The Role of Cannabinoids in CNS Development: Focus on Proliferation and Cell Death

The active principles of Cannabis sativa are potential treatments for several diseases, such as pain, seizures and anorexia. With the increase in the use of cannabis for medicinal purposes, a more careful assessment of the possible impacts on embryonic development becomes necessary. Surveys indicate that approximately 3.9% of pregnant women use cannabis in a recreational and/or medicinal manner. However, although the literature has already described the presence of endocannabinoid system components since the early stages of CNS development, many of their physiological effects during this stage have not yet been established. Moreover, it is still uncertain how the endocannabinoid system can be altered in terms of cell proliferation and cell fate, neural migration, neural differentiation, synaptogenesis and particularly cell death. In relation to cell death in the CNS, knowledge about the effects of cannabinoids is scarce. Thus, the present work aims to review the role of the endocannabinoid system in different aspects of CNS development and discuss possible side effects or even opportunities for treating some conditions in the development of this tissue.


Introduction
Cannabis sativa has been used for over 5000 years in a medicinal and recreational way (Pertwee 2006). The use of cannabis as a medicinal herb was first reported in China, approximately 2727 B.C. Since then, cannabis has been used as a medicinal agent with antinociception, anti-inflammatory, anticonvulsant, and antiemetic effects. Endogenous molecules modulating endocannabinoid system, the endocannabinoids, such as anandamide (AEA) and 2-arachidonoylglycerol (2-AG), have also been shown to have analgesic and antiemetic effects, increase appetite and inhibit the growth of tumors.
Over 100 different types of phytocannabinoids have been isolated from Cannabis sativa. The main components of the plant are CBD, with no psychoactive properties, 1 3 and THC, which is the main psychoactive component of the plant (Parker 2017). There are several other constituents much less studied, such as cannabigerol (CBG), cannabichromene (CBC), cannabidivarin (CBDV), and tetrahydrocannabivarin, among others (Russo 2019). In addition to phytocannabinoids, there are several other phytomolecules in cannabis, such as terpenes and flavonoids. Under several circumstances, the combination of these compounds leads to an improved effect compared to an isolated molecule, thus denominated the entourage effect (Koltai and Namdar 2020), also described for endocannabinoids (Pacher et al. 2020).
The use of phyto-, endo-or synthetic cannabinoids is becoming common in medicine due to the wide amount of data showing its beneficial effects for the treatment of several diseases (Baron 2018;Levinsohn and Hill 2020;Sarris et al. 2020). Its therapeutic use can be carried out through various mediators of the endocannabinoid system, including synthetic agonists and antagonists of the main cannabinoid receptors (CB1 and CB2) and modulators of enzymes such as fatty acid acyl hydrolase (FAAH) (Bonini et al. 2018). Nevertheless, the medicinal use is mainly due to the manipulation of specific cannabis compounds. The use of highly purified cannabidiol (CBD) treatment has been approved in the US for the treatment of Lennox-Gastaut syndrome, primarily for its anticonvulsant effects (Huestis et al. 2019). It was shown that a single dose of CBD is well tolerated with very few adverse effects and has no drug accumulation (Tayo et al. 2020). Corroborating these data, another study evaluated the effects of CBD and its adverse effects, despite being well tolerated; the use of CBD might cause diarrhea, nausea, headache and drowsiness as the main side effects. In addition to being well tolerated in a single dose, daily doses over 3 months were well tolerated in the treatment of epilepsy and multiple sclerosis (Taylor et al. 2018). Currently, an oromucosal spray composed of tetrahydrocannabinol (THC) and CBD is approved (in European and non-European countries) with the aim of alleviating the symptoms of multiple sclerosis. In this case, CBD and THC correspond to at least 90% of the extract, but they have other constituents that can contribute to the effects. Therefore, it is indicated that the THC/ CBD spray is well tolerated and has few side effects, with no sequelae, although there are side effects such as headache, drowsiness, loss of attention, euphoria and dizziness (Stott et al. 2013). They have also been studied in the treatment of neurodegenerative diseases, epilepsy and psychiatric illnesses, such as anxiety and schizophrenia (Billakota et al. 2019;Das et al. 2019;Sarris et al. 2020). Since some cannabinoids might be associated with anxiolytic effects even the considered "recreational use" could be related to a medicinal effect. Therefore, based on scientific evidence, medicinal cannabinoid use is becoming more common in people's daily lives.
The endocannabinoidome is involved in cell fate, proliferation, migration, axonal growth and synaptogenesis, participating throughout embryonic development, intercellular communication and neural development (Alexandre et al. 2020;Garcia-Arencibia et al. 2019;Heimann et al. 2021;Leonelli et al. 2005). In effect, interference in the endocannabinoid system during development might cause complications, such as intrauterine growth restriction, miscarriages, and risk of neonatal morbidity (Campos et al. 2017;Fernández-Ruiz et al. 2000;Galve-Roperh et al. 2006;Harkany et al. 2008;Maia et al. 2020;Prenderville et al. 2015). Exposure to cannabinoids could modify developmental events, such as proliferation, migration, differentiation and survival, in addition to changing the maturation of neurotransmitters (Navarrete et al. 2020). Several studies have shown that early exposure to THC and/or to increased levels of endocannabinoids during development might lead to several behavioral changes (Almada et al. 2020a, b;Campos et al. 2017;Philippot et al. 2019). Therefore, changes due to exogenous substances might impact CNS development.
As mentioned above, the cannabinoid field has dedicated much effort to study and establish the role of the endocannabinoid system in the physiology and pathology of the mature CNS. However, the early effects of cannabinoids in CNS development have been much less explored. Hence, the present review aims to provide information regarding the role of the cannabinoid system during early CNS development, especially on proliferation and cell death.

Search Strategy and Selection Criteria
A detailed search in Medline database was conducted for all published reviews and research data. In the present review the inclusion criteria were scientific articles or reviews mainly from 2015 to 2021, with CNS or proliferation, differentiation, cell fate, cell death, (endo) cannabinoids in the title and abstract. Although older data and reviews were included due to the unique character and little information on some subjects. Articles from both in vitro and in vivo studies were included. For in vivo studies we selected only those that used vertebrate animals, to have a closer comparison with humans.

Endocannabinoids
The endocannabinoid system is highly conserved among animal species, including humans. The main endocannabinoids, AEA and 2-AG, are lipophilic neuromodulators released independently from synaptic vesicles, thus differing 1 3 from the classical neurotransmitters. They are produced on demand by the action of calcium-dependent phospholipases and function as retrograde messengers (Fig. 1). Beyond the classic lipid endocannabinoids, a new class of peptides known as hemopressins act as modulators of cannabinoid receptors, especially as inverse agonists of CB1 receptors (Heimann et al. 2021). RVD hemopressin, an N-terminally extended form of hemopressin, acts as a negative allosteric modulator of CB1 receptors; however, it is a positive allosteric modulator of CB2 receptors. In addition to modulating classical cannabinoid receptors, it has been described that hemopressins can modulate other receptors and components of the endocannabinoid system (Riquelme-Sandoval et al. 2020).
Endocannabinoids are degraded by specific enzymes. FAAH is primarily a postsynaptic integral membrane protein that degrades AEA into arachidonic acid (AA) and ethanolamine, and it is the main catabolic enzyme for the fatty acid amide lipid (FAA) class (Fig. 2). Later, the FAAH-2 enzyme was also discovered in humans, and it has the ability to hydrolyze AEA and palmitoylethanolamide (PEA), contributing to the degradation of NAEs (Kaczocha et al. 2010;Wei et al. 2006). AEA can also be degraded by N-acylethanolamine amidase (NAAA) in AA and ethanolamide (Macarrone 2015). Knockout (KO) mice for FAAH exhibit increased endogenous cannabinoid activity and a reduced perception of pain. These effects are probably due to high levels of AEA and other NAEs, such as oleoylethanolamide (OEA) and PEA, which were 15 times higher (pmol/g) in comparison to the wild type (Cravatt et al. 2001). It has also been shown that pharmacological inhibition of FAAH increases AEA levels, which regulates nociception, confirming the analgesic effect observed in KO FAAH animals (O'Sullivan 2016). Monoacylglycerol lipase (MAGL), an endocannabinoid hydrolytic enzyme, is primarily a presynaptic cytosolic enzyme that degrades 2-AG in AA and glycerol (Fig. 1). In the brain, most (~ 85%) degradation of 2-AG is carried out by MAGL.
However, there is also a smaller participation of the α/βhydrolase domain-containing 6 (ABHD6) and α/β-hydrolase domain-containing 12 (ABHD12) enzymes ( Fig. 2) (Saario et al. 2005). These main hydrolases act in different subcellular compartments, in which ABHD12 has an extracellular orientation, while ABHD6 has a cytoplasmic and integral membrane orientation (Fig. 1). However, MAGL is localized in the cytoplasm and the soluble/peripheral membrane. This distinction in the location of hydrolase enzymes possibly explains the difference in the participation of 2-AG degradation, as ABHD6 and ABHD12 are responsible for ~ 4% and ~ 9%, respectively (Blankman et al. 2009;Gulyas et al. 2004). Muccioli et al. (2007) described another MAGL in the mouse microglial cell line BV-2. The authors verified that although this cell line did not express the original MAGL mRNA, the hydrolysis of 2-AG still occurred (Muccioli et al. 2007). Finally, FAAH can also hydrolyze 2-AG to a lesser extent into AA and glycerol (Murataeva et al. 2014). There is also the occasional participation of cyclooxygenase-2 (COX-2), lipoxygenase-12 and -15 (LOX-12/-15) and cytochrome p450 oxygenase that metabolize both AEA and 2-AG ( Fig. 2) (Parker 2017;Pinho Costa et al. 2011;Redmond et al. 2014).
Endocannabinoid components are present since the early stages of CNS development, and modulation of cannabinoid system may impact CNS development. Accordingly, an increase in the comprehension of cannabinoid function during CNS development, may lead to development of strategic cannabinoid interventions for pathological conditions.

Cannabinoid Receptors
Classically, cannabinoid molecules act through two G protein-coupled receptors, CB1 and CB2 receptors (CB1R and CB2R), regulating several intracellular pathways. These receptors are mainly coupled to the inhibitory G protein, which, in turn, decreases the activation of adenylyl cyclase, reducing the formation of cAMP. Moreover, they are mainly expressed in presynaptic terminals, and can block voltagedependent Ca 2+ channels and lead to the opening of K + channels. Thus, a classical response is the decrease of neurotransmitter release, such as GABA, glutamate, serotonin, acetylcholine and dopamine (López-Moreno et al. 2008). Although both CB1R and CB2R are present in CNS, CB1R are considered to be the most abundant metabotropic receptors in the CNS of mammals (Kendall and Yudowski 2017). Signaling in the endocannabinoid synapse. Glutamate released by presynaptic cells stimulates ionotropic and metabotropic receptors in postsynaptic cells, leading to an increase in Ca 2+ influx/ concentration. Ca 2+ and/or Gq stimulate PLC, which leads to the production of DAG, the substrate for DAGL to form 2-AG, while NAPE-PLD hydrolyses NAPE to form AEA. Endocannabinoids diffuse through the synaptic cleft, reaching the CB1 and CB2 receptors TRPV1 and GPR55, among others. The classical downstream pathway of the CB receptor occurs through Gi-protein: inhibition of AC activity, modulation of K + and Ca 2+ channels, and subsequent inhibition of NT release. The signaling is ended by endocannabinoid degradation. FAAH metabolizes AEA into AA and ethanolamine, while COX-2 can eventually generate prostamide from AEA. MAGL, ABHD12, and ABHD6 degrade 2-AG in AA and glycerol, while COX-2 is metabolized in PGE2-G. The thickness of arrows from metabolizing enzymes represents the importance in the control of AEA and 2-AG availability Although, CB1R is present at high levels in the plasma membrane of presynaptic terminals, it is also present in postsynaptic terminals in neurons, astrocytes and oligodendrocytes (Busquets-Garcia et al. 2018;Han et al. 2012). Interestingly, CB1 might also be found in subcellular compartments (Zou and Kumar 2018). In the lysosome, the CB1R increases the cytoplasmic calcium concentration through the release of internal calcium stores in addition to the lysosomal calcium per se due to the increase in lysosomal permeability (Zou and Kumar 2018). In the mitochondria, it acts directly to inhibit cellular respiration and the production of cAMP, regulating energy metabolism (Bénard et al. 2012). Finally, CB1R in the endosome can regulate β-arrestin signaling pathways (Kendall and Yudowski 2017;Zou and Kumar 2018). The above data reveal the relevance of CB1R in the functionality of the CNS. Meanwhile, they are also related to different kinds of CNS disorders, such as Huntington disease, multiple sclerosis, Alzheimer's disease, anorexia, post-traumatic stress, schizophrenia, and eating and alcohol use disorders (Cristino et al. 2020;Kendall and Yudowski 2017).
CB1R can activate several intracellular pathways, such as the signaling pathways of the three subfamilies of the mitogen-activated protein kinase MAPK, the kinase 1/2 regulated by extracellular signal (ERK1/2), the c-Jun-N-terminal (JNK) kinase and p38, which are involved in the regulation AEA synthesis-An increase in Ca 2+ stimulates NAPE-PLD to hydrolyze NAPE, producing AEA. PLC catalyzes the formation of phospho-NAE, and the phosphatases PTPN22 and INPP5D then convert phospho-NAE to AEA. NAPE is catalyzed by sPLA2 or ABHD4, generating lyso-NAPE. Then, ABHD4 catalyzes the formation of GP-NAE, which is converted to AEA by the phosphodiesterase GDE1. 2-AG synthesis-PLC mediates the hydrolysis of PIP 2 , which produces DAG, which is converted to 2-AG by DAGLα/β. Phosphatidic acid (PA) is hydrolyzed by PA phosphohydrolase, producing DAG, which is converted by DAGLα/β into 2-AG. b. AEA degradation-AEA is degraded primarily by FAAH into AA and ethanolamine, while COX-2 degrades into prostamide. 2-AG degradation-2-AG is degraded primarily by MAGL into AA and glycerol. ABHD6 and ABHD12 also degrade 2-AG into AA and glycerol, although they are less common. COX-2 degrades 2-AG into PGE2-G 1 3 of main development processes such as cell proliferation, control of cell cycle and cell death. Furthermore, it has been observed that CB1R can exert a protective effect through the activation of the PI3K/AKT pathway, which is important for cell growth and controlling cell death (Galve-Roperh et al. 2002;Del Gómez et al. 2000;Zou and Kumar 2018).
CB2R is also present in the mature CNS, even though it is less abundant than CB1R. CB2R is found in microglia, astrocytes and neurons, and it is involved in synaptic function control and plasticity (Campos et al. 2017;Freund et al. 2003;Kendall and Yudowski 2017), and it has been frequently associated with inflammatory processes (Kendall and Yudowski 2017;Mccoy 2016). It modulates neuronal excitability through the regulation of Ca 2+ -activated Cl − channels in prefrontal cortical pyramidal neurons (Zou and Kumar 2018). It has also been described that CB2R activation inhibits acute nociception, inflammatory hyperalgesia, and neuropathic sensory hypersensitivity (Ibrahim et al. 2006). CB2R is also related to neuropsychiatric disorders, including alcoholism, eating disorders, depression, schizophrenia, and autism spectrum disorders (Chen et al. 2017;Cortez et al. 2020).
Cannabinoids can interact with other receptors, such as the family of nuclear receptors activated by peroxisome proliferators (PPARs), also targeted by their derivatives PEA and OEA (O'Sullivan 2016). Cannabinoids might also activate the G55 protein-coupled receptor (GPR55), considered an orphan receptor, and GPR18, which has been associated with the endocannabinoid system, although it is not known whether CB1R or CB2R ligands can activate or block GPR18 (Morales and Reggio 2017). Finally, cannabinoids can still regulate some transient receptor potential (TRP) channels and ionotropic receptor members. It has been identified that vanilloid TRP (TRPV1, TRPV2. TRPV3, TRPV4), anquirin TRP (TRPA1) and melastatin TRP (TRPM8) can be modulated by cannabinoids (Muller et al. 2019).
Endogenous substances, such as AEA and 2-AG, mainly act on CB1R and CB2R. AEA can be considered a partial agonist of CB1R and CB2R; it is less effective than 2-AG and THC and has a lower affinity for CB2R than CB1R. AEA can also target TRPV1 and TRPA1 channels, PPAR-α and γ, GPR55 and GPR119 (Maccarrone 2017;Muller et al. 2019;Reggio 2010;Solymosi and Kofalvi 2016). 2-AG acts as a full agonist for the CB1R and CB2R and can also bind to PPAR-α and γ, TRPV1 and GPR55 (Baggelaar et al. 2018;Campos et al. 2017). However, the phytocannabinoid THC mimics endocannabinoids as a partial agonist at CB1R and CB2R (Pacher et al. 2020). On the other hand, CBD acts as a very weak inverse agonist of the CB1R and CB2R and as an agonist of the TRPV1 receptor. PEA and OEA lack affinity at CB1R and CB2R but evoke pharmacological effects and may act indirectly on them (Borrelli et al. 2015;Brown 2007). There are other nonendocannabinoid compounds that can also bind to nonclassical cannabinoid receptors, such as PEA and OEA, which are part of N-acylethanolamine. PEA is a component found in foods with anti-inflammatory, analgesic, anti-epileptic and neuroprotective effects by acting through PPAR (O'Sullivan 2016). OEA comes from dietary fat digested in small intestine enterocytes and causes satiety and reduces body weight gain. OEA signaling dysfunction can contribute to overweight and obesity (Tsuboi et al. 2018).
Although cannabinoid receptors act as individual monomers, studies have shown that CB1R and CB2R can form heteromers. CB1R can form heterodimers with receptors for serotonin, angiotensin, opioids, somatostatin, dopamine, and GPR55, among others (Martínez-Pinilla et al. 2014;Turu and Hunyady 2010;Viñals et al. 2015;Zou et al. 2017). On the other hand, the formation of heterodimers by CB2 may be associated with an interaction with CB1, GPR55, serotonin receptor 5HT1A or the chemokine receptor CXCR4 (Coke et al. 2016). This oligomerization can affect receptor signaling, cannabinoid receptor trafficking and ligand binding. Thus, some "nonclassical" responses associated with cannabinoid receptor stimulation could be explained by this oligomerization. Further studies are still necessary to expand knowledge regarding this mechanism (Mackie 2005; Morales and Reggio 2017).

Cannabinoid in Early Development: Pregnancy
The vast amount of information about the possible therapeutical properties of cannabis impacts legalization and increases medicinal usage in several countries around the world (Bonini et al. 2018;Levinsohn and Hill 2020). However, for sensitive subgroups such as pregnant women, the consumption could be hazardous. The use of marijuana has increased in recent years, and approximately 3.9% of pregnant women and 7.6% of nonpregnant women of reproductive age use marijuana (Brown 2017). A study in the United States found that approximately 7.5% of pregnant women aging 18-25 years use cannabis (Grant 2018). Specifically, in California, a study found that approximately 7% of pregnant women surveyed used cannabis, and an impressively 19% of women aged 18-24 years used cannabis in this period (Young-Wolff 2017). Cannabis usage during pregnancy is associated with restricted intrauterine growth, premature labor and fetal neurodevelopmental disorders (Grant 2018). Side-effects for adolescents and young women might also increase stroke risk and promote proliferation of breast cancer cells (Huang et al. 2020). Therefore, exposure to cannabinoids throughout the gestational period and/or lactation may result in developmental changes in fetuses and neonates.

3
It is general acknowledged that AEA is present since the origin of blastocysts and in the early embryonic stages, being important for the maintenance of pregnancy. Increased AEA levels impair embryo development and implantation while lower levels of endocannabinoids stimulate growth and differentiation via CB1 (Paria et al. 2001). Indeed, anandamide (3 mg/kg) administered on pregnant rats from gestational day 7 to postnatal day 21 exhibit a higher number of stillborn and pups with low body weight than control rats. In the case of females treated with AEA, there is an increase in the expression of CB1 receptors in the brain but not in the offspring (Amlani et al. 2017). Interestingly, the placenta of people who have suffered miscarriage has high expression of the CB1 receptor and low expression of FAAH (Trabucco et al. 2009), which raises the hypothesis that hyperactivity of the cannabinoid system in placental tissue is involved in dysfunctions of the placenta with consequent abortion.
During pregnancy, the uterus undergoes morphological and physiological changes, in addition to apoptosis and necrosis, which lead to regression of the decidual tissue. Accordingly, AEA induces cell death in cultured rat decidual cells in a dose-and time-dependent manner through CB1 receptor, as AM251 blocked cell death, while CB2 and TRPV1 antagonists AM630 and capsazepine had no effect, respectively. In addition, pretreatment with methylβ-cyclodextrin (MβCD), a cholesterol depletory, increased cell viability and decreased LDH release in the presence of 2-AG (25 µM), suggesting the involvement of cholesterolrich lipid rafts (Fig. 3) (Fonseca et al. 2009).
The levels of 2-AG were higher than those of AEA in the uterus and brain (Wang et al. 2007). In a later study, 2-AG also induced decidual cell death in rats. It is possible that the cell death caused by both AEA and 2-AG is related to COX-2, as AEA-derived oxidative metabolites and the oxygenation of 2-AG by COX-2 may be related to some of these effects (Fig. 3) (Fonseca et al. 2010). Moreover, 2-AG can also induce endoplasmic reticulum stress-induced apoptotic cell death in the placenta via CB2 receptor, and PERK-ATF4-CHOP. In this case, stress is probably related to the increase in ROS production induced by 2-AG (Fig. 3) (Almada et al. 2020a, b). According to this idea, BeWo cells, a human cytotrophoblastic cell lineage, undergo apoptosis when exposed to THC or synthetic cannabinoids (JWH-018, JWH-122, UR-144) by a mechanism that depends on increased ROS and endoplasmic reticulum stress (Fig. 3) (Lojpur et al. 2019;Maia et al. 2020). Furthermore, in human granulosa cells (hGCs) and the COV434 granulosa cell line, AEA treatment caused an increase in caspase 2/7 activities, inducing apoptosis in both cells, (Costa et al. 2021). The data together suggest that cell death induced by the activation of endocannabinoid receptors depends, in part, on the imbalance between cellular antioxidant defects and ROS production. However, more evidence needs to be gathered to support this hypothesis.
It has been previously shown a degree of correlation between the levels of uterine AEA and the status of autophagy in blastocysts. Autophagy activation and AEA levels are higher at the beginning of pregnancy in mice; however, they decreased near the time of implantation (Lee et al. 2011). When the synthetic agonist methanandamide (at 28 nM) was administered to pregnant mice, it enhanced autophagy in preimplanted mouse embryos, prolonging autophagy activation and causing cell death by apoptosis. Therefore, although AEA and autophagy are important for embryo implantation, increased levels of this endocannabinoid lead to changes in development and survival, extending the activation of autophagy that can lead to cell death (Oh et al. 2013).
Thus, the increase in stillborn may be related to a possible increase in the expression and activity of CB1 receptors in the placenta of pregnant women treated with AEA, which causes placental dysfunction by increasing the death of trophoblastic cells. However, more work needs to be done in this area to better elucidate this point. Therefore, identifying the roles of endocannabinoids in neurodevelopment is crucial to evaluate the possible side effects of the modulation of the cannabinoid system.

Endocannabinoid System in the CNS Development
The endocannabinoid system is present and functional from early stages of nervous system development through several distinct roles. The bioavailability of both 2-AG and AEA varies throughout brain development. Analysis by gas chromatography/mass spectrometry showed that AEA levels are limited in the brains of rats in midgestation, increasing from the perinatal to the adult period (Harkany et al. 2008). Indeed, 2-AG levels increase during the CNS differentiation (Bisogno et al. 2003;Keimpema et al. 2010). DAGLα and DAGLβ are expressed at embryonic day 10 (E10) in axons in mice, crossing the floor plate of the spinal cord, and at E14, DAGLα and DAGLβ are present in the retinal nerve fiber layer and in the optic nerve (Bisogno et al. 2003). They can also be seen in the cerebellum, dendritic field and deep cerebellar nuclei (Bisogno et al. 2003). Moreover, the expression of both DAGLα and DAGLβ enzymes is increased in axonal tracts during the middle and final gestational periods of mice restricting the levels of 2-AG (MacCarrone et al. 2014). Accordingly, DAGL is important for axonal growth, orderlying projection of ganglion cell axons (Brittis et al. 1996).
FAAH is also present during brain development. Neural progenitors express FAAH at postnatal day 2 in mice. It has been observed in hippocampus radial glia during late gestation and postnatal periods in vitro and in vivo . FAAH is also expressed in undifferentiated neural cells, participating in astrogliogenesis and neural progenitor differentiation in vivo (Basavarajappa et al. 2009). Curiously, FAAH KO adult mice showed enhanced hippocampal proliferation and astroglial differentiation .
MAGL distribution changes during synaptogenesis, accumulating in growth cones that oppose to postsynaptic neurons, but not to collateral axons that have not selected a target (Keimpema et al. 2010). Moreover, MAGL changes the axonal growth rate, modulates direction and determines the size of axons (Keimpema et al. 2010). However, when MAGL is inhibited, it can trigger neurite outgrowth in neurons in both cells that emit a tunted quiescent axon shortly after plating and neurons with an established primary neurite. After synaptic wiring, DAGL and MAGL are redistributed in neurons, with DAGL becoming exclusively present in somatodendritic domains of neurons and MAGL becoming located in the presynaptic neuronal site (Basavarajappa et al. 2009;Keimpema et al. 2010;MacCarrone et al. 2014).

Early Stages of CNS Development: Cell Fate and Proliferation
The endocannabinoid system appears to be involved in all stages of CNS development since neurulation. Exposure to ∆ 9 -THC during developmental stages 3-8 before neurogenesis and somitogenesis on chicken embryos, which corresponds to 2-3 gestational weeks in humans, causes deregulation of intrinsic factors important for CNS development. For example, the Sox2 expression domain appeared to be reduced in treated embryos, while Pax6 was downregulated in nascent neural tubes. Moreover, ∆ 9 -THC caused abnormal formation of neural plaques, interfering with the initial stages of the CNS (Psychoyos et al. 2008;De Salas-Quiroga et al. 2015).
Cell fate and proliferation of neural progenitors are also influenced by cannabinoids. In two human-induced pluripotent stem cell (hiPSC) lines, ∆ 9 -THC, as well as two synthetic CB agonists (THJ-018 and EG-18), decreased the PAX6 neural progenitor marker while increasing the Hu/C marker of early differentiated neurons (Miranda et al. 2020). In addition, an increase in GFAP-positive cells and a decrease in the synaptophysin marker were shown, suggesting that these cannabinoid agonists anticipate progenitor commitment, promoting glial cell fate (Miranda et al. 2020). Finally, it is still uncertain whether all physiological processes might be altered by cannabis during development.
CB1 and CB2 receptors are expressed by embryonic stem cells and neural progenitor cells, which regulate proliferation and cell fate. They usually show opposite patterns of expression in neuronal progenitors, with CB1 increasing while CB2 decreasing during neuronal differentiation (Galve-Roperh et al. 2013). CB1 and CB2 are also detected in oligodendrocyte progenitors (Molina-Holgado et al. 2002). CB2 receptors are present in neural progenitors during embryonic stages until adulthood. The CB1 receptor is expressed at high levels in differentiated neurons and at lower levels in glial cells in the hippocampus, basal ganglia and cortex. CB2 during development is present in microglial and macrophage cells (Zurolo et al. 2010).
Activation of the CB1 receptor on embryonic neural stem cells increases the differentiation of progenitor cells into neurons, suggesting an intrinsic development program during neuronal development. In R1 mouse embryonic stem cells in culture, cell fate default in deep layer pyramidal neurons is accompanied by increased expression of cannabinoid system components, such as CB1, DAGL, NAPE-PLD, FAAH, 2-AG and AEA (Paraíso-Luna et al. 2020). R1 cells that develop in the presence of an MAGL inhibitor (JZL-184, 1 µM) or in the presence of CB1 agonists HU-210 (100 nM) or Δ 9 -THC (2 µM) have a greater number of deep layer pyramidal neurons than upper layer pyramidal neurons, an effect that depends on the ERK and AKT pathways. Interestingly, brain organoids exposed to CB1 agonists have a higher density of pyramidal neurons in the deep layer than neurons in the upper layers (Paraíso-Luna et al. 2020). Together, these data demonstrate that CB1 signaling is involved in the development and maturation of deep layer pyramidal neurons. Endocannabinoids can also inhibit neuronal differentiation in vitro in cortical embryonic neural progenitors (Rueda et al. 2002).
Several data in the literature correlate the endocannabinoid system with the proliferation of neural progenitors in different experimental paradigms. Intranasal instillation of 2-AG (1-10 µM) or WIN 55212-2 (10 µM) increases the proliferation of neural progenitors in the olfactory epithelium of neonate mice (Hutch and Hegg 2016). The same is seen in neuronal stem cells treated with eicosapentaenoic acid, which increase 2-AG levels and activate CB1 and CB2 receptors and the p38 MAPK pathway. Accordingly, neuronal progenitors from CB1 KO showed a decrease in cell renewal and proliferation (Campos et al. 2017).
The PI3K/AKT pathway is also important for the proliferation of granular cerebellar precursor cells, shown for HU-210 through phosphorylation of AKT and GSK3β, leading to translocation of β-catenin to the nucleus, and an increase in the expression of cyclin D1 mRNA (Trazzi et al. 2010). Alternatively, activation of CB1 in the hippocampus modulates the fate of neural progenitor cells, promoting proliferation, similarly as in SVZ neurospheres. CB1 also inhibits differentiation, and all these effects are inhibited by the CB1 antagonist SR141716 (Jiang et al. 2005). Although CB1 is more abundant in the brain, CB2 also plays a role in proliferation. In the lineage of striatum neural progenitors ST14A, the 24-h administration of 2-AG (5 µM) or CB2 selective agonist (JHW133, 300 nM) increases cell proliferation in culture, which is blocked by the selective CB2 inhibitor (AM630, 300 nM) as well as U73122 and wortmannin, PLC and PI3K blockers, respectively (Cottone et al. 2021). Thus, cell proliferation of striatal progenitors in culture depends on the activation of the CB2 receptor and on the activation of the PLC and PI3K pathways. Likewise, activation of the CB2 receptor by the agonist HU-308 (50 nM) for 16 h induces the proliferation of BiB5 cells in culture, a lineage of rat neural progenitor cells, in a PI3K/AKT/ mTORC1-dependent manner (Palazuelos et al. 2012). The same proliferative profile dependent on CB2 and mTORC1 activation is found in the ventricular and subventricular areas in embryonic cortical slices (E14.5) of mice (Palazuelos et al. 2012). In turn, the CB2 receptor participates in cell proliferation in cultured neurospheres derived from the post-natal rat dentate DG (P1-3), however in a manner dependent on the activation of the CB1 receptor (Rodrigues et al., 2017). In addition, blocking the CB2 receptor with AM630 (1 µM) also inhibits BDNF-induced cell proliferation in culture neurospheres derived from the DG and SVZ of postnatal (P1-3) mice (Ferreira et al., 2018). Interestingly, oligodendrocyte progenitor proliferation is promoted by 2-AG through the PI3K/AKT/mTOR pathway (Gomez et al. 2015). Thus, the proliferative action triggered by CB2 receptors in neural and glial progenitors seems to depend on the PLC/PI3K/AKT/mTORC1 pathway.
CBD can also influence cell proliferation and migration in the hCMEC/D3 cell line. It is a microvascular endothelial cell line of the brain and represents a model of the blood-brain barrier. CBD at a concentration of 1 μM causes cell proliferation through activation of TRPV2. In addition, CBD also has an effect on cell motility by inducing cell migration in brain endothelial cells, and this effect was abrogated by using a TRPV2 antagonist. Finally, the study also found that CBD triggers tubulogenesis (Luo et al. 2019).
The brain is usually the main organ studied in the CNS. Most of the information about cannabinoids in development is related to it. However, there are other tissues that are being used to study the cannabinoid system, such as the retina. In chicken embryo retinas, CB1 and CB2 expression was positive in all retinal layers evaluated by western blot at E5, E7, E9, E14 and post-hatched day 7. In addition, expression of the MAGL enzyme was also visualized at all ages and in all layers of the retina (da Silva Sampaio et al. 2018). Later, we verified that the cannabinoid receptor agonists WIN-55,212-2 (0.5-1 μM), URB597 1 μM (FAAH inhibitor), and URB602 (50-100 μM, a MAGL inhibitor) inhibited the proliferation of retinal progenitors in culture. These data were corroborated by the lower incorporation of [ 3 H]-thymidine incorporation and, in the case of WIN, by the lower number of PCNA + cells in culture (Freitas et al. 2019). In rats, the presence of retinal CB1 receptor was first detected at E13, which is present throughout the rest of the developmental process (Buckley et al. 1997). During postnatal development, the CB2 receptor, FAAH, DAGLα and MAGL were also identified (Lu et al. 2000;Yazulla 2010). As cannabinoids cross the blood-brain barrier, it is very likely that they are able to cross the blood-retinal barrier. In the developmental context, these data indicate that exposure to cannabinoids during pregnancy could disrupt the correct proliferation of neural progenitors in the retina, which would ultimately lead to dysfunctions of the visual system in the adult individual.

Differentiation
The endocannabinoid system also influences neuronal differentiation. The activation of CB1 by 2-AG (1 µM) and Δ 9 -THC (3 µM) in cortical neurons derived from hiPSC cells leads to a reduction in neurite outgrowth after 24 h of exposure in an ERK-and AKT-dependent manner (Shum et al. 2020). Likewise, neurons differentiated from hiPSCs in the presence of Δ 9 -THC or synthetic cannabinoids (THJ-018 and EG-018) show functional abnormalities in voltagedependent calcium channels when stimulated with extracellular potassium. Many of these neurons do not or are slow to respond to potassium stimulation (Miranda et al. 2020). This fact may have a direct implication on neuronal connectivity and functionality during CNS development. Accordingly, pregnant Sprague Dawley rats treated with 2 mg/kg/ day Δ 9 -THC (from gestational day 5 until gestational day 20) led to sensorimotor dysfunctions and hyperactivity in newborn male rats after acute stimulation with Δ 9 -THC, and both behaviors were associated with an increase in dopamine release in the nucleus accumbens. In addition, dopaminergic neurons fired action potentials more frequently in a spontaneous or evoked manner, showing a depolarized membrane potential and a reduced voltage threshold (Frau 2019). Acute prenatal exposure to Δ 9 -THC also modifies dopamine responses in pups, with an increase in the frequency of spontaneous and evoked action potentials dependent on CB1 (Frau 2019). Dopaminergic hyperfunction was also observed in dopaminergic neurons derived from hiP-SCs treated throughout the differentiation period with low doses of AEA (1 µM). On the other hand, treatment with high concentrations of AEA (10 µM) or Δ 9 -THC (10 µM) weakens the functionality of dopaminergic neurons, as indicated by the decrease in ionic currents and synaptic activity (Stanslowsky et al. 2017). In utero exposure to Δ 9 -THC also leads to cognitive and behavioral impairment in adult rats, which is related to increased neuronal hyperexcitability and reduced number and dendritic complexity of cholecystokinin-expressing interneurons (Bara et al. 2018;de Salas-Quiroga et al. 2020;Vargish 2017). Collectively, these results suggest that prenatal exposure to Δ 9 -THC influences neural connectivity and excitability and modify dopaminergic system functionality, altering the intrinsic properties of neurons and endowing them with a hyperexcitable phenotype, a clinical feature of several psychiatric disorders, such as schizophrenia. These alterations in the dopaminergic system could also probably lead to addiction and/or hyperactivity.

Cell Death
Many studies address the role of cannabinoids in cell death in pathological conditions, mainly cancer and neurodegenerative diseases. However, few studies have aimed to verify the possible negative effects that cannabinoid modulation might induce during development. In addition to recreational use, marijuana has been administered among pregnant women for its antiemetic effect (Flament et al. 2020). Thus, it is important to understand the risks that the drug might cause to the developing embryo, especially because it has been shown that Δ 9 -THC can cross the placental barrier and thus reach the fetus (Alexandre et al. 2020).
Synthetic cannabinoids are often more potent than Δ 9 -THC, causing several adverse effects, such as psychosis, arrhythmia, myocardial infarction, and even death (Drummer et al. 2019;Shanks and Behonick 2016). One study in six patients who used synthetic cannabinoids found major intoxication and later failures of multiple organ systems, including hepatic and renal failure and psychogenic effects in the brain (Armstrong et al. 2019). Although the mechanisms related to all these effects have yet to be elucidated, it is possible that other noncannabinoid receptors might be involved, with the influence of serotoninergic, dopaminergic, GABAergic and glutamatergic receptors (Giorgetti et al. 2020). Therefore, the literature regarding the effects of synthetic cannabinoid-induced death, as well as its mechanisms, has yet to be further investigated.
Δ 9 -THC induces cell death in cortical neuron culture cells via CB1R, which is blocked by the CB1R antagonist AM251 and by pertussis toxin (PTX), a G i/0 inhibitor. Furthermore, Δ 9 -THC led to an increase in the release of cytochrome c and, consequently, activation of caspase-3 in a PTX-sensitive manner. These data indicate that Δ 9 -THCinduced cell death involves the activation of CB1R through cytochrome c release and caspase-3 activation (Campbell 2001). Another study demonstrated that Δ 9 -THC could be neurotoxic in hippocampal neurons using P1 Sprague Dawley rats. Δ 9 -THC caused neuronal death via CB1R, an event blocked by inhibitors of phospholipase A2 (PLA2) and COX-2 inhibitors such as aspirin and indomethacin. Thus, it is suggested that Δ 9 -THC increases arachidonic acid through PLA2, activating COX-2 and leading to the formation of ROS, causing neuronal death (Fig. 4) (Chan et al. 1998). A recent study demonstrated that a high concentration of CBD (10 µM) increases caspase 3 immunolabeling and cell death in human iPSCs (Miranda et al. 2020). As mentioned, CBD did not change Ki67 detection, a marker of the cell cycle, suggesting that the CBD effect is restricted to cell death (Miranda et al., 2020). However, the combination of CBD and Δ 9 -THC would have a broader effect on CNS development. These studies raise concerns about the consequences of cannabinoid exposure during pregnancy and CNS alterations.
Synthetic cannabinoids also influence CNS cell survival. Synthetic agonist CP-55,940, a full mix agonist for CB1 and CB2 receptors, is cytotoxic in forebrain cultures of mice at 15 days of gestation. Treatment with CP-55,940 decreased cell viability, leading to apoptosis through caspase-3. This effect was reduced in the presence of AM251, a CB1 receptor antagonist, but not by AM630 (CB2 receptor antagonist). Furthermore, AM630 did not cause cytotoxicity (Fig. 4) (Tomiyama and Funada 2014). The same cytotoxic effect was observed in the NG 108-15 neuroblastoma cell line, with the agonists CP-55940, CP-47497 (specific for CB1R) and CP-47497-C8 (found in spice herbal blends), as AM251 (but not AM630) suppressed some of these cytotoxic effects. Therefore, these data show a clear CB1R-mediated apoptosis in forebrain cultures and in the neuroblastoma lineage NG 108-15 (Tomiyama and Funada 2011). Dorsal forebrain organoids derived from human-induced pluripotent stem cells (hiPSCs) treated with WIN-55212-2 were selectively enriched for cells exhibiting both cleaved PARP (one of several known cellular substrates of caspases) and DNA fragmentation, which occurs when endonucleases cleave chromatin into nucleosomal units and is therefore a marker of apoptosis (Notaras et al. 2021).
In the retinal model, WIN 55212-2 induces cell death during development. In E7C2 (C2-second day of culture) chick embryo cultures, treatment with WIN (0.5, 1, 5 µM) induced a significant decrease in the number of living retinal cells in a dose-dependent manner (Freitas et al. 2019). This effect was inhibited by both AM251 and AM630, indicating that the visualized cell death is mediated by CB1 and CB2. Fig. 4 Apoptosis induced by cannabinoids in the CNS. CP-55,940 activates CB1R and causes apoptosis via caspase-3 in forebrain cultures. Δ9-THC, through CB1R, leads to cytochrome c release and caspase-3 activation, inducing apoptosis in cultured cortical neuron cells. Δ9-THC increases AA through PLA2, activating COX-2 and culminating in an increase in ROS, causing hippocampal neu-ral death. P2X7 activation increases the activity of NAPE-PLD and DAGL, leading to an increase in 2-AG and AEA. 2-AG, AEA and WIN activate CB1R and CB2R signaling through P2X7R and cytoplasmic Ca 2+ , causing an increase in superoxide and cytochrome c release and leading to apoptosis in the retina during development 1 3 WIN still induced an increase in ROS and mitochondrial stress, which could be involved in the cell death response (Fig. 4) (Freitas et al. 2019). Interestingly, WIN-induced cell death was completely inhibited by the selective antagonist for P2X7R, A438079. P2X is an ATP purinoreceptor that has been associated with cannabinoids, and specifically, P2X7 is more related to the effects of inflammation, cell death and pain. Thus, it was seen that P2X7R is related to WIN-induced death of retinal cells, corroborating a huge amount of data revealing the important role of P2X7 in cell death regulation (Kanellopoulos and Delarasse 2019). Interestingly, even though P2X7R mediated WIN-induced cell death, the regulation of cell proliferation (mentioned before) by WIN was P2X7R-independent, showing the electivity of P2X7R signaling to promote cell death. Although the involvement of P2X7 in cell death is well-known, this was the first study showing the relationship of cannabinoidinduced cell death with the P2X7 receptor. The mechanisms involved in this event are still widely unknown. One possibility is that CB1 could induce ATP release, which, in turn, stimulates P2X7 receptors, allowing an intracellular calcium increase and an increase in ROS and mitochondrial stress, leading to cell death. However, whether dying cells are induced by P2X7 is unknown. In cultured astrocytes, as well as in microglia, P2X7 promotes an increase in 2-AG production by enhancing DAGL and inhibiting MAGL activity in a calcium-dependent manner (Witting et al. 2004). Therefore, another raised question relates to CB1 activation by WIN that could possibly initiate positive feedback that increases endocannabinoid availability in a deadly pathway. Finally, the data also showed a decrease in glial, but not neuronal, cell markers, with a decrease in proliferation and cell viability. Thus, glial progenitors could undergo a reduction in proliferation together with an increase in cell death, leading to an increase in the neuronal population. Therefore, it is possible that endocannabinoids could favor neuronal differentiation in the retina.
The relationship between endocannabinoids and cell death in several tissues, including the CNS, is still scarce. AEA induces apoptosis in lymphocytes, pheochromocytoma and neuroblastoma. It has also been shown that AEA, but not 2-AG, PEA or AA, induces cell death in PC12 cells, activating p38 MAPK, JNK and ERK1/2 through cytochrome c release. Cell death observed in PC12 cells is dependent on lipid rafts but not on cannabinoid receptors (CB1, CB2 or TRPV1) (Sarker and Maruyama 2003;Sarker et al. 2003). Interestingly, the same response was detected for C6, Neuro-2a, HEK 293, CHO, HVSM cells, Jurkat, HL-60 and nerve growth factor (NGF)-differentiated PC12 (d-PC12) cells . Another study found that AEA induces apoptosis in CHP100 neuroblastoma cells and in C6 glioma cells via TRPV1. Furthermore, in C6 cells but not in CHP100 cells, MβCD reduced AEA-induced apoptosis.
It was also demonstrated that cholesterol depletion through MβCD enhanced the signaling and binding of the CB1 receptor, stimulating MAPK and causing the reduction of cytochrome c release from mitochondria, showing that this is probably the mechanism involved in the protection caused by MβCD in C6 cells (Bari et al. 2005).
CBD (10 μM) is also neurotoxic when used in neural progenitors, inducing cell death in hiPSC cultures after 21-22 days in vitro. A lower concentration of CBD (1 μM) was associated with a consistently lower density culture, but did not significantly differ from the control for cell death, suggesting that it could affect the proliferation and/or survival of neuronal progenitors and differentiated neurons (Miranda et al. 2020).
Most studies focus more on the influence of cannabinoids on cell death by apoptosis, and little is known about their effects on processes such as autophagy. Basal autophagy is necessary for the degradation and removal of damaged proteins and organelles from the cell itself via lysosomes, protecting against apoptosis in many situations (Galluzzi et al. 2018). Autophagy is essential for many developmental and physiological processes, and dysfunction is related to several diseases (Costa et al. 2016). However, depending on the context, autophagy can promote apoptosis (Filomeni et al. 2015). Increased accumulation of lipofuscin was described in the CA3 region of the hippocampus of CB1 receptor KO mice. The accumulation of this pigment is usually associated with age-related oxidative stress and deficits in autophagy. In agreement, upregulation of the levels of autophagy markers, such as p62 and LC3-II, was observed in CB1 KO mice. Therefore, it was shown in a study that CB1 receptor functionality affects lysosomal activity and autophagy (Piyanova et al. 2013). In a model of experimental autoimmune encephalomyelitis, the synthetic CB2 selective agonist HU-308 stimulated autophagy and inhibited the activation of the NLRP3 inflammasome, preserving the spinal cord of C57BL/6 mice (Shao et al. 2014).
There is also a possibility that the protective effects of the cannabis-based product Sativex® (27 mg/ml THC + 25 mg/ ml CBD) observed in PK −/− /Tau VLW mice, a model for complex neurodegenerative diseases, might be related to an increase in autophagy. Sativex® improves behavior, dopamine metabolism, oxidative stress, glial function and tau/amyloid pathology through reduction of free radicals, increase in mitochondrial activity and stimulation of autophagy. Moreover, mutated tau overexpression may impair autophagy. Thus, some of these protective effects may be related to the increase in autophagy induction and to the possible improvement of the process of autophagosomelysosome fusion or lysosomal degradative activity (Casarejos et al. 2013).
It is largely unknown whether cannabinoids might cause cell death by necrosis other than in cancer cells, especially 1 3 in the CNS. AEA (25-100 μmol/L) induces cell death by necrosis, in addition to proliferation inhibition, in primary hepatic stellate cells of humans and rats. However, the AEA effect was not mediated by CB1 and CB2 receptors, but was probably dependent on lipid rafts, as MβCD prevented AEA binding to primary hepatic stellate cells, in addition to inhibiting ROS formation, intracellular calcium release and cell death (Siegmund et al. 2005).
Cell death triggered by the activation of CB1 and CB2 receptors is well established in tumor cells and in deciduous tissue cells (Fonseca et al. 2017). Multiple mechanisms follow this event, which are dependent on autophagy, reticulum stress and oxidative stress. Evidence suggests that activation of cannabinoid receptors leads to cell death in developing CNS cells, although the mechanisms remain unclear. In any case, these findings need further attention as future consequences of exposure to exogenous cannabinoids during crucial stages of embryonic development.

Conclusion
Endocannabinoids are important and abundant modulators in CNS physiology and development (Campos et al. 2017;Parker 2017). A great amount of work shows a neuroprotective effect associated with endocannabinoids through classical CB1R and CB2R in different CNS pathologies. Accordingly, medicinal cannabinoids are turning into a recent reality in several countries. Moreover, a large number of studies have demonstrated the antitumoral effects against transformed cell lines (Garcia-Arencibia et al 2019). However, as we hope to be clearly pointed out in the present review, knowledge about the role of cannabinoid substances in the early developmental CNS, especially for cell death, is still scarce. The signaling pathways have been widely explored in the oncological research field and even in the mature CNS in the context of pathological or physiological conditions. Although rare, the data available indicate that endocannabinoids regulate cell fate, proliferation, differentiation and cell death, actively participating in the formation of the mature CNS. Thus, exogenous cannabinoid exposure could impair CNS development, possibly leading to a dysfunctional nervous system. Therefore, during pregnancy, the use of recreational cannabis must be avoided, and the medicinal/synthetic cannabinoid field must consider all these elements.
Author Contributions ECB, RB, GRF, KCC: Conception and design, collection, reading and interpretation of scientific papers from databases, manuscript writing, final approval of the manuscript. LFM, LSS, RAMR: Manuscript writing, text review, final approval of manuscript.
Funding Work in the laboratories of the authors was supported by grants from the Conselho Nacional de Desenvolvimento Científico e Tecnológico ( Data availability Not applicable.

Conflict of interest The authors have no financial interests to disclose.
Ethical approval Not applicable.

Consent to participate Not applicable.
Consent to publish Not applicable.