Cannabinoids modulate Olig2 and polysialylated neural cell adhesion molecule expression in the subventricular zone of post-natal rats through cannabinoid receptor 1 and cannabinoid receptor 2

Reagents

The selective agonist of CB1, arachidonyl-2-chloroethylamide (ACEA), was purchased from Tocris (Bristol, UK), as was the selective CB2 agonist, JWH-056 (JWH). The non-selective CB1/CB2 agonist (+)WIN 55,212-2 (WIN) was obtained from Sigma (Madrid, Spain). The selective antagonists of CB1, SR141716A (SR1), and CB2, SR144528 (SR2), were kindly provided by Sanofi-Aventis (Montpellier, France). All other reagents were obtained from standard suppliers.

Animals

Pregnant Wistar rats were obtained from Harlan-Interfauna Ibérica (Barcelona, Spain). For the localization of CB1 and CB2 in the post-natal SVZ, non-treated male progeny were killed at P7 and P15. For the experiments with cannabinoids, male pups were exposed to agonists or agonists and antagonists from P1 to P14 and the animals were killed at P15. The pups were maintained with their dams on a 12 h light/dark cycle in our in-house colony (Hospital Nacional de Parapléjicos, Toledo, Spain). Male littermates were included as control animals for each treatment and all litters were reduced to a fixed number of 10 pups.

Cannabinoid treatment of animals

Animals received a daily subcutaneous injection of cannabinoids in the neck from P1 to P14. Cannabinoids were always injected at noon with a G30 non-pyrogenic and sterile needle. Each treatment was performed in a separate litter and control littermates were injected with vehicle alone [either 1% bovine serum albumin and 0.2% dimethylsulphoxide in saline solution for WIN and WIN plus SR1 or SR2, or 1% bovine serum albumin and 0.2% ethanol in saline solution for ACEA]. Four neonatal males were injected with WIN, ACEA or WIN plus SR2 and five were treated with WIN plus SR1 or JWH. The doses of cannabinoids administered were based on those previously reported to effectively promote remyelination in a murine model of multiple sclerosis (Arevalo-Martin et al., 2003). To avoid potential habituation, the doses of the drugs administered increased as the treatment progressed, in accordance with the following regimes: 2.5 mg/kg of WIN up to P5, 3.75 mg/kg on P6–10 and 5 mg/kg on P11–14; the doses of ACEA were maintained at 1.25 mg/kg until P5 and rose to 1.9 mg/kg on P6–10 and 2.5 mg/kg on P11–14; SR1 or SR2 was administered at 2 mg/kg until P5, 3 mg/kg on P6–10 and 4 mg/kg on P11–14; JWH was injected at 2.5 mg/kg until P5, 3.75 mg/kg on P6–10 and 5 mg/kg on P11–14. Animals were handled in accordance with the guidelines published in the NIH Guide for the Care and Use of Laboratory Animals, the principles presented in the Guidelines for the Use of Animals in Neuroscience Research by the Society for Neuroscience and following the European Union guidelines (Council Directive 86/609/EEC). Experimental procedures were approved by our institutional animal use and care committee, the Ethical Committee for Animal Welfare at the National Paraplegics Hospital (CEBA). Special care was taken to reduce the number of animals used to the minimum required for statistical accuracy.

Immunohistochemistry

Rats were deeply anaesthetized with pentobarbital (100 mg/kg, Normon Veterinary Division, Madrid, Spain) and perfused transcardially with 0.1 m phosphate buffer and then with 4% paraformaldehyde in 0.1 m phosphate buffer or with a mixture of 4% paraformaldehyde and 1% glutaraldehyde in 0.1 m phosphate buffer for electron microscopy. The isolated brains were post-fixed in the same fixative for 4 h and stored in non-freezing Olmos solution at −18 °C until use. Serial coronal vibratome sections (50 µm) were obtained and subjected to immunohistochemistry using antibodies against: CB1 (1 : 300; 216401, Calbiochem, Barcelona, Spain or AB5636P, Chemicon, Madrid, Spain); the C-terminal region of CB2 (1 : 500; AB5642P, Chemicon); the N-terminal region of CB2 (1 : 500; PA1-746, Affinity Bioreagents, Madrid, Spain); glial fibrillary acidic protein (GFAP) (1 : 1000; Sigma, clone GA5); nestin (1 : 100; 556309, BD Biosciences, Madrid, Spain); phospho-histone H3 to specifically label mitotic nuclei (1 : 1000; 06-570, Upstate, Barcelona, Spain); PSA-NCAM (1 : 500, ABC0019, Abcys, Paris, France); anti-olig2 (1 : 30 000; from Dr Stiles' laboratory, a generous gift from Dr Rowitch); myelin basic protein (MBP) (1 : 1000; Sternberger Monoclonals); anti-vimentin phosphorylated at Ser55, clone 4A4 (1 : 1000; a generous gift from Dr Martínez-Cerdeño); the radial glia marker anti-3CB2 (1 : 300; a generous gift from Dr de la Rosa; Prada et al., 1995); or anti-aldolase C/zebrin II (1 : 1000; a generous gift from Dr Hawkes). For double-labelling experiments and optical density measurements, primary antibodies were detected with Alexa-conjugated anti-mouse and anti-rabbit secondary antibodies (Molecular Probes, Eugene, OR, USA), which were visualized by confocal microscopy (TCS 4D; Leica, Madrid, Spain). To estimate cell number, the primary antibodies were detected with secondary peroxidase-conjugated (Bio-Rad, Hercules, CA, USA) or biotinylated (Jackson Immunoresearch Laboratories, West Grove, PA, USA) antibodies that were then incubated with avidin-peroxidase (Sigma). After developing with diaminobenzidine (Sigma), sections were mildly counterstained with toluidine blue to visualize the SVZ. Sections processed for phospho-histone H3 and phospho-vimentin 4A4 were developed using nickel-intensified diaminobenzidine staining as described previously (DonCarlos et al., 2003). Negative controls were performed by omitting the primary antibody or by pre-absorbing the CB1 antibody with the immunogenic peptide derived from residues 1–14 of the human, mouse and rat CB1 protein (Cayman Chemical, Madrid, Spain).

Electron microscopy

For electron microscopy, brains were post-fixed overnight in 4% paraformaldehyde and 1% glutaraldehyde, rinsed in phosphate buffer and coronal sections (50 µm) were obtained using a vibrating microtome. CB1 was visualized by immunocytochemistry carried out in the absence of detergent and following diaminobenzidine staining. The dorsolateral SVZ was dissected out, post-fixed in 1% osmium tetroxide in 0.1 m phosphate buffer with 5% glucose, dehydrated in acetone and embedded in Spurr resin (TAAB, UK). Semithin sections (3 µm) were stained with 1% toluidine blue, re-embedded in Spurr and 60 nm sections were mounted onto grids and examined under a Jeol 1200EXII electron microscope.

Quantification of the cell number in the subventricular zone

The complex mixture and distribution of cells in the SVZ make it difficult to count cells other than by nuclear staining. In a thick section of the SVZ, it was not easy to identify the number of cells with cytoplasmic or membrane immunoreactivity in an unbiased manner. Therefore, we only quantified the number of cells when the immunostainings were directed against nuclear markers except in the case of 4A4, in which the cytoplasmic staining was easy to distinguish and was restricted to a small number of cells.

In any type of quantification, at least three histological sections lying between the coordinates +1.3 and 0 mm from Bregma (Paxinos & Watson, 2005) were assessed per rat. These coordinates were macroscopically distinguishable as a region immediately posterior to the disappearance of the ventral interhemispheric fissure and the appearance of the medial septum (approximately +1.3 cm from Bregma) and immediately anterior to the crossing of the anterior commissure (approximately 0 cm from Bregma). The SVZ was defined as the area of high nuclear density between the lateral ventricle, striatum and floor of the corpus callossum. This area was delimitated using a slight counterstaining of cell nuclei with toluidine blue or Hoescht.

Optical density measurements

Cannabinoid receptor 2 expression in the post-natal subventricular zone

We studied the expression of CB2 in the neonatal rat SVZ by reverse transcriptase-polymerase chain reaction and immunohistochemistry.

Western blot analysis

The SVZ was dissected and lysed in Tris-buffered saline, pH 7.6, containing 10% glycerol, 1% Nonidet P-40, 1 mm EDTA, 1 mm EGTA, 1 mm sodium orthovanadate, 2 mm NaF, 5 mm dithiothreitol and a protease cocktail. Cell lysates were separated and transferred to membranes, and the blots probed with anti-CB2 (AB5642P, Chemicon, 1 : 750) at 4 °C as described by Molina-Holgado et al. (2002).

Statistical analysis

Student's t-test or anova followed by Student's t-test was performed with prism 3.0 software. P-values less than 0.05 were considered significant.

Cannabinoid receptor 1 expression in cells of the subventricular zone

Post-natal day 7

The presence of CB1 in the embryonic and post-natal rat SVZ has been analysed previously by in-situ hybridization and through the binding of a radiolabelled ligand, without achieving cellular resolution (Berrendero et al., 1998; Buckley et al., 1998). Using specific antibodies, we found intense CB1 immunoreactivity in cells within both the ventricular zone and SVZ at P7. At this developmental stage the walls of the ventricles were mainly covered by radial glia, elongated cells with irregular nuclei and a long basal process. Dual immunohistochemistry for CB1 and the radial glia marker 3CB2 demonstrated that the vast majority of these radial glia express CB1 (Fig. 1A–C and Supplementary material, Tables S1 and S2). In contrast, very few cells coexpressed CB1 and PSA-NCAM, a marker of migrating cells (Fig. 1D–F and supplementary Tables S1 and S2). Cells that express PSA-NCAM are considered to be neuroblasts and these small cells form long chains that migrate towards the olfactory bulb in the rostral migratory stream. These chains of neuroblasts are surrounded by astrocyte processes or oligodendrocyte precursors that migrate individually without associating with other migratory cells or astrocytic feet (Doetsch et al., 1997; Menn et al., 2006). The ensheathing of PSA-NCAM cells by the thin processes of the radial glia/B-cells sometimes makes it difficult to clearly ascertain whether CB1 and PSA-NCAM are expressed by the same cell or whether CB1 is rather in the surrounding cellular processes. Accordingly, we performed electron microscopy studies to more accurately define the CB1-immunoreactive cells using the ultrastructural criteria of SVZ cell types described previously (Doetsch et al., 1997; Tramontin et al., 2003; Merkle et al., 2004). Briefly, type A cells, corresponding to the dark cells in toluidin stainings, show the smallest cross-sectional area and present an elongated cell body with few processes, a very thin cytoplasm and elongated nuclei with abundant lax chromatin. Type B cells have irregular contours that intermingle with the neighbouring cells, an irregular nucleus that frequently contains invaginations, light cytoplasm with few free ribosomes and abundant intermediate filaments and dense bodies. Type C cells are larger, less elongated and more electron-lucent than type A cells but more electron-dense than type B cells. Their nuclei contain deep invaginations and mostly lax chromatin. Type C cells have a large reticulated nucleolus and show in their cytoplasm a large Golgi apparatus and no apparent bundles of intermediate filaments. Ependymocytes have interdigitated processes, microvilli and cilia in the ventricular surface, spherical nuclei with non-clumped chromatin and electron-lucent cytoplasm with apical basal bodies. Radial glia have long basal processes and some anatomical features of cells in the astroglial lineage, including intermediate filaments and glycogen granules.

We confirmed, in both semithin and ultrathin sections, that antibodies against CB1 did not stain any cells with the morphological properties of neuroblasts/A-cells. However, CB1 expression was found in many radial glia cells (Fig. 2A) and astrocyte-like cells, forming tubes around clusters of neuroblasts (Fig. 2B). In addition, CB1 is located in disperse cells at P7 that we could not clearly assign to the B-, radial glia- or even the C-cell type. Indeed, these cells displayed morphological characteristics that were apparently intermediate between these cell types. Although only a small number of these cells were observed, we performed double labelling for CB1 and aldolase C/zebrin II, a marker that does not stain C-cells (Marshall & Goldman, 2002). As expected, the cells that expressed CB1 largely overlapped with the cell population that expressed zebrin II (supplementary Fig. S1), confirming that CB1 is essentially absent from the populations of C- and A-cells.

Post-natal exposure to an agonist of cannabinoid receptor 1 modulates cell proliferation in the subventricular zone

The reduction in the number of proliferating cells in the CB1 knockout mice has implicated this cannabinoid receptor in the control of SVZ cell proliferation (Jin et al., 2004). When we estimated the cells immunostained for phospho-histone H3, a marker of all cells undergoing mitosis, exposure of post-natal rats to cannabinoids from P1 to P14 did not appear to affect the proliferation of cells in the SVZ at P15 (Fig. 3A and B). However, due to the slow cell cycle of the stem cells, any change in their proliferation may be masked by the larger number of transit amplifying cells that proliferate much more rapidly. Therefore, we attempted to specifically study the effect of cannabinoids on the proliferation of radial glia. In order to achieve this, we used the 4A4 antibody that specifically recognizes vimentin phosphorylated by the cdc2 kinase at Ser55 and that has been described as a marker of radial glial proliferation (Noctor et al., 2004). Chronic post-natal treatment with the selective CB1 agonist ACEA increased the number of 4A4-positive cells in the SVZ in P15 rats (Fig. 3C–E). We studied the whole population of radial glia by measuring the area occupied by the rat radial glia marker 3CB2 in the SVZ at P15. In rats treated with ACEA, we did not observe a significant increase in the presence of this marker when compared with their control littermates, although a trend towards an increase in 4A4 immunostaining was observed in the animals exposed to ACEA (P = 0.247, Fig. 3F–H).

Post-natal treatment with a cannabinoid receptor 1 agonist increases gliogenesis in the subventricular zone

Cannabinoid receptor 1 has recently been shown to be involved in gliogenesis within the subgranular zone (Aguado et al., 2006). Here, we studied the formation of gliogenic precursors in the post-natal rat SVZ by counting the number of Olig2-positive cells and the capacity of the neural precursor cells to migrate analysed by measuring the expression of PSA-NCAM.

Olig2 is a basic helix-loop-helix transcription factor (Lu et al., 2000) that is expressed early in glial precursors. It has recently been shown to both prevent neuronal differentiation and to drive neural precursor cells towards oligodendroglial fates in the post-natal and adult SVZ (Hack et al., 2005; Marshall et al., 2005). Post-natal treatment of rats with ACEA increased the number of Olig2-positive cells in the SVZ at P15 (Fig. 4A–C). Glial progenitors of the SVZ migrate through the dorsolateral SVZ to the white matter, where they differentiate into oligodendroglial cells (Levison & Goldman, 1993). PSA-NCAM is expressed by these cells (Menn et al., 2006) and is necessary for the migration of glial progenitors in the normal and injured brain (Wang et al., 1994; Nait-Oumesmar et al., 1999; Picard-Riera et al., 2002). Unlike the effects observed on Olig2, we did not observe changes in the expression of PSA-NCAM in the dorsolateral SVZ after treatment with ACEA (Fig. 4D–F).

Cannabinoid receptor 2 is expressed in the post-natal subventricular zone

There is increasing evidence that CB1 is not the only receptor that mediates the effects of cannabinoids in the central nervous system. Thus, we wondered whether CB2 might also be present in the post-natal rat SVZ. Hence, we analysed the expression of CB2 mRNA at P7 and P15 in the dissected SVZ by reverse transcriptase-polymerase chain reaction with primers designed to hybridize in two separate exons. CB2 transcripts were clearly identified in the isolated SVZ at both ages, indicating that this receptor may be present in this tissue (Fig. 5A and B).

Subsequently, we used two different antibodies to localize CB2 in the SVZ, one generated against the N-terminal region (1–33 amino acids, Affinity Bioreagents) and the other raised against the C-terminal domain of the rat CB2 (18 amino acids, Chemicon). Both antibodies stained equivalent populations in the SVZ, both at P7 and P15, and these antibodies recognized cells with a location distinct from that obtained with the CB1 antibodies. Most cells recognized by the CB2 antibodies were isolated from the ventricle lumen and they were also labelled with PSA-NCAM antibodies (Fig. 5C–K). In addition, we found some GFAP-positive cells that displayed CB2 immunoreactivity at P15 (Fig. 5L–N). Omission of the primary antibodies abolished this staining and pre-absorption of C-terminal antibody with its immunogenic peptide led to a strong decrease in immunogenicity (supplementary Fig. S2). In addition, in western blots of post-natal SVZ, this antibody recognized bands corresponding to the molecular weights previously described for CB2 (supplementary Fig. S2) (Van Sickle et al., 2005; Gong et al., 2006). This was also demonstrated using the antibody raised against the N-terminal region of CB2 (supplementary Fig. S2).

Post-natal exposure to a cannabinoid receptor 2 agonist increases polysialylated neural cell adhesion molecule expression in the subventricular zone

In contrast to what we had observed following ACEA treatment, the activation of CB2 with the selective agonist JWH did not significantly affect the number of Olig2-positive cells in the SVZ (Fig. 6A–C). Indeed, JWH increased the expression of PSA-NCAM by more than two-fold in this region (Fig. 6D–F).

Post-natal cannabinoid treatment modulates myelin basic protein expression in the external capsule

Olig2 is a transcription factor that is necessary for oligodendrocyte formation and the posterior myelination of the white matter (Lu et al., 2000; Hack et al., 2005; Marshall et al., 2005; Yue et al., 2006). In contrast, PSA-NCAM is a molecule that facilitates the migration of oligodendrocyte precursors (Wang et al., 1994; Nait-Oumesmar et al., 1999; Picard-Riera et al., 2002). As the first weeks of post-natal development represent an intense period of myelination (Bjelke & Seiger, 1989; Hamano et al., 1996), we looked for a possible correlation between the changes in the expression of these factors and the modification of MBP expression, which can be considered as a good index of myelin synthesis (Cohen & Guarnieri, 1976). We focused on the external capsule, which is one of the first regions inside the subcortical white matter to be myelinated (Fig. 7). In spite of their effects on glial precursors (Olig2 population) or the migrating phenotype of the cells (PSA-NCAM expression), the activation of CB1 (ACEA treatment) or CB2 (JWH) alone did not exert a significant influence on MBP expression, although JWH treatment did induce a trend towards its up-regulation (P = 0.095).

We then explored the effect of a post-natal exposure to WIN, a synthetic cannabinoid that activates both CB1 and CB2. We found that exposure to WIN increased MBP immunostaining in the external capsule of P15 rats when compared with control littermates (Fig. 7A, D and G). When the rats were treated simultaneously with WIN and either a specific CB1 (SR1) or specific CB2 (SR2) antagonist, the increase in MBP induced by WIN was not only inhibited but, moreover, MBP expression decreased to levels below those of control littermates (Fig. 7E–G).