Mechanisms of Endocannabinoid Transport in the Brain

Abstract

The endocannabinoids (eCBs) 2-arachidonoylglycerol and anandamide are among the best studied lipid messengers in the brain. By activating cannabinoid receptors in the central nervous system (CNS), eCBs tune synaptic function thereby influencing a variety of physiological and behavioral processes. Extensive research conducted over the last few decades has considerably enhanced our understanding of the molecular mechanisms and physiological functions of the eCB system. It is now well-established that eCBs are synthesized by postsynaptic neurons and serve as retrograde messengers that suppress neurotransmitter release at central synapses. While the detailed mechanisms by which eCBs gate synaptic function and behavioral processes are relatively well characterized, the mechanisms governing eCB transport at central synapses remain ill defined. Recently, several studies have begun to unravel the mechanisms governing intracellular and intercellular eCB transport. In this Review, we will focus on new advances in the mechanisms of intracellular and synaptic eCB transport in the CNS.

Introduction

Since the discovery of cannabinoid receptors (CBRs) and their endogenous lipid agonists, the endocannabinoid (eCB) system has been the focus of extensive research. It is now well-established that eCBs are ubiquitous retrograde messengers that depress neurotransmitters release at both excitatory and inhibitory synapses in the central nervous system (CNS) (Alger, 2012; Kano, Ohno-Shosaku, Hashimotodani, Uchigashima, & Watanabe, 2009; Katona & Freund, 2012). By modulating synaptic strength and plasticity throughout the CNS, eCBs are implicated in the regulation of an array of cognitive and physiological processes including learning and memory (Kruk-Slomka, Dzik, Budzynska, & Biala, 2017; Mechoulam & Parker, 2013), pain (Paulsen & Burrell, 2019; Schlosburg, Kinsey, & Lichtman, 2009; Zogopoulos, Vasileiou, Patsouris, & Theocharis, 2013), stress (Lutz, Marsicano, Maldonado, & Hillard, 2015; Morena, Patel, Bains, & Hill, 2016), reward (Parsons & Hurd, 2015; Sagheddu, Muntoni, Pistis, & Melis, 2015), and feeding behaviors (Bermudez-Silva, Viveros, McPartland, & Rodriguez de Fonseca, 2010; Tarragon & Moreno, 2019). In addition to controlling synaptic function in the mature brain, eCBs also regulate synaptic maturation and neurogenesis during development (Berghuis et al., 2007; Harkany et al., 2007). Consequently, dysregulation of eCB signaling is implicated in psychiatric (e.g., anxiety and depression) (Hillard, Weinlander, & Stuhr, 2012; Mechoulam & Parker, 2013), neurological (e.g., Epilepsy, Alzheimer disease and Huntington’s disease) (Chen et al., 2012; Cristino, Bisogno, & Di Marzo, 2020; Dvorzhak, Semtner, Faber, & Grantyn, 2013), and neurodevelopment disorders such as autism (Foldy, Malenka, & Sudhof, 2013). Consequently, pharmacological modulation of the eCB system constitutes a potential therapeutic strategy to treat psychiatric and neurological disorders (Cristino et al., 2020).

The eCB system is composed of two G-protein coupled receptors (GPCRs), commonly named cannabinoid type 1 and 2 receptors (CB1Rs and CB2Rs, respectively), their endogenous lipid agonists, 2-arachidonoylglycerol (2-AG) and N-arachidonoylethanolamine (anandamide or AEA) and the enzymes involved in their synthesis and degradation. CB1Rs are highly expressed throughout the CNS (Egertova & Elphick, 2000; Herkenham et al., 1991; Matsuda, Bonner, & Lolait, 1993) and consequently the behavioral and physiological effects of eCBs are largely mediated by CB1Rs. Activation of these receptors, which are predominantly localized on synaptic terminals, plays a major role in controlling neurotransmitter release and synaptic plasticity (Castillo, Younts, Chavez, & Hashimotodani, 2012). Although CB2Rs are mainly expressed in peripheral tissues and immune cells including brain microglia (Munro, Thomas, & Abu-Shaar, 1993; Van Sickle et al., 2005), some evidence suggests neuronal CB2R expression in discrete brain areas (Jordan & Xi, 2019; Liu et al., 2017; Van Sickle et al., 2005; Zhang et al., 2014). However, it should be noted that the expression of CB2R in the healthy brain remains controversial, although its expression in neurons can be induced following neuronal insults (Haider et al., 2020; Viscomi et al., 2009).

The predominant mechanism by which eCBs control synaptic function is via retrograde signaling (Fig. 1). eCBs are synthesized and released by postsynaptic neurons in response to neuronal activation. Once released, eCBs traverse the synaptic cleft, activate pre-synaptic CB1Rs and suppress neurotransmitter release (Fig. 1) (Alger, 2012; Chevaleyre, Takahashi, & Castillo, 2006; Freund, Katona, & Piomelli, 2003; Kano et al., 2009; Katona & Freund, 2012). In addition to retrograde signaling, eCBs also serve as autocrine messengers that control intrinsic neuronal excitability by activating postsynaptic CB1Rs (Bacci, Huguenard, & Prince, 2004; Marinelli, Pacioni, Cannich, Marsicano, & Bacci, 2009), thereby gating several membrane currents (Gantz & Bean, 2017). The magnitude and spatiotemporal profile of eCB signaling is tightly regulated by the synthesis (Hashimotodani et al., 2013; Tanimura et al., 2010), degradation (Hashimotodani, Ohno-Shosaku, & Kano, 2007), and transport of eCBs (Haj-Dahmane et al., 2018). The detailed mechanisms regulating eCBs synthesis and degradation are well established and were the subject of previous reviews (Ahn, McKinney, & Cravatt, 2008; Mechoulam & Parker, 2013). This review will discuss our current understanding of the mechanisms underlying eCB transport and its impact upon synaptic eCB signaling.

Figure 1. Model of retrograde 2-AG signaling at central synapses.

(Step 1) Depolarization of postsynaptic neurons triggers Ca²⁺ influx through voltage-dependent Ca²⁺ channels. (Step 2) The increase in intracellular Ca²⁺ activates various enzymes involved in eCB synthesis, including PLC. (Step 3) Activation of GPCRs coupled to the G_q/11 family, including metabotropic glutamate receptor type 1 and 5 (mGluR1/5), triggers PLC activation. (Step 4) PLC hydrolyzes PIP₂ into DAG. (Step 5) DAGLα converts DAG into 2-AG. (Step 6) 2-AG is released in the synaptic cleft. Once released, 2-AG translocates across the synaptic cleft via distinct potential mechanisms discussed in this review. (Step 7) 2-AG activates presynaptic CB1Rs resulting in inhibition of voltage-dependent Ca²⁺ channels and the suppression of neurotransmitter release. (Step 8) 2-AG is then inactivated via hydrolysis into arachidonic acid (AA) and glycerol by MAGL localized in presynaptic terminals. (Step 9) MAGL in astrocytes also contributes to the termination of synaptic 2-AG signaling. (Step 10) AEA released from microglia via macrovesicles may also contribute to eCB signaling at central synapses.

Synaptic eCB signaling

The first evidence that eCBs function as retrograde messengers originated from electrophysiological studies of depolarization-induced suppression of inhibition (DSI), a form of short-term synaptic plasticity discovered by Alger and Marty at GABA synapses in the hippocampus and cerebellum, respectively (Llano, Leresche, & Marty, 1991; Pitler & Alger, 1994). These studies established that activation of postsynaptic neurons triggers calcium (Ca²⁺)-dependent eCB synthesis and release, which in turn activate presynaptic CB1Rs and inhibit GABA release, thereby inducing DSI (Ohno-Shosaku, Maejima, & Kano, 2001; Wilson & Nicoll, 2001). Concurrent to the discovery of DSI, Kreitzer and Regehr reported that eCBs also mediate depolarization-induced suppression of excitation (DSE), a form of short-term plasticity at glutamate synapses in the cerebellum (Kreitzer & Regehr, 2001). Following these seminal publications, eCB-mediated DSI and DSE have been reported throughout the brain (for review, see (Ohno-Shosaku & Kano, 2014)).

At most synapses, Ca²⁺ influx during neuronal activation stimulates Ca²⁺-dependent phospholipase Cβ (PLC_β), leading to the hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP₂) and the generation 1,2-diacylglycerol (DAG) (Fig. 1). DAG is then converted into 2-AG by diacylglycerol lipase α (DAGLα) localized on the inner leaflet of the postsynaptic plasma membrane (Yoshida et al., 2006). 2-AG is subsequently released and acts as a retrograde messenger by activating presynaptic CB1Rs to mediate DSI (Hashimotodani et al., 2007; Hashimotodani, Ohno-Shosaku, Maejima, Fukami, & Kano, 2008; Kim & Alger, 2004) and DSE (Hashimotodani et al., 2013; Pan et al., 2009; Uchigashima et al., 2007). Pharmacological inhibition or genetic deletion of DAGLα profoundly reduces the magnitude of DSI (Hashimotodani et al., 2007; Hashimotodani et al., 2008) and DSE (Hashimotodani et al., 2013), thus establishing its role as the key biosynthetic enzyme for 2-AG in the brain. 2-AG signaling is terminated by hydrolysis into arachidonic acid and glycerol by the enzyme monoacylglycerol lipase (MAGL), located in presynaptic terminals and surrounding astrocytes (Fig. 1) (Blankman, Simon, & Cravatt, 2007; Dinh, Freund, & Piomelli, 2002; Viader et al., 2015). Inhibition of 2-AG degradation facilitates and enhances the duration of DSI and DSE (Pan et al., 2009; Straiker et al., 2009; Zhong et al., 2011). In the case of both DSI and DSE, the transient activation of CB1Rs induces direct G-protein (likely through the βγ subunits) mediated inhibition of voltage-dependent Ca²⁺ channels (VDCC) (Brown, Brenowitz, & Regehr, 2003; Wilson & Nicoll, 2001) and subsequent depression of neurotransmitter release (Fig. 1).

In addition to mediating short-term synaptic plasticity (i.e. DSI and DSE), eCBs also induce presynaptic forms of long-term depression (eCB-LTD) at excitatory (Gerdeman, Ronesi, & Lovinger, 2002; Haj-Dahmane & Shen, 2010; Robbe, Kopf, Remaury, Bockaert, & Manzoni, 2002) and inhibitory synapses (Chevaleyre & Castillo, 2003; Marsicano et al., 2002). Examination of the mechanisms underpinning eCB-LTD reveals that activation of CB1Rs is required during the induction but not the expression of the LTD. The expression of eCB-LTD is predominantly mediated by G_i/o-dependent inhibition of the adenylate cyclase/cAMP/PKA signaling pathway (Chevaleyre et al., 2006; Haj-Dahmane & Shen, 2010; Heifets & Castillo, 2009) and may involve functional alteration of the presynaptic proteins Rab3B/RIM1a (Chevaleyre, Heifets, Kaeser, Sudhof, & Castillo, 2007; Tsetsenis et al., 2011) or inhibition of P/Q-type VDCCs (Mato, Lafourcade, Robbe, Bakiri, & Manzoni, 2008). Since these early studies, eCB-mediated short and long-term synaptic plasticity has been reported throughout the brain (for review, see (Araque, Castillo, Manzoni, & Tonini, 2017)). As such, eCBs are ubiquitous retrograde messengers critically involved in finetuning synaptic plasticity.

The synthesis of eCBs can also be induced by activation of numerous neurotransmitter receptors, in particular G_q/11-coupled GPCRs. Stimulation of these receptors, including group I metabotropic glutamate receptors (mGluR1/5s) (Maejima et al., 2005), M₁ muscarinic receptors (Kim, Isokawa, Ledent, & Alger, 2002), α₁-adrenergic receptors (Haj-Dahmane & Shen, 2014), and orexin receptors (Haj-Dahmane & Shen, 2005) mobilizes 2-AG via activation of PLCβ. The PLCβ-driven 2-AG serves both as a retrograde signal that induces transient or long-lasting depression of neurotransmitter release and as an autocrine messenger that regulates the intrinsic excitability of postsynaptic neurons by gating membrane channels (Gantz & Bean, 2017).

Besides “on demand” eCB synthesis, there is growing consensus that 2-AG and AEA are tonically synthesized in the brain. In the absence of neuronal activation, pharmacological inhibition or genetic deletion of MAGL or the AEA hydrolyzing enzyme fatty acid amide hydrolase (FAAH) increases brain 2-AG and AEA levels, respectively (Booker et al., 2012; Cravatt et al., 2001; Hashimotodani et al., 2007; Tanimura et al., 2010). Inhibition of these enzymes potentiates synaptic transmission via CB1R-dependent mechanisms, indicating that tonic eCB signaling controls basal synaptic transmission (Haj-Dahmane et al., 2018; Hentges, Low, & Williams, 2005; Robbe et al., 2002). However, the precise physiological roles and the mechanisms controlling tonic eCB synthesis remain poorly understood.

Unlike classical neurotransmitters, which are released by vesicle exocytosis and diffuse across the synaptic cleft to activate their receptors (Sudhof, 2013), eCBs are not stored in synaptic vesicles for future release. Instead, the current model of eCB signaling posits that eCBs are released into the synaptic cleft immediately upon their synthesis (Alger, 2012). In the following sections, we will discuss the potential mechanisms that mediate the release, intracellular and synaptic transport of eCBs at central synapses.

Membrane transport of eCBs

The rapid nature of retrograde eCB signaling (Wilson, Kunos, & Nicoll, 2001) necessitates efficient eCB transport across the lipid bilayer away from their biosynthetic enzymes, followed by translocation across the synaptic cleft to activate presynaptic CB1Rs. The mechanisms underlying the membrane transport of eCBs have been extensively investigated, leading to two prevailing models in which eCBs are transported either by simple diffusion (Fasia, Karava, & Siafaka-Kapadai, 2003; Glaser et al., 2003) or facilitated diffusion by a putative transmembrane transporter (Beltramo et al., 1997; Di Marzo et al., 1994). The simple diffusion model is supported by the observation that eCBs are uncharged lipids that readily partition into and diffuse across protein-free bilayers (Kaczocha, Lin, et al., 2012; Lynch & Reggio, 2006; Tian, Guo, Yao, Yang, & Makriyannis, 2005). Furthermore, eCB transport across cell membranes is observed universally in all cell-types regardless of tissue origin, and the kinetics of eCB uptake appear to be non-saturable (Fowler, 2013; Glaser et al., 2003; Kaczocha, Hermann, Glaser, Bojesen, & Deutsch, 2006) (for a critical discussion of eCB transport studies, see (Glaser, Kaczocha, & Deutsch, 2005)). Evidence supporting the existence of a putative membrane transporter is based largely upon the inhibition of eCB transport by structural analogs of eCBs as well as its temperature-dependence (Beltramo et al., 1997; Chicca, Marazzi, Nicolussi, & Gertsch, 2012; Di Marzo et al., 1994; Fegley et al., 2004; Moore et al., 2005; Piomelli et al., 1999). Further support for the existence of a putative membrane transporter originates from studies demonstrating that the site of action of eCB transport inhibitors is predominantly at the plasma membrane (Ligresti et al., 2010; Oddi et al., 2005). However, it is worth noting that these inhibitors target additional proteins involved in eCB transport and metabolism (e.g., FAAH and FABP5), which has made it challenging to distinguish between the contributions of the putative membrane transporter versus other proteins that modulate eCB uptake (Alexander & Cravatt, 2006; Glaser et al., 2003; Kaczocha et al., 2006; Kaczocha, Vivieca, Sun, Glaser, & Deutsch, 2012). More recently, Gertsch and colleagues identified small molecule inhibitors that are selective for the putative eCB membrane transporter (Chicca et al., 2012; Nicolussi, Chicca, et al., 2014; Nicolussi, Viveros-Paredes, et al., 2014). Future studies that leverage such pharmacological tools will be necessary to characterize the putative eCB membrane transporter and determine the precise contributions of this putative protein and simple diffusion toward eCB membrane transport.

Intracellular eCB Transport

The cytosol presents an energetic barrier that limits intracellular lipid (e.g., fatty acid and eCB) diffusion to various cellular organelles (Weisiger, 1996, 2002) including the endoplasmic reticulum and lipid droplets/adiposomes, the sites of AEA hydrolysis and accumulation (Cravatt et al., 1996; Deutsch & Chin, 1993; Giang & Cravatt, 1997; Glaser et al., 2003; Kaczocha, Glaser, Chae, Brown, & Deutsch, 2010; Oddi et al., 2008). The spatial separation between its site of signaling at the plasma membrane and intracellular inactivation suggests the existence of mechanisms that rapidly translocate AEA across the cytosol to FAAH for hydrolysis. Indeed, over the last decade, several classes of proteins have been identified as intracellular carriers for AEA and 2-AG (for a comprehensive review, see (Maccarrone, Dainese, & Oddi, 2010)). Fatty acid binding proteins (FABPs) were first to be identified as intracellular carriers for AEA (Kaczocha, Glaser, & Deutsch, 2009). The mammalian brain expresses three members of the FABP family: FABP3, FABP5, and FABP7 (Furuhashi & Hotamisligil, 2008). AEA displays highest affinity for FABP7, followed by FABP5 and FABP3, and a similar pattern is observed for 2-AG (Elmes et al., 2015; Kaczocha, Vivieca, et al., 2012). Accordingly, overexpression of FABP5 and FABP7, but not FABP3, enhances the cellular uptake of AEA (Kaczocha et al., 2009). The adult brain abundantly expresses FABP5 while FABP7 undergoes postnatal downregulation (Owada, Yoshimoto, & Kondo, 1996), suggesting that FABP5 may serve as the major FABP that regulates brain AEA metabolism. Indeed, pharmacological or genetic inhibition of FABP5 reduces the cellular uptake and metabolism of AEA, and elevates its levels in the brain (Bjorklund, Blomqvist, Hedlin, Persson, & Fowler, 2014; Kaczocha et al., 2015; Kaczocha et al., 2014; Kaczocha, Vivieca, et al., 2012; Yu, Levi, Casadesus, Kunos, & Noy, 2014). FABP5 deletion also produces CB1R-mediated antinociceptive effects (Berger et al., 2012; Kaczocha et al., 2015; Kaczocha et al., 2014; Peng et al., 2017). Compared to AEA, relatively little is known about the regulation of 2-AG transport by FABPs although a recent study reported that FABP5 inhibition elevates 2-AG levels and modulates 2-AG signaling in the dorsal raphe nucleus (DRn) (Haj-Dahmane et al., 2018).

FAAH-like anandamide transporter (FLAT), a truncated variant of FAAH that lacks its membrane anchoring N-terminus, was identified as another intracellular carrier for AEA (Fu et al., 2011). FLAT expression was reported throughout the brain and in numerous peripheral tissues including the liver, intestine, and pancreas. In contrast to membrane-bound FAAH, FLAT is a peripheral membrane protein that is largely devoid of catalytic activity but retains the ability to drive AEA accumulation in cells, presumably by facilitating its transport from the plasma membrane to intracellular FAAH (Fu et al., 2011). The FLAT-selective inhibitor ARN272 elevates tissue AEA levels and produces CB1R-mediated antinociceptive effects in a variety of pain models, suggesting a role for FLAT in AEA inactivation in vivo (Fu et al., 2011). Although the contribution of FLAT to AEA signaling and its potential as a therapeutic target warrants further investigation, it is noteworthy that follow up studies failed to detect FLAT expression in mammalian cells or tissues (Bjorklund et al., 2014; Leung, Elmes, Glaser, Deutsch, & Kaczocha, 2013). Consequently, the functional relevance of FLAT as an eCB carrier remains to be clarified (Fowler, 2014).

Maccarrone and colleagues identified heat shock protein 70.2 (Hsp70) and serum albumin as additional intracellular AEA transport proteins (Oddi et al., 2009). Compared to Hsp70, AEA bound to albumin with ~5-fold higher affinity, which is consistent with its known high affinity for AEA (Bojesen & Hansen, 2003). Overexpression of Hsp70 in SH-SY5Y neuroblastoma cells led to a 5-fold increase in AEA uptake, consistent with its function as an intracellular AEA carrier. To date, the relative contributions of Hsp70 and albumin toward eCB transport and inactivation in vivo have not been investigated.

Sterol carrier protein-2 (SCP2) is a lipid chaperone that also binds AEA (Liedhegner, Vogt, Sem, Cunningham, & Hillard, 2014). In silico docking approaches demonstrated that SCP2 exhibits greater affinity for AEA compared to 2-AG and expectedly, AEA was able to compete with cholesterol for binding to SCP2. Overexpression of SCP2 increased the cellular uptake of AEA, demonstrating a functional role for SCP2 in AEA accumulation (Liedhegner et al., 2014). The influence of SCP2 upon AEA metabolism in vivo is not known, however deletion of SCP2 in concert with the liver-specific FABP1 isoform did not alter brain AEA levels (Martin et al., 2019).

Employing a photoactivatable chemical probe based upon the structure of AEA, Cravatt and colleagues performed a proteome-wide interaction analysis and identified nucleobindin-1 (NUCB1) as a novel AEA binding protein (Niphakis et al., 2015). In addition to AEA, 2-AG and arachidonic acid were also identified as NUCB1 ligands. Pharmacological or genetic inhibition of NUCB1 markedly elevated AEA levels in cells (Niphakis et al., 2015). Interestingly, previous work has demonstrated that NUCB1 is largely a Golgi apparatus resident protein (Lavoie, Meerloo, Lin, & Farquhar, 2002) and while expressed in brain neurons, NUCB1 appears to exclusively localize to the Golgi apparatus (Tulke et al., 2016). Consequently, the precise mechanism(s) through which NUCB1 regulates cellular AEA metabolism and the contribution of this protein towards eCB signaling and inactivation in vivo awaits elucidation.

Outside of the brain, retinol-binding protein 2 was recently identified as an intestinal binding protein for 2-AG but not AEA (Lee et al., 2020). Moreover, the liver-specific FABP1 binds to AEA and 2-AG and its deletion results in elevated brain AEA and 2-AG levels (Huang et al., 2016; Martin et al., 2016). Mechanistically, FABP1 deletion elevates circulating levels of arachidonic acid, which is postulated to diffuse into the brain and increase the levels of AEA and 2-AG precursors leading to augmented AEA and 2-AG biosynthesis (Martin et al., 2016). FABP2, an FABP isoform expressed in the intestine, was likewise demonstrated to interact with AEA and 2-AG albeit with low affinity (Lai, Katz, Bernard, Storch, & Stark, 2020). Although not a binding protein, caveolae-mediated endocytosis has been proposed as another mechanism that contributes to AEA internalization (McFarland, Bardell, Yates, Placzek, & Barker, 2008; McFarland et al., 2004). Accordingly, pharmacological or genetic inhibition of components of the endocytic machinery reduced AEA cellular accumulation. Collectively, multiple proteins have been identified as intracellular AEA and 2-AG carriers and in a subset of cases their inhibition was shown to augment eCB signaling in vivo.

Extracellular and synaptic transport of eCBs

Once released, eCBs must traverse the synaptic cleft to rapidly activate CB1Rs and control synaptic transmission and plasticity. Similar to the cytosol, the hydrophilic property of the synaptic cleft presents an energetic barrier that impedes efficient eCB diffusion, suggesting the existence of mechanisms that facilitate extracellular/synaptic eCB transport. The discovery that FABPs function as intracellular eCB carriers (Kaczocha et al., 2009) raises the possibility that this family of proteins may also contribute to synaptic eCB transport. In order to serve as extracellular carriers, FABPs must be secreted into the extracellular milieu. Interestingly, results from previous studies have reported the secretion of FABPs from various cell-types through nonconventional mechanisms (Ertunc et al., 2015; Josephrajan et al., 2019; Villeneuve et al., 2018). Furthermore, proteomic analysis identified FABP5 as a component of human cerebrospinal fluid (Chiasserini et al., 2014), supporting a potential extracellular role for this protein. Consistent with this notion, we recently demonstrated that FABP5 is released by primary astrocytes in vitro, localizes to synapses, and is indispensable for retrograde 2-AG signaling at glutamate synapses of the DRn (Haj-Dahmane et al., 2018). Pharmacological inhibition or genetic deletion of FABP5 blocked 2-AG mediated DSE as well as G_q/11-driven retrograde 2-AG signaling. Inhibition of FABP5 also resulted in a blockade of tonic 2-AG-mediated control of excitatory synaptic transmission in the DRn (Haj-Dahmane et al., 2018). These effects were not mediated by impaired presynaptic CB1R function nor a deficit in 2-AG synthesis, thereby supporting the model that FABP5 may serve as a synaptic carrier for 2-AG at glutamate synapses.

Extracellular vesicles (EVs) have been proposed as an additional mechanism that mediates the extracellular and synaptic transport of eCBs. Thus, in an early study conducted in microglia, Gabrialli et al. demonstrated that microvesicles released from activated microglia were enriched in AEA. Importantly, addition of EVs isolated from activated microglia to a primary culture of GABA neurons inhibited GABAergic synaptic transmission through CB1R-dependent mechanisms (Gabrielli et al., 2015). Such findings have led to the suggestion that EVs can mediate the extracellular transport of eCBs released from microglia and enable microglia to modulate synaptic function. Further support for the concept that EVs contribute to eCB signaling stems from the recent demonstration that cocaine-driven retrograde eCB transport in midbrain dopamine neurons requires the release of non-synaptic EVs via a sigma 1 receptor (Sig1R)-dependent mechanism. Mechanistically, the secretion of EVs involves the dissociation of intracellular Sig1R from ADP-ribosylation factor 6 (ARF6), a G-protein that regulates EV trafficking (Nakamura et al., 2019). Inhibition of Sig1R or ARF6 suppressed cocaine-driven retrograde 2-AG mediated control of GABAergic synaptic transmission. Interestingly, while inhibition of EV release impaired cocaine-induced 2-AG signaling, it was dispensable for tonic eCB release, suggesting that distinct mechanisms may govern these temporally-distinct modes of eCB transport.

Future directions

Although some recent progress has been made in elucidating potential mechanisms underlying synaptic eCB transport in the brain, significant gaps still remain. For the FABP5-mediated extracellular/synaptic eCB transport model, key questions that require further investigation include the cellular origin of FABP5 and its mechanism(s) of secretion. For instance, although astrocytes maintained in primary culture secrete FABP5 while neurons do not (Haj-Dahmane et al., 2018), it is currently not known whether astrocytes constitute the major source of synaptic FABP5 in vivo. Furthermore, because neuronal activity drives eCB mobilization and release, it remains to be elucidated whether FABP5 secretion likewise requires neuronal and/or astrocytic activation. Previous studies have identified several nonconventional mechanisms underlying the secretion of peripherally-expressed FABPs (Ertunc et al., 2015; Josephrajan et al., 2019; Villeneuve et al., 2018); whether similar mechanisms govern FABP5 secretion in the brain will require further investigation. To date, FABP5 has been implicated in controlling retrograde 2-AG signaling (Haj-Dahmane et al., 2018), however it remains to be determined whether this function extends to the synaptic transport of AEA, which mediates retrograde signaling in the striatum and amygdala (Adermark & Lovinger, 2007; Gerdeman et al., 2002; Ramikie et al., 2014). In addition to FABP5, FABP3 and FABP7 are expressed in the brain and could in theory contribute to synaptic eCB transport. Furthermore, as noted above, several additional intracellular carriers have been identified and it is conceivable that a subset of these proteins (or as yet to be identified carriers) could contribute to synaptic eCB transport.

Additional studies are also required to address several key questions regarding the microvesicle and non-synaptic EV-mediated extracellular transport of eCBs. For instance, while activated microglia release AEA-containing microvesicles that stimulate CB1Rs when applied to neuronal cultures in vitro, the precise role of this mode of transport in gating synaptic eCB signaling in vivo remains to be determined. In addition, it is still not known whether this mechanism underlies the transport of microglial 2-AG. Similarly, although non-synaptic EVs transport 2-AG released from midbrain neurons following cocaine and GPCR activation, it is currently not known whether this mechanism contributes to retrograde 2-AG and AEA transport induced by neuronal activation. Importantly, given the slow kinetics of EV release (Cashikar & Hanson, 2019; Sung et al., 2020; Verweij et al., 2018) and the rapid time scale of eCB-mediated short-term synaptic plasticity (i.e. DSI and DSE) (Kreitzer & Regehr, 2001; Wilson et al., 2001), it is conceivable that EVs may be best suited to gate eCB-mediated synaptic function that necessitates eCB release over longer time scales. Additionally, it remains to be determined whether EVs represent a common mechanism for 2-AG and AEA transport that extends to other brain areas. The finding that cocaine application triggers EV release in the midbrain raises the question as to whether other drugs of abuse that enhance eCB signaling employ this mode of 2-AG transport.

There is growing consensus that eCBs mediate crosstalk between neurons and astrocytes by activating astrocytic CB1Rs (Araque et al., 2017). Furthermore, astrocytic MAGL contributes to the metabolism of 2-AG following its release by postsynaptic neurons (Viader et al., 2015). This raises the intriguing and yet unexplored possibility that overlapping or distinct mechanisms may govern 2-AG transport to these distinct cell-types. Clearly, future studies are needed to further define the precise mechanisms underlying synaptic eCB transport in the brain.

Abbreviations

eCB

Endocannabinoid

CNS

Central nervous system

AEA

Anandamide

2-AG

2-arachidonoylglycerol

DSE

Depolarization-induced suppression of excitation

DSI

Depolarization-induced suppression of inhibition

LTD

Long-term depression

CB1R

Cannabinoid receptor 1

CB2R

Cannabinoid receptor 2

DAGLα

Diacylglycerol lipase α

DAG

Diacylglycerol

AA

Arachidonic acid

FAAH

Fatty acid amide hydrolase

MAGL

Monoacylglycerol lipase

FABP

Fatty acid binding protein

EV

Extracellular vesicle

VDCC

Voltage-dependent Ca²⁺ channels

GPCR

G-protein coupled receptor

PLC

Phospholipase C

PIP₂

Phosphatidylinositol 4,5-bisphosphate

FLAT

FAAH-like anandamide transporter

Hsp70

Heat shock protein 70.2

SCP-2

Sterol carrier protein-2

NUCB1

Nucleobindin-1

Sig1R

Sigma 1 receptor

ARF6

ADP-ribosylation factor 6