Intracellular labeling of neurons in the medial accessory olive of the cat: III. Ultrastructure of axon hillock and initial segment and their GABAergic innervation

The gamma-aminobutyric acid (GABA) synaptic input of identified axons in the cat inferior olive was studied by use of combination of intracellular labeling with horseradish peroxidase and postembedding gold-immunocytochemistry. With this technique olivary cells were physiologically identified and light microscopically reconstructed, and the horseradish peroxidase reaction product and the immunogold labeling were subsequently simultaneously visualized for electron microscopic investigation with the use of serial ultrathin sections. The axons of cell type I (characterized by dendrites which radiate away from the cell body) originated from the soma, whereas those of type II neurons (characterized by dendritic trees which curve back towards the soma) were derived from a primary dendrite. The axons of olivary neurons stand out by the length of their axon hillock (up to 21 microns) and initial segment (up to 40 microns). The hillock forms various spiny appendages which were located within glomeruli together with dendritic spines of other olivary neurons. Axonal spines of type II neurons were more numerous and complex looking than those of type I. The axonal spines, the shaft of the axon hillock, and the transition between the hillock and initial segment were primarily innervated by GABAergic terminals (65%) but non-GABAergic terminals (35%) were present as well. The terminals apposed to the axons of type I neurons contacted mainly the axonal shafts, whereas most of the terminals adjacent to the axons of type II neurons established synaptic contacts with the axonal spines. The initial segments were largely devoid of synaptic input. Distally, the initial segment acquired a myelin sheath.     

Gamma-aminobutyric acid in the medial rat nucleus accumbens: ultrastructural localization in neurons receiving monosynaptic input from catecholaminergic afferents

Neurons containing gamma-aminobutyric acid (GABA) in the medial portion of the adult rat nucleus accumbens were characterized with respect to their ultrastructure, sites of termination, and catecholaminergic input. Antisera against GABA-conjugates and the catecholamine-synthesizing enzyme, tyrosine hydroxylase (TH), were localized within single sections by means of peroxidase-antiperoxidase (PAP) and immunoautoradiographic labeling methods. Peroxidase reaction product indicating GABA-like immunoreactivity (GABA-LI) was seen in medium-size (15-20 microns) perikarya containing either round and unindented or invaginated nuclear membranes. The cells with invaginated nuclei were few in number and usually exhibited more intense peroxidase reaction product in sections collected at the same distance from the surface of the tissue. Reaction product for GABA was also detected in proximal (1.5-3.0 microns) dendrites, axons, and terminals. Terminals with GABA-LI formed symmetric junctions on perikarya, proximal dendrites, and dendritic spines of neurons that usually lacked detectable immunoreactivity. Many of the GABAergic terminals also were apposed directly to other unlabeled terminals and to terminals exhibiting either peroxidase labeling for GABA or immunoautoradiographic labeling for TH. Many of the unlabeled terminals associated with the GABAergic axons formed asymmetric junctions on dendritic spines. From 138 TH-labeled, principally dopaminergic terminals that were examined in the medial nucleus accumbens, 4% were associated with the somata of GABAergic neurons and another 14% formed symmetric junctions with proximal dendrite showing GABA-LI. The remaining TH-immuno-reactive terminals either lacked recognizable densities or formed symmetric synapses on unlabeled dendrites and spines. A few of the unlabeled dendrites, as well as those containing GABA-LI, received symmetric synapses from both catecholaminergic and GABAergic terminals. We conclude that in the medial portion of the rat nucleus accumbens, GABA is localized to two morphologically distinct types of neurons, one or both of which receive monosynaptic input from catecholaminergic afferents, and that GABAergic terminals form symmetric synapses on other principally non-GABAergic neurons. The results also support earlier physiological evidence showing that GABA may modulate the output of other GABAergic and non-GABAergic neurons through presynaptic associations.     

Inhibitory contacts on dendritic spines of the dentate fascia

GABA-containing axon terminals were observed in the distal two-thirds of the dentate molecular layer to contact spines and dendrites of the granule cells. These contacts have the morphological characteristics of inhibitory synapses: they contain pleomorphic vesicles and have symmetrical junctional specializations. Convergence of an asymmetrical, non-GABAergic and a symmetrical, GABAergic synapse on one spine was often observed.     

NOS-positive local circuit neurons are exclusively axo-dendritic cells both in the neo- and archi-cortex of the rat brain

Neuronal nitric oxide synthase (nNOS)-containing neurons and axon terminals were examined in the rat somatosensory and temporal neocortex, in the CA3/a-c areas of Ammon's horn and in the hippocampal dentate gyrus. In these areas, only nonpyramidal neurons were labeled with the antibody against nNOS. Previous observations suggested that all nNOS-positive nonpyramidal cells are GABAergic local circuit neurons, which form exclusively symmetric synapses. In agreement with this, nNOS-positive axon terminals in the hippocampal formation formed symmetric synapses exclusively with dendritic shafts. In the neocortex, in contrast, in addition to the nNOS-positive axon terminals that formed synapses with unlabeled spiny and aspiny dendrites and with nNOS-positive aspiny dendrites, a small proportion of the nNOS-positive axon terminals formed symmetric synapses with dendritic spines. These results suggest that nNOS-positive local circuit neurons form a distinct group of axo-dendritic cells displaying slightly different domain specificity in the archi- and neocortex. However, nNOS-positive cells show no target selectivity, because they innervate principal cells and local circuit neurons. Afferents to the NOS-positive cells display neither domain nor target selectivity, because small unlabeled terminals formed synapses with both the soma or dendrites of nNOS-positive neurons and an adjacent unlabeled dendrite or spine in both the hippocampal formation and in neocortex.     

Synaptic relationships between axon terminals from the mediodorsal thalamic nucleus and gamma-aminobutyric acidergic cortical cells in the prelimbic cortex of the rat

Although the reciprocal interconnections between the prefrontal cortex and the mediodorsal nucleus of the thalamus (MD) are well known, the involvement of inhibitory cortical interneurons in the neural circuit has not been fully defined. To address this issue, we conducted three combined neuroanatomical studies on the rat brain. First, the frequency and the spatial distribution of synapses made by reconstructed dendrites of nonpyramidal neurons were identified by impregnation of cortical cells with the Golgi method and identification of thalamocortical terminals by degeneration following thalamic lesions. Terminals from MD were found to make synaptic contacts with small dendritic shafts or spines of Golgi-impregnated nonpyramidal cells with very sparse dendritic spines. Second, a combined study that used anterograde transport of Phaseolus vulgaris leucoagglutinin (PHA-L) and postembedding gamma-aminobutyric acid (GABA) immunocytochemistry indicated that PHA-L-labeled terminals from MD made synaptic junctions with GABA-immunoreactive dendritic shafts and spines. Nonlabeled dendritic spines were found to receive both axonal inputs from MD with PHA-L labelings and from GABAergic cells. In addition, synapses were found between dendritic shafts and axon terminals that were both immunoreactive for GABA. Third, synaptic connections between corticothalamic neurons that project to MD and GABAergic terminals were investigated by using wheat germ agglutinin conjugated to horseradish peroxidase and postembedding GABA immunocytochemistry. GABAergic terminals in the prelimbic cortex made symmetrical synaptic contacts with retrogradely labeled corticothalamic neurons to MD. All of the synapses were found on cell somata and thick dendritic trunks. These results provide the first demonstration of synaptic contacts in the prelimbic cortex not only between thalamocortical terminals from MD and GABAergic interneurons but also between GABAergic terminals and corticothalamic neurons that project to MD. The anatomical findings indicate that GABAergic interneurons have a modulatory influence on excitatory reverberation between MD and the prefrontal cortex.     

Differential synaptic distribution of the AMPA-GluR2 subunit on GABAergic and non-GABAergic neurons in the basolateral amygdala

The cellular and ultrastructural distribution patterns of the AMPA glutamate receptor subunit, GluR2, were determined in the rat basolateral amygdala. GluR2 immunoreactivity was widely and uniformly distributed in the basolateral nucleus, with both pyramidal and non-pyramidal neurons labelled. In fact, double label immunohistochemical analyses demonstrated that over 90% of the GABAergic interneurons were labelled for GluR2. Electron microscopic analyses further confirmed the presence of GluR2 in the soma and dendrites of GABAergic interneurons as well as in the soma, spines and dendritic shafts of pyramidal cells. As in our parallel study in the rat hippocampus, immunogold analyses revealed that GluR2 immunoreactivity was frequently preferentially located at asymmetric synapses on both pyramidal cell spines and shafts, as well as the dendritic processes and soma of GABAergic interneurons. However, the number of immunogold particles per labelled synapse on GABAergic neurons was significantly lower than at similar labelled asymmetric synapses on spines of presumed pyramidal cells. Given that the presence of GluR2 within the AMPA receptor complex decreases calcium flux, these data indicate that GABAergic local circuit neurons might possess AMPA receptors with higher calcium permeability on average than pyramidal cells, as has been suggested for hippocampus. Such cell class-specific differences in the subunit representation and resultant channel properties of AMPA receptors have implications for response properties as well as selective vulnerability of neurons within the basolateral nucleus of the amygdala.     

Dendroaxonic synapses in the substantia gelatinosa glomeruli of the spinal trigeminal nucleus of the cat

The glomeruli in the substantia gelatinosa layer of the spinal trigeminal nucleus of the cat contain three kinds of dendritic processes. One of these, the type 2 dendrite, contains large synaptic vesicles in its spine heads and in its shaft. The type 2 dendrite receives axodendritic synapses from primary trigeminal afferent (C) axons and an occasional axodendritic synapse from small axonal (P) endings with small synaptic vesicles. The type 2 dendrites in turn form dendroaxonic synapses on the C endings. The dendroaxonic synapse and the axodendritic synapse of the C ending typically occur in reciprocal pairs. The axodendritic synapse usually lies in the depths of scalloped depressions in the surface of the C ending while the dendroaxonic synapse is found on the rim of the depression. Type 1 spines, i.e., dendritic spines receiving axodendritic synapses from the primary ending and lacking synaptic vesicles, also receive dendrodendritic synapses from type 2 dendrites. The types 2 dendrite with its large, rounded synaptic vesicles is considered to be excitatory at its dendroaxonic and dendrodendritic synapses. The type 2 dendrites course from glomerulus to glomerulus receiving their excitatory input through the axodendritic synapses of C axons. A type 2 dendrite, in response to C axon excitation would activate type 1 spines directly through their dendrodendritic synapses (C→2→1) and indirectly by increasing transmitter release at the axodendritic synapses of the C axonal endings through their dendroaxonic synapses (2→C→1). The type 2 dendrites could serve two functions. First, they may prolong transmitter release from the axodendritic synapses of C axonal endings beyond the time of arrival of incoming action potentials because of the reciprocal pairing of dendroaxonic and axodendritic synapses (C?2). Second, they may extend the spatial range of the excitatory output of active primary afferent axons to type 1 spines of glomeruli whose primary afferent axons may be inactive (C→2→1).     

Spatial relationships between the terminations of somatic sensory motor pathways in the rostral brainstem of cats and monkeys. II. Cerebellar projections compared with those of the ascending somatic sensory pathways in lateral diencephalon

In order to classify the presynaptic elements contacting the principle class of globus pallidus neurons, electron microscopic examination of serial sections made from a medially located large globus pallidus neuron, labeled with intracellular horseradish peroxidase, was undertaken. In addition, the use of labeled and light microscopically reconstructed material allowed us to quantitatively determine the distribution of each bouton type along the soma and dendrites. Six types of presynaptic terminals contacting the labeled cell have been recognized. Type 1 endings, the most numerous (84%), make symmetrical contacts on all portions of the cell, except spines, contain large pleomorphic, and a few large dense-core vesicles. Type 2 endings are filled with small spherical-to-ellipsoidal synaptic vesicles. They make asymmetrical contacts only with higher-order dendrites and account for 12% of synaptic contacts onto the labeled neuron. Type 3 endings are large, contain sparsely distributed large pleomorphic vesicles, and make two symmetrical synapses per bouton, one onto a spine head and the other onto the underlying dendritic shaft. They are infrequent (0.2%), being found only in association with dendritic spines. Type 4 endings contain large pleomorphic synaptic vesicles and no dense-core vesicles. They make symmetrical contacts with the short primary dendrites. Type 5 endings contain a mixture of small clear pleomorphic vesicles and numerous large dense-core vesicles. They contact only the cell body and the short primary dendrites, making up 20% of somatic synaptic contacts but less than 1% of contacts onto dendrites. Type 6 boutons contain oval and flattened synaptic vesicles and establish symmetrical contacts with higher-order dendritic branches and the cell body.     

Reticulo-olivary circuits: An intracellular HRP study in the rat

Intracellular recordings were obtained from medullary reticular neurons subsequent to electrical stimulation of the ipsilateral or contralateral inferior cerebellar peduncle (ICP) and/or the midbrain. After recording physiological data, the neurons were intracellularly injected with horseradish peroxidase (HRP). Thirty-four HRP filled neurons were subjected to light microscopic analysis. They could be divided into two general groups: those which extend dendritic processes into the neuropil of the inferior olivary complex (n = 19); and those that have no anatomical relationship to the inferior olive (n = 15). These two populations of reticular neurons differ in their distribution, morphological characteristics and physiological responses. Neurons which extend dendritic processes into the inferior olive are located within 200 microns of the dorsal border of this nuclear complex, between the exiting fibers of the XIIth nerve and the raphe. The cell bodies are located in the nucleus reticularis gigantocellularis and are fusiform or multipolar in shape. Their dendrites extend for long distances in the mediolateral direction; are thin and relatively spine-free except at their distal tips where spines and varicose appendages are evident. Physiologically, midbrain stimulation elicits a fast rising hyperpolarization which is identified as an inhibitory postsynaptic potential. However, only rarely is a response observed subsequent to stimulation of either the ipsilateral or contralateral ICP. Dendrites from 4 neurons from the first group of reticular cells were analyzed at the ultrastructural level. Based on random and serial thin sections, the following features were noted: they contain numerous mitochondria when compared to olivary dendrites; they contribute to the postsynaptic elements within olivary synaptic clusters (glomeruli); and they exhibit focal clusters of synaptic vesicles although conventional synaptic complexes have not been observed. Reticular neurons of the second group, those that do not extend dendritic processes into the inferior olive, are located either lateral to the XIIth nerve or at distances greater than 200 microns from the dorsal border of the inferior olivary complex. Their cell bodies and dendrites are comparable morphologically to the reticular neurons whose dendrites do arborize in the inferior olive. However, rarely are the distal tips of their dendrites characterized by spines or varicose appendages. Physiologically, this population of reticular neurons respond to midbrain stimulation with a low amplitude, short latency depolarizing potential which is interrupted by a hyperpolarizing potential.(ABSTRACT TRUNCATED AT 400 WORDS)     

Synaptic organization of the mushroom body calyx in Drosophila melanogaster

The calyx neuropil of the mushroom body in adult Drosophila melanogaster contains three major neuronal elements: extrinsic projection neurons, presumed cholinergic, immunoreactive to choline acetyltransferase (ChAT-ir) and vesicular acetylcholine transporter (VAChT-ir) antisera; presumed gamma-aminobutyric acid (GABA)ergic extrinsic neurons with GABA-like immunoreactivity; and local intrinsic Kenyon cells. The projection neurons connecting the calyx with the antennal lobe via the antennocerebral tract are the only source of cholinergic elements in the calyces. Their terminals establish an array of large boutons 2-7 microm in diameter throughout all calycal subdivisions. The GABA-ir extrinsic neurons, different in origin, form a network of fine fibers and boutons codistributed in all calycal regions with the cholinergic terminals and with tiny profiles, mainly Kenyon cell dendrites. We have investigated the synaptic circuits of these three neuron types using preembedding immuno-electron microscopy. All ChAT/VAChT-ir boutons form divergent synapses upon multitudinous surrounding Kenyon cell dendrites. GABA-ir elements also regularly contribute divergent synaptic input onto these dendrites, as well as occasional inputs to boutons of projection neurons. The same synaptic microcircuits involving these three neuron types are repeatedly established in glomeruli in all calycal regions. Each glomerulus comprises a large cholinergic bouton at its core, encircled by tiny vesicle-free Kenyon cell dendrites as well as by a number of GABAergic terminals. A single dendritic profile may thereby receive synaptic input from both cholinergic and GABAergic elements in close vicinity at presynaptic sites with T-bars typical of fly synapses. ChAT-ir boutons regularly have large extensions of the active zones. Thus, Kenyon cells may receive major excitatory input from cholinergic boutons and considerable postsynaptic inhibition from GABAergic terminals, as well as, more rarely, presynaptic inhibitory signaling. The calycal glomeruli of Drosophila are compared with the cerebellar glomeruli of vertebrates. The cholinergic boutons are the largest identified cholinergic synapses in the Drosophila brain and an eligible prospect for studying the genetic regulation of excitatory presynaptic function.     

Spino-olivary termination on spines in cat medial accessory olive

Three functional regions of the inferior olive, the caudal medial accessory olive (cMAO) and the caudal and rostral dorsal accessory olive (DAO) receive input from the spinal cord. The present study determined how spinal inputs to cMAO interact with olivary neurons. These inputs were labeled by injections in cat lumbosacral of wheat germ agglutinin conjugated to horseradish peroxidase. The tracer was visualized with tetramethylbenzidine. The morphology of the labeled spino-olivary terminals and the relationship between these terminals and postsynaptic elements were determined. Spino-olivary terminals in cMAO displayed the morphological characteristics classically associated with excitatory synapses. Almost three quarters synapsed on spines, most of which contacted other spines, forming spine clusters. The majority of postsynaptic spines also received convergent input from apparently excitatory, nonlumbosacral afferents. This postsynaptic organization provides several possible benefits for the putative role of cMAO in the control of posture. An earlier study demonstrated that in DAO, almost three quarters of lumbosacral, spino-olivary terminals synapse on dendrites (Molinari: Neuroscience 27:425-435, 1988). Thus, lumbosacral afferents appear to differ fundamentally in the way in which they interact with neurons in cMAO and DAO. These results suggest that the way information is processed may be as important in determining the functional differences between olivary regions as what information is received.     

Immunocytochemical analysis of GABA-positive and calretinin-positive horizontal cells in the tiger salamander retina

By using immunocytochemical techniques, we demonstrate that there are two distinct, nonoverlapping populations of horizontal cells (HCs) in the tiger salamander retina: GABA-positive cells account for about 72% and GABA-negative (calretinin-positive) cells account for 28% of the total HC somas. The calretinin-positive HCs have relatively sparse and thick dendrites: soma diameter of 19.72 +/- 0.29 microm, and soma density of 140 +/- 13 cells/mm(2), morphological features very much like the A-type HCs described in the accompanying article. The GABA-positive HCs have thinner dendritic and coarse axon-terminal-like processes of higher density: soma diameter of 18 +/- 0.18 microm, and soma density of 364 +/- 18 cells/mm(2), features that very much resemble the B-type HCs and B-type HC axon terminals in the accompanying article. By using double and triple immunostaining techniques we found that only 18% of the non-GABAergic HC dendritic clusters contact rods, whereas the remaining 82% of the dendritic clusters contact cones. This is consistent with the physiological finding in the accompanying article that the A-type HCs are cone-dominated. On the other hand, 32% of GABAergic HC dendrites contact rod pedicles and 68% contact cone pedicles, consistent with the physiological finding that B-type HCs and B-type HC axon terminals receive mixed rod/cone inputs. Detailed confocal microscope analysis shows that 4% rods, 6% principal double cones/single cones, and 100% accessory double cones contact calretinin-positive HCs, and 79% rods, 100% principal double cones, 14% accessory double cones, and 82% single cones contact GABAergic HCs. These results suggest that GABAergic and non-GABAergic HC input/output synapses differ and they may mediate different functional pathways in the outer retina.     

Localization of glutamic-acid-decarboxylase-immunoreactive axon terminals in the inferior olive of the rat, with special emphasis on anatomical relations between GABAergic synapses and dendrodendritic gap junctions

Immunocytochemical and electron microscopic methods were used to examine the GABAergic innervation of the inferior olivary nucleus in adult rats. This neuronal system was visualized with an antibody against glutamic acid decarboxylase (GAD, EC 4.1.1.15), the GABA-synthesizing enzyme. A GAD-positive reaction product was encountered only in short segments of preterminal axons and in axon terminals. Their relative number per unit area of neuropil was very similar in all olivary subnuclei. Despite this homogeneity in density, obvious intraregional differences existed. Some regions were strongly immunoreactive (the "c" subgroup, the beta nucleus, and the mediolateral outgrowth of the medial accessory olive), whereas others were weakly labeled (the dorsomedial cell column and the central zones of the medial accessory and principal olives). The strongly immunoreactive areas contained the largest and most intensively labeled axon terminals. Areas of weak labeling were filled with small, weakly immunoreactive nerve terminals. Thus, variations in size and in intensity of labeling create a specific pattern of GABA innervation, revealed by an almost continuous gradient between the above-mentioned extremes. The GAD-positive axon terminals established conventional synapses with dendrites (94% of the samples) or with cell bodies (6%). The vast majority of these synapses were type II (84%) and only a small proportion formed type I synaptic contacts (16%), regardless of the nature of the postsynaptic element. Immunoreactive terminals were also involved in the complex synaptic arrangements--the glomeruli, which characterize the olivary neuropil. Within these formations, olivary neurons were electrotonically coupled through dendrodendritic gap junctions. There was a constant association between GAD-positive axon terminals and small dendritic appendages linked by gap junctions. This association was revealed not only by the systematic presence of immunolabeled terminals directly apposed to the dendritic appendages but, more importantly, by the frequent presence of type II synapses straddling both elements. These synapses were in close proximity to the low-resistance pathways represented by the gap junctions. The strategic location of these GABA synapses is discussed in relation to recent findings indicating the possibility of a synaptic modulation of the electrical coupling: the release of GABA, by increasing nonjunctional membrane conductance, could shunt the coupling between olivary neurons. The functional decoupling of selected gap junctions would be responsible for the spatial organization of the olivary electrotonic coupling.     

Ultrastructure of commissural neurons of the hilar region in the hippocampal dentate gyrus

Previous studies have described the polymorph neurons in the hilus of the dentate gyrus at the light microscopic level and have indicated that many of those neurons are the cells of origin for both ipsilateral associational and commissural projections to the dentate gyrus. Because previous studies have not described the ultrastructural characteristics of the hilar neurons, we identified these features of the commissural neurons in the hilus. The method of retrograde transport of horseradish peroxidase (HRP) was utilized with a silver staining technique for HRP intensification. Two populations of labeled commissural neurons were observed in electron microscopic preparations of the contralateral hilus. One type consisted of cells with somata that exhibited round or oval nuclei with no intranuclear inclusions and formed symmetric axosomatic synapses. The main dendrites of those neurons were thick and tapering. In contrast, the other type of labeled neuronal soma had infolded nuclei containing intranuclear rods or sheets, displayed both symmetric and asymmetric axosomatic synapses, and had dendrites that were less thick and generally aspinous. In those same preparations, labeled commissural axon terminals formed synapses with dendrites and dendritic spines in the hilus and molecular layer and iwth somata in the granule cell layer. From the results of this study it appears that there are two distinct populations of commissural hilar neurons: one type resembles the morphology of the spiny CA3 pyramidal neuron, a type of excitatory projection cell, and the other type is similar to the dentate gyrus basket cell, a local circuit neuron associated with GABAergic inhibition. This latter cell type provides further support for the notion that some commissural neurons are inhibitory.     

Simultaneous Demonstration of Serotonin-immunoreactive Terminals and GABAergic Neurons in the VPL Nucleus of the Cat Thalamus

Pre-embedding immunoperoxidase (for serotonin) and postembedding immunogold (for γ-aminobutyric acid; GABA) labelling were combined at light and electron microscopic levels to demonstrate the neuronal targets of serotonin (5-HT) afferents in the ventral posterior lateral nucleus (VPL) of the cat thalamus. 5-HT-immunoreactive fibres and terminal varicosities were found in close proximity to GABA-immunoreactive interneurons and non-GABAergic relay neurons. Ultrastructurally, the vast majority of 5-HT terminals made close membrane contacts without overt membrane specializations with GABAergic axon terminals, GABAergic presynaptic dendrites and GABAergic somata. A very small number of 5-HT terminals formed typical asymmetrical synapses with GABAergic presynaptic dendrites and with dendritic shafts of relay cells. Some 5-HT terminals participated with the presynaptic dendrites in triadic synaptic arrangements. These findings suggest a dual innervation pattern by 5-HT afferents in VPL and the release of 5-HT in large part at sites not associated with morphologically detectable synapses.     

Synaptic reorganisation in the rat striatum after dopaminergic deafferentation: An ultrastructural study using glutamate decarboxylase immunocytochemistry

The ultrastructure of GABAergic and non-GABAergic synapses in the adult rat neostriatum was examined 6-8 months after unilateral removal of the nigrostriatal dopaminergic pathway by 6-hydroxydopamine injection into the medial forebrain bundle. GABAergic profiles were identified by preembedding glutamate decarboxylase (GAD) immunocytochemistry performed on parasagittal vibratome sections. In three representative fields of the striatum, the nature and number of boutons and their postsynaptic partners were determined and the differences between the striata ipsi- and contralateral to the lesion analyzed. The percentage of GAD-immunoreactive boutons was increased from 23% on the intact side to 28% on the lesioned side. In addition, the GABAergic boutons underwent significantly more multiple contacts with several independent postsynaptic profiles, preferentially with dendritic spines. This could reflect a lesion-induced sprouting of local GABAergic axon terminals. On the other hand, although the vast majority of GABAergic boutons underwent synaptic contacts with dendrites (77% vs. 80%), the number of boutons per dendrite or per dendritic circumference remained unchanged. Thus, the higher frequency of GABAergic boutons may simply reflect the loss of the dopaminergic nerve endings, without a heterosynaptic replacement by GABAergic boutons. The deafferentation also induced structural changes of the postsynaptic profiles. Some dendritic spines had a shortened neck; others were completely integrated in the dendrite which now contained a spine apparatus and was contacted by boutons with the features of axospinous synapses. The spine retraction resulted in a quantitative decrease in the number of spines. Analysis of the synaptic curvature revealed that only spines with a flat contact zone were lost. Concurrently, the number of dendrites was increased, of the GAD-containing in particular, suggesting that the denrites of GABAergic interzeurons tend to elongate and/or arborize. Taken together, the results of the present study show that the dopaminergic denervation caused a remodeling of the postsynaptic neurons. The relative increase of the number of GABAergic boutons and their synaptic contacts suggests that an altered wiring of the intrinsic GABAergic system contributes to the changes in the striatal output activity. © 1995 Wiley-Liss, Inc.     

Ultrastructural evidence for presynaptic mu opioid receptor modulation of synaptic plasticity in NMDA-receptor-containing dendrites in the dentate gyrus

Physiological studies have demonstrated that long-term potentiation (LTP) induction in N-methyl-D-aspartate (NMDA) receptor containing dentate granule cells following lateral perforant path stimulation is opioid dependent, involving mu-opioid receptors (MORs) on gamma-aminobutyric acid (GABA)-ergic neurons. To determine the cellular relationships of MORs to postsynaptic NMDA receptor-containing dendrites, immunoreactivity (-I) against MOR and the NMDA receptor subunit 1 (NMDAR1) was examined in the outer molecular layer of the dentate gyrus using electron microscopy. MOR-I was predominantly in axons and axon terminals. NMDAR1-I was almost exclusively in spiny dendrites, but was also in a few terminals. Using immunogold particles to localize precisely NMDAR1, one-third of the NMDAR1-I was detected on the dendritic plasmalemma; in dendritic spines plasmalemmal immunogold particles were near synaptic densities. Many MOR-labeled axons and terminals contacted NMDAR1-labeled dendrites. MOR-labeled terminals formed symmetric (inhibitory-type) synapses on NMDAR1-labeled dendritic shafts or nonsynaptically contacted NMDAR1-labeled shafts and spines. MOR-labeled axons often abutted NMDAR1-containing dendritic spines which received asymmetric (excitatory-type) synapses from unlabeled terminals. Occasionally, MOR-labeled terminals and dendrites were apposed to NMDAR1-containing terminals. These results provide anatomical evidence that endogenous enkephalins or exogenous opioid agonists could inhibit GABAergic terminals that modulate granule cell dendrites, thus boosting depolarizing events in granule cells and facilitating the activation of NMDA receptors located on their dendrites.     

Subdivisions in the Multiple GABAergic Innervation of Granule Cells in the Dentate Gyrus of the Rat Hippocampus

The sources of GABAergic innervation to granule cells were studied to establish how the basic cortical circuit is implemented in the dentate gyrus. Five types of neuron having extensive local axons were recorded electrophysiologically in vitro and filled intracellularly with biocytin (Han et al., 1993). They were processed for electron microscopy in order to reveal their synaptic organization and postsynaptic targets, and to test whether their terminals contained GABA. (1) The hilar cell, with axon terminals in the commissural and association pathway termination field (HICAP cell), formed Gray's type 2 (symmetrical) synapses with large proximal dendritic shafts (n= 18), two-thirds of which could be shown to emit spines, and with small dendritic branches (n= 6). Other boutons of the HICAP neuron were found to make either Gray's type 1 (asymmetrical) synapses (n= 4) or type 2 synapses (n= 6) with dendritic spines. Using a highly sensitive silver-intensified immunogold method for the postembedding visualization of GABA immunoreactivity, both the terminals and the dendrites of the HICAP cell were found to be immunopositive, whereas its postsynaptic targets were GABA-immunonegative. The dendritic shafts of the HICAP cell received synapses from both GABA-negative and GABA-positive boutons; the dendritic spines which densely covered the main apical dendrite in the medial one-third of the molecular layer received synapses from GABA-negative boutons. (2) The hilar cell, with axon terminals distributed in conjunction with the perforant path termination field (HIPP cell), established type 2 synapses with distal dendritic shafts (n= 17), most of which could be shown to emit spines, small-calibre dendritic profiles (n= 2) and dendritic spines (n= 6), all showing characteristics of granule cell dendrites. The sparsely spiny dendrites of the HIPP cell were covered with many synaptic boutons on both their shafts and their spines. (3) The cell with soma in the molecular layer had an axon associated with the perforant path termination field (MOPP cell). This GABA-immunoreactive cell made type 2 synapses exclusively on dendritic shafts (n= 20), 60% of which could be shown to emit spines. The smooth dendrites of the MOPP cell were also restricted to the outer two-thirds of the molecular layer, where they received both GABA-negative and GABA-positive synaptic inputs. (4) The extensive axonal arborization of the dentate basket cell terminated mainly on somata (n= 26) and proximal dendrites (n= 9) in the granule cell layer, and some boutons made synapses on somatic spines (n= 6); all boutons established type 2 synapses. (5) The dentate axo-axonic cell established type 2 synapses (n= 14) exclusively on axon initial segments of granule cells in the granule cell layer, and on initial segments of presumed mossy cells in the hilus. The results demonstrate that granule cells receive inputs from the local circuit axons of at least five distinct types of dentate neuron terminating in mutually exclusive domains of the cell's surface in four out of five cases. Four of the cell types (HICAP cell, MOPP cell, basket cell, axo-axonic cell) contain GABA, and the HIPP cell may also be inhibitory. The specific local inhibitory neurons terminating in conjunction with particular excitatory amino acid inputs to the granule cells (types 1 – 3) are in a position to interact selectively with the specific inputs on the same dendritic segment. This arrangement provides a possibility for the independent regulation of the gain and long-term potentiation of separate excitatory inputs, through different sets of GABAergic local circuit neurons. The pairing of excitatory and inhibitory inputs may also provide a mechanism for the downward reseating of excitatory postsynaptic potentials, thereby extending their dynamic range.