Diacylglycerol lipase α manipulation reveals developmental roles for intercellular endocannabinoid signaling

Abstract

Endocannabinoids are small signaling lipids, with 2-arachidonoylglycerol (2-AG) implicated in modulating axonal growth and synaptic plasticity. The concept of short-range extracellular signaling by endocannabinoids is supported by the lack of trans-synaptic 2-AG signaling in mice lacking sn-1-diacylglycerol lipases (DAGLs), synthesizing 2-AG. Nevertheless, how far endocannabinoids can spread extracellularly to evoke physiological responses at CB₁ cannabinoid receptors (CB₁Rs) remains poorly understood. Here, we first show that cholinergic innervation of CA1 pyramidal cells of the hippocampus is sensitive to the genetic disruption of 2-AG signaling in DAGLα null mice. Next, we exploit a hybrid COS-7-cholinergic neuron co-culture system to demonstrate that heterologous DAGLα overexpression spherically excludes cholinergic growth cones from 2-AG-rich extracellular environments, and minimizes cell-cell contact in vitro. CB₁R-mediated exclusion responses lasted 3 days, indicating sustained spherical 2-AG availability. Overall, these data suggest that extracellular 2-AG concentrations can be sufficient to activate CB₁Rs along discrete spherical boundaries to modulate neuronal responsiveness.

Figure 1. DAGLα localization in the fetal cholinergic basal forebrain.

(a) Schema of 2-AG signaling during fetal development and in adulthood. Note the lack of both MGL in growth cones and glial 2-AG inactivation during development, allowing 2-AG spread. (b,b₁) CB₁R mRNA in the medial septum (ms) at embryonic day (E)18.5 and postnatal day 1 (P1). (b₂,b₃) The lack of hybridization signal in sense control experiments (neocortex, P1) confirmed detection specificity. (c,c₁) A subpopulation of ChAT⁺ septal neurons was immunoreactive for CB₁Rs in neonates. (c₂–c₄) In the neonatal hippocampus, cholinergic afferents likely harbored CB₁Rs given the lack of physical signal separation for ChAT and CB₁R immunoreactivities. Solid and open arrowheads pinpoint the presence and lack of co-localization for select marker combinations, respectively. (c₅) Colocalization coefficients for cholinergic (ChAT⁺) and CB₁R⁺ boutons in the CA1 subfield of the neonatal hippocampus. Data on dual-labeled terminal subsets were expressed as the percentage of all cholinergic or CB₁R-containing synapses per field of view. (d–d₁) A subpopulation of p75^NTR+ neurons expressed DAGLα. p75^NTR+ neurons, assumed to acquire cholinergic phenotype, were generally situated proximal to DAGLα⁺ cells (arrowheads). (d₂–d₄) Representative images of cholinergic neurites coursing along DAGLα⁺ neurons and of adjacent p75^NTR+-DAGLα cell pairs in the medial septum. (d₅) Quantitative analysis of the portion of p75^NTR+ processes that apposed DAGLα⁺ perikarya (representative configuration: d₃,d₄), and of p75^NTR+ perikarya in the vicinity (<10 μm) of a DAGLα⁺ cell. (e,e₁) p75^NTR+ fibers (dashed lines) were found interspersed with DAGLα⁺ processes in the E18.5/P1 corpus callosum. Arrowheads indicate physical separation between fibers. (e₂) Distances between p75^NTR+ and DAGLα⁺ parallel fibers. Abbreviations: cc, corpus callosum; ml, midline; ls, lateral septum; pyr, pyramidal cells. Data were expressed as means ± s.e.m.; Scale bars = 100 μm (b₁), 20 μm (c), 10 μm (c₂,d₂–d₄), 5 μm (e₁), 2 μm (c₄).

Figure 2. Genetic deletion of DAGLα alters cholinergic innervation of the adult mouse hippocampus.

(a,a₁) DAGLα knock-out mice (DAGLα-KO) presented altered cholinergic afferentation of the hippocampus, manifesting as the loss of perisomatic “baskets” around CA1 pyramidal cells (*). Note the complete lack of DAGLα staining in the DAGLα-KO mouse, confirming staining specificity and genotype. (a₂) Paired high-resolution images of complete (in wild-type) or fragmented (in DAGLα-KO and DAGLβ-KO) perisomatic “baskets” (arrowheads) in the CA1 pyramidal layer. (b,b₁) The number of cholinergic perisomatic “baskets” decreased significantly in the CA1 pyramidal layer of DAGLα-KO but not DAGLβ-KO mice (b). In contrast, the cumulative density of ChAT⁺ presynaptic profiles did not change (b₁), emphasizing disrupted synapse targeting rather than impaired synaptogenesis. Grey and white circles differentiate knock-out animals on C57Bl6 background (n = 2–3/group) obtained from Tanimura et al. and Gao et al., respectively. (c). Particle profiling tools revealed no appreciable difference amongst the size of ChAT⁺ synaptic profiles from DAGL knock-outs. Abbreviations: n, nucleus; n.s., non-significant; Pyr, pyramidal layer; **p < 0.01, *p < 0.05. Scale bars = 20 μm (a,a₁), 5 μm (a₂).

Figure 3. DAGLα overexpression does not affect COS-7 physiology and induces 2-AG accumulation.

(a–b₂) Endogenously produced DAGLα in non-transfected COS-7 (“parent”) cells is undetectable by indirect immunohistochemistry (open arrowheads; a–a₂). In contrast, DAGLα, when highly expressed in COS-7 cells transfected with a DAGLα-V5 vector, accumulates in the plasma membrane (solid arrowheads; b–b₂). (c) COS-7-DAGLα cells expressed DAGLα protein and (c₁) synthesized 2-AG in a tetrahydrolipstatin (THL)-sensitive fashion. (d–d₃) DAGLα overexpression did not influence COS-7 cell morphology (d,d₁), including their surface area (d₂) and density in culture (d₃). Data were normalized to mean values from COS-7 “parent” cells. Abbreviations: n.s. = non significant. Scale bars = 20 μm (b,d₁).

Figure 4. Extracellular 2-AG promotes spherical exclusion of cholinergic growth cones (GCs).

(a) Cholinergic neurons from C56Bl6 mice co-cultured with COS-7-DAGLα cells allowed measurement of GC exclusion triggered by extracellular 2-AG. (a₁) Schema of the in vitro method and the parameters determined and plotted in (b₂–f). (b,b₁) Representative images of cholinergic GCs apposing control (“parent”) or DAGLα-overexpressing COS-7 cells (COS-7-DAGLα). Arrows point to filopodia. Dotted lines indicate the membrane surface of COS-7(-DAGLα) cells with arrowheads pinpointing their proximal segment facing the GCs. (b₂) Percentage of GCs that were unable to make contact upon DAGLα overexpression. (b₃) The distance (in μm) of cholinergic GCs from COS-7(-DAGLα) cells. GCs gradually approached COS-7 cells in control. DAGLα overexpression occluded this response, suggesting GC repulsion by extracellular 2-AG. DAGLα overexpression did not affect the angle at which the GC faced the proximal surface of the COS-7 cell (c), neurite outgrowth per se (d) or the distance of cholinergic somata from COS-7 cells (e). Upon making contact, cholinergic neurites were no longer affected by DAGLα overexpression (f). Data were averaged from 3 (1,2DIV) or 4 (3DIV) separate experiments; n = 25–77 cells/group. Abbreviations: DIV, days in vitro; N, neuron; n, nucleus. Data were expressed as means ± s.e.m. except for (b₃) showing a combination of individual data points and mean values (horizontal lines); **p < 0.01, *p < 0.05, ⁺p < 0.1. Scale bars = 10 μm (a), 3 μm (b).

Figure 5. Pharmacological inhibition of DAGLα or CB₁Rs prevents chemorepulsion of cholinergic growth cones (GC).

(a) Representative photomicrographs of cholinergic neurite-COS-7(-DAGLα) cell contacts upon inhibiting DAGLs (THL, O-3841) or CB₁R antagonism (AM 251, O-2050) at 3DIV. Arrows point to cholinergic neurites. Dotted lines indicate the membrane surface of COS-7(-DAGLα) cells. (b,c) Disrupting excess 2-AG signaling prevented chemorepulsion of cholinergic GCs. None of the ligands affected the angle of approach (d), neurite outgrowth (e), or cholinergic-to-COS-7 cell distance (f). Data were expressed as means ± s.e.m. except for (c) where individual data points and mean values (horizontal lines) were plotted. Data were averaged from duplicate samples from 2 independent experiments; n = 12–30 cells/group. **p < 0.01, *p < 0.05, ⁺p < 0.1. Scale bar = 20 μm (a).