Aspiny neurons and their local axons in the neostriatum of the rat: a correlated light and electron microscopic study of Golgi-impregnated material

Summary Three types of neuron with smooth (aspiny) dendrites could be distinguished in the Golgi-impregnated rat neostriatum. Examples of each type of aspiny neuron were found with local axon collaterals within the neostriatum and these were selected for gold-toning and examination in the electron microscope. One type of aspiny neuron had an elongated, usually spindle-shaped, medium-size soma with two, or rarely three, primary dendrites originating from opposite poles of the cell; one example of this type of neuron had two separate axons. The second type of aspiny neuron had a nearly round, medium-size soma with four primary dendrites that branched profusely quite close to the cell body. A third type of aspiny neuron had a very large polygonal-shaped cell body. Afferent axon terminals were found in synaptic contact with the dendrites and cell bodies of all three types of aspiny neuron.Axon collaterals of each type of neuron displayed varicosities which, when examined in the electron microscope, were frequently found to be boutons making synaptic contact. All such synaptic contacts had symmetrical membrane specializations and the most common postsynaptic targets were dendritic shafts, sometimes spine-bearing. Dendritic spines themselves also received synapses from each type of neuron. No axosomatic synapses involving boutons of identified axons were found. One example of a synapse between an axon collateral of an aspiny neuron and one of the same neuron's dendrites (an autapse) was demonstrated by electron microscopy.It is concluded that the synaptic terminals of at least four types of neuron, the three aspiny types described here and the medium-size densely spiny neuron, participate in local circuit interactions in the neostriatum.     

Ultrastructure of Golgi-impregnated and gold-toned spiny and aspiny neurons in the monkey neostriatum

Summary Golgi-impregnated, gold-toned spiny and aspiny neurons in the monkey neostriatum were deimpregnated and examined at the electron microscope level.Spiny type I neurons have relatively large nuclei with few indentations and aggregates of chromatin under the nuclear membrane which in some regions give the appearance of a dark rim. The small quantity of surrounding cytoplasm is poor in organelles.Aspiny type I neurons have eccentric, highly indented nuclei. The relatively large proportion of cytoplasm is rich in organelles especially Golgi apparatus and rough endoplasmic reticulum which often appears in stacks.Synapses with symmetric membrane densities are common on the somata of spiny type I neurons. Those on the proximal and distal dendritic shafts are few in number and asymmetric, and those on spines more frequent and primarily asymmetric. Aspiny type I neurons have few synapses on their cell bodies. Proximal and distal dendrites, however, are contacted by numerous profiles which contain small round vesicles and make both symmetric and asymmetric synapses. The same axon terminals also synapse with dendritic spines of spiny neurons, indicating that an input, most likely of afferent origin, is shared by both cell types. Other less frequently occurring profiles forming symmetric membrane densities also contact the dendrites of aspiny and spiny neurons. The axon hillocks and initial segments of both neuronal types receive a synaptic input, which is more common on spiny cells.Results offer unequivocal evidence for the differences in the ultrastructure of these two most common categories of medium-size neostriatal neurons, which may help in their proper identification in standard material, as well as information on the types and distributions of synaptic inputs onto these neurons. Moreover, the findings clarify some controversies in the literature probably originating from observations on a mixed population of cells of medium size.     

Distribution of DARPP-32 in the basal ganglia: an electron microscopic study

Summary DARPP-32, a dopamine and cyclic AMP-regulated phosphoprotein, has been studied by light and electron microscopical immunocytochemistry in the rat caudatoputamen, globus pallidus and substantia nigra. In the caudatoputamen, DARPP-32 was present in neurons of the medium-sized spiny type. Immunoreactivity for DARPP-32 was present in dendritic spines, dendrites, perikaryal cytoplasm, most but not all nuclei, axons and a small number of axon terminals. Immunoreactive axon terminals in the caudatoputamen formed symmetrical synapses with immunolabelled dendritic shafts or somata. Neurons having indented nuclei were never immunoreactive. In the globus pallidus and substantia nigra pars reticulata, DARPP-32 was present in myelinated and unmyelinated axons and in axon terminals. The labelled axon terminals in these regions formed symmetrical synaptic contacts on unlabelled dendritic shafts or on unlabelled somata. These data suggest that DARPP-32 is present in striatal neurons of the medium-sized spiny type and that these DARPP-32-immunoreactive neurons form symmetrical synapses on target neurons in the globus pallidus and substantia nigra. The presence of DARPP-32 in these striatal neurons and in their axon terminals suggests that DARPP-32 mediates part of the response of medium-size spiny neurons in the striaturn to dopamine D-l receptor activation.     

Anatomy and physiology of substantia nigra and retrorubral neurons studied by extra- and intracellular recording and by horseradish peroxidase labeling

The question of origin of the excitatory and inhibitory responses that occur in neostriatal neurons following electrical stimulation of the substantia nigra is complicated by the possible spread of stimulus currents to numerous unspecifiable systems of neuronal elements. The present work begins to address this problem through the study of conduction properties of specific nigral and perinigral neurons in the cat. Neurons of pars compacta of substantia nigra and of the retrorubral area were found to have similar latencies for antidromic activation, whether from caudate nucleus stimuli (6.8–8 ms) or medial forebrain bundle stimulation (2.4–6.4 ms).The soma-dendritic features of both pars compacta and retrorubral neurons (revealed by intracellular injection of horseradish peroxidase) resembled the sparsely-branched, medium-sized substantia nigra neurons known from Golgi studies to have long dendrites with scattered and mainly distally-located spine-like appendages. Two types of pars compacta neurons were found; one with an ascending axon lacking collateral branches, and another with a descending axon that issued collaterals which terminated in the compacta, in pars reticulata, and possibly in retrorubral areas. Despite failure to detect as ascending axonal trajectory for this latter neuron, both types of pars compacta cells responded antidromically to stimulation of the caudate nucleus or medial forebrain bundle.The conduction time for impulse propagation in axons of pars compacta or retrorubral neurons suggests that either may mediate at least some of the excitatory responses that are known to occur in neostriatal neurons following stimulation of the substantia nigra in the cat. However, these conduction times are not compatible with the production of other excitatory responses which are commonly observed in the cat striatum at latencies shorter than 6 to 7 ms following stimulation of the substantia nigra.     

A second type of striatonigral neuron: a comparison between retrogradely labelled and Golgi-stained neurons at the light and electron microscopic levels

In a light and electron microscopic examination of the neostriata of rats that had received injections of horseradish peroxidase into the ipsilateral substantia nigra, two morphologically distinct types of horseradish peroxidase-labelled neurons were observed. In confirmation of previous findings, one type was of medium-size and was characterized by Golgi-staining and gold-toning as the densely spinous type. The second type of neuron was in contrast, larger, had an indented nucleus and numerous cytoplasmic organelles. The synaptic input to the perikarya of the latter neurons consisted of numerous boutons containing large round and oval vesicles. The boutons formed symmetrical synaptic contacts and were similar to those of the local axon collaterals of medium-size densely spiny striatonigral neurons.In an attempt to establish what type of Golgi-impregnated neuron the second type of horseradish peroxidase-labelled neuron was, seventeen Golgi-stained or gold-toned neurons were examined in the electron microscope. Three of them were very similar in their ultrastructural features and synaptic input to the horseradish peroxidase-labelled neurons. All three were of a similar morphological appearance in the light-microscope and characteristically had long (up to 700 μm), essentially smooth dendrites. Both the large horseradish peroxidase-labelled neurons and the Golgi-impregnated neurons with long dendrites have so far only been found in the most ventral regions of the neostriatum.It is concluded that there are at least two morphologically distinct types of striatonigral neurons.     

The postsynaptic targets of substance P-immunoreactive terminals in the rat neostriatum with particular reference to identified spiny striatonigral neurons

Summary Substance P-immunoreactive boutons were examined in the electron microscope in sections of the rat neostriatum that contained retrogradely labelled striatonigral neurons and/or Golgi-impregnated medium-size densely spiny neurons. The postsynaptic targets of the immunoreactive boutons were characterized on the basis of ultrastructural features, their projection to the substantia nigra and/or their somato-dendritic morphology. Substance P-immunoreactive axonal boutons formed symmetrical synaptic specializations. Of a total of 233 randomly identified synaptic boutons 72.5% made contact with dendritic shafts, 15% with dendritic spines and 10.7% with perikarya. The ultrastructural characteristics of some of the postsynaptic neuronal perikarya were consistent with their identification as striatal interneurons. Similarly, the observation of some of the substance P-containing terminals in contact with spines, spine-bearing dendritic shafts and perikarya with the ultrastructural characteristics of medium-size densely spiny neurons suggested that one of the targets of substance P-positive terminals are striatal projection neurons. Direct evidence for this was obtained in sections from rats that had received injections of horseradish peroxidase conjugated with wheatgerm agglutinin in the substantia nigra. The perikarya of retrogradely labeled striatonigral neurons were found to receive symmetrical synaptic input from substance P-positive boutons. Ultrastructural analysis of Golgi-impregnated medium-size densely spiny neurons, some of which were also retrogradely labeled from the substantia nigra, demonstrated directly that this class of neuron was postsynaptic to the substance P-immunoreactive boutons. The combination of Golgi-impregnation with substance P-immunocytochemistry made it possible to study the pattern or topography of the substance P-positive input to medium size densely spiny neurons. The substance P-containing boutons made contact predominantly with perikarya and dendritic shafts. This pattern of input is markedly different from that of other identified inputs to medium-size densely spiny neurons.     

Morphology of intracellularly stained spiny neurons in rat striatal grafts.

Two to six months after implantation of fetal striatal primordia into the kainic acid-lesioned neostriatum of adult rats, spiny neurons in the grafts were stained intracellularly with biocytin. To determine whether the spiny neurons in the grafts differentiate morphologically as in the host neostriatum, the intracellularly stained spiny neurons in the grafts were studied with light and electron microscopy and compared with that of spiny neurons in the host neostriatum. The spiny neurons in the grafts had ovoid or polygonal cell bodies with dendrites radiating in all directions. The somata were smooth and the dendrites, except for their most proximal portions, were rich in spines. All these features resembled the appearance of spiny neurons in the intact neostriatum. However, quantitative studies showed that the somata of spiny neurons in the grafts were larger than those in the host neostriatum (projected cross-sectional areas of 230 +/- 64.6 microns 2 in the grafts and 158 +/- 28.9 microns 2 in the host) and the spine density of graft neurons was lower than that of host neurons. Cells near the border of the grafts had dendrites extending both into the graft and into the host neostriatum. In these cells, the dendrites in the grafts had fewer spines than the dendrites in the host tissue. The axons of spiny neurons in the grafts had very large and dense intrastriatal collateral arborizations, which occupied a much larger volume than that of the dendritic domain of the parent cells. The local axonal arborizations of each of these cells filled almost the entire graft. In some cells, axonal branches were traced outside the grafts and were seen to enter the internal capsule fascicles. Unlike spiny neurons in the normal adult neostriatum, the spiny cells of the graft could have nuclear indentations. With this exception, the ultrastructural features of spiny neurons in the grafts were very similar to those in the hosts. Many unlabeled boutons made synapses on identified spiny neurons in the grafts. Terminals with small round vesicles made synaptic contacts on dendritic shafts and dendritic spines, while terminals with flattened or pleomorphic vesicles contacted somata, dendrites, and dendritic spines. Labeled axon collaterals of graft neurons made symmetrical synapses on somata, dendrites and spines in the grafts and in the host neostriatum. In the grafts, more than 60% of the axon terminals contacted dendritic shafts. The proportion of axosomatic and axospinous synapses varied substantially from cell to cell.(ABSTRACT TRUNCATED AT 250 WORDS)     

Neuronal types in the claustrum of man

Summary Neuronal types of the human clastrum have been investigated by means of a transparent Golgi technique which enables one to study the characteristics of not only the cellular processes but also the marking features of the nuclei, the cellular organelles, and the paraplasmic substances of various types of nerve cells.Five varieties of neurons have been distinguished:Type I represents a class of spiny nerve cells varying to a certain extent in size and shape. These cells contain fine and widely dispersed lipofuscin granules which can only faintly be tinged by aldehydefuchsin.Type II cells are large aspiny neurons. Their cell bodies contain a great number of deeply stained coarse pigment granules.Type III cells are large aspiny neurons devoid of pigment deposits.Type IV is a small pigment-laden aspiny neuron.Type V is a small aspiny neuron devoid of lipofuscin granules.The pattern of pigmentation revealed by the different types of nerve cells turns out to be highly characteristic. It can well be used for classification of the various types of nerve cells which occur within the reaches of the claustrum.     

Synaptic connections of morphologically identified and physiologically characterized large basket cells in the striate cortex of cat

Neurons were studied in the striate cortex of the cat following intracellular recording and iontophoresis of horseradish peroxidase. The three selected neurons were identified as large basket cells on the basis that (i) the horizontal extent of their axonal arborization was three times or more than the extent of the dendritic arborization; (ii) some of their varicose terminal segments surrounded the perikarya of other neurons. The large elongated perikarya of the first two basket cells were located around the border of layers III and IV. The radially-elongated dendritic field, composed of beaded dendrites without spines, had a long axis of 300-350 microns, extending into layers III and IV, and a short axis of 200 microns. Only the axon, however, was recovered from the third basket cell. The lateral spread of the axons of the first two basket cells was 900 microns or more in layer III and, for the third cell, was over 1500 microns in the antero-posterior dimension, a value indicating that the latter neuron probably fulfills the first criterion above. The axon collaterals of all three cells often branched at approximately 90 degrees to the parent axon. The first two cells also had axon collaterals which descended to layers IV and V and had less extensive lateral spreads. The axons of all three cells formed clusters of boutons which could extend up a radial column of their target cells. Electron microscopic examination of the second basket cell showed a large lobulated nucleus and a high density of mitochondria in both the perikarya and dendrites. The soma and dendrites were densely covered by synaptic terminals. The axons of the second and third cells were myelinated up to the terminal segments. A total of 177 postsynaptic elements was analysed, involving 66 boutons of the second cell and 89 boutons of the third cell. The terminals contained pleomorphic vesicles and established symmetrical synapses with their postsynaptic targets. The basket cell axons formed synapses principally on pyramidal cell perikarya (approximately 33% of synapses), spines (20% of synapses) and the apical and basal dendrites of pyramidal cells (24% of synapses). Also contacted were the perikarya and dendrites of non-pyramidal cells, an axon, and an axon initial segment. A single pyramidal cell may receive input on its soma, apical and basal dendrites and spines from the same large basket cell.(ABSTRACT TRUNCATED AT 400 WORDS)     

Morphology of cortically projecting basal forebrain neurons in the rat as revealed by intracellular iontophoresis of horseradish peroxidase

The intracellular horseradish peroxidase technique was employed to study the morphology of basal forebrain neurons that were identified as cortically projecting by antidromic invasion from the cerebral cortex. Four neurons were examined in detail; they were located at different rostrocaudal levels within the basal forebrain. Their somata were large, 30-50 microns in longest dimension, and gave rise to three to eight primary dendrites, which ramified into third- to fifth-order dendrites. The longest observed dendrite in each neuron terminated at a distance of 600-900 microns from the soma. The sizes of soma and dendritic field of the two most rostrally located cells were smaller than those of the other two cells located more caudally. Dendritic spines were seen in all four cortically projecting basal forebrain neurons. Spines had shafts of variable lengths, and usually had spherical or elongated heads. The density of spines varied among the four neurons; one neuron, a type II cortically projecting basal forebrain neurons as defined physiologically by Reiner et al., had a much greater number of dendritic spines than the other three neurons, which were type I neurons. No somatic spines were observed. Presumptive axons were identified in three of the four cortically projecting basal forebrain neurons. These axons originated from either the soma or a primary dendrite, and two of them gave off local collaterals, which displayed occasional bouton-like swellings. The above observations confirm and extend previous findings that cortically projecting neurons in the basal forebrain are large multipolar cells, and provide evidence to support the conclusion that these cells, although somewhat variable in size, generally have extensive dendrites which display frequent spines.     

The spiny stellate neurons in layer IV of the human auditory cortex. A Golgi study

The spiny stellate neurons have been studied by the Golgi method in the auditory koniocortex and parakoniocortex of man. Spiny stellate cells are a consistent though not very common component of layer IV. They are not confined to specific sublayers but occur at all depths of layer IV, and also in layer IIIc. Spiny stellate cells in the auditory areas show a great variety of their dendritic arborization pattern. The presence of all intermediate forms between small pyramidal cells--which constitute the dominant cell type in layer IV and which display an extraordinary heteromorphism--and spiny stellate cells shows the close kinship between both neuronal types. The morphology and distribution of spines along the dendrites of spiny stellate neurons are similar to those of the small pyramidal cells of the same layer. The axons, which were impregnated only in their proximal portions, mostly descend, giving rise to recurrent ascending collaterals, but initially ascending axons do also occur. Spiny stellate neurons are present in the different cytoarchitectonic areas examined, and thus they are not confined to the auditory koniocortex.     

The section-Golgi-impregnation procedure--3. Combination of Golgi-impregnation with enzyme histochemistry and electron microscopy to characterize acetylcholinesterase-containing neurons in the rat neostriatum

Three morphologically distinct types of neuron that contain acetylcholinesterase have been distinguished by Golgi-impregnation of sections of the rat neostriatum that had been incubated to reveal acetylcholinesterase activity. The neuron that stained most intensely for acetylcholinesterase was a large cell, with smooth or sparsely spiny dendrites; the axon of one these neurons was partially impregnated by the Golgi stain and had local axon collaterals (type 1). Another acetylcholinesterase-containing neuron had a small to medium-size cell body with long sparsely spiny dendrites emerging from opposite poles (type 2). The third type of neuron that contained acetylcholinesterase was medium to large size and had many primary, sparsely spiny dendrites that branched frequently (type 3). Examination of the same Golgi-impregnated, acetylcholinesterase-stained neurons that had been studied in the light microscope by electron microscopy allowed us to distinguish several other differences between the three types of neuron. Whereas all three types had acetylcholinesterase reaction product in the endoplasmic reticulum and along the nuclear envelope, only neurons of type 1 displayed reaction product in the Golgi apparatus. All three types of neuron received synaptic input, mainly along their dendrites. It is concluded that the combination of Golgi-impregnation with histochemical procedures that demonstrate endogenous enzyme activity can be applied to reveal the morphological characteristics, synaptic input and local synaptic output of neurons with specific biochemical properties.     

Large aspiny cells in the matrix of the rat neostriatum in vitro: physiological identification, relation to the compartments and excitatory postsynaptic currents.

1. Large aspiny neurons (20-60 microns diam) in the neostriatum were studied in an in vitro rat slice preparation by whole-cell recording to reveal physiological identification from medium-sized spiny projection cells (10-20 microns diam), relation to the patch and matrix compartments, and excitatory synaptic inputs. Recorded cells were identified by intracellular biocytin staining. Compartmental identification was made by calbindinD28K immunohistochemistry in fixed slices. 2. Large stained neurons were morphologically heterogeneous and had aspiny or sparsely spiny dendrites and dense local axonal branches. They were located in the matrix or on the patch-matrix border. Axonal branches of the large aspiny cells were preferentially distributed in the matrix and gave off terminal boutons there. Some of the secondary dendrites arising from stem dendrites running along the border, however, crossed compartment boundaries and made fine branches in a patch. 3. Large aspiny cells had less negative resting membrane potentials and lower thresholds for spike generation than medium spiny cells. They showed longer-duration and larger-amplitude afterhyperpolarizations (AHPs) than medium spiny cells. During hyperpolarizing current pulses, apparent resistance slowly reduced, and a prominent sag was observed in the voltage record, which was absent in medium spiny cells. The large aspiny cells showed no spontaneous firing but had a tendency to fire repetitive spikes in response to depolarizing current pulses, although spike interval tended to increase in later spikes. Spike frequency of large aspiny cells increased less with current intensity than that of medium spiny cells. 4. Most large aspiny cells were considered to belong to a single physiological class, although one large aspiny cells showed shorter-duration AHPs than both most other large aspiny cells and medium spiny cells, and little spike-frequency adaptation. 5. Excitatory postsynaptic currents (EPSCs) of large aspiny cells induced by intrastriatal stimulation had two components. An early, linear component was blocked by 10 microM 6-cyano-7-nitro-quinoxaline-2,3-dione (CNQX), a selective antagonist of non-N-methyl-D-aspartate (NMDA) receptors. A later component with a nonlinear current-voltage (I-V) relationship was blocked by 50 microM DL-2-amino-5-phosphonovaleric acid (DL-APV), a selective antagonist of NMDA receptors. 6. From these results, four conclusions can be drawn. 1) Most large aspiny neostriatal cells in the matrix, although they take heterogeneous shapes, belong to one physiological class with long-duration AHPs and a strong time-dependent component of anomalous rectification.(ABSTRACT TRUNCATED AT 400 WORDS)     

Electrophysiological and morphological properties of cell types in the chick neostriatum caudolaterale

The neostriatum caudolaterale, in the chick also referred to as dorsocaudal neostriatal complex, is a polymodal associative area in the forebrain of birds that is involved in sensorimotor integration and memory processes. We have used whole-cell patch-clamp recordings in chick brain slices to characterize the principal cell types of the neostriatum caudolaterale. Electrophysiological properties distinguished four classes of neurons. The morphological characteristics of these classes were examined by intracellular injection of Lucifer Yellow. Type I neurons characteristically fired a brief burst of action potentials. Morphologically, type I neurons had large somata and thick dendrites with many spines. Type II neurons were characterized by a repetitive firing pattern with conspicuous frequency adaptation. Type II neurons also had large somata and thick dendrites with many spines. There was no clear morphological distinction between type I and type II neurons. Type III neurons showed high-frequency firing with little accommodation and a prominent time-dependent inward rectification. They had thin, sparsely spiny dendrites and extensive local axonal arborizations. Electrophysiological and morphological properties indicated them as being interneurons. Type IV neurons had a longer action potential duration, a larger input resistance, and a longer membrane time constant than the other classes. Type IV neurons had small somata and short dendrites with few spines. The long axon collaterals of neurons in all spiny cell classes (types I, II, IV) followed similar patterns, suggesting that neurons from all these types can contribute to the projections of the neostriatum caudolaterale to sensory, limbic and motor areas.The electrophysiological and anatomical characterization of the major classes of neurons in the caudal forebrain of the chick provides a framework for the investigation of sensorimotor integration and learning at the cellular level in birds.     

A golgi study of the periaqueductal gray matter in the cat. Neuronal types and their distribution

Summary In Golgi material, the neurons of the periaqueductal gray matter (PAG) of the cat have been classified into five types, according to the following criteria: number of dendrites per cell, characteristics of secondary arborization, frequency of spines and axon caliber. Type 1 cells, which are multipolar and rich in spines are the most frequent, and are probably intranuclear neurons. Type 4 cells have a short axon which ends in the PAG, but they differ from Type 1 in that their dendritic ramification is of a different type and there are few spines. Type 2 and 3 neurons have a thick axon which runs outside the PAG, and dendrites rich in spines. Type 2 cells have more primary dendrites, while Type 3 neurons have dendrites which may spread outside the PAG. Type 5 cells have dendrites with few spines and no secondary ramification. Their thick and long axon projects outside the PAG. Type 2, 3 and 5 cells have been considered projective neurons. The various neuron types are present in every area of the PAG, although in the ventral region there is a predominance of Type 2 and 5 neurons, in the dorsal regions of Type 2 and 3 cells, and in lateral regions of Type 3 and 5 cells. Local intrinsic circuits have been observed in which both the interneurons and the projective, with early axonic collaterals, are involved. The prevalence of neurons to which an afferent role has been attributed (Type 2 and 3 cells) compared with efferent cells (Type 5), is in agreement with hodological studies which indicate that the PAG receives multiple and numerous afferents in comparison with the relatively scarce efferent fibers. These projections can be intensely and deeply elaborated and modulated by means of local intrinsic circuits.     

Immunogold demonstration of GABA in synaptic terminals of intracellularly recorded, horseradish peroxidase-filled basket cells and clutch cells in the cat's visual cortex

To identify the putative transmitter of large basket and clutch cells in the cat's visual cortex, an antiserum raised against GABA coupled to bovine serum albumen by glutaraldehyde and a postembedding, electron microscopic immunogold procedure were used. Two basket and four clutch cells were revealed by intracellular injection of horseradish peroxidase. They were identified on the basis of the distribution of their processes and their synaptic connections. Large basket cells terminate mainly in layer III, while clutch cells which have a more restricted axon, terminate mainly in layer IV. Both types of neuron have a small radial projection. They establish type II synaptic contacts and about 20-30% of their synapses are made with the somata of other neurons, the rest with dendrites and dendritic spines. Altogether 112 identified, HRP-filled boutons, the dendrites of three clutch cells and myelinated axons of both basket and clutch cells were tested for the presence of GABA. They were all immunopositive. The postsynaptic neurons received synapses from numerous other GABA-positive boutons in addition to the horseradish peroxidase-filled ones. Dendritic spines that received a synapse from a GABA-positive basket or clutch cell bouton also received a type I synaptic contact from a GABA-negative bouton. A few of the postsynaptic dendrites, but none of the postsynaptic somata, were immunoreactive for GABA. The fine structural characteristics of the majority of postsynaptic targets suggested that they were pyramidal and spiny stellate cells. These results provide direct evidence for the presence of immunoreactive GABA in identified basket and clutch cells and strongly suggest that GABA is a neurotransmitter at their synapses. The laminar distribution of the synaptic terminals of basket and clutch cells demonstrates that some GABAergic neurons with similar target specificity segregate into different laminae, and that the same GABAergic cells can take part in both horizontal and radial interactions.     

Quantitative electron microscopic study of immunoreactive somatostatin axons in the rat neostriatum

Various features of immunoreactive somatostatin axons including bouton size, synaptic length, the type of synapse formed (symmetric or asymmetric) and postsynaptic target, were examined at the ultrastructural level in the caudate nucleus. These features were compared to those of unlabeled axons in the surrounding caudate neuropil. Results showed that immunoreactive somatostatin axons make relatively short-surfaced, symmetric contacts, mostly with dendritic shafts whereas the majority of unlabeled axons form long-surfaced, asymmetric synapses with dendritic spines. Observations indicate that immunoreactive somatostatin axons belong to a sparse and homogeneous population of axons, have features corresponding to those of intrinsic caudate neurons, and synapse with caudate spiny cells. These findings are consistent with earlier speculation that immunoreactive somatostatin axons in caudate arise from a population of aspiny interneurons which have previously been identified to contain the peptide.