CB1 Cannabinoid Receptors Increase Neuronal Precursor Proliferation through AKT/Glycogen Synthase Kinase-3β/β-Catenin Signaling

The endocannabinoid system is involved in the regulation of many physiological effects in the central and peripheral nervous system. Recent findings have demonstrated the presence of a functional endocannabinoid system within neuronal progenitors located in the hippocampus and ventricular/subventricular zone that participates in the regulation of cell proliferation. It is presently unknown whether the endocannabinoid system exerts a widespread effect on neuronal precursors from different neurogenic regions, and very little is known about the signaling by which it regulates neuronal precursor proliferation. Herein, we demonstrate the presence of cannabinoid CB1 receptors in granule cell precursors (GCPs) during early cerebellar development. Activation of CB1 receptors by HU-210 promoted GCP proliferation in vitro, an effect that was prevented by a selective CB1 antagonist. Accordingly, in vivo experiments showed that GCP proliferation was increased by chronic HU-210 treatment and that in CB1-deficient mice cell proliferation was significantly lower than in wild-type littermates, indicating that the endocannabinoid system is physiologically involved in regulation of GCP proliferation. The pro-proliferative effect of cannabinoids in GCPs was mediated through the CB1/AKT/glycogen synthase kinase-3β/β-catenin pathway. Involvement of this pathway was also observed in cultures of neuronal precursors from the subventricular zone, suggesting that this pathway may be a general mechanism by which endocannabinoids regulate proliferation of neuronal precursors. These observations suggest that endocannabinoids constitute a new family of lipid signaling cues that may exert a widespread effect on neuronal precursor proliferation during brain development.

Cannabinoid drugs such as ⌬ 9 -tetrahydrocannabinol, the principal psychoactive constituent of marijuana, act via signaling pathways consisting of endogenous cannabinoids (endocannabinoids) and their receptors. Among endocannabinoids, the best characterized are arachidonylethanolamide or anandamide and 2-arachidonylglycerol, although additional candidates have been proposed. Two G-protein-coupled cannabinoid receptors, designated CB 1 and CB 2 , have been cloned (1,2). The CB 1 receptor is highly expressed in the central nervous system and is also present in peripheral and extraneuronal sites (3,4). In contrast, the CB 2 receptor is almost restricted to the immune system (2).
Endocannabinoid signaling pathways have been implicated in a broad range of neurobiological processes, including movement control, cognition, learning and memory, pain relief, and in promoting neuronal survival after cerebral ischemia or trauma (5)(6)(7)(8)(9). In addition, pharmacological and gene knockout studies point to a role for endocannabinoid signaling in promoting brain development (10). For example, the expression of endocannabinoids and cannabinoid receptors appears early in the developing brain (10 -13), and the perinatal exposure to synthetic or plant-derived cannabinoids was shown to modify the maturation of neurotransmitter systems and their related behaviors (14). Recent studies have demonstrated the presence of a functional endocannabinoid system in neuronal progenitor cells of the ventricular (VZ) 3 and subventricular zone (SVZ) and subgranular zone of the hippocampal dentate gyrus, where it increases cell proliferation (15)(16)(17)(18)(19)(20). Consistent with the proliferation-promoting function of CB 1 receptors, impaired proliferation is observed during cortical development in the VZ/SVZ of CB 1 knock-out mice (16). Conversely, inhibition of the activity of the fatty acid amide hydrolase, an enzyme involved in the breakdown of endocannabinoids (18), elicits an increase in cell proliferation.
The brain regions that have the highest densities of CB 1 receptors are the hippocampal formation, basal ganglia, and cerebellum (21). The majority, if not all, of the cannabinoid receptors in the cerebellum are located on axon terminals of cerebellar granule cells (22), glutamatergic neurons that project to cerebellar Purkinje cells. In rodents, cerebellar granule cells are generated during the first two postnatal weeks from progenitor cells in the outermost layer of the cerebellar cortex, the external granule layer (EGL). The regulation of cerebellar granule precursor proliferation, differentiation, and survival is controlled by a number of extracellular signaling cues (23). Whereas the function of the endocannabinoid system has been extensively studied in differentiated cerebellar granule cells (24 -27), its potential role in the regulation of cerebellar granule precursor (GCP) proliferation/survival has not been addressed so far.
The activation of the CB 1 receptor is coupled with the inhibition of adenylyl cyclase, the inhibition of voltage-dependent Ca 2ϩ channels, and the activation of G-protein-regulated inwardly rectifying K ϩ channels (4,28). In addition, several signaling pathways have been shown to be regulated by the cannabinoid receptors. For instance, the CB 1 receptor has been shown to regulate different members of the mitogen-activated protein kinase family, such as extracellular signal-regulated kinase (ERK) (29,30), c-Jun N-terminal kinase (31,32), and p38 (32,33). The CB 1 receptor can also activate the phosphatidylinositol 3-kinase/protein kinase B (PI3K/AKT) signaling pathway (34,35). Although the mechanisms by which the endocannabinoid system exerts most of its effects have been elucidated (4), very little is known about the signaling mechanism underlying cannabinoid-regulated neuronal precursor cell proliferation.
In the present work, we examined the presence of cannabinoid receptors in GCP during early cerebellar development; we investigated the role of the cannabinoids in the regulation of GCP proliferation and identified the pathway downstream from the CB 1 receptor implicated in the regulation of neuronal precursor proliferation.

EXPERIMENTAL PROCEDURES
Animals and Treatments-Experimental animals were C57BL/ 6J mice, wild-type (CB 1 ϩ/ϩ ) and homozygous CB 1 knock-out mice (CB 1 Ϫ/Ϫ ) (36). Experiments were performed in accordance with the Italian and European Community law for the use of experimental animals and were authorized by the bioethical committee of the University of Bologna. Pregnant mothers were isolated in single cages when pregnancy became evident. C57BL/6J, CB 1 ϩ/ϩ , and CB 1 Ϫ/Ϫ pups received two subcutaneous injections (injection volume 20 ml/kg) either of HU-210 (100 g/kg, diluted in phosphate-buffered saline) or phosphate-buffered saline for 3 days, starting at postnatal day 4 (P4). Six hours after the last administration of either HU-210 or phosphate-buffered saline, mice received a single subcutaneous injection (150 g/g body weight) of BrdUrd (5-bromo-2-deoxyuridine, Sigma) and were sacrificed after 2 h. Pups were killed by decapitation, and the cerebellum was rapidly dissected and fixed by immersion in Glyo-Fixx (Thermo Electron Corp., Waltham, MA) for 24 h.
Antisense Experiment-To silence the expression of ␤-catenin, experiments were performed by using an antisense oligonucleotide (5Ј-ggAGTTTAACCACAACAGGCAGTcc-3Ј) and, as a control, a sense oligonucleotide (5Ј-ggACTGCCTGTTGTG-GTTAAACTcc-3Ј) (41). Both oligonucleotides (Sigma-Aldrich) were phosphorothioated to make them more resistant to RNase attack. Cultures of GCPs were exposed 2 h after plating to different concentrations of the oligonucleotides for 24 h.
Immunocytochemistry and Determination of ␤-Catenin Nuclear Localization in Neurosphere Cultures-Neurospheres were harvested on microscope slides by cytospin centrifugation (212 ϫ g, 5 min, Shandon, Thermo, Dreieich, Germany). Specimens were fixed in 4% paraformaldehyde for 20 min. Blocking was done in 5% goat serum for 30 min followed by incubation with anti-␤-catenin (1:100, BD Transduction Laboratories) antibody. Detection was done with Cy3-conjugated anti-mouse antibody (1:200, Jackson ImmunoResearch Laboratories). Nuclear staining was obtained using Hoechst 33342 as previously described (40). Neurospheres were examined at 20ϫ magnification using an Eclipse TE 2000-S microscope (Nikon). Immunofluorescence analysis was performed in a blinded fashion by numerically coding each slide. Digital images were captured using NIS-Elements AR software (Nikon). Only cells with both cellular and nuclear integrity were analyzed. Cellular and nuclear integrity was assessed morphologically. Intact nuclei were defined as well circumscribed oval bodies as delineated by Hoechst staining. To assess ␤-catenin nuclear translocation, Hoechst and ␤-catenin images of the same cell were processed using NIS-Elements AR software (Nikon). The perimeter of the nucleus was traced using the Hoechst (blue) counterstaining as a guide to define the nuclear area of each cell, and the intensity of Cy3 staining corresponding to the ␤-catenin signal was quantified by determining the number of positive (bright) nuclear pixels. Approximately 150 cells were analyzed from each slide (2 slides for each condition; 3 experiments).
Immunohistochemistry and Cell Count-Cerebella from C57BL/6J, CB 1 ϩ/ϩ , and CB 1 Ϫ/Ϫ mouse pups were embedded in paraffin and cut with a microtome in 8-m-thick sections. One of eight sections was processed for BrdUrd or anti-cleaved caspase-3 (1:100, Cell Signaling Technology) immunohistochemistry as previously described (42,43). Cell count was done in the external granular layer (Fig. 6D) for BrdUrd-positive cells and in both the external granular layer and internal granule layer for cleaved caspase-3-positive cells, in three lobuli (II, III, and VI) of each sampled section. Cell number was expressed as cells/mm 2 .
Laser Capture Microdissection-Frozen 12-m-thick sections from the cerebellum of P6 C57BL/6J mice were cut with a cryostat, fixed in 70% ethanol, stained briefly with cresyl violet, and dehydrated with xylene. The oEGL of the cerebellum was microdissected from two-three sections using a laser-capture microscope (Eclipse TE 2000-U, Nikon) with the following parameters: spot size, 30 m; power, 85 milliwatts; and duration, 750 -1200 s.
Statistics-Data are expressed as mean Ϯ S.E., and statistical significance was assessed by two-way analysis of variance (ANOVA), followed by Bonferroni post hoc test or by the twotailed t test. Significance was set to p Յ 0.05. Statistical analysis was performed using GraphPad Prism 4.0.

Expression of CB 1 and CB 2 Receptors in Cerebellar Granule
Cell Precursors in Culture-Recent evidence suggests that endocannabinoid signaling plays a role in neurogenesis in the subventricular zone and subgranular layer of the hippocampal dentate gyrus (see the introduction). However, it is presently unknown whether cannabinoids are able to promote proliferation of cerebellar GCPs. To determine whether GCPs express cannabinoid receptors, we examined CB 1 protein and gene expression in cultures of GCPs from mouse pups using CB 1 antibody immunocytochemistry and quantitative real-time PCR (RT-qPCR). About 98% of total cells were faintly but clearly double labeled with antibodies for CB 1 and N-CAM (a neuronal marker) 24 h after cell plating (DIV 1) (Fig. 1A), indicating that CB 1 receptors were expressed by neuronal precursors. On the other hand, CB 1 immunoreactivity was not localized in glial fibrillary acidic protein-positive cells (data not shown), indicating that CB 1 -positive cells were not astrocytes. Actively dividing cells, recognizable by the condensed chromosomes typical of the late M phase of the cell cycle (white arrows in Fig. 1A), were immunoreactive for the CB 1 receptor, indicating that dividing GCPs were, at this stage, already expressing CB 1 receptors. Because the expression of the Ki-67 protein is strictly associated with cell proliferation, double immunostaining for CB 1 and Ki-67 observed in GCP cultures at DIV 1 further confirmed the expression of CB 1 receptor in proliferating GCPs (Fig. 1C, arrowhead). Differentiated cerebellar granule cells after 7 days in culture (DIV 7) showed an intense CB 1 -membrane staining overlapping the N-CAM stain (Fig. 1A), which is consistent with a high expression of CB 1 receptors in differentiated neurons in the adult cerebellum (21). As expected, GCP cultures from CB 1 -deficient mice (36) did not show immunostaining for the CB 1 receptor (Fig. 1B). Confirming the results obtained by CB 1 immunocytochemistry, RT-qPCR showed that CB 1 receptors were expressed in GCPs at DIV 1, even though their level of expression was lower (10 times less) compared with differentiated cerebellar granule cells at DIV 7 (Fig. 1D).
To further characterize the developmental pattern of CGPs in culture, we examined the expression of two genes, N-Myc and GABA A receptor, in cultures at DIV 1 and DIV 7. Although N-myc is a marker of proliferating GCPs (45), the GABA A receptor is expressed by mature CGCs (46). Accordingly, we found that N-Myc mRNA expression was higher at DIV 1 that at DIV 7 and that the opposite occurred for GABA A receptor (Fig. 1E).
Because neural stem cells from the SVZ appear to express both CB 1 and CB 2 receptors (47), we wondered whether GCPs also express CB 2 receptors. We found CB 2 receptor transcripts were present in DIV 1 and DIV 7 granule cell cultures but at levels much lower than those of the CB 1 receptors (Fig. 1D). Although CB 2 receptor expression slightly increased from DIV 1 to DIV 7, it remained at a level that was 10 4 times lower than that of CB 1 receptors (Fig. 1B). Pharmacological studies suggest the existence of additional receptors for cannabinoids. For example, arachidonylethanolamide and the CB 1 -selective antagonist rimonabant (SR141716A) bind to the TRPV1 vanilloid receptor, which is also named transient receptor potential vanilloid channel 1 (48). We found that no transcripts for TRPV1 were detectable in either DIV 1 or DIV 7 granule cell cultures (data not shown). Based on the low expression levels of the CB 2 receptors and absence of TRPV1 receptors, cannabinoids should act on GCPs mainly through CB 1 receptors.

Activation of the CB 1 Receptor Increases Proliferation of Cerebellar Granule
Precursors-To test whether CB 1 and/or CB 2 receptor agonists affect GCP proliferation, we determined the effects of different concentrations of HU-210 (a CB 1 and CB 2 agonist) and JWH-133 (a CB 2 -selective agonist) on proliferation of GCPs. Cell proliferation was evaluated through BrdUrd immunohistochemistry. Under control conditions, proliferating GCPs represented ϳ8 -10% of total cell number. In cultures treated with HU-210 for 24 h, the number of proliferating cells increased up to 40 -50% at concentrations of 0.5-2.5 M (Fig.  2B). By contrast, treatment with JWH-133 had no effect on cell proliferation even at the highest tested concentration (Fig. 2B), suggesting that the effect of HU-210 on cell proliferation was specifically mediated by CB 1 receptors.
Because there is evidence that the activation of CB 1 receptors in differentiated granule cells induces apoptotic cell death (24), we counted the number of apoptotic cells in cultures treated either with HU-210 or JWH-133. Apoptotic cells were recognized based on the pyknotic appearance of Hoechst-stained nuclei (white arrow in Fig. 2A). Under control conditions, apoptotic cells represented ϳ6 -7% of total cell number. HU-210 induced an increase in apoptotic cell death (up to ϩ70 and 90%) at the two highest tested concentrations (1.0 and 2.5 M) ( To confirm that the increase in GCP proliferation was specifically due to CB 1 activation, we used SR141716A (rimonabant), a selective antagonist of CB 1 receptors. GCPs were incubated with SR141716A (2 M) for 24 h alone or in the presence of HU-210. As shown in Fig. 2, while treatment with SR141716A alone was ineffective on GCP proliferation, the effect of HU-210 on GCP proliferation was prevented by cotreatment with SR141716A (Fig. 2D). To confirm that activation of CB 1 receptors is necessary to induce an increase in GCP proliferation, we generated GCP cultures from CB 1 -deficient mice (36) and their wild-type littermates. We found that HU-210 was unable to increase GCP proliferation in CB 1 -deficient neuronal progenitors (Fig. 2E), supporting the direct impact of CB 1 receptor activation on GCP proliferation.
Intracellular Signaling Involved in HU-210-induced Cerebellar Granule Precursor Proliferation-The phosphatidylinositol 3-kinase (PI3K)/AKT signaling plays a role in self-renewal of neural stem cells (49) and can mediate mitogenic signaling during corticogenesis (50). Because cannabinoids appear to regulate cell proliferation of oligodendrocytes and neuronal precursors via G i/o , PI3K, and AKT (34,47), we sought to establish whether this pathway is involved in the CB 1 -mediated regulation of GCP proliferation.
We first analyzed the phosphorylation of AKT and its downstream target, the glycogen synthase kinase-3␤ (GSK-3␤) (51) by Western blot analysis. AKT is activated by phosphorylation of two critical residues, namely threonine 308 and serine 473 (52). We found that HU-210 increased phosphorylation of AKT, at both Thr-308 and Ser-473 residues, and of GSK-3␤ and that this action was prevented by the PI3K inhibitor wortmannin (Fig. 3, A-C). Wortmannin treatment by itself reduced AKT Thr-308 and Ser-473 and GSK-3␤ phosphorylation, suggesting that basal AKT and GSK-3␤ phosphorylation in this cell system mainly depends on PI3K. We evaluated the number of BrdUrd-positive GCPs in the presence of wortmannin plus HU-210 and found that HU-210 was unable to increase cell proliferation when PI3K was inhibited (Fig. 3D). We found that exposure to wortmannin alone slowed down GCP proliferation rate, which was decreased by ϳ20%, as compared with un- treated cells (Fig. 3D). Similar results were obtained using another selective PI3K inhibitor LY294002 (Fig. 3D). These data clearly show that CB 1 receptors exert their effects on proliferation through PI3K. We analyzed the effect of two inhibitors of AKT, BML-257, an inhibitor of AKT membrane translocation (53), and Akti-1/2, a selective non-ATP-competitive inhibitor (54), on HU-210-induced GCP proliferation. We found that both inhibitors completely blocked the effect of HU-210 on GCP proliferation (Fig. 3E). These data show that cannabinoids exert their effects on GCP proliferation through the PI3K/AKT pathway.
Because the mitogen-activated protein kinase/extracellular signal-regulated protein kinase (ERK) pathway appears to be modulated by cannabinoids in hippocampal neuronal precursors (20), we examined the possibility that in GCPs CB 1 activation affects the phosphorylation levels of ERK1/2. We found that HU-210 administration did not modify the phosphorylation levels of ERK1/2 detected by immunoblot (data not shown), indicating that ERK is not involved in the cannabinoid effect on GCP proliferation.
There is evidence that CB 1 receptors can interact with the receptor tyrosine kinase signaling (55). Mechanisms for CB 1 receptor-receptor tyrosine kinase transactivation include cleavage of membrane-bound precursor proteins such as EGF (55). We found that treatment with PD158780, a selective EGF receptor inhibitor, did not reduce the HU-210-induced GCP proliferation (data not shown), suggesting that transactivation of the EGF pathway is not involved in the CB 1 -induced cell proliferation.
HU-210 Induced ␤-Catenin Accumulation through PI3K/ AKT Activation-In the mitogenic signaling pathways Wnt and fibroblast growth factor-2, the phosphorylation-induced inactivation of GSK-3␤ leads to nuclear translocation of ␤-catenin (56,57). In the nucleus, ␤-catenin binding to TCF/LEF converts these transcriptional repressors into activators and results in up-regulation of a variety of genes important in a wide variety of developmental events (58).
GCPs treated with HU-210 for 1.5 h showed an increase in ␤-catenin immunoreactivity in the nuclear compartment (Fig.  4A). Quantification of ␤-catenin protein levels in the nuclear fraction showed an increase by ϳ100% following HU-210 treatment (Fig. 4B). Co-treatment with BML-257 completely prevented the effect of HU-210 on ␤-catenin nuclear accumulation (Fig. 4B), suggesting the involvement of AKT in ␤-catenin nuclear translocation. In the nucleus ␤-catenin interacts with transcription factors of the LEF/TCF family to induce changes in the expression of cell-cycle genes, such as cyclin D1 (59). Consistent with this, we found that cyclin D1 mRNA expression was significantly increased in GCPs after HU-210 treatment (Fig. 4C).
To define whether ␤-catenin is an essential gene by which CB 1 induces GCP proliferation, in cultures of GCPs, we suppressed its expression with antisense oligonucleotides (AS), using sense oligonucleotides (S) as control (41). We found that 24-h treatment with AS but not with S markedly decreased the expression of ␤-catenin in a dose-dependent manner (Fig. 4D). We then evaluated GCP proliferation after 24 h of co-treatment with HU-210 and different concentrations of AS or S. Co-treatment with AS, but not with S, completely blocked the effect of HU-210 on GCP proliferation (Fig.  4E), indicating the involvement of ␤-catenin in the cannabinoid effect on GCP proliferation.
To establish whether the CB 1 -dependent activation of the AKT/GSK-3␤/␤-catenin system, observed in GCPs, is also present in other neuronal precursor types, we used clonally expanded neurospheres derived from neuronal precursors from the SVZ of newborn mice. As previously reported (18), we found that HU-210 increased neurosphere generation (data not shown) and neuronal precursor proliferation (Fig. 5A). The effect of CB 1 activation on nuclear accumulation of ␤-catenin was subsequently examined. Treatment with HU-210 for 1.5 h increased the nuclear accumulation of ␤-catenin (Fig. 5, B and C) up to 60% (Fig. 5D). This effect was completely prevented by the selective CB 1 antagonist SR141716A (Fig. 5D), ruling out the involvement of the CB 2 receptors or other nonspecific mechanisms. To explore the mechanism by which HU-210 induces ␤-catenin nuclear translocation in precursors from the SVZ, we examined whether nuclear translocation of ␤-catenin was dependent on AKT signaling. To this purpose, we cotreated neurospheres with BML-257 and HU-210. BML-257 completely inhibited the nuclear translocation of ␤-catenin (Fig. 5D), indicating that in precursors from the SVZ CB 1 -induced nuclear translocation of ␤-catenin was also mediated by the AKT signaling pathway.

HU-210 Promotes Cerebellar Granule Precursor Proliferation
during Cerebellar Development-In mouse cerebellum, the production of granule neurons lasts from birth to the second postnatal week, with a peak during the first postnatal week (60). Newborn cells derive from GCPs located in the EGL. The EGL can be subdivided into two morphologically distinct zones, the outer EGL (oEGL) (Fig. 6, A and B), which is formed by actively dividing GCPs, and the inner EGL (iEGL) (Fig. 6, A and B), which mainly contains pre-migratory postmitotic cells (61,62).
To confirm results obtained in culture, we dissected by the laser capture technique the oEGL from the cerebellum of P6 mouse pups and analyzed CB 1 expression by qRT-PCR. We found that CB 1 receptors were expressed by GCPs located in the oEGL (Fig. 6C). The expression levels were lower than in whole cerebellar extracts (Fig. 6C), which is in line with the  lower levels of CB 1 receptors found in vitro in undifferentiated (DIV 1) versus differentiated (DIV 7) granule cells (Fig.  1, A and D).
To establish whether activation of CB 1 receptors during cerebellar development induces an increase in neuronal proliferation similar to that observed in vitro, mouse pups were injected with vehicle or HU-210 (100 g/kg, intraperitoneal; two daily injections for 3 days). To evaluate cell proliferation, animals received one BrdUrd injection at the end of treatment. Two hours after BrdUrd injection, most of the BrdUrd-positive cells were located in the oEGL and only scattered BrdUrd-positive cells were present in the iEGL (Fig. 6D). Estimate of the density of BrdUrd-positive cells in the EGL showed that HU-210treated mice had more (ϩ50%) BrdUrd-positive cells than control mice (Fig. 6, D and E).
The CB 1 Ϫ/Ϫ mouse is an ideal model to test the role of the endocannabinoid/CB 1 system on GCP proliferation. To establish the physiological role of this system we compared granule cell proliferation in neonate CB 1 Ϫ/Ϫ mice and wild-type littermates. We found that in mice without CB 1 receptors GCP proliferation was significantly smaller (Ϫ18%) than in wild-type littermates (Fig. 6, D and E), indicating that the endocannabinoid system is physiologically involved in regulation of GCP during cerebellar development. HU-210 treatment in neonate CB 1 Ϫ/Ϫ mice had no effect on GCP proliferation (Fig. 6E), confirming that the CB 1 receptor is involved in the HU-210-promoted neuronal proliferation.
Because there is evidence that cannabinoid administration increases apoptotic cell death (63), we counted the number of apoptotic cells in the cerebellum of HU-210-treated and untreated mice. Apoptotic cells were recognized based on cleaved caspase-3 immunostaining. At this stage of development, there were very few apoptotic cells in both the EGL and IGL of untreated mice. Estimate of the density of apoptotic cells in these two layers showed no significant differences between HU-210-treated and untreated mice (Fig. 6F), indicating that the dose of HU-210 used here has no adverse effects on cell survival. No differences were found in apoptotic cell death among CB 1 Ϫ/Ϫ , HU-210-treated CB 1 Ϫ/Ϫ mice, and wild-type littermates (Fig. 6F), suggesting that lack of the CB 1 receptor does not compromise cell survival.
Activation of the CB 1 Receptor in Vivo Promotes ␤-Catenin Nuclear Translocation in Cerebellar Granule Precursors through the AKT/GSK-3␤ Pathway-To establish whether CB 1 receptors activate the AKT/GSK-3␤ pathway also in vivo, the levels of p-AKT (Ser 473 ) and p-GSK-3␤ were quantified by Western blot, in cerebellar homogenates (Fig. 7A), and visualized by immunostaining, in cerebellar sections (Fig. 7, B and C), from HU-210-treated and untreated mice. We found an increase in the phosphorylation levels of both AKT and GSK-3␤ (Fig. 7, A-C), which confirms results obtained in vitro.
We then analyzed ␤-catenin expression and found that it was considerably higher in cerebellar extracts from mice treated with HU-210 with respect to untreated mice (Fig. 7D). To examine more in detail the increase of ␤-catenin at the cellular level, cerebellar sections were immunostained with a ␤-catenin antibody (Fig. 7, E and F). We found that ␤-catenin expression was considerably higher in the EGL of mice treated with HU-210 (Fig. 7F) compared with untreated mice (Fig. 7E), which is in agreement with the increased phosphorylation levels of GSK-3␤ in treated mice. Although in untreated animals ␤-catenin had mainly an extranuclear location (Fig. 7E), in HU-210treated mice it was present both at the nuclear and extranuclear level (Fig. 7F, arrows).

DISCUSSION
The process of neurogenesis is modulated by numerous neurobiological factors, including the endocannabinoid system (15)(16)(17)(18)(19)(20). Although it is well established that the endocannabinoid system is involved in the modulation of precursor proliferation in the hippocampus and VZ/SVZ, it is presently unknown whether it may exert a more widespread action during brain development, modulating the proliferation of neuronal precursors in various brain neurogenic regions. Moreover, very little is known about the molecular mechanisms by which endocannabinoids modulate neuronal precursor proliferation. Our results demonstrate that cannabinoids modulate proliferation also of cerebellar neuronal precursors, via the CB 1 receptor. Our study additionally shows that, in neuronal precursors from both the cerebellum and SVZ, CB 1 receptors promote proliferation through the PI3K/ AKT/GSK-3␤/␤-catenin pathway, suggesting that this pathway plays a pivotal role in the CB 1 -dependent modulation of neuronal proliferation. 1 immunoreactivity is abundant in the adult cerebellum (21), where CB 1 receptors reside primarily in the presynaptic terminals of parallel fibers arising form granule cells located in the internal granular layer (21). Here we addressed the question whether cannabinoid receptors are expressed by the precursors of cerebellar granule cells. We report, for the first time, that GCPs express CB 1 receptors, similarly to the precursors of the hippocampus and VZ/SVZ (15)(16)(17)(18)(19)(20). CB 2 receptors appear to be expressed by differentiated cerebellar granule cells, though at a lower level than CB 1 receptors (64). This is in agreement with our finding of very low levels of CB 2 receptors in cultures of granule cells (see Fig. 1B). Administration of HU-210, a potent agonist of CB 1 and CB 2 receptors, increased GCP proliferation rate. This effect was mediated by CB 1 receptors, because it was prevented by cotreatment with SR141716A, a selective CB 1 receptor antagonist. This conclusion is strengthened by the observation that in CB 1 -deficient mice treatment with HU-210 was unable to increase GCP proliferation. Although CB 2 receptors have been recently described to promote proliferation of neural stem cells derived from the cortex of mouse embryos (47), we found that a CB 2 -selective agonist did not affect GCP proliferation. Taken together these data suggest that the cannabinoid system regulates GCP proliferation through CB 1 receptors.

CB 1 Receptors Are Expressed by Cerebellar Granule Cell Precursors and Their Activation Promotes Proliferation-CB
The fact that GCP proliferation, neither in CB 1 Ϫ/Ϫ granule cultures nor in control cultures treated with SR141716A, was impaired suggests that endocannabinoid production is a noncell-autonomous process for cerebellar granule cell precursors.

CB 1 Receptors Modulate Cerebellar Granule Precursor Proliferation through the PI3K/AKT/GSK-3␤/␤-Catenin Pathway-
We found that HU-210 treatment increased phosphorylation of AKT and GSK-3␤ and that inhibition of PI3K and AKT suppressed the CB 1 -mediated proliferation increase of GCPs. These data indicate the involvement of the PI3K/AKT/GSK-3␤ pathway in regulation of GCP proliferation. Our results are in agreement with evidence in other cellular systems, showing that cannabinoids regulate cell proliferation of oligodendrocytes and cortical neuronal precursors via G i/o , PI3K, and AKT (34,47). We found that ERK phosphorylation levels did not change following activation of CB 1 receptors, suggesting that this kinase is not involved in the CB 1 -mediated regulation of GCP proliferation. The CB 1 -dependent proliferation increase of embryonic hippocampal neuronal precursors requires ERK activation (20), suggesting that the pro-proliferative pathways downstream from CB 1 receptors may include ERK in some types of precursor cells.
␤-Catenin, an important mediator of the canonical Wntsignaling pathway (56, 65) is a multifunctional protein, the stability of which is mainly regulated by GSK-3␤. This is consistent with our results showing that increased phosphorylation of GSK-3␤ was accompanied by increased levels of ␤-catenin within the nuclear compartment. The observation that inhibition of AKT prevented ␤-catenin nuclear translocation following HU-210 treatment suggests that ␤-catenin belongs to a CB 1 receptor-driven signaling pathway formed by PI3K/AKT/GSK-3␤/␤-catenin.
When phosphorylated by GSK-3␤, ␤-catenin is ubiquitinated and becomes degraded by the proteasome pathway. When GSK-3␤ is inactivated through phosphorylation, ␤-catenin is stabilized and accumulates in the cytosol and can be translocated to the nucleus where it functions as a transcriptional regulator (66,67). Recent evidence shows that ␤-catenin, besides being involved in a variety of functions (68,69), plays an important role in regulating proliferation of neural stem cells. In these cells, ␤-catenin acts downstream from the canonical Wnt-signaling pathway (65, 70 -72) and appears to increase proliferation by decreasing cell cycle exit (71). A recent study suggests that fibroblast growth factor-2 regulates neural stem cell proliferation via ␤-catenin signaling (73). The finding that inhibition of ␤-catenin expression prevented the CB 1 -induced proliferation increase of GCPs indicates that ␤-catenin regulates proliferation of GCPs by acting downstream from a cannabinoid-signaling pathway. This mechanism appears to be shared by different types of neuronal precursors, because ␤-catenin nuclear translocation following activation of CB 1 receptors was present not only in GCPs but also in neuronal precursors derived from SVZ. Our data additionally suggest that cyclin D1 up-regulation may be one of the mechanisms by which ␤-catenin regulates proliferation of neuronal precursors.
Taken together, our results suggest a plausible mechanism for the regulation of neuronal precursor proliferation by cannabinoids, via ␤-catenin. As summarized by Fig. 8, CB 1 receptor activation increases PI3K/AKT activity. Following AKT-mediated phosphorylation of GSK-3␤, ␤-catenin is stabilized and translocates to the nucleus where it functions as a transcriptional regulator, modulating the expression of genes, such as cyclin D1, involved in the regulation of cell proliferation.
Role for the Endocannabinoid System on GCP Proliferation during Cerebellar Development-Current findings in vivo show that (i) HU-210 administration to neonate mice increased GCP proliferation, (ii) proliferation was impaired in CB 1 Ϫ/Ϫ mice, and (iii) proliferation did not increase in CB 1 Ϫ/Ϫ mice following HU-210 treatment. All these data indicate that the endocannabinoid system plays a role in the control of neuronal precursor proliferation during cerebellar development through CB 1 receptors. Previous studies show that endocannabinoid signaling is instrumental for cortical and hippocampal neurogenesis (16,19). Our findings in the cerebellum are in agreement with these studies and additionally suggest that endocannabinoids may have a widespread effect in the regulation of brain development. Concerning the impact of the endocannabinoid system on cerebellar development, our data show that granule cell proliferation was relatively mildly impaired in CB 1 Ϫ/Ϫ mice (Ϫ18%), suggesting that, although this system physiologically contributes to the regulation of cerebellar neurogenesis, it may not be the major actor in this process. Yet, the finding that agonists of CB 1 receptors are able to powerfully increase neuronal precursor proliferation (Fig. 2, A and B) (15)(16)(17)(18)(19)(20) appears of relevance in the context of brain pathophysiology, because it provides a rational basis for studies aimed at establishing whether CB 1 receptor agonists may be employed in brain pathologies characterized by defects in neurogenesis/neurodegeneration. In this connection, it seems important to note that in mice treated with HU-210 apoptotic cell death was not affected in the cerebellum (current study) and hippocampus (20), suggesting that agonists of cannabinoid receptors can be employed to increase neurogenesis without concomitant aversive effects on cell survival.
In conclusion, our results support the notion that endocannabinoids constitute a new group of lipid signaling cues involved in the control of neuronal precursor proliferation and show that they may participate in the control of neurogenesis during early phases of brain development.