Molecular and morphological configuration for 2-arachidonoylglycerol-mediated retrograde signaling at mossy cell-granule cell synapses in the dentate gyrus

Abstract

2-Arachidonoylglycerol (2-AG) is the endocannabinoid that mediates retrograde suppression of neurotransmission in the brain. In the present study, we investigated the 2-AG signaling system at mossy cell (MC)-granule cell (GC) synapses in the mouse dentate gyrus, an excitatory recurrent circuit where endocannabinoids are thought to suppress epileptogenesis. First, we showed by electrophysiology that 2-AG produced by diacylglycerol lipase α (DGLα) mediated both depolarization-induced suppression of excitation and its enhancement by group I metabotropic glutamate receptor activation at MC-GC synapses, as they were abolished in DGLα-knock-out mice. Immunohistochemistry revealed that DGLα was enriched in the neck portion of GC spines forming synapses with MC terminals, whereas cannabinoid CB(1) receptors accumulated in the terminal portion of MC axons. On the other hand, the major 2-AG-degrading enzyme, monoacylglycerol lipase (MGL), was absent at MC-GC synapses but was expressed in astrocytes and some inhibitory terminals. Serial electron microscopy clarified that a given GC spine was innervated by a single MC terminal and also contacted nonsynaptically by other MC terminals making synapses with other GC spines in the neighborhood. MGL-expressing elements, however, poorly covered GC spines, amounting to 17% of the total surface of GC spines by astrocytes and 4% by inhibitory terminals. Our findings provide a basis for 2-AG-mediated retrograde suppression of MC-GC synaptic transmission and also suggest that 2-AG released from activated GC spines is readily accessible to nearby MC-GC synapses by escaping from enzymatic degradation. This molecular-anatomical configuration will contribute to adjust network activity in the dentate gyrus after enhanced excitation.

Figure 1.

Retrograde suppression of excitation is absent at MC–GC synapses in the dentate gyrus of DGLα-knock-out mice. A, Sample traces and average time courses of the amplitudes of MC–GC EPSCs of wild-type (WT; open circles) and DGLα-knock-out (KO) mice (blue diamonds) before and after depolarization (from −70 to 0 mV, for 3 s). B, Top, Retrograde suppression at MC–GC synapses after combined depolarization (5 depolarization pulses from −70 to 0 mV with 100 ms duration at 1 Hz) with bath application of DHPG (10 μm) in wild-type mice (filled circles). Note that depolarization alone failed to induce noticeable suppression (open circles). Bottom, No suppression was induced by either depolarization alone (blue diamonds) or combined depolarization with DHPG application (filled diamonds) at MC–GC synapses of DGLα-KO mice. Protocols of depolarization and DHPG application were the same as in wild-type mice. Sample EPSC traces (a–d) were obtained at the time points indicated in the graphs. Calibration: A, 100 pA, 5 ms (WT) and 50 pA, 5 ms (DGLα-KO); B, 20 pA, 5 ms (WT) and 50 pA, 5 ms (DGLα-KO). **p < 0.001 (two-factor repeated-measures ANOVA followed by post hoc Bonferroni test).

Figure 2.

DGLα is highly expressed in GCs and MCs of the dentate gyrus. Fluorescent in situ hybridization (FISH) for DGLα mRNA is shown. A, B, Single FISH in the mouse brain (A) and dentate gyrus (DG; B). Note the strong signals in the pyramidal cell layer of the hippocampus (Hi) and the GC layer (GL) of the DG. Hilar cells also express DGLα mRNA in the polymorphic layer (PL) of the DG. C, D, Double FISH for DGLα (red) and VGluT1 (C, green) or GAD67 (D, green) mRNAs. Hilar cells expressing DGLα mRNA at high levels (C, D, arrows) coexpress VGluT1 mRNA, but not GAD67 mRNA, indicating prominent expressions of DGLα mRNA in MCs. Low levels of DGLα mRNA are detected in some cells expressing GAD67 mRNA (D, arrowheads). Cb, Cerebellum; CPu, caudate–putamen; Cx, neocortex; Md, midbrain; MO, medulla oblongata; Th, thalamus, ML, molecular layer. Scale bars: A, 1 mm; B–D, 20 μm.

Figure 3.

Distribution of MC terminals (A–F) and immunofluorescence for DGLα (G–K) in the inner molecular layer of the dentate gyrus. A, B, Double immunofluorescence for VGluT1 (red) and calretinin (green). VGluT1-positive boutons are densely distributed and exclusively colabeled for calretinin in the inner molecular layer (IML; B, arrowheads). C, D, Triple immunofluorescence for VGuT2 (red), VIAAT (blue), and calretinin (green). E, F, Double immunofluorescence for VAChT (red) and calretinin (green). Note that VGluT2/VIAAT-positive terminals from the supramammillary nucleus (C, D, arrowheads) or VAChT-positive cholinergic terminals (E, F, arrowheads) are negative for calretinin and sparsely distributed in the MC-recipient IML. G–J, Immunofluorescence for DGLα. G, Single immunofluorescence reveals dense punctate immunolabeling for DGLα all over the molecular layer (ML). H, Double immunofluorescence for DGLα (red) and somatodendritic marker MAP2 (green) in the IML. Tiny punctate labeling for DGLα is distributed along MAP2-labeled dendrites. I, Double immunofluorescence for DGLα (red) and terminal marker synaptophysin (green) in the IML. DGLα is apposed to synaptophysin-labeled nerve terminals (arrowheads). J, Triple labeling for mGluR5 (J₁,₃, red), DGLα (J₂,₄, red), and F-actin (J>₃,₄, green) in the IML. Note that F-actin (arrows) is overlapped well with mGluR5 but aligned side by side with DGLα. K, Summary bar graph showing the mean correlation coefficient of two fluorescent signals: between F-actin and mGluR5 (left), between F-actin and DGLα (middle), and between F-actin and MAP2 (right). The numbers and error bars indicate the mean correlation coefficient and SEM, respectively. Seven sets of images were analyzed for each pair. GL, GC layer. Scale bars: A, C, E, G, 10 μm; B, D, F, H–J, 2 μm.

Figure 4.

Preferential distribution of DGLα on dendritic spines and shafts of GCs with the highest accumulation in the spine neck. A, B, Preembedding silver-enhanced immunogold electron microscopy. Metal particles for DGLα are located on the cell surface of dendritic spines (Sp; blue) and shafts (Dn; yellow) of GCs but not on that of MC terminals (MCT). Note two serial images in B₁ and B₂, in which the spine neck portion is consistently labeled. C, Bar graph showing the mean labeling density for DGLα, i.e., the mean number of metal particles per 1 μm of the plasma membrane in each element of wild-type (WT) and DGLα-knock-out (KO) mice. The numbers in and above parentheses on each column indicate the total length of measured plasma membrane or the mean labeling density, respectively. D–F, Histograms showing intraspine distribution of DGLα. The abscissa indicates the distance (in nanometers) from the edge of the PSD (D) or from the spine–dendrite border (E) and the normalized distance from the edge of the PSD (0%) to the spine–dendrite border (100%) (F). The ordinates indicate the percentage of metal particles falling in each bin. The total number of 71 metal particles was collected from 54 GC spines. Scale bars, 500 nm.

Figure 5.

CB₁-expressing MC terminals exist at high densities in the inner molecular layer. A–D, Triple immunofluorescence for CB₁ (red), VGluT1 (blue), and calretinin (A, B, green) or VIAAT (C, D, green) in the dentate gyrus. All images were taken from semithin cryosections (200 nm in thickness). B and D are high-power images in the inner molecular layer (IML). VGluT1-positive glutamatergic terminals are densely distributed, and most of them are positive for calretinin (B, arrowheads) and express low levels of CB₁ (B, D, arrowheads). In contrast, VIAAT-positive inhibitory terminals are sparsely distributed, and some of them express high levels of CB₁ (C, D, arrows). Double arrowheads in B indicate calretinin-positive and VGluT1-negative elements; these elements are mostly nonterminal portions of MC axons, as they are also negative to VIAAT (data not shown). E, A schematic drawing representing the distribution of centers of CB₁/VGluT1-positive terminals (blue spots) and CB₁/VIAAT-positive terminals (yellow spot), as reconstructed from the image in D. Concentric circles having the radius of 0.5, 1.0, 1.5, 2.0, 2.5, and 3.0 μm are overlaid on two selected blue spots. Note that blue spots included in given concentric circles outnumber yellow spots. F, Graph showing the average number of CB₁/VGluT1-positive terminals or CB₁/VIAAT-positive terminals in concentric circles with the radius ranging from 0 to 3.0 μm. n = 108 CB₁/VGluT1-posotive terminals. Error bars indicate SD. ML, Molecular layer; GL, GC layer. Scale bars: A–C, 5 μm; D, 2 μm.

Figure 6.

Immunoelectron microscopy showing the distribution of CB₁ in MC axons and its gradient toward the terminal portion. A, A schematic illustration showing the definition of the terminal (synaptic vesicle-containing portion), juxtaterminal (≤0.5 μm from the edge of terminal), and extraterminal portions (> 0.5 μm) of MC axons. Black lines represent the PSD on GC spines and the active zone of MC terminals. B, Preembedding silver-enhanced immunogold electron microscopy. Metal particles for CB₁ are selectively localized on the surface of MC terminals (MCT) and axons (Ax). Arrows indicate the border between the terminal and juxtaterminal portions, whereas arrowheads indicate the edge of the PSD. Sp, GC spine. C, Bar graph showing the mean labeling density for CB₁, i.e., the mean number of metal particles per 1 μm of the axolemma. The number on the top of each column indicates the mean labeling density. n = 14 axons. Scale bar, 200 nm.

Figure 7.

In situ hybridization showing distinct cellular expression of MGL mRNA in the dentate gyrus. A, Fluorescent in situ hybridization (FISH) and isotopic in situ hybridization (RI-ISH; inset) for MGL mRNA in parasagittal brain sections. B, FISH for MGL mRNA in the hippocampus formation. Principal neurons are labeled at moderate levels in the dentate GC layer (GL) and hippocampal pyramidal cell layer (Py). Note that medium-sized (arrows) or small (arrowheads) cells express MGL mRNA at high or low levels, respectively. C–E, Double FISH for MGL mRNA (red) with GAD67 (C, green), GLAST (D, green), or VGluT1 (E, green) mRNA. Note that MGL mRNA is expressed at high levels in inhibitory interneurons labeled for GAD67 mRNA (C, arrows) and at low levels in astrocytes labeled for GLAST mRNA (D, arrowheads). In contrast, MGL mRNA is negative in MCs identified by VGluT1 mRNA labeling (E, double arrowheads). Cb, Cerebellum; CPu, caudate–putamen; Cx, neocortex; DG, dentate gyrus; GP, globus pallidus; Hi, hippocampus; Md, midbrain; MO, medulla oblongata; Ob, olfactory bulb; SNr, substantia nigra pars reticulata; SC, superior colliculus; Th, thalamus; CA1, CA1 subregion of the hippocampus; Or, stratum oriens; Ra, stratum radiatum; LMo, lacunosum moleculare; ML, molecular layer; PL, polymorphic layer. Scale bars: A, 1 mm; B, 200 μm; C–E, 20 μm.

Figure 8.

MGL is expressed in astrocytes and some inhibitory terminals in the dentate gyrus and hippocampus. A–C, Immunofluorescence for MGL in parasagittal brain sections of wild-type (WT; A, C) and MGL-knock-out (KO; B) mice. C, An enlarged view of the hippocampus formation. D, E, Double immunofluorescence for MGL (red) and astrocytic marker GFAP (D, green) or GLAST (E, green) in the inner molecular layer (IML) of the dentate gyrus. Note the overlap of MGL with GFAP in perikarya and shaft processes of astrocytes (D, arrows) and with GLAST in their peripheral processes (E, arrows). The arrowhead in E indicates intense MGL in GLAST-immunonegative elements (putative inhibitory terminals). F, Triple immunofluorescence for MGL (red), VIAAT (F1, green), and CB₁ (F3, blue) in the IML. Note that intense MGL is found in many VIAAT-positive inhibitory terminals, which include both CB₁-positive (arrowheads) and CB₁-negative (double arrowheads) terminals. G, Triple immunofluorescence for MGL (red), VGluT1 (G1, green), and calretinin (G3, blue) in the IML. Note that there is no MGL immunoreactivity in MC terminals colabeled for VGluT1 and calretinin (arrows). H–J, Double immunofluorescence for MGL (red) with CNPase (H, green), Iba1 (I, green), or MAP2 (J, green). Note that there is no MGL labeling in CNPase-positive oligodendrocytes, Iba1-positive microglia, or MAP2-positive neuronal dendrites. K, L, Double immunofluorescence for MGL (red) and VGluT1 (green) in the hippocampal CA3 (K) and CA1 (L). M, Triple immunofluorescence for MGL (red), VIAAT (green), and CB₁ (blue) in the hippocampal CA1. N, Double immunofluorescence for MGL and GLAST in the hippocampal CA1. In the CA3, intense immunoreactivity is detected in VGluT1-labeled mossy fiber terminals (K). In the CA1, MGL immunoreactivity is detected in VGluT1-labeled excitatory terminals of Schaffer collaterals (L, arrows), VIAAT-labeled inhibitory terminals including both CB₁-positive (M, arrowheads) and CB₁-negative (M, double arrowheads) ones, and GLAST-labeled astrocytes (N, arrowheads). Cb, Cerebellum; CPu, caudate–putamen; Cx, neocortex; DG, dentate gyrus; GP, globus pallidus; Hi, hippocampus; Md, midbrain; MO, medulla oblongata; Ob, olfactory bulb; SNr, substantia nigra pars reticulata; SC, superior colliculus; Th, thalamus; Su, subiculum; Or, stratum oriens; Py, pyramidal cell layer; Ra, stratum radiatum; LMo, lacunosum moleculare; ML, molecular layer; GL, GC layer; PL, polymorphic layer; Lu, stratum lucidum. Scale bars: A, B, 1 mm; C, 200 μm; D, 20 μm; E–G, L–N, 2 μm; H–K, 5 μm.

Figure 9.

MGL is selectively localized in glial elements and inhibitory terminals in the inner molecular layer of the dentate gyrus. A, B, Preembedding silver-enhanced immunogold electron microscopy. Metal particles are distributed inside perikarya and lamellate processes of glial cells (Gl; purple) but not in MC terminals (MCT) or GC spines (Sp). C, Two serial images showing consistent MGL labeling in inhibitory terminals (InT) forming symmetrical synapses (arrows) with dendritic shafts (Dn). D, Bar graph showing the number of metal particles for MGL per 1 μm² of the cytoplasmic area in each subcellular element of wild-type (WT; open bars) and MGL-knock-out (KO; filled bars) mice. The numbers in and above parentheses on the top of each column indicate the total area analyzed and the mean labeling density, respectively. Scale bars, 500 nm.

Figure 10.

Serial electron microscopic analysis for subcellular elements contacting to GC spines. A, B, Four serial electron micrographs showing two adjacent MC–GC synapses (A) and their 3D reconstructed images (B). Two GC spines, Sp-1 and Sp-2, are innervated by MC terminal MCT-1 or MCT-2, respectively (black arrowheads in A, red regions in B). Note that Sp-1 is also contacted nonsynaptically by MCT-2 (white arrowheads in A, white dashed line in B). Each spine is also contacted by nonterminal axons (Ax) and glial elements (Gl). C, GC spines (Sp) are also contacted nonsynaptically by inhibitory terminals (InT) forming symmetrical synapses (arrow) onto dendritic shafts (Dn). D, Bar graph showing the frequency distribution of GC spines (n = 57) having nonsynaptic contact with zero, one, two, and three MC terminals. Note that the majority (80.7%) of GC spines contact with one to three noninnervating MC terminals. E, Bar graph showing the percentage of the total surface area of GC spines (n = 15) covered with presynaptic (Pre), postsynaptic (Post), and glial elements. The numbers and error bars indicate the mean percentage and SEM. N.D., Elements that could not be identified. A circular graph shows the percentage of GC spine surface area covered with respective presynaptic elements, i.e., innervating MC terminals, noninnervating MC terminals, InT, and Ax. Scale bars, 200 nm.

Figure 11.

The molecular–anatomical configuration of 2-AG-mediated retrograde signaling and cross talk at MC–GC synapses in the dentate gyrus. The molecular–anatomical configuration of 2-AG-mediated signaling at MC–GC synapses includes (1) wide distribution of DGLα in dendritic spines and shafts of GCs with the highest accumulation at the spine neck portion, (2) accumulation of CB₁ in the terminal portion of MC axons, (3) spatial proximity among MC–GC synapses that causes frequent contact of GC spines to other MC terminals making synaptic contacts with other GC spines, and (4) incomplete surrounding of GC spines by MGL-expressing elements (astrocytes and some inhibitory terminals). These characteristics provide the framework fundamental to 2-AG-mediated retrograde signaling and also suggest 2-AG-mediated intersynaptic cross talk among local MC–GC synapses. This molecular–anatomical configuration will be important to adjust network activity in the dentate gyrus after enhanced excitability.