This Review examines the functional role of eCB signalling in behaviours relevant to the aetiology of anxiety and stress-related disorders. We begin by outlining the foundational properties of eCB signalling and its regulation of avoidance and defensive responses to stress. We then highlight emerging tools and methodologies that enable precise dissection of circuit-level eCB dynamics and delve into recent findings on the cellular and circuit mechanisms by which eCBs modulate stress-induced defensive responses. Finally, we integrate these insights into theoretical frameworks of anxiety and stress adaptation and explore the therapeutic promise of targeting the eCB system for these highly prevalent disorders.
The eCB signalling system is comprised of molecular machinery for the synthesis, intercellular and intracellular movement, and degradation of eCB ligands as well as their receptor targets and downstream signalling cascades (Fig. 1). Central levels of the eCB ligand N-arachidonoylethanolamine (AEA) are regulated by the sequential actions of a calcium-dependent N-acyltransferase activity, which generates N-arachidonoyl-phosphatidylethanolamine, and an N-acyl-phosphatidylethanolamine (NAPE)-specific phospholipase D (NAPE-PLD), an enzyme that converts a broad range of NAPEs into N-acylethanolamines (NAEs), including AEA, and phosphatidic acid. Indeed, some mutant mouse lines with disruptions in the Napepld gene (which encodes NAPE-PLD) show ~15-50% reductions in brain AEA (but see ref. ). Moreover, a small-molecule inhibitor of NAPE-PLD decreased brain AEA by ~50% in wild-type mice but not in Napepld mice. Alternative pathways involving sequential actions of phospholipase C (PLC) to generate phospho-AEA from NAPE and protein tyrosine phosphatase non-receptor type 22 to yield AEA may also contribute to brain AEA production. By contrast, the other major eCB ligand, 2-arachidonoylglycerol (2-AG), is produced in the brain exclusively by diacylglycerol lipases (DAGLα or DAGLβ), which convert diacylglycerol species into monoacylglycerols, including 2-AG. Importantly, 2-AG is produced by a multi-molecular complex including group 1 metabotropic glutamate receptors and DAGLα. Whereas neuronal and synaptically active 2-AG is produced almost exclusively by DAGLα, DAGLβ is expressed at low levels by some neurons and robustly in non-neuronal cells in the brain, including astrocytes and especially microglia.
Once formed, eCBs need to traverse cell membranes to access receptor targets. Although potential eCB membrane transport mechanisms have been well described, the molecular identification of selective eCB membrane transport proteins remains elusive. Despite this, alternative mechanisms that facilitate intercellular and intracellular movement of eCBs have been suggested recently. For example, fatty acid-binding protein 5 (FABP5) is an important mediator of intracellular transport of AEA and delivery to the endoplasmic reticulum for its metabolism. FABP5 has also been implicated in the synaptic release of eCBs. More recent data suggest that astrocyte-secreted FABP5 is a critical mediator of synaptic eCB transport. α-Synuclein and extracellular vesicles have also been implicated in the release of eCBs from postsynaptic membranes and microglia.
Degradation of AEA occurs via fatty acid amide hydrolase (FAAH), which converts NAEs into free fatty acids and ethanolamine, and genetic disruption of Faah or pharmacological FAAH inhibition results in large increases in brain and peripheral AEA levels, along with other NAE species. Moreover, a common missense functional variant in FAAH (leading to a C385A substitution) results in reduced enzyme levels and, consequently, increased brain AEA levels in a humanized mouse model and circulating AEA in humans who are heterozygous or homozygous for the A form, relative to those who are homozygous for the C form. By contrast, 2-AG is degraded by monoacylglycerol lipase (MGL, encoded by Mgll), and Mgll deletion and pharmacological MGL inhibition cause dramatic increases in brain and peripheral 2-AG levels. 2-AG can also be degraded by αβ-hydrolase domain-containing enzyme 6 (ABHD6).
The primary molecular targets of eCBs are cannabinoid receptors (cannabinoid type 1 (CB1) and cannabinoid type 2 (CB2)), but other molecular targets, including TRPV1 for AEA, and ion channels, including GABA receptors for 2-AG, have been suggested. The CB1 receptor, the primary target of eCBs within the brain, is a G protein-coupled receptor (GPCR) that couples to inhibitory G-proteins and signals through a variety of pathways, including ERK, β-arrestin, FAK and cAMP-PKA, and can affect cellular membrane potential via indirect modulation of potassium and calcium conductances. Lastly, GPCR kinase-mediated C-tail phosphorylation of CB1 receptors results in receptor internalization and desensitization in response to ligand binding.
The distribution of eCB signalling elements has been extensively studied in laboratory mice and rats. The preferential high expression of many eCB signalling elements within limbic forebrain regions, such as the prefrontal cortex (PFC), amygdala, hippocampus and key subcortical nodes, such as the ventral striatum, central amygdala and hypothalamus, all of which are regions heavily implicated in affective behaviour modulation, is likely to have subserved the major impact of eCB signalling systems on innate and conditioned defensive behaviours and stress adaptation. NAPE-PLD is expressed in the cerebral cortex, thalamus, hypothalamus and hippocampus. Detailed analysis of hippocampal NAPE-PLD expression revealed strong labelling in the CA3 and dentate gyrus within presynaptic neuronal elements, especially in mossy terminals of granule cells, as well as in parvalbumin-expressing interneurons. Within the ventromedial hypothalamus, NAPE-PLD was localized primarily presynaptically, but one study found expression postsynaptically in the hypothalamus. By contrast, FAAH is located exclusively postsynaptically within soma and dendrites of principal neurons. FAAH expression is highest in limbic cortical-like regions, including the PFC and basolateral amygdala (BLA) complex, the hippocampus and the cerebellum. Lower levels are also found in the thalamus, septum and basal ganglia nuclei. Regarding central 2-AG metabolism, DAGLα is expressed within postsynaptic dendrites and the annulus surrounding the synaptic spines that form asymmetric glutamatergic and invaginating perisomatic GABAergic synapses in cortical-like brain regions, with high protein expression observed in the cortex, cerebellum, hippocampus, dorsal and ventral striatum, and amygdala. DAGLα is also expressed at lower levels in the thalamus and hypothalamus. DAGLβ is expressed within the cortex, hippocampus, cerebellum and substantia nigra at low levels. MGL is expressed within excitatory and inhibitory presynaptic elements and astrocytes and expressed at high levels in the cortex, hippocampus, amygdala, basal ganglia, superior colliculus and cerebellum.
CB1 receptors are expressed at high levels throughout the brain, with the highest expression in limbic cortical regions, the amygdala, hippocampus, basal ganglia circuitry, including dorsal and ventral striatum, and cerebellum, with lower levels in the thalamus and hypothalamus, but they are not readily detected in the brainstem. Within cortical-like brain regions (including the BLA), CB1 receptors are highly expressed in a subpopulation of large CCK-expressing perisomatic-targeting interneurons, a proportion of which form invaginating synapses and also express presynaptic VGLUT3 (refs. ). CB1 receptors are also expressed by parvalbumin-positive interneurons in the striatal regions. CB1 receptors are expressed at lower levels by principal neurons of the limbic neocortex and hippocampus. At the cellular level, most CB1 receptors are expressed within presynaptic axon terminals, where they are located in close proximity to presynaptic proteins, including bassoon. CB1 receptors are also expressed at low levels postsynaptically, where they can regulate ion channel conductance. Expression of CB1 receptors by astrocytes and their localization to some intracellular organelles, including mitochondria, have also been observed (Fig. 1). CB2 receptor expression is restricted to activated or reactive microglia in the mouse and human brain.
Retrograde suppression of presynaptic neurotransmitter release is the primary and most well-established function of eCB signalling. This effect is mediated by 2-AG, which is synthesized by DAGLα in the postsynaptic compartment and crosses the synaptic cleft to engage presynaptic CB1 receptors. 2-AG is temporally regulated by presynaptic MGL (Fig. 1). Postsynaptic 2-AG synthesis can be triggered by calcium-driven, receptor-driven and calcium-assisted receptor-driven mechanisms. Although DAGLα demonstrates calcium sensitivity in vitro, the precise molecular mechanism that links postsynaptic depolarization and calcium influx to 2-AG synthesis that mediates short-term retrograde synaptic suppression remains unclear. By contrast, receptor-driven 2-AG signalling is mediated via activation of an array of Gq-coupled GPCRs (mGluR1, mGluR5 and others), downstream activation of PLCβ to liberate diacylglycerols, and ultimate liberation of 2-AG by DAGLα. Cooperative calcium-assisted receptor-driven 2-AG release relies on the calcium sensitivity of PLCβ, which increases diacylglycerol and subsequent 2-AG production. As noted above, once formed, a variety of transport and efflux mechanisms involving FABP5, α-synuclein and exosomes may contribute to functional release of 2-AG, although these findings require further confirmation.
Postsynaptic depolarization-induced retrograde calcium-driven 2-AG signalling (termed depolarization-induced suppression of inhibition and depolarization-induced suppression of excitation (DSE)) can exert short-term transient (on the order of seconds) suppression of presynaptic glutamate and GABA release via activation of CB1 receptors, whereas receptor-driven and calcium-assisted receptor-driven 2-AG release can contribute to long-term synaptic depression (LTD), which, in many cases requires additional factors, such as increases in presynaptic activity, presynaptic protein synthesis or presynaptic NMDA receptor activation. Both forms of synaptic depression can be prolonged via inhibition of 2-AG degradation by neuronal and astrocytic MGL, and MGL inhibition can also increase 2-AG spillover and constitutive engagement of CB1 receptors (also known as tonic 2-AG signalling) at some synapses. Prolonged direct CB1 receptor activation can also trigger LTD via cAMP-PKA-RIM1α-dependent mechanisms, although the signalling pathways mediating eCB LTD may differ by region and synapse. Somatodendritic 2-AG signalling has been less well-studied but involves calcium-driven 2-AG release from postsynaptic cells and autocrine activation of somatic CB1 receptors, leading to enhanced potassium channel conductance and transient membrane hyperpolarization within subsets of GABA and glutamatergic principal neurons in the cortex. Interestingly, constitutive presynaptic CB1 receptor signalling in the absence of ligand has been suggested to control tonic GABA release at some hippocampal synapses. Lastly, 2-AG activation of CB1 receptors on astrocytes can result in paradoxical increases in presynaptic neurotransmitter release via CB1 receptor-mediated increases in astrocytic calcium levels and release of gliotransmitters, including adenosine, that act on presynaptic receptors to increase neurotransmitter release probability. Although the role of AEA in the regulation of synaptic and cellular signalling is less well understood, AEA has been implicated in some forms of LTD mediated via TRPV1 receptors in cortical and subcortical brain regions. Despite these data, the precise mechanisms and sites that subserve AEA production and its mode of synaptic transmission (that is, retrograde, anterograde, volume) remains a major open question in the field.