GABAergic and Cortical and Subcortical Glutamatergic Axon Terminals Contain CB1 Cannabinoid Receptors in the Ventromedial Nucleus of the Hypothalamus

Background Type-1 cannabinoid receptors (CB1R) are enriched in the hypothalamus, particularly in the ventromedial hypothalamic nucleus (VMH) that participates in homeostatic and behavioral functions including food intake. Although CB1R activation modulates excitatory and inhibitory synaptic transmission in the brain, CB1R contribution to the molecular architecture of the excitatory and inhibitory synaptic terminals in the VMH is not known. Therefore, the aim of this study was to investigate the precise subcellular distribution of CB1R in the VMH to better understand the modulation exerted by the endocannabinoid system on the complex brain circuitries converging into this nucleus. Methodology/Principal Findings Light and electron microscopy techniques were used to analyze CB1R distribution in the VMH of CB1R-WT, CB1R-KO and conditional mutant mice bearing a selective deletion of CB1R in cortical glutamatergic (Glu-CB1R-KO) or GABAergic neurons (GABA-CB1R-KO). At light microscopy, CB1R immunolabeling was observed in the VMH of CB1R-WT and Glu-CB1R-KO animals, being remarkably reduced in GABA-CB1R-KO mice. In the electron microscope, CB1R appeared in membranes of both glutamatergic and GABAergic terminals/preterminals. There was no significant difference in the percentage of CB1R immunopositive profiles and CB1R density in terminals making asymmetric or symmetric synapses in CB1R-WT mice. Furthermore, the proportion of CB1R immunopositive terminals/preterminals in CB1R-WT and Glu-CB1R-KO mice was reduced in GABA-CB1R-KO mutants. CB1R density was similar in all animal conditions. Finally, the percentage of CB1R labeled boutons making asymmetric synapses slightly decreased in Glu-CB1R-KO mutants relative to CB1R-WT mice, indicating that CB1R was distributed in cortical and subcortical excitatory synaptic terminals. Conclusions/Significance Our anatomical results support the idea that the VMH is a relevant hub candidate in the endocannabinoid-mediated modulation of the excitatory and inhibitory neurotransmission of cortical and subcortical pathways regulating essential hypothalamic functions for the individual's survival such as the feeding behavior.


Introduction
The hypothalamus plays a crucial role in regulating energy balance and food intake [1]. The ventromedial nucleus (VMH) is placed in the tuberal region of the hypothalamus and is associated with several homeostatic and behavioral functions, including regulation of appetite, energy balance, sexual behavior, anxiety, thermogenesis, cardiovascular functions and pain [2,3]. Functionally, the dorsomedial VMH participates in the regulation of energy homeostasis, whereas the ventrolateral VMH controls female reproduction [2,4].
The VMH has been proposed as a satiety nucleus that provides a strong excitatory input to arcuate neurons, contributing to the activation of anorexigenic neuronal pathways [9,10]. The endocannabinoid system is implicated in endocrine regulation and energy balance. The derivatives of Cannabis sativa are well known to regulate food intake and the endocannabinoid system controls neuronal signaling in hypothalamic networks [11]. Although low levels of cannabinoid receptors are present in the hypothalamic nuclei [12,13], their efficiency is higher than in other brain regions [14]. Moreover, specific cannabinoid receptor binding is found in several hypothalamic areas, including the VMH, which also expresses high levels of CB 1 R mRNA [13]. Although the overall CB 1 R immunolabeling intensity is much lower in the hypothalamus than in other brain regions, the VMH, in particular, exhibits a moderate CB 1 R immunostaining [15].
The levels of the two main endocannabinoids, anandamide and 2-arachydonoylglycerol (2-AG), in the hypothalamus are higher during fasting and lower following food intake reaching a critical point that favors a motivational state for food intake [11,[16][17][18]. The administration of anandamide into the VMH also stimulates appetite in rats [19]. In contrast, both chronically-treated animals with CB 1 R antagonists [11,20,21] and CB 1 R null mice [11,20,22] display an anorexigenic phenotype. Furthermore, activation of presynaptic CB 1 R inhibits the excitatory and inhibitory neurotransmission in neuronal circuits involved in eating behaviors [11,18,[23][24][25]. Indeed, Glu-CB 1 R-KO conditional mice that do not express CB 1 R in neurons of cortical origin exhibit a hypophagic phenotype after food deprivation very similar to the full CB 1 R-KO. On the contrary, GABA-CB 1 R-KO mutants that lack CB 1 R in forebrain GABAergic neurons are hyperphagic under the same experimental conditions [26].
Taken together, it is well established that the endocannabinoid system exerts a neuronal modulation through the activation of presynaptic CB 1 R localized on both excitatory and inhibitory pathways in distinct brain networks regulating homeostatic and behavioral functions including food intake. In view of the described observations that both the endocannabinoid system and the VMH play a role in ingestive behaviors, the aim of this study was to analyze the CB 1 R contribution to the molecular architecture of the excitatory and inhibitory synaptic terminals in the mouse VMH. For this purpose, preembedding immunocytochemical techniques for light and high resolution electron microscopy were used. Highly specific CB 1 R antibodies were applied to the VMH of conditional mutant mice with a selective deletion of CB 1 R mainly from cortical glutamatergic (Glu-CB 1 R-KO) and mainly from forebrain GABAergic neurons (GABA-CB 1 R-KO) [27,28]. Mutants with the lack of CB 1 R in all the cells of the body (CB 1 R-KO mice) were also examined [29].

Immunolocalization of CB 1 R in the VMH
In the light microscope, the CB 1 R immunoreactivity was uniformly distributed throughout the entire VMH of CB 1 R-WT (Fig. 1A) with a somehow similar appearance in the Glu-CB 1 R-KO mice (Fig. 1B). At higher magnification, the pattern consisted of abundant small immunoreactive dots densely packed within the oval-shaped VMH (Fig. 1A', B'). However, CB 1 R staining decreased drastically in the VMH of GABA-CB 1 R-KO mice (Fig. 1C), particularly in the dorsomedial part ( Fig. 1C'), suggestive of the presence of CB 1 R in GABAergic profiles. The immunolabeling fully disappeared in the VMH of CB 1 R-KO mice (Fig. 1D, D').
Then, we analyzed the ultrastructural distribution of CB 1 R in the dorsomedial region of the VMH using a preembedding immunogold method for electron microscopy (Fig. 2). CB 1 R immunoparticles were typically localized away from the active zones on preterminal or synaptic terminal membranes making synapses with dendrites or dendritic spines. They showed characteristic features of excitatory (asymmetric synapses with obvious postsynaptic densities, abundant clear and spherical synaptic vesicles) and inhibitory (symmetric synapses with more pleomorphic synaptic vesicles) synapses ( Fig. 2A, B). 24.062.9% and 28.967.5% of the synaptic terminals making asymmetric and symmetric synapses, respectively, were CB 1 R immunopositive in the VMH of CB 1 R-WT mice (Fig. 3A). In this case, CB 1 R density was 0.42 immunoparticles/ mm membrane in terminals making asymmetric synapses and 0.47 immunoparticles/ mm in terminals making symmetric synapses (Fig. 3B). There were no statistically significant differences in these parameters between terminals with asymmetric or symmetric synapses in the CB 1 R-WT mice.
To define the contribution of cortical glutamatergic and GABAergic synaptic terminals to the intrinsic CB 1 R pattern in the VMH, conditional CB 1 R mutant mice lacking the receptor either in cortical glutamatergic (Glu-CB 1 R-KO) or in forebrain GABAergic neurons (GABA-CB 1 R-KO) were used. CB 1 R was still observed in VMH axon terminals making synapses with dendritic and spiny elements of both mutant strains ( Fig. 2C-H). In Glu-CB 1 R-KO mice, CB 1 R immunopositive terminals made asymmetric ( Fig. 2C, D) and symmetric synapses (Fig. 2E). Also, CB 1 R immunonegative asymmetric synaptic terminals were found in the Glu-CB 1 R-KO mutants (Fig. 2C, E), suggesting the presence of CB 1 R in cortically-derived axon terminals. In GABA-CB 1 R-KO tissue, CB 1 R immunoparticles decorated presynaptic membrane profiles forming asymmetric (Fig. 2F, G) but not symmetric synapses (Fig. 2H). The immunolabeling was specific as the CB 1 R pattern disappeared in the VMH of CB 1 R-KO mice (Fig. 2I, J).
We next semiquantified the CB 1 R immunolabeled excitatory axonal boutons to determine the contribution of cortical axons to the pattern of CB 1 R in the VMH. For this purpose, only typical excitatory terminals with abundant clear and spherical vesicles, forming asymmetric synapses with thick postsynaptic densities were taken into account. In this case, 21.362.5% and 27.260.7% of the asymmetric synapses were CB 1 R immunopositive in the VMH of Glu-CB 1 R-KO and CB 1 R-WT mice, respectively (Fig. 4C). However, this difference was not statistically significant (x 2 = 0.4189, p = 0.5175). Finally, the percentage of CB 1 R immunolabeled asymmetric synapses was very low in CB 1 R-KO mice (Fig. 4C).
Taken together, these observations indicate that CB 1 R is localized in GABAergic as well as in cortical and subcortical glutamatergic inputs to the VMH.

CB 1 R is localized in excitatory and inhibitory presynaptic boutons in the VMH
The main finding of this study was the localization of CB 1 R in VMH presynaptic terminals impinging on postsynaptic dendrites and spines of CB 1 R-WT, Glu-CB 1 R-KO and GABA-CB 1 R-KO mice. Furthermore, an extensive analysis of the proportion of immunolabeled profiles identified the contribution of CB 1 R to GABAergic and cortical and subcortical glutamatergic inputs to the VMH.
The dense network of synaptic connections constitutes the anatomical basis for the neuroendocrine and vegetative functions regulated by the hypothalamus. The proportion of CB 1 R immunolabeled synaptic terminals in the VMH of mice lacking CB 1 R in neurons of cortical origin (Glu-CB 1 R-KO) was identical to WT animals (,20%), indicating that CB 1 R probably was in excitatory synaptic terminals of intrinsic hypothalamic neurons. However, although the difference was not statistically significant, the analysis of synaptic terminals forming asymmetric synapses showed a slight decrease of glutamatergic synaptic profiles with CB 1 R in Glu-CB 1 R-KO compared to CB 1 R-WT mice. Altogether, these results indicate that CB 1 R localizes mostly in subcortical excitatory axon terminals [8,22,30,31] and to a lesser extent in excitatory synaptic boutons of cortical origin [1,8,27,31].
The absence of CB 1 R in forebrain GABAergic neurons (GABA-CB 1 R-KO) caused a reduction of the CB 1 R immunolabeled synaptic terminals (12.4%) indicating that CB 1 receptors are also a molecular component of the GABAergic axon boutons in the VMH. For GABA-CB 1 R-KO mutants, DLX mice lead also to recombination in hypothalamic dopaminergic neurons [32]. However, it is unlikely the presence of CB 1 R in dopaminergic synaptic terminals in the VMH of the GABA-CB 1 R-KO mutants as there is no tyrosine hydroxylase immunoreactivity in the VMH [32]. Overall, our findings can be interpreted as for the presence of

Functional significance
This investigation has demonstrated that CB 1 receptors in GABAergic and glutamatergic afferents explain the CB 1 R pattern in the VMH. The density of CB 1 R immunoparticles was rather low in GABAergic and glutamatergic boutons in the VMH (,0.40-0.50 particles/ mm) as compared to the density found in other brain regions [33,34], particularly in inhibitory synaptic terminals [35]. However, CB 1 R efficiency in the activation of GTP-binding proteins appears to be much higher in the hypothalamus than in other brain regions [14], which may have a functional significance. Physiologically, the identification of CB 1 R in glutamatergic and GABAergic synaptic terminals in the VMH could be regarded as a potential neuronal substrate for the Figure 2. Ultrastructural localization of CB 1 R in the mouse VMH. Preembedding immunogold method for electron microscopy. A, B: In CB 1 R-WT, CB 1 R immunoparticles (arrows) are localized on membranes of presynaptic terminals (Ter) making asymmetric (white arrowheads) and symmetric synapses (black arrowheads) with dendritic spines (Sp) or dendrites (Den). C-E: In Glu-CB 1 R-KO, CB 1 R immunoparticles (arrows) localize to asymmetric synaptic terminals (Ter) presumably of excitatory subcortical neurons (observe the thick postsynaptic density marked with white arrowheads in D) as well as in inhibitory terminals (Ter) with symmetric synapses (black arrowheads in E). Notice CB 1 R immunonegative axon terminals (Ter) establishing asymmetric synapses (white arrowheads in C, E) with a dendrite (Den) or a spine (Sp). F-H: In GABA-CB 1 R-KO, CB 1 R immunolabeling (arrows) is in excitatory synaptic terminals (Ter) (see asymmetric synapses with white arrowheads in F and G) impinging on dendritic elements (Den). Observe in H a CB 1 R immunonegative synaptic terminal (Ter) making a symmetric synapse (black arrowheads) with a dendrite (Den). I, J: CB 1 R immunolabeling disappears in CB 1 R-KO mice indicating the specificity of the CB 1 R antibody used in the study. Note CB 1 R immunonegative synaptic terminals (Ter) making asymmetric (white arrowheads in I) and symmetric (black arrowheads in J) synapses with a dendritic spine (Sp) and a soma (Som), respectively. Scale bars: 0.4 mm. doi:10.1371/journal.pone.0026167.g002 effects of cannabinoids on eating behaviors. Actually, Glu-CB 1 R-KO conditional mice exhibit a hypophagic behavior after food deprivation very similar to the full CB 1 R-KO. On the contrary, GABA-CB 1 R-KO mutants are hyperphagic under the same experimental conditions [26].
As a conclusion, the VMH may be a good hub candidate in the endocannabinoid-mediated modulation of the excitatory and inhibitory neurotransmission regulating food intake behaviors. These anatomical data contribute to the understanding of the complex regulation of energy balance by the endocannabinoid system.

Ethics Statement
The protocols for animal care and use were approved by the appropriate Committee at the Basque Country University

CB 1 R mutant lines
Mutant animals were obtained and genotyped as previously described [26,27,29]. CB 1 R-KO mice were generated and genotyped as described [29]. Conditional CB 1 R mutant mice were obtained by crossing the respective Cre-expressing mouse line with CB1 f/f mice [36], using a three-step breeding protocol [27].
Generation of CB1 f/f; NEX-Cre mice (here Glu-CB 1 R-KO). CB1 f/f; NEX-Cre mice were obtained by crossing CB1 f/f with NEX-Cre mice [37,38]. The helix-loop-helix transcription factor NEX is a marker of embryonic neuronal progenitors, which will develop into mature cortical glutamatergic neurons [39], whereas, in the adult brain, NEX is expressed in mature glutamatergic cortical neurons, but not in cortical GABAergic interneurons and to a much lesser extent in subcortical regions [40]. Cre expression under the control of the regulatory sequences of NEX in transgenic mice (NEX-Cre mice) as generated by knock-in into the NEX locus, leads to the specific deletion of ''floxed'' alleles in forebrain neurons [37].

Animal treatment
12 wild-type, Glu-CB 1 R-KO, GABA-CB 1 R-KO and CB 1 R-KO mice (3 of each condition) were used in this study. Mice were deeply anesthetized by intraperitoneal injection of ketamine/ xylazine (80/10 mg/kg body weight) and were transcardially perfused at room temperature (RT, 20-25uC) with phosphatebuffered saline (PBS 0.1M, pH 7.4) for 20 seconds, followed by the fixative solution made up of 4% formaldehyde (freshly depolymerized from paraformaldehyde), 0.2% picric acid and 0.1% glutaraldehyde in phosphate buffer (PB 0.1 M, pH 7.4) for 10-15 minutes. Then, brains were removed from the skull and postfixed in the fixative solution for approximately one week at 4uC. Afterwards, brains were stored at 4uC in 1:10 diluted fixative solution until used.

CB 1 R immunocytochemistry for light microscopy
Coronal hypothalamic sections were cut at 50 mm in a vibratome and collected in 0.1 M PB at RT. Sections were preincubated in a blocking solution of 10% bovine serum albumin (BSA), 0.1% sodium azide and 0.5% triton X-100 prepared in Tris-HCl buffered saline (TBS 1X, pH 7.4) for 30 minutes at RT. Then, they were incubated in a primary polyclonal goat anti-CB 1 R antibody (2 mg/ml, Frontier Science co. Ltd, 1-777-12, Shinko-nishi, Ishikari, Hokkaido, Japan) prepared in the blocking solution, on a shaker for 2 days at 4uC. The CB 1 R antibody recognizes 31 aminoacids of the C-terminus part (NM007726) of the mouse CB 1 R. After several washes in 1% BSA and 0.5% triton X-100 in TBS, tissue sections were incubated in a secondary biotinylated horse anti-goat IgG (1:200, Vector Laboratories, Burlingame, CA, USA) prepared in the washing solution for 1 hour on a shaker at RT. The VMH sections were washed in the washing solution described above and processed by a conventional avidin-biotin peroxidase complex method (ABC, Elite, Vector Laboratories, Burlingame, CA, USA). Tissue was incubated in the avidin-biotin complex (1:50) prepared in the washing solution for 1 hour at RT. Then, sections were washed and incubated with 0.05% diaminobenzidine in 0.1 M PB with 0.5% triton-X100 and using 0.01% hydrogen peroxide as a cromogen, for 5 minutes at RT. Finally, tissue was mounted, dehydrated in graded alcohols (50u, 70u, 96u, 100u) to xylol and coverslipped with DPX. Sections were observed and photographed with a light microscope Zeiss Axiophot. Figure compositions were made at 300 dots per inch (dpi). Labeling and minor adjustments in contrast and brightness were made using Adobe Photoshop (CS, Adobe Systems, San Jose, CA, USA).

Preembedding immunogold method for electron microscopy
Coronal hypothalamic vibrosections were cut at 50 mm and collected in 0.1 M PB at RT. Sections were preincubated in a blocking solution of 10% BSA, 0.1% sodium azide and 0.02% saponin in TBS for 30 minutes at RT. Then, they were incubated in a primary polyclonal goat anti-CB 1 R antibody (2 mg/ml, Frontier Science co. Ltd, 1-777-12, Shinko-nishi, Ishikari, Hokkaido, Japan) prepared in the blocking solution but with 0.04% saponin, on a shaker for 2 days at 4uC. After several washes with 1% BSA in TBS, tissue sections were incubated in a secondary 1.4 nm nano-gold antigoat antibody (1:100, Fab' fragment, Nanoprobes Inc., Yaphank, NY, USA) prepared in the same solution as the primary antibody for 3 hours on a shaker at RT. Then, tissue was washed overnight at 4uC and postfixed in 1% glutaraldehyde for 10 minutes. After several washes in double distilled water, gold particles were silver-intensified with a HQ Silver Kit (Nanoprobes Inc., Yaphank, NY, USA) for 12 minutes in the dark. Then, tissue was extensively washed in double distilled water and in 0.1 M PB and osmicated in 1% osmium tetroxide for 20 minutes. After washing in 0.1 M PB, sections were dehydrated in graded alcohols (50u, 70u, 96u, 100u) to propylene oxide and embedded in Epon resin 812. 80 nm ultrathin sections were collected on mesh nickel grids, stained with lead citrate for 20 minutes and examined in a PHILIPS EM208S electron microscope. Tissue preparations were photographed by using a digital camera coupled to the electron microscope. Figure compositions were made at 300 dots per inch (dpi). Labeling and minor adjustments in contrast and brightness were made using Adobe Photoshop (CS, Adobe Systems, San Jose, CA, USA).
Specificity of the immunostainings was assessed by incubation of the CB 1 R antibody in CB 1 R-KO VMH tissue in the same conditions as above.
Statistical analysis of CB 1 R in the VMH 50-mm-thick hypothalamic sections from each animal condition (n = 3 each) showing good and reproducible silver-intensified gold particles were cut at 80 nm. Image-J (version 1.43 m, NIH, USA) was used to measure the membrane length. Electron micrographs (18,000-28,000X) were taken from grids (132 mm side) containing silver-intensified gold particles; all of them showed a similar labeling intensity indicating that selected areas were at the same depth. Furthermore, to avoid false negatives, only ultrathin sections in the first 1.5 mm from the surface of the tissue block were examined. Positive labeling was considered if at least one immunoparticle was within approximately 30 nm from the plasmalemma. Metal particles on synaptic membranes were visualized and counted.
Percentages of CB 1 R positive profiles and density of immunoparticles were analyzed and displayed as mean 6 S.E.M. using a statistical software package (GraphPad Prism 4, GraphPad Software Inc, San Diego, USA). Group differences were compared by chisquare test, p,0.05 (percentages of CB 1 R positive profiles) and Mann Whitney test, p,0.05 (CB 1 R density).