Integrating endocannabinoid signaling, CCK interneurons, and hippocampal circuit dynamics in behaving animals

Abstract

The brain's endocannabinoid signaling system modulates a diverse range of physiological phenomena and is also involved in various psychiatric and neurological disorders. The basic components of the molecular machinery underlying endocannabinoid-mediated synaptic signaling have been known for decades. However, limitations associated with the short-lived nature of endocannabinoid lipid signals had made it challenging to determine the spatiotemporal specificity and dynamics of endocannabinoid signaling in vivo. Here, we discuss how novel technologies have recently enabled unprecedented insights into endocannabinoid signaling taking place at specific synapses in behaving animals. In this review, we primarily focus on cannabinoid-sensitive inhibition in the hippocampus in relation to place cell properties to illustrate the potential of these novel methodologies. In addition, we highlight implications of these approaches and insights for the unraveling of cannabinoid regulation of synapses in vivo in other brain circuits in both health and disease.

Keywords: CB1 receptor; behavior; brain state; cholecystokinin; disease; endocannabinoid; hippocampus; inhibition; interneuron; translational.

Figure 1.. Characteristics of eCB dynamics in vivo

(A) Cartoon depicting GRAB_eCB2.0 as a GPCR conjugated to green fluorescent protein (GFP) with and without agonist. Upon ligand binding, the sensor increases GFP fluorescence. (B) Active neurons (cells 2 and 5) in the CA1 principal cell layer (CA1PC) are associated with an increased eCB signal, denoted as a green haze, that does not spill over onto neighboring neurons. See (C) for the associated calcium and eCB traces taken from regions of interest segmented around PC soma during two-photon imaging. (C) Calcium and eCB traces from segmented neurons in (B). eCB traces are tightly coupled to calcium activity, both spatially and temporally, but there is a slight lag in the eCB trace, consistent with activity-dependent production. (D) Mean calcium and eCB signals from the CA1PC layer during physiological (locomotion; left) and pathological (seizure; right) activity with pharmacological interrogation of the signal. Drugs to block the synthesis and degradation of AEA and 2-AG were used to determine if the eCB signal depended on either or both of these endogenous ligands. Under both physiological and pathological conditions, inhibiting 2-AG synthesis and degradation, but not AEA, was associated with the reduction and augmentation of the eCB signal, respectively, highlighting a key role for 2-AG in activity-dependent eCB dynamics. Notably, seizures were associated with a spreading wave that accompanied postictal flattening on the local field potential. This signal was ~350 times greater than physiological eCB signal changes.

Figure 2.. Basket cell dichotomy in vivo

(A) Top: illustration of CCKBC and PVBC phase preferences during oscillations. Yellow lines symbolize field potentials. Markers indicate the relative timing of action potential firing (PVBC: orange; CCKBC: blue; PC: black). While PVBCs fire in the descending phase of the extracellular theta, CCKBCs fire in the ascending phase. During gamma oscillations, PVBCs fire in a strongly phase-locked manner after the trough of each gamma cycle, consistent with their role of driving gamma oscillations. By contrast, CCKBCs fire less consistently at variable phases. Bottom: illustration of cell type-specific activity dynamics. Curves in color symbolize average DF/F signals in an in vivo calcium imaging experiment. Vertical lines indicate the time of various events associated with brain state transitions. During locomotion (yellow shading), PVBCs are recruited while CCKBCs are suppressed. By contrast, after stopping (dashed vertical lines), CCKBCs are recruited. CCKBCs are suppressed before SPW-Rs, which emerge during immobility (gray shading) from a non-theta brain state, while PVBCs are recruited during the SPW-R. During sleep, barrages (BARR) of CA2 spiking recruit CCKBCs, while PVBCs are suppressed. When comparing SPW-Rs that are recorded before or after a goal-oriented learning session, PVBCs became more activated, while CCKBCs became more inhibited around SPW-Rs after learning. When sensory cues were presented during the task, CCKBCs were recruited compared with other INs, including PVBCs. (B) Top: illustration of an in vitro paired recording experiment to assess DSI. Note that in CCKBCs, but not in PVBCs, IPSPs in postsynaptic PCs are suppressed following PC depolarization (solid lines) compared with baseline (dashed lines). This suppression is mediated by eCBs, as indicated by its sensitivity to CB₁ receptor antagonists. Middle: illustration of in vivo imaging experiments to assess DSI. GRAB_eCB2.0 imaging shows eCB release following calcium transients in PCs (left). All-optical interrogation of IPSPs shows suppressed IPSP following PC depolarization (right). CCKBCs were optogenetically stimulated using blue light, while a genetically encoded voltage indicator was imaged to record changes in the membrane potential of postsynaptic PCs. Blue bar illustrates photoactivation of presynaptic CCKBCs. Optically evoked IPSPs were detected when the PC was quiet before the light pulse (dashed line), while IPSPs were suppressed if the light pulse followed a spontaneous plateau-driven complex spike in the PC (solid line). The depolarization associated with such events may release eCBs and induce DSI. IPSPs are not drawn to scale with action potentials. Bottom: immediate-early gene expression in PCs leads to BC type-specific inhibitory plasticity: while Fos activation leads to increased PVBC and decreased CCKBC synapse strength, Npas4 activation leads to increased inhibition by CCKBCs. Red nuclei symbolize activated cells (dark colors) compared with neighboring inactive cells (light colors). Round markers symbolize synapses.

Figure 3.. Outstanding questions regarding multi-scale integration of eCB signaling in behaving animals

The panel numbers correspond to the outstanding questions section within the main text. The illustrated drawings are schematic depictions of key topics discussed and referenced in the main text. 1. When and where is AEA synthesized, where does it act, and how long does its activity persist in vivo? Schematic shows a representative image of a neuron expressing NAPE-PLD in its axon terminals. 2. Does the co-release of GABA with glutamate and/or neuropeptides (CCK or VIP or both) occur in vivo following physiological level of activity? Does glutamate co-release trigger retrograde eCB signaling in VGLUT3 CCKBCs? 3. What are the postsynaptic molecules that interact with the receptor tyrosine kinase ErbB4 to determine the specific targets of CCKBC synapses? Slitrk3 is a potential candidate. 4. What are the in vivo roles of CB₁ receptor- and CCK-expressing, dendritically projecting INs (CCKdIN) in regulating dendritic information processing? 5. Do CCKBCs preferentially regulate distinct subclasses of PCs such as the superficial versus deep CA1 PCs in behaving animals? 6. How does eCB signaling regulate axonal pathfinding and shape spontaneous GABAergic network dynamics (GABAergic GDPs) in developing neuronal circuits in vivo? 7. What are the physiologic and pathologic roles of astrocytes and microglia in the eCB system? Astrocytes contain CB₁ receptors, as well as eCB degradation enzymes MAGL and FAAH. Meanwhile, microglia likely contain CB₂ receptors and eCB the synthesis and degradation enzymes DAGLβ and ABHD12. What are the roles of these receptors and machinery in shaping eCB signaling during behavior? 8. Does CB₁-mediated tonic inhibition of GABA release exist in vivo, and does it influence the probability of GABA release in a functionally relevant manner? The schematic illustration depicts how blockade of the intrinsic, ligand-free activity of the CB₁ receptor with an inverse agonist leads to a robust increase in the unitary evoked IPSCs originating from a presynaptic CCKBC and recorded from a postsynaptic PC. By contrast, blocking eCB ligand synthesis does not have similar effect. 9. The schematic depicts iLTD that is absent when CB₁ receptors are blocked. What are the physiological conditions (e.g., theta-burst firing) that can induce long-term depression of cannabinoid-sensitive inhibition, and what are its functional effects in vivo? 10. Can physiologically occurring dendritic plateau potentials trigger DSE in behaving animals, and does it play a role in BTSP? 11. How does neuronal hyperactivity, and the resulting increase in 2-AG, interact with synaptic signaling (through CB₁ activation) and inflammatory signaling (through prostaglandin synthesis) in disease processes?

Key features of the eCB system

(A) Molecular structures of 2-AG and AEA. (B) 2-AG is formed from its precursor diacylglycerol (DAG) by diacylglycerol lipase (DAGL). It is broken down into arachidonic acid (AA) by monoacylglycerol lipase (MAGL). AEA is formed from its precursors by the enzyme N-arachidonoyl phosphatidylethanolamine-specific phospholipase D (NAPE-PLD). It is broken down into AA by fatty acid amide hydrolase (FAAH). (C) Expression of CB₁ receptors within the reward circuit. HIPP, hippocampus; PFC, prefrontal cortex; GP, globus pallidus; dlStr, dorsolateral striatum; VP, ventral pallidum; NAc, nucleus accumbens; VTA, ventral tegmental area; BLA, basolateral amygdala; CeA, central nucleus of the amygdala; BNST, bed nucleus of the stria terminalis. Reproduced, with permission, from Curran et al. (D) Cannabinoid agonist application leads to the illustrated whole-organismal effects collectively referred to as the cannabinoid triad. (E) Pathways leading to CB₁ receptor-mediated suppression of GABA release from CCKBCs. Phasic inhibition leads to eCB-mediated short-term depression (eCB-STD) and can be divided into calcium (Ca²⁺)-driven, receptor-driven, and synaptically driven, based on experimental protocols that have been used to successfully elicit eCB-STD.^–, DSI is a form of calcium-driven eCB-STD. In this process, a depolarization-dependent increase in intracellular calcium at the postsynaptic cell leads to the production and subsequent release of 2-AG, which then activates presynaptic CB₁ receptors to suppress GABA release.^– Receptor-driven eCB-STD occurs when G_q/11-coupled receptors stimulate phospholipase Cb1 (PLCb1), which leads to the production of 2-AG after several intermediate steps.^, Although this form of eCB-STD depends on postsynaptic PLCb1 stimulation and can occur without a postsynaptic increase in calcium, it is believed to be more physiologically relevant when it occurs alongside increased intracellular calcium.^, Finally, synaptically driven eCB-STD occurs through repeated electrical stimulation of excitatory inputs, leading to 2-AG production through the same mechanisms described above.^, Dotted arrow indicates incompletely identified enzymatic pathway. (F) Schematic of a paired recording showcasing DSI. Stimulation of postsynaptic PC leads to suppression of unitary IPSCs that are evoked by the presynaptic CCKBC. (G) Schematic of a paired recording showcasing tonic inhibition. Application of a CB₁ receptor antagonist/inverse agonist (AM251) leads to an increase in the probability of GABA release and thus in the average amplitude of the unitary evoked IPSCs following presynaptic spikes. Dotted trace depicts silent GABAergic synapses that are occasionally found at this synaptic connection (see outstanding questions).

Cellular and molecular BCs dichotomy

(A) PVBCs and CCKBCs exhibit different firing patterns (top) and receive different levels of excitation (middle). PVBC firing leads to synchronous postsynaptic IPSCs, whereas CCKBC firing leads to asynchronous postsynaptic IPSCs (bottom). Inserts illustrate excitatory synaptic potentials evoked by electrical stimulation (black: PC, in color: PVBC or CCKBC). Thin lines symbolize individual trials, and bold lines symbolize averages. (B) Embedding of cortical INs classified by transcriptome, rendered using the Allen Brain Cell Atlas. Individual neurons in the CTX-MGE-GABA (red) and CTX-CGE-GABA (blue) classes are plotted. GABAergic INs are classified based on the location of their birth in the embryonic brain, before migrating to their destination during development. Cells born in the CGE are molecularly distinct from cells born in the MGE. Top: the embedding is colored by class or supertype. Bottom: the embedding is colored by the mRNA level of selected genes. Dashed circles indicate regions of the embedding likely containing PV (orange) and CCK (blue) INs. Note that Pvalb and Sncg clearly delineate non-overlapping populations, indicating that PV and CCK cells are transcriptionally distinct. Genes associated with CCKBCs, such as Cck and Cnr1, however, while quantitatively highly expressed in CCK INs, are also expressed at lower levels across several IN types, limiting their usefulness as markers (see also Box 2).