery from the blood circulation. However, during periods of intense neural activity, when energy demand exceeds glucose supply, this also occurs through mobilization of their glycogen stores, suggesting that astrocyte glycogen may offer some protection against hypoglycemic neural injury [61]. Importantly, stored glycogen provides fuel to support brain functions not only in pathological conditions (i.e. hypoglycemia or hypoxia) but it also supplies neuronal energy demands during burst of synaptic activity and synaptic plasticity [62]. Therefore one may speculate that a modulation of glycogen metabolism by cannabinoids and/or leptin would also interfere with the fueling of neuronal energy needs. This could in turn influence higher brain functions, such as memory formation and consolidation, and associated synaptic plasticity in which astroglial glycogen mobilization is required [63,64]. Indeed, besides its ability to regulate the organism's energy balance, leptin has numerous other functions in the brain. Particularly, it was shown that leptin deficiency may alter some neuronal functions such as memory process, which can be restored by leptin injection [65,66]. This is suggestive of a protective action of leptin in the brain. Whether this is dependent upon modification of glycogen storage or not remains unexplored. However it is tempting to speculate that while leptin acts in the periphery to stimulate energy expenditure and fatty acid oxidation [23], it concomitantly modulates central energy stores in order to preserve neuronal function from a possible hypoglycaemic injury. Notably, astroglial CB₁ receptors have recently been shown to mediate the cannabinoid effect on spatial working memory and long-term depression at hippocampal synapses [10]. Furthermore, it has been also demonstrated that energy derived from glycogen degradation is important for the astrocytic glutamate uptake [67,68]. As this constitutes a main mechanism affecting glutamate neurotransmission [41,68], a cannabinoid-mediated modulation of astroglial glycogen processing could additionally impact excitatory synaptic activity. As both endocannabinoids [1,2,69] and leptin [43,70,71] have emerged as important regulators of synaptic plasticity, density and morphology, the roles of astroglial CB₁ and leptin receptors should be further considered when examining modulation of synaptic transmission.

Altogether our data demonstrate that the astroglial CB₁ receptor regulates ObR expression and signaling through an exclusive astrocytic mechanism. This was evidenced using astrocytes isolated from both the cortex and the hypothalamus, suggesting that the astroglial CB₁ receptors are required to maintain leptin-mediated functional responses, independently of the brain structure studied. Considering the complex and active role of astrocytes in brain functions as well as in sensing peripheral metabolic information, cannabinoid-dependent alterations of astroglial responses to leptin might therefore drastically impact diverse physiological processes, including the regulation of the organism's energy balance, brain energy storage and synaptic transmission and therefore participate to the development of pathological conditions, spanning from obesity to learning and memory deficits.

Conflict of interest

The authors wish to confirm that there are no known conflict of interest associated with this publication and there has been no significant financial support for this work that could have influenced its outcome.

Appendix A. Supplementary materials

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.molmet.2013.08.001.

Appendix A. Supplementary materials

Supplementary material