Cannabinoid CB1 receptor gene inactivation in oligodendrocyte precursors disrupts oligodendrogenesis and myelination in mice

Cannabinoids are known to modulate oligodendrogenesis and developmental CNS myelination. However, the cell-autonomous action of these compounds on oligodendroglial cells in vivo, and the molecular mechanisms underlying these effects have not yet been studied. Here, by using oligodendroglial precursor cell (OPC)-targeted genetic mouse models, we show that cannabinoid CB1 receptors exert an essential role in modulating OPC differentiation at the critical periods of postnatal myelination. We found that selective genetic inactivation of CB1 receptors in OPCs in vivo perturbs oligodendrogenesis and postnatal myelination by altering the RhoA/ROCK signaling pathway, leading to hypomyelination, and motor and cognitive alterations in young adult mice. Conversely, pharmacological CB1 receptor activation, by inducing E3 ubiquitin ligase-dependent RhoA proteasomal degradation, promotes oligodendrocyte development and CNS myelination in OPCs, an effect that was not evident in OPC-specific CB1 receptor-deficient mice. Moreover, pharmacological inactivation of ROCK in vivo overcomes the defects in oligodendrogenesis and CNS myelination, and behavioral alterations found in OPC-specific CB1 receptor-deficient mice. Overall, this study supports a cell-autonomous role for CB1 receptors in modulating oligodendrogenesis in vivo, which may have a profound impact on the scientific knowledge and therapeutic manipulation of CNS myelination by cannabinoids.


INTRODUCTION
During developmental CNS myelination, oligodendrocyte progenitor cells (OPCs) proliferate, migrate, and differentiate into mature myelinating oligodendrocytes (OLs), which generate the myelin sheath internode and, thereby, interact with axons to organize the nodal, paranodal, and juxtaparanodal regions [1,2]. Thus, the OL developmental program is temporally and spatially controlled by a high number of extracellular signals that coordinately regulate essential intracellular signaling pathways and their downstream transcriptional programs. Alterations in essential genes modulating OPC differentiation and/or OL maturation cause myelination defects, which is translated into neuronal dysfunction and, eventually, behavioral alterations in mice [2][3][4][5]. Despite the identification of a high number of these regulatory signals, the whole molecular network that controls developmental oligodendrogenesis and CNS myelination has remained incomplete.
Almost two decades after the first evidence supporting that cannabinoid compounds modulate OL development and CNS myelination [6,7], several pharmacological studies have shown that synthetic cannabinoids [8], phytocannabinoids [9], and endocannabinoids (eCBs) [10] modulate oligodendrogenesis and CNS myelination. Elevated levels of the endocannabinoid 2-arachidonoylglycerol (2-AG) upon pharmacological inhibition of the 2-AG-degrading enzyme monoacylglycerol lipase (MAGL) enhances OL development in cultured OPCs [11], both at embryonic stages in mice [10] and in the Theiler's murine encephalomyelitis virus (TMEV) progressive mouse model of multiple sclerosis (MS) [12,13]. Moreover, administration of the phytocannabinoid Δ 9 -tetrahydrocannabinol (THC) [9] or of synthetic cannabinoids [8] promotes OPC differentiation and developmental CNS myelination, as well as OL regeneration and functional CNS remyelination upon cuprizone-induced demyelination [14]. Moreover, the phytocannabinoid cannabidiol (CBD) prevents hypoxia/ischemia-induced hypomyelination in newborn rats [15]. Of note, collectively, these studies have only been based on systemic pharmacological approaches. Owing to the abundant expression of CB 1 receptors in various neuronal and glial cell populations [16], the cell-autonomous action of cannabinoid compounds in OPCs to modulate OL development in vivo has remained unexplored. Thus, deciphering the cellular neurobiology of the eCB system may help to identify the cellular targets of cannabinoids under physiological or pathophysiological settings.
Here, by using genetic mouse models aimed to inactivate CB 1 receptor gene expression selectively in OPCs, we show that CB 1 receptors exert an essential function in modulating OPC differentiation and oligodendrogenesis during postnatal myelination in vivo. We found that selective depletion of CB 1 receptor signaling in OPCs, by altering the RhoA/ROCK signaling pathway, prevents cell differentiation, perturbs oligodendrogenesis and postnatal myelination, and causes hypomyelination as well as motor and cognitive defects at young adult ages. Moreover, pharmacological inactivation of ROCK in vivo overcomes the defects in oligodendrogenesis and functional CNS myelination of CB 1 receptor deficient OPCs. This study supports an essential role for CB 1 receptors in modulating OPC functions and CNS myelination, which may contribute to understanding the complex molecular network that controls CNS myelination.

Selective CB 1 receptor gene inactivation in OPCs in vivo
To interrogate a cell-autonomous role for CB 1 receptors in the modulation of oligodendrogenesis and CNS myelination in vivo, we generated a new mouse line by crossing the CB 1 f/f [17] mouse line with the Ng2-CreERT2 [18] and Rosa-stop-Ai6 [19] mouse lines. This would conceivably deplete CB 1 receptors gene (Cnr 1 ) expression selectively in OPCs upon tamoxifen (TAM) administration, and ZsGreen1 fluorescently label OPCs to track OL differentiation along the process of CNS myelination (Fig. S1A). We found approximately 77.7% of recombination efficiency in Ng2/Ai6-CB 1 KO and Ng2/Ai6-CB 1 HET mice by quantifying the percentage of oligodendroglial-lineage Olig2 + cells that expressed the Ai6 recombinant transgene by immunofluorescence analysis in the developing corpus callosum (CC) (Fig. S1B). We verified the recombination in the CB 1 receptor locus by genomic DNA analysis of Ng2/Ai6-CB 1 KO and Ng2/Ai6-CB 1 HET CC FAC-sorted cells at postnatal day 10 (P10) (Fig. S1C, D) [20]. We also confirmed CB 1 protein depletion in Ng2/Ai6-CB 1 KO CC isolated cells at P10 by immunofluorescence (Fig. S1E).

CB 1 receptor gene inactivation in OPCs disrupts postnatal oligodendrogenesis
To study the role for CB 1 receptors in oligodendrogenesis, we induced TAM-driven recombination in Ng2/Ai6-CB 1 KO and Ng2/ Ai6-CB 1 HET mice at P6 and P7 and performed a differentiationstate analysis of the OL-lineage populations by immunofluorescence in the CC at P15 and at a young adult age (P60) (Fig. 1A). Analysis revealed a higher proportion of NG2 + OPCs and a reduced proportion of CC1 + OLs within the recombinant Ai6 + population in the CC of Ng2/Ai6-CB 1 KO mice compared to their controls (Ng2/Ai6-CB 1 HET), thus pointing to a blockade of OPC differentiation ( Fig. 1B and Fig. S2A). Similar data were obtained when analyzing the Olig2 + cell population in the Ng2-CB 1 KO mouse line at P15 (Fig. 1C and Fig. S2B). Ng2-CB 1 KO mice also showed a reduced density of CC1 + OLs compared to their respective controls (Ng2-CB 1 WT mice) at both ages (Fig. 1C). We did not find differences in Olig2 + cell densities between Ng2-CB 1 KO mice and their control Ng2-CB 1 WT littermates at both ages (Fig. 1C), indicating that CB 1 receptor gene inactivation in OPCs disrupts cell differentiation without affecting OPC cell survival. We provided further support of these data by generating another OPC-specific mouse line upon crossing Pdgfrα-Cre [21] and CB 1 f/f animals. Pdgfrα-CB 1 KO mice also showed a reduced CC1 + OL cell density as well as a reduced proportion of CC1 + cells among Olig2 + cells in the developing CC compared to their control Pdgfrα-CB 1 WT mice at P15 (Fig. 1D and Fig. S2C). Taken together, these results indicate that CB 1 receptors regulate OPC differentiation during postnatal development in a cell autonomous manner.
Next, we studied the maturation state of the OL population in the CC of OPC-CB 1 receptor-deficient mice. Immunofluorescence analysis of OL (CC1) and mature myelinating OL markers (glutathione S-transferase P, GSTπ; myelin-associated glycoprotein MAG) showed a reduced density of GSTπ + CC1 + or MAG + mature myelinating OLs (Fig. S3A, B), together with a reduced proportion of myelinating GSTπ + OLs within CC1 + OLs cells, in Ng2-CB 1 KO mice compared to their Ng2-CB 1 WT control mice at P15 and P60 (Fig. 1E). We also found a reduced proportion of CC1 + GSTπ + cells within the Ai6 + population in the CC of Ng2/Ai6-CB 1 KO mice compared to their Ng2/Ai6-CB 1 HET controls ( Fig. 1B and Fig. S3C), thus suggesting that CB 1 receptors modulate not only OPC differentiation but also OL maturation. Altogether, these findings indicate that CB 1 receptor ablation in OPCs prevents cell differentiation and disrupts oligodendrogenesis during the critical period of postnatal myelination.

CB 1 receptor gene inactivation in OPCs disrupts postnatal CNS myelination
To address whether the observed defects in oligodendrogenesis shown by Ng2-CB 1 KO and Pdgfrα-CB 1 KO mice impact the myelination process, we analyzed myelination in the CC of these mice and their respective controls at P15. Western blot analysis of dissected CC extracts revealed reduced levels of myelin-associated proteins, such as MAG, myelin oligodendrocyte glycoprotein (MOG), and myelin basic protein (MBP) from Ng2-CB 1 KO and Pdgfrα-CB 1 KO mice compared to their controls (Ng2-CB 1 WT and Pdgfrα-CB 1 WT mice respectively) ( Fig. 2A). Fluoromyelin staining, as well as MBP and proteolipid protein (PLP) immunofluorescence analysis in the developing CC, supported the hypomyelinated phenotype of Ng2-CB 1 KO mice compared to their respective controls (Fig. 2B, C). Moreover, ultrastructural analysis by electron microscopy showed a reduced density of myelinated axons in CC of Ng2-CB 1 KO mice compared to their controls (Fig. 2D). These results indicate that CB 1 receptor deficiency in OPCs at early postnatal ages disrupts developmental CNS myelination. CB 1 receptor gene inactivation in OPCs causes hypomyelination in young adult mice We next studied whether CB 1 receptor deficiency in OPCs at early postnatal ages leads to alterations of myelin levels at P60. Tamoxifen-driven recombination was induced at P6 in Ng2-CB 1 KO and their controls, and myelination was evaluated in the CC at P60. Sudan black staining unveiled a hypomyelinated CC in adult Ng2-CB 1 KO mice compared to controls ( Fig. 2E and Fig. S4B). Western blot and real time-PCR analyses revealed reduced protein and mRNA levels of myelin-associated proteins, such as MAG, MOG and MBP, in CC extracts from Ng2-CB 1 KO mice compared to their controls ( Fig. 2F and Fig. S4A). Furthermore, we also found reduced myelin-related protein levels in other CNS areas of Ng2-CB 1 KO mice, such as the cerebellum, the hippocampus, and the medial prefrontal cortex, compared to Ng2-CB 1 WT mice (Fig. 2F). Then, we analyzed whether the hypomyelinated phenotype observed upon OPC-CB 1 receptor depletion impacts node/ paranode density or structure. Immunofluorescence analysis of nodal (Nav1.6) and paranodal (Caspr) markers in the CC at P60 revealed an altered nodal and paranodal length, with equal node density in Ng2-CB 1 KO mice compared to their controls (Fig. 2G), confirming a defect in myelination in Ng2-CB 1 KO mice.      . B Immunofluorescence analysis of proteolipid protein (PLP) and MBP in Ng2-CB 1 KO and Ng2-CB 1 WT mice at P15. C Fluoromyelin staining in the CC at P15. D Electron microscopy analysis and quantification of myelinated axon density in Ng2-CB 1 KO and Ng2-CB 1 WT mice at P15. E Representative images of Sudan black staining in the CC of Ng2-CB 1 KO and Ng2-CB 1 WT mice at P60. G Immunofluorescence analysis and quantification of node density and paranode (Caspr) and node (Nav1.6) length in the CC of Ng2-CB 1 KO and Ng2-CB 1 WT at P60. Data are shown as mean ± SEM. n = 3-5 for A, n = 3 for D, E, n = 3-4 for E, F, and n = 4 for G, independent data points used per each experimental group. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 vs Ng2-CB 1 WT mice, by two-tailed unpaired Student's t-test for A, D, E, F and G. Scale bars, 80 µm for B, 100 µm for C, 4 µm for D, 600 µm for F, and 3 µm for G.
A. Sánchez-de la Torre et al.

CB 1 receptors modulate OPC differentiation through the RhoA/ROCK signaling axis
The Ras homolog family member A (RhoA) is a multifunctional small GTPase protein that has recently emerged as a central control point of OPC differentiation and OL maturation [22,23]. RhoA, by stabilizing actin fibers, regulates cytoskeletal reorganization, thereby modulating the morphological changes necessary for OPC differentiation and OL maturation. Thus, activation of the RhoA/Rho-associated protein kinase (ROCK) signaling axis impedes OPC differentiation, and RhoA/ROCK inactivation promotes OPC differentiation in vitro [22,24,25], or under inhibitory conditions, such as in the presence of myelin [26], or following spinal cord injury [27] or hypoxia [28] in mice. Thus, we studied whether the aforementioned deficits in cell differentiation of CB 1 receptor deficiency in OPCs may be caused by an altered RhoA signaling. Western blot analysis revealed increased RhoA protein levels and increased activation of the RhoA downstream target ROCK, paralleled by a reduced activation of cofilin, in CC extracts from Ng2-CB 1 KO mice (Fig. 4A). Immunofluorescence analysis in the developing CC corroborated the increased RhoA protein levels in Ng2/Ai6-CB 1 KO OPCs compared to their Ng2/Ai6-CB 1 HET controls (Fig. 4B). In line with these observations, acute THC administration (3 mg/kg, i.p.) to Ng2-CB 1 WT mice reduced RhoA protein and ROCK activation levels, and increased cofilin activation levels, in CC extracts (Fig. 4C). Immunofluorescence analysis in the developing CC of Ng2-dsRed mice confirmed the reduced RhoA protein levels in THC-treated Ng2-dsRed + OPCs compared to Vehtreated mice (Fig. 4D).
To address the cell-autonomous action of THC in targeting RhoA selectively in OPCs through CB 1 receptors activation to drive oligodendrogenesis, we administered THC or Veh to Ng2-CB 1 KO mice. THC failed to alter RhoA protein levels and activation of its downstream targets in CC extracts from Ng2-CB 1 KO mice compared to Veh-treated animals (Fig. 4C). Similarly, immunofluorescence (Fig. 4E, G) and western blot (Fig. 4F) analyses revealed that THC enhanced OL differentiation and increased myelin-associated protein levels in the CC of Ng2-CB 1 WT or Ng2/ Ai6-CB 1 HET control mice, but not in their Ng2-CB 1 KO or Ng2/Ai6-CB 1 KO littermates. These results indicate a cell-autonomous action of THC in inducing OPC differentiation and myelination in vivo Data are shown as mean ± SEM. n = 8-25 independent data points used per each experimental group. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 vs Ng2-CB 1 WT mice, by two-tailed unpaired Student's t-test, twotailed unpaired Student's t-test with Welch's correction or Man Whitney test.
A. Sánchez-de la Torre et al.
through CB 1 receptors activation, at least in part, by targeting RhoA protein levels.
Modulation of protein stability by proteasomal degradation plays an important role in RhoA biological functions [29]. To explore the possibility that CB 1 receptors target proteasomal degradation to modulate RhoA protein levels we used a reversible proteasome inhibitor, MG-132. Thus, pretreatment with MG-132 prevented the THC-induced reduction in RhoA protein levels in CC extracts (Fig. 5A), as well as the THC-induced OL differentiation determined by immunofluorescence analysis in the CC of Ng2/Ai6-CB 1 WT A. Sánchez-de la Torre et al. mice (Fig. 5B). We excluded the transcriptional modulation of RhoA by CB 1 receptors in OPCs, as we did not find differences in RhoA mRNA levels in CC extracts from acutely THC-treated mice compared to Veh-treated controls (Fig. 5C). THC-treated mice also exhibited a reduction in RhoA activity in CC extracts compared to Veh-treated controls (Fig. 5C), in line with the differences observed at RhoA protein levels. These results indicate that CB 1 receptors modulate OPC differentiation, at least in part, by regulating RhoA proteasomal degradation.
To further study the molecular mechanism of CB 1 receptormediated RhoA proteasomal degradation, we performed electroporation experiments in P2 pups from THC or Veh-treated mothers by using siRNAs against the three main E3 ubiquitin-protein ligases reported so far to target RhoA for proteasomal degradation, namely Smurf-1, Cullin-3 (CUL-3), and SCF FBXL19 [30][31][32][33]. First, we observed that maternal THC administration during the early postnatal period indirectly induced OPC differentiation in pupsĆ C (Fig. 5D). Then, we electroporated a control GFP reporter plasmid and quantified by immunofluorescence the percentage of CC1 + OLs within GFP + Olig2 + cells. This analysis revealed an enhanced oligodendrogenesis in siC-electroporated pups of THCtreated mothers compared to SiC-electroporated pups from Vehtreated mothers (Fig. 5E), confirming that maternal THC administration induced OPC differentiation in control pups. Moreover, we found that Smurf-1 and CUL-3 silencing prevented the THCinduced OPC differentiation in the developing CC (Fig. 5E), indicating that CB 1 receptor-mediated modulation of OPC differentiation requires E3 ubiquitin ligase-mediated RhoA proteasomal degradation.

ROCK inactivation overcomes the defects in oligodendrogenesis and functional myelination of CB 1 receptor deficient OPCs
To confirm the involvement of the RhoA/ROCK axis in the CB 1 receptor-mediated modulation of OPC differentiation, we analyzed the effect of inhibiting ROCK pharmacologically in vivo by administering the selective inhibitor Y-27632. CB 1 receptor deficiency was induced in Ng2-CB 1 KO and Ng2/Ai6-CB 1 KO and control mice, followed by Y-27632 or Veh administration (Fig. 6A). Immunofluorescence analysis in the developing CC confirmed a reduced proportion of CC1 + OLs within the Ai6 + population in Veh-treated Ng2/Ai6-CB 1 KO mice compared to Veh-treated Ng2/ Ai6-CB 1 HET animals (Fig. 6B). Importantly, ROCK inhibition induced OPC differentiation in Ng2/Ai6-CB 1 KO mice, rescuing CB 1 receptornull OPC deficiency up to the levels found in Veh or Y-27632treated Ng2/Ai6-CB 1 HET littermates (Fig. 6B). Immunofluorescence (Fig. S5) and western blot (Fig. 6C) analysis of myelin-associated proteins evidenced that ROCK blockade also restored Ng2-CB 1 KO myelin protein levels up to the levels of Ng2-CB 1 WT-Veh or Y-27632-treated mice at both, postnatal and adult ages. Notably, we also found that ROCK pharmacological blockade overcame the motor and cognitive deficits observed in Ng2-CB 1 KO mice at adult ages in the open field (Fig. 6D), beam walking (Fig. 6E), elevated plus-maze (EPM, Fig. 6F) and novel object recognition (NOR, Fig. 6G) tests. Altogether, these findings show that ROCK pharmacological inactivation overcomes the defects in oligodendrogenesis and functional CNS myelination of CB 1 receptor deficient OPCs during postnatal development in mice.

DISCUSSION CB 1 receptors modulate OPC differentiation cellautonomously
Here, by using new conditional mouse models, we provide the first evidence for a cell-autonomous role of CB 1 receptors in modulating OPC differentiation and OL development during postnatal myelination in mice. Our observations support the relevance of eCB signaling in the control of postnatal myelination through the activation of CB 1 receptors located on OPCs. However, mouse OPCs express CB 1 and CB 2 receptors at developmental stages [9], respond to THC administration in vivo (present study), and pharmacologically blocking either of the two receptors prevents THC effects [9]. Moreover, we show that THC administration to Ng2-CB 1 KO mice fails to modulate oligodendrogenesis and postnatal myelination in the CC, addressing that CB 1 receptor expression in OPCs is strictly required for cannabinoid modulation of developmental oligodendrogenesis. Thus, the involvement of CB 2 receptors may be substantiated by the existence of a functional interaction between both receptors in OPCs, such as the CB 1 -CB 2 heteromers observed in neuronal cells [34], or the non-cell autonomous modulation of oligodendrogenesis by CB 2 receptors. Therefore, our study also provides evidence for a cellautonomous action of THC administration in modulating OPC differentiation and OL development in the CC during postnatal myelination in mice, and suggests that previous pharmacological studies based on the use of synthetic cannabinoids [8], phytocannabinoids [14] or eCBs [10] may have also involved the targeting of OPCs cell-autonomously to enhance OL development during postnatal myelination.
Our results show that genetic inactivation of CB 1 receptors in OPCs not only blocks cell differentiation, but also affects OL maturation and CNS myelination in the CC. These data are in line with previous findings in which administration of cannabinoid agonists induce OPC differentiation and enhance OL maturation and CNS myelination [8,10,15]. As premature and mature OLs derived from CB 1 receptor deficient OPCs shall inherit genetic depletion of CB 1 receptor, we cannot exclude the possibility that some of the observed maturation defects arise from an altered OPC differentiation. Thus, assessing the selective action of CB 1 receptors in modulating OL maturation and myelin formation would require the generation of selective premature/mature OL Cre-dependent mouse models. Tamoxifen-driven recombination was induced at postnatal day 6 (P6-P7) in Ng2-CB 1 KO, Ng2-CB 1 WT, Ng2/Ai6-CB 1 KO, or Ng2/Ai6-CB 1 HET mice, followed by Δ 9 -Tetrahydrocannabinol (THC, 3 mg/kg) administrations the following day, for 1 time C, D or for 2 E, or 4 F, G consecutive days. A Western blot analysis for Ras homolog gene family member A (RhoA), phosphorylated RhoA/Rho-associated protein kinase 2 (pROCK 2 ), and pCofilin of dissected corpus callosum (CC) extracts from Ng2-CB 1 KO and Ng2-CB 1 WT at P15. Quantification of optical density (O.D.) B Immunofluorescence analysis of RhoA expression in the recombinant Ai6 + population in Ng2/Ai6-CB 1 KO, or Ng2/Ai6-CB 1 HET mice at P15. C Western blot analysis of RhoA, pROCK2 and pCofilin protein levels of dissected CC extracts from THC or Veh-treated Ng2-CB 1 KO or Ng2-CB 1 WT mice, at 6 hours after THC administration. D Immunofluorescence analysis of RhoA in Ng2-dsRed + OPCs at 6 h after THC or Veh administration. E-G Ng2-CB 1 KO, Ng2-CB 1 WT, Ng2/Ai6-CB 1 KO, or Ng2/Ai6-CB 1 HET mice were administered with THC or Veh for 2 E or 4 F, G days. E Immunofluorescence analysis and quantification of the percentage of CC1 + cells among the recombinant Ai6 + population. Western blot (F) or immunofluorescence (G) analysis of myelin-related proteins such as myelinassociated glycoprotein (MAG), myelin oligodendrocyte glycoprotein (MOG), and myelin basic protein (MBP) or proteolipid protein (PLP). Data are shown as mean ± SEM. n = 3 for E, n = 3-4 for A, C, n = 4 for F, independent data points were used for each experimental group. *p ≤ 0.05, **p ≤ 0.01 and ***p ≤ 0.001 vs Ng2-CB 1 WT or vehicle-treated groups, by two-tailed unpaired Student's t-test for A, two-tailed unpaired Student's or Mann Whitney test for C and F, and two-way ANOVA followed by Turkey´s multiple comparison for E. Scale bars, 20 µm for B, D and E, and 100 µm for G.
A. Sánchez-de la Torre et al.

CB 1 receptor-mediated RhoA proteasomal degradation in OPCs
The CB 1 receptor is one of the most abundant G protein-coupled receptors (GPCRs) in the mammalian brain. To date, other GPCRs, such as GPR17 [35], GPR37 [36], GPR56 [37], and GPR126 [38] have been shown to modulate OL development and/or CNS myelination. Thus, our present study adds evidence for GPCRs in modulating oligodendrogenesis and myelination. Mechanistically, GPR37 regulates OPC differentiation and CNS myelination via cAMPdependent Ras-ERK1/2 activation [39], while GPR56 modulates OPC proliferation via Gα12/13-RhoA, without affecting OL differentiation. Here, we observed that the CB 1 receptor-mediated modulation of OPC differentiation occurs by inducing Smurf-1/CUL-3-mediated RhoA proteasomal degradation. Thus, CB 1 receptors may modulate RhoA proteasomal degradation by targeting both, the inactive (GDP-bound) and the active (GTP-bound) forms of RhoA by CUL 3 and Smurf-1 respectively [29]. Indeed, pharmacological activation of CB 1 receptor has been shown to induce RhoA proteasomal degradation in migrating pyramidal neurons during mouse corticogenesis [40]. Therefore, it is plausible to speculate that CB 1 receptor-mediated RhoA degradation under those settings may involve Smurf-1 or CUL-3 actions (Fig. S6). Moreover, CB 1 receptors may exert opposing effects regarding RhoA activation. CB 1 receptors activation promote bone marrow-derived  .). B Immunofluorescence analysis and quantification of the percentage of Ai6 + cells that expressed the OL marker CC1 at 48 hours after THC administrations. C Relative levels of RhoA protein levels, activity, or mRNA levels in CC extracts at 6 hours after THC. D Maternal Tetrahydrocannabinol (THC) administration induces oligodendrocyte development in postnatal pups. THC (3 mg/kg) or Veh were administered to CD1 mouse mothers the day their pups were 1 day old, once a day for 8 consecutive days, and oligodendrogenesis was analyzed in corpus callosum at postnatal days 15. Immunofluorescence analysis and quantification of the percentage of CC1 + OLs among the Olig2 + population. E Electroporation experiments with siRNAs against Cullin-3 (CUL-3), Smurf-1 or SCF FBXL19 , and with a GFP reporter plasmid control in Pups from Veh or THC-treated mothers. Quantification of CC1 + OLs among GFP + Olig2 + cells in the subcortical white matter at P30. Data are shown as mean ± SEM. n = 3 for A and D, n = 3-4 for B, n = 3-5 for C, and n = 4-9 for E, independent data points were used for each experimental group. *p ≤ 0.05, **p ≤ 0.01 and ***p ≤ 0.001 vs vehicle-treated groups; # p < 0.05 and ### p < 0.001 vs THC-treated group, by one-way ANOVA followed by Tukey´s multiple comparisons for A and B, two-tailed unpaired Student's t-test for C and D, and by one-way ANOVA Kruskal-Wallis followed by Dunn´s test for E. Scale bar, 20 µm for B, D, E.
A. Sánchez-de la Torre et al.
macrophage migration or phagocytosis [41][42][43] by increasing RhoA activity and subsequent ROCK activation via Gαi/o, but also via Gα12/13, in neurons [44] or platelets [45]. Conversely, CB 1 receptors also inhibit carcinoma cell migration [46] by reducing RhoA activity, indicating that the differential CB 1 receptor-modulation of RhoA signaling may depend on factors such as the cell type and the pathophysiological context.

CB 1 receptors in OPCs and motor and cognitive development
Here, we also reveal that CB 1 receptor deficiency in OPCs at early postnatal ages affects the number and maturation state of the OL population in the CC in young adult mice, which is also associated with hypomyelination in several CNS regions and important behavioral deficits. In line with these results, increasing evidence has linked deregulated developmental myelination to impaired functional performance or neuropsychiatric alterations in adult mice [3,47], such as those related with motor function [48,49], memory [50], and anxiety-like behaviors [51][52][53]. Thus, it is plausible that the altered behavioral traits found in Ng2-CB 1 KO mice is due to a reduced number of myelinating OLs, and, thereby, to reduced myelin levels in CNS regions responsible for motor, memory, anxiety-like behaviors, which, in turn, would impact neuronal and whole-body functionality. In fact, alterations of the eCNB system during embryonic, postnatal, or adolescent ages evokes multiple long-lasting behavioral alterations in mice that persist in adulthood [54,55]. Moreover, THC exposure during embryonic or postnatal development has been linked to psychiatric disorders, such as depression and anxiety [56,57], spontaneous behavior, or habituation [58,59], thus suggesting that, in addition to the neuronal component, at least part of these behavioral defects are mediated by the restricted OPC CB 1 receptor population.
Conversely, on top of the observed behavioral alterations in complete CB 1 receptor-deficient mice [54], the development of   Fig. 6 Pharmacological inactivation of ROCK overcomes the defects in oligodendrogenesis and myelination of CB 1 deficient OPC. A Tamoxifen-driven recombination was induced at postnatal day 6 (P6-P7) in Ng2-CB 1 KO, Ng2-CB 1 WT, Ng2/Ai6-CB 1 KO and Ng2/Ai6-CB 1 HET mice followed by administrations of the ROCK inhibitor Y-27632 (10 mg/kg), and tissue was analyzed at P11 or P60. B Immunofluorescence analysis and quantification of the percentage of CC1 + oligodendrocytes among the recombinant Ai6 + Olig2 + population at P11. C Western blot analysis of the myelin associated proteins, myelin-associated glycoprotein (MAG), myelin oligodendrocyte glycoprotein (MOG), and myelin basic protein (MBP) in corpus callosum extracts at P60. Quantification of optical density (O.D.). D-G Behavioral analysis of Ng2-CB 1 KO and Ng2-CB 1 WT mice at P60. D Open field test. Representative images of mouse trajectories. Quantification of total distance travelled. E Beam walking test. Quantification of missteps. F Elevated plus maze (EPM) test. Quantification of the time spent in open arms. G Novel object recognition (NOR) test. Quantification of the discrimination index. Data are shown as mean ± SEM. n = 6-9 for B, n = 3 for C, 16-24 for D-G, independent data points were used for each experimental group. *p ≤ 0.05, **p ≤ 0.01 vs Veh-treated Ng2/Ai6-CB 1 HET or Ng2-CB 1 WT mice; # p ≤ 0.05, ## p ≤ 0.01 vs Veh-treated Ng2/Ai6-CB 1 KO mice or Ng2-CB 1 KO mice by two-way ANOVA followed by Tukey´s multiple comparisons. Scale bar, 20 µm in B.
A. Sánchez-de la Torre et al. mouse models with CB 1 receptors selectively deleted in dorsal telencephalic glutamatergic neurons or forebrain GABAergic neurons have contributed to dissect the populations of neuronal CB 1 receptors responsible for the modulation of motor, memory, anxiety-like behaviors [60][61][62]. To sum up, this is the first study that supports the necessary role for CB 1 receptors located on OPCs in the modulation of motor and cognitive development in mice, therefore contributing to understand the complex control of motor and cognitive development exerted by the eCB system.
The therapeutic potential of targeting OPC CB 1 receptors in demyelinating disorders A number of studies have pointed out important similarities in the effects of molecules that modulate OPC differentiation during postnatal myelination and during remyelination [63,64]. Specifically, cannabinoids modulate oligodendrogenesis under demyelinating conditions, enhancing OL regeneration, and functional CNS remyelination following cuprizone-induced demyelination [14], or in the TMEV animal model of progressive MS [12,13], thus suggesting that CB 1 receptors also modulate OPC differentiation under demyelinating conditions. In MS, there is a failure in OPC differentiation, which limits their remyelinating potential. Although the cause of this failure is not completely known, several studies have denoted the presence of inhibitory signals in the demyelinated CNS that prevent OPC to become mature myelinating cells. Indeed, inhibition of OPC differentiation or OL maturation by myelin debris [26], or by astrocytic chondroitin sulfate proteoglycans [27,65], are mediated by RhoA/ROCK axis modulation [23]. Moreover, elevated levels of the eCB 2-AG upon pharmacological inhibition of MAGL reduces astrocytic chondroitin sulfate proteoglycan production and enhances OL differentiation under inhibitory conditions [12]. Thus, it is plausible that, following demyelination, CB 1 receptor-mediated modulation of RhoA signaling controls OPC differentiation and functional remyelination in a cell-autonomous manner though OPC CB 1 receptors, but also non-cell-autonomously through astroglial CB 1 receptors. In any event, the CB 1 receptor-evoked control of RhoA signaling could be potentially targeted to promote functional recovery, which would open new avenues to the therapeutic manipulation of currently intractable demyelinating diseases.

Immunofluorescence
Brain tissue was processed as previously described [66]. Briefly, mice were perfused transcardially with 4% paraformaldehyde (PFA), and brains postfixed overnight in 4% PFA and treated with 30% sucrose before freezing. 30-μm-thick coronal free-floating brain cryosections were washed in PBS, blocked with 10% goat serum, and incubated with the indicated primary antibodies (overnight at 4°C). When needed, antigen retrieval immunostaining was performed with citric acid (10 mM, pH 6, 65°C for 30 min) or with retrieve-all antigen unmasking system 3. Acidic (Biolegend) at 92°C for 10 min. For immunofluorescence analysis in vitro, CC extracts from P10 Ng2/Ai6-CB 1 KO and Ng2/Ai6-CB 1 WT mice administered with tamoxifen (37.5 mg/kg) at P6 and P7, was dissociated, and individual cells were attached to glass slides by cytospin. After air-drying, slides were treated with 4% PFA for 10 minutes. The primary antibodies used are as follows

Confocal microscopy
Optical sections (z = 0.5 µm) of confocal epifluorescence images were sequentially acquired using a confocal laser-scanning microscope TCS-SP8 (20×, 40× and 63×, Leica DMI6000 B instrument) and Leica Application Suite X (LAS X) software. Images were acquired in the CC of each animal, a minimum of six correlative slices from a 1-in-10 series located between +0.7 and −0.5 mm from bregma were analysed. Cell counts were performed blindly using ImageJ software (NIH) in the corpus callosum (CC) and data are presented as the mean cell number per mm 2 or as a percentage of positive cells. For CB 1 inmunostaining of cytospined cells, the microscope was configured to capture 16 images following the z axis, which resulted in 8 µm thick stacks.

Electron microscopy
Mice were transcardially perfused with PBS followed by a fixative solution containing 4% PFA, 2.5% glutaraldehyde, and 0.5% NaCl in 0.1 M PB, as described [67]. Brains were postfixed overnight at 4°C and stored in 1% paraformaldehyde. Vibratome sections (100 µm) containing the CC were cut in the coronal plane on a vibratome (VT1000S, Leica (Wetzlar, Germany) and incubated in 1% OsO4, then embedded in epoxy resin overnight to polymerize at 60˚C, and then trimmed and glued onto epoxy resin capsules. Semi-thin sections (500 nm) were cut using a Power Tome ultramicrotome (RMC Boeckeler) and stained with 1% toluidine blue. Ultrathin (60 nm) sections were cut with a diamond knife (Diatome), collected on nickel mesh grids, and stained with 4% uranyl acetate and 2.5% lead citrate for electron microscope visualization. Electron microscopy images of the rostral CC were taken from randomly selected fields with a Jeol JEM Plus electron microscope at the Service of Analytical and High-Resolution Microscopy in Biomedicine of University of the Basque Country UPV/EHU. The mean number of myelinated axons was analyzed in 10 non-serial electron micrographs per animal taken systematically at a magnification of 5,000×.

Sudan black
Floating sections were mounted on to TESPA-coated glass slides, dehydrated in 70% ethanol and stained with 0.5% Sudan Black in 70% ethanol for 20 min. Excess staining was removed by washing the slides in 70% ethanol and finally rinsed with water. Samples were observed under light microscopy in a Zeiss Axioplan2 microscope.

Western blot
CC tissue was microdissected from 500-μm-thick coronal sections and proteins were extracted using RIPA buffer (SDS 0.1%, Sodium deoxycholate 0.5%, NP40 1%, NaCl 150 mM, Tris-HCl 50 mM pH8 in PBS) containing PMSF, protease inhibitors, and sodium orthovanadate. Protein samples were separated on 12% acrylamide (Bio-Rad) gels and transferred to polyvinylidene difluoride (PVDF membranes (Millipore). Membranes were placed in blocking buffer (5% w/v BSA in TBS-T + Azida 0.02%) and probed with primary antibodies overnight at 4°C. The primary antibodies used are as follows: anti-MAG (1:1000 After incubation with corresponding HRP-conjugated secondary antibody proteins were visualized using an enhanced chemiluminescence substrate mixture (ECL Plus; GE Healthcare; Santa Cruz Biotechnology; 1:5000). Band intensity of films was quantified using Adobe Photoshop software. Protein levels were normalized to the internal control α-tubulin. See full length uncropped original western blots in supplemental material.

Quantitative PCR
RNA was isolated using RNeasy Plus kit (Quiagen). cDNA was obtained with Transcriptor (Roche). Real-time quantitative PCR (qPCR) assays were performed using the FastStart Master Mix with Rox (Roche, Basel, Switzerland) and probes were obtained from the Universal Probe Library Set (Roche). Amplifications were run in a 7900 HT-Fast Real-Time PCR System (Applied Biosystems). The sequence of primers used are as follows:

Genomic recombination analysis
Recombination was tested by PCR with genomic DNA isolated from sorted Rosa-Ai6 cells by the use of the GenElute mammalian genomic DNA Miniprep Kit (Sigma-Aldrich), as previously described [68]. PCR was performed using Taq DNA polymerase (Thermo Fisher Scientific, Inc.) For the recombined CB1 allele we used the forward primer 5′-GCT GTCTCTGGTCCTCTTAAA-3′ (G50) and the reverse primer 5′-CTCCTGT ATGCCATAGCTCTT-3′ (G53) resulting in a 600 bp fragment.

RhoA activity essay
Dissected corpus callosum extracts were processed for active RhoA quantification with the G-LISA kit (Cytoskeleton, Cat#BK124) following manufacturer's guide.

Behavioral assessments
All tests were conducted during the light cycle, with uniform lighting conditions and white noise in an isolated room. Animals were acclimated to the room for 45-60 min before testing. Beam walking test: To evaluate fine motor coordination, mice were trained to cross a narrow wood beam (100 cm length, 10 and 7 mm width) [69]. Open-field test: Mice were placed in the center of an open-field arena (70 × 70 × 40 cm) and allowed to freely explore it for 5 min. Behavior was recorded with a video camera placed above and the video tracking software SMART 3.0 (Panlab, Spain) was used for analysis [70]. Novel Object Recognition test (NOR): The day before the test, mice were habituated to an empty open-field arena for 10 min. The day of the test, mice were allowed to freely explore the arena with two identical objects for 10 min and, after 2 hours, mice could explore the arena containing a familiar object and a novel object. Discrimination index was calculated as the difference in exploration time between the novel and familiar object divided from the total exploration time and expressed as a percentage. Preference Index was calculated as the percentage of exploration time spent examining the novel object over the total exploration time. Modified Y-Maze test: Mice were allowed to explore the maze with one of the three arms closed for 3 min. After an inter-trial interval of 5 min, mice could freely explore all three arms of the maze for 3 min, and the time spent in each arm was registered. Actimeter/ Locomotor activity test: Spontaneous motor activity was evaluated using an automated actimeter (Acti-Track; Panlab, Barcelona, Spain). This consisted of a 22.5 × 22.5 cm area with 16 surrounding infrared beams coupled to a computerized control unit. After 1 min of habituation, activity was recorded for a period of 5 min and data were collected with Acti-Track v2.7 software (Panlab, Barcelona, Spain). Elevated plus-maze test (EPM): Mice were placed in the center of the maze (two open arms and two closed arms of 30 ×7 cm arranged orthogonally 60 cm above the floor) and allowed to explore it for 5 min. The test was conducted under red light conditions.

Statistics
The n number of animals per group or experiments per condition is indicated in every case. The declared group size is the number of independent data points, and that statistical analysis was done using these independent data points. Studies were designed to generate groups of equal size, using randomisation and blinded analysis. The numbers illustrated represent the animals used in each of the experiments, after considering any unexpected loss of data or exclusion criterion. In some cases, experimental losses may be determined by animals receiving the wrong treatment, infections unrelated to the experiment, sampling errors (e.g., inadequate calibration of equipment, software error during acquisition), or other human error (e.g., forgetting to switch on equipment). Power analysis was conducted with IBM SPSS software (IBM France, Bois-Colombes, France). Sample sizes were based on our prior studies where similar sample sizes were adequately powered to detect significant differences. All variables were first tested for both, normality (Shapiro test or D'Agostino & Pearson normality test with p > 0.05) and homogeneity of variances (Brown-Forsythe test with p > 0.05). When comparing two groups we use unpaired two-tailed t-test for normal distribution (unpaired t-tests were Welch-corrected if needed) or Mann-Whitney test when they did not distribute normally. For comparisons of more than two groups, if data were found to be normally distributed, one-way ANOVA followed by Tukey's post hoc test was carried out. The post hoc tests A. Sánchez-de la Torre et al.
were conducted only if F in ANOVA achieved P < 0.05 and there was no significant variance in homogeneity. If data were found to not be normally distributed, then a Kruskal-Wallis one-way ANOVA with uncorrected Dunn's post hoc test was carried out. For data from Figs. 4E and 6, two-way ANOVA followed by Tukey's post hoc test was carried out. Differences with p < 0.05 between group means were considered statistically significant. All data analyses were done using GraphPad Prism 7.00.

DATA AVAILABILITY
All data generated or analyzed during this study are available in the supporting information.