Production of an antiserum to rat brain glutamate (GAD)/cysteine sulfinate (CSD) decarboxylase

Characterization of the cDNA coding for rat brain cysteine sulfinate decarboxylase: brain and liver enzymes are identical proteins encoded by two distinct mRNAs

Cysteine sulfinate decarboxylase (CSD) is considered as the rate-limiting enzyme in the biosynthesis of taurine, a possible osmoregulator in brain. Through cloning and sequencing of RT-PCR and RACE-PCR products of rat brain mRNAs, a 2,396-bp cDNA sequence was obtained encoding a protein of 493 amino acids (calculated molecular mass, 55.2 kDa). The corresponding fusion protein showed a substrate specificity similar to that of the endogenous enzyme. The sequence of the encoded protein is identical to that encoded by liver CSD cDNA. Among other characterized amino acid decarboxylases, CSD shows the highest homology (54%) with either isoform of glutamic acid decarboxylase (GAD65 and GAD67). A single mRNA band, approximately 2.5 kb, was detected by northern blot in RNA extracts of brain, liver, and kidney. However, brain and liver CSD cDNA sequences differed in the 5' untranslated region. This indicates two forms of CSD mRNA. Analysis of PCR-amplified products of genomic DNA suggests that the brain form results from the use of a 3' alternative internal splicing site within an exon specifically found in liver CSD mRNA. Through selective RT-PCR the brain form was detected in brain only, whereas the liver form was found in liver and kidney. These results indicate a tissue-specific regulation of CSD genomic expression.

Immunocytochemical localization of cysteine sulfinate decarboxylase in astrocytes in the cerebellum and hippocampus: a quantitative double immunofluorescence study with glial fibrillary acidic protein and S-100 protein

Immunocytochemistry of cysteine sulfinate decarboxylase was performed with a new rabbit antiserum that we have recently produced and characterized using as antigen an 11,000-fold purified fraction isolated from rat liver. This antiserum precipitated cysteine sulfinate decarboxylase enzymatic activity, labeled one band (mol. wt 51,000) on immunoblots of crude tissue extracts and did not stain any cells in peripheral tissues devoid of cysteine sulfinate decarboxylase. According to these criteria, this antiserum appeared to be specific for cysteine sulfinate decarboxylase. Numerous cells were immunolabeled in the cerebellum and the hippocampus. Most notable was the labeling of the small cells surrounding the Purkinje cells and sending radial fibers up to the pial surface of the cerebellar cortex or the staining of small star-shaped cells with thin immunolabeled processes abutting on blood vessels. Identified nerve cells such as the Purkinje cells and granule cells in the cerebellum or the pyramidal and granule cells in the hippocampus were devoid of any immunoreactivity. simultaneous double immunofluorescence was carried out using anti-glial fibrillary acidic protein or anti-S-100 monoclonal antibodies. Cysteine sulfinate decarboxylase as well as glial fibrillary acidic protein- or S-100-immunopositive cells were plotted independently for the same section. Quantitative analysis of the maps indicated that the overwhelming majority of cysteine sulfinate decarboxylase-immunolabeled cells were positive for the established astrocytes markers, glial fibrillary acidic protein or S-100. Between 82 and 98% of cysteine sulfinate decarboxylase-immunolabeled cells were also glial fibrillary acidic protein-positive, depending upon the layer. Cysteine sulfinate decarboxylase immunostaining was localized within the cytoplasm, while that of glial fibrillary acidic protein was linked to the cytoskeleton. Since both labels could not be fully superposed, some double immunolabeled cells may have escaped our analysis. More than 94% up to 99% of cysteine sulfinate decarboxylase-immunolabeled cells were simultaneously S-100-immunopositive. Our quantitative data establish that cysteine sulfinate decarboxylase is strictly localized in astrocytes in the cerebellum and in the hippocampus. This finding suggests that taurine is synthesized by astrocytes in the brain and accordingly may play a role in relation to glial function, possibly within the framework of glial-neuronal interactions.

Neuronal and synaptic organization of the centromedian nucleus of the monkey thalamus: a quantitative ultrastructural study, with tract tracing and immunohistochemical observations

The ultrastructure of the centromedian nucleus of the monkey thalamus was analysed qualitatively and quantitatively and projection neurons, local circuit neurons, and synaptic bouton populations identified. Projection neurons were mostly medium-sized, with oval-fusiform or polygonal perikarya, few primary dendrites, and frequent somatic spines; local circuit neurons were smaller. Four basic types of synaptic boutons were distinguished: (1) Small- to medium-sized boutons containing round vesicles (SR) and forming asymmetric contacts, identified as corticothalamic terminals. (2) Heterogeneous medium-sized boutons with asymmetric contacts and round vesicles, similar to the so-called large round (LR) boutons, which were in part of cortical origin. (3) Heterogeneous GAD-positive small- to medium-sized boutons, containing pleomorphic vesicles and forming symmetric contacts (F1 type), which included pallidothalamic terminals. (4) Presynaptic profiles represented by GAD-positive vesicle-containing dendrites of local circuit neurons. Complex synaptic arrangements, serial synapses and triads with LR and SR boutons engaging all parts of projection neuron dendrites and somata, were seen consistently, whereas classical glomeruli were infrequent. LR and SR boutons also established synapses on dendrites of local circuit neurons. F1 boutons established synapses on projection neuron somata, dendrites and initial axon segments. Compared to other previously studied motor-related thalamic nuclei, differences in synaptic coverage between proximal and distal projection neuron dendrites were less pronounced, and the density of synapses formed by local circuit dendrites on projection neuron dendrites was lower. Thus, compared to other thalamic nuclei, the overlap of different inputs was higher on monkey centromedian cells, and centromedian inhibitory circuits displayed a different organization.

Specificity of cysteine sulfinate decarboxylase (CSD) for sulfur-containing amino-acids

Cysteine sulfinate decarboxylase (CSD) which decarboxylates cysteine sulfinic acid (CSA) to form hypotaurine is thought to be involved in the biosynthesis of taurine. It was recently localized in astrocytes in the cerebellum and hippocampus by immunocytochemistry. Another sulfur-containing amino-acid (SCAA), homocysteic acid (HCA), was also found in astrocytes in these regions. We therefore investigated the specificity of CSD vs CSA and HCA as well as the related analogs homocysteine sulfinic acid (HCSA) and cysteic acid (CA). CSD was immunotrapped from brain and liver tissue supernatant using a specific CSD antiserum and Protein-A Sepharose. It was then incubated with the L-form of the various SCAA. Reaction products were identified and quantified by pre-column o-phthalaldehyde derivatization HPLC. CA and HCA from 2.5 to 25 mM inhibited the formation of hypotaurine from CSA (0.25 mM). Moreover, the inhibition curves were parallel for liver and brain CSD. CA or HCA (25 mM) elicited a near-total inhibition. HCSA did not produce a significant inhibition up to 25 mM. Incubation with 25 mM CSA or CA led to the formation of hypotaurine and taurine, respectively. The ratio of formation of taurine to that of hypotaurine was similar for CSD from liver and brain. In contrast no homotaurine, the decarboxylated reaction product of HCA, could be detected following incubation with 25 mM HCA. According to the sensitivity of the HPLC analysis this indicates that the decarboxylation of HCA, if any, was 130-fold and 50-fold less than that of CSA by CSD from liver and brain, respectively, in our experimental conditions. Similarly, following incubation with HCSA, no new peak appeared on the chromatogram when compared to a blank sample. These results show that CSD from either brain or liver has a high specificity for CSA and CA, which are the SCAA involved in the biosynthesis of taurine. HCA is an inhibitor of CSD but does not appear to be a substrate for CSD in vitro. HCSA is neither a substrate nor an inhibitor of CSD in vitro. Accordingly, CSD is unlikely to play a role in the metabolism of HCA or HCSA in vivo.

Development of the GABAergic phenotype in murine spinal cord-dorsal root ganglion cultures

Murine spinal cord and dorsal root ganglion GABAergic neurons, derived from 12-day-old fetuses, were examined autoradiographically, biochemically and immunocytochemically in vitro to determine the timecourse of appearance and maturation of this phenotype and the extent and mode of its innervation of target neurons. Specific 3H-GABA uptake into spinal cord neurons was the first property to develop and was present at the earliest time studied, one day in vitro. Immunocytochemical localization of glutamic acid decarboxylase (GAD) revealed positively stained neurons beginning at four days. At five days in vitro, electron microscopic immunocytochemistry revealed GAD-immunoreactive (GAD-IMR) boutons investing neuronal perikarya as well as neuronal processes. By one week in vitro, GAD-IMR neurons constituted 27% of the total population and GAD-IMR boutons could be seen contacting every cell with a neuronal morphology. The mode of investment of target neurons by GAD-IMR boutons was not circumscribed to either soma or dendrites but usually involved the entire neuronal perimeter and did not change with time in culture. Three morphologically distinct types of GAD-IMR neurons were evident: a small, bipolar type; a medium-sized multipolar neuron which was the most common and a large, multipolar type, resembling a motoneuron. A small population (8%) of dorsal root ganglion neurons was found to contain GAD both biochemically and immunocytochemically but was never invested by GAD-IMR boutons. GAD activity in vitro paralleled in vivo levels with maximal activity being reached at four weeks in vitro and 10 days postnatally in the intact mouse spinal cord. Murine spinal cord GABAergic neurons are a morphologically diverse and abundant neuronal population with extensive, precocious innervation of all other neuronal phenotypes in vitro suggesting that GABA has a widespread influence over other developing neuronal systems in the murine spinal cord.

The level of GAD67 protein is highly sensitive to small increases in intraneuronal gamma-aminobutyric acid levels

Increases (> 2.5-fold) in GABA levels in rat brain lead to a large decrease in the level of the 67-kDa form of glutamate decarboxylase (GAD67) through a mechanism involving either a change in GAD67 protein stability or a change in GAD67 mRNA translation. In the present study, brain levels of GABA were manipulated by treating rats with various doses of gamma-vinyl-gamma-aminobutyric acid (GVG), and the dependence of total GAD activity and levels of GAD67 and GAD65 protein on the levels of GABA was analyzed. Initial studies showed that both GABA and GAD67 protein levels reached new steady-state levels after two to four daily injections; GABA increased 1.5- (30 mg of GVG/kg) and fourfold (150 mg of GVG/kg), and GAD67 protein content decreased by 30 and 70%. To assess the sensitivity of GAD67 to GABA, rats were injected with eight different doses of GVG (15-150 mg/kg) for 5 days. With increasing doses of GVG, we observed a gradual increase in both whole-tissue and synaptosomal GABA levels and a gradual decrease in GAD67 protein and GAD activity. The levels of GAD65 remained constant at all GVG doses. GAD67 was remarkably sensitive to GABA. The synaptosomal GAD67 level decreased approximately 12% and the whole-neuron GAD67 level decreased approximately 3% for each 1% increase in nerve terminal GABA content when it was close to its physiological level. Our results clearly demonstrate that GAD67 is tightly controlled by intraneuronal GABA, and we suggest that this regulatory mechanism has important implications for the physiological regulation of GABAergic function in the mammalian brain.

Production and characterization of a new specific antiserum against the taurine putative biosynthetic enzyme cysteine sulfinate decarboxylase

We have shown previously that cysteine sulfinate decarboxylase (CSD), the putative biosynthetic enzyme of taurine in the brain, is identical to the liver enzyme according to biochemical, kinetic, and immunochemical criteria. In the present work, CSD was purified in its native form from rat liver. The purification was performed in eight steps, which included conventional chromatography (diethylaminoethyl cellulose, hydroxylapatite), followed by HPLC (hydrophobic, adsorption, and ion-exchange HPLC). The purification factor was 11,000, and the final yield was around 2%. The procedure led to the enrichment of a protein, the molecular mass of which was 51,000 daltons as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The final fraction was more than 90% homogeneous. By using this fraction as the antigen, an antiserum was raised in rabbit that (a) quantitatively immunoprecipitated CSD activity from liver and brain extract, and (b) immunolabeled one band (51,000 daltons) on immunoblots of partially purified fractions from liver. Enrichment of CSD specific activity and that of the protein immunolabeled by the antiserum for a given step, e.g., hydrophobic HPLC, were consistently parallel. The antiserum was used to carry out CSD immunocytochemistry in cerebellum. Numerous small cells were labeled in the Purkinje cell layer, the granular layer, and the white matter. In the molecular layer, Bergmann radial fibers were immunostained. The Purkinje and stellate cells were devoid of any labeling at the cell body and terminal levels. The antiserum appears to be specific for CSD and suitable for immunocytochemical visualization of CSD in the brain.

Effects of excitatory amino acids on immunocytochemically identified populations of neurons in turtle retina

Excitatory amino acids play an important role in visual processing in the retinas of many species, but little is known about the identity of the specific postsynaptic cell types and the pharmacology of their receptors. To investigate which specific cell types were affected by excitatory amino acids, we examined the effects of exogenous aspartate, glutamate, kainic acid, N-methyl-D-aspartate, and MK-801 on retinal neurons. Specific populations of neurons were labelled using antibodies directed against glucagon, enkephalin, neurotensin, gamma-aminobutyric acid, glutamic acid decarboxylase, serotonin, glycine, glutamate or aspartate. We analyzed a combination of long-term in vivo injections (seven days following an intraocular injection of kainic acid) and short term in vitro incubations. There were changes in the labelling intensity and sometimes in the relative localization of all of the antigens in the drug treated retinas. Some observations suggested that the drugs were altering neurotransmitter metabolism. Differential responses were seen in specific cell types within the populations of neurons with neurotensin-, glutamate-, aspartate-, glycine, gamma-aminobutyric acid-, and glutamic acid decarboxylase-like immunoreactivity. The immunocytochemical approach used in these studies was able to determine specific retinal cell types which were influenced by particular excitatory amino acids. The broad extent of cell types influenced and the potential metabolic effects suggest that excitatory amino acids and their receptors play a complex role in visual processing.

Fine structure of the ventral lateral nucleus (VL) of the Macaca mulatta thalamus: cell types and synaptology

Ultrastructure of the major cerebellar territory of the monkey thalamus, or VL as delineated in sagittal maps by Ilinsky and Kultas-Ilinsky (J. Comp. Neurol. 262:331-364, '87), was analyzed by using neuroanatomical tracing, immunocytochemical, and quantitative morphometric techniques. The VL nucleus contains nerve cells of two types. Multipolar neurons (PN) retrogradely labeled with wheat germ agglutinin-horseradish peroxidase (WGA-HRP) from the precentral gyrus display a tufted branching pattern of the proximal dendrites and have a range of soma areas from 200 to 1,000 microns2 (mean 535.2 microns2, SD = 159.5). Small glutamic acid decarboxylase (GAD) immunoreactive cells (LCN) exhibit sizes from 65 to 210 microns2 (mean 122.5 microns2, SD = 32.8) and remain unlabeled after cortical injections. The two cell types can be further distinguished by ultrastructural features. Unlike PN, LCN display little perikaryal cytoplasm, a small irregularly shaped nucleolus, and synaptic vesicles in proximal dendrites. The ratio of PN to LCN is 3:1. The LCN dendrites establish synaptic contacts on PN somata and all levels of dendritic arbor either singly or as a part of complex synaptic arrangements. They are also presynaptic to other LCN dendrites. Terminals known as LR type, i.e., large boutons containing round vesicles, are the most conspicuous in the neuropil. They form asymmetric contacts on somata and proximal dendrites of PN as well as on distal dendrites of LCN. The areas of these boutons range from 0.7 to 12 microns2 and the appositional length on PN dendrites ranges from 1.1 to 14 microns. All LR boutons except the largest ones become anterogradely labeled from large WGA-HRP injections in the deep cerebellar nuclei. These boutons are also encountered as part of triads and glomeruli, but very infrequently since the latter complex synaptic arrangements are rare. The most numerous axon terminals in the neuropil are the SR type, i.e., small terminals (mean area 0.42 micron2) containing round vesicles. The SR boutons become anterogradely labeled after WGA-HRP injections in the precentral gyrus. They form distinct asymmetric contacts predominantly on distal PN and LCN dendrites; however, their domain partially overlaps that of LR boutons at intermediate levels of PN dendrites. The SR boutons are components of serial synapses with LCN dendrites which, in turn, contact somata and all levels of dendritic arbors of PN. They also participate in complex arrangements that consist of sequences of LCN dendrites, serial synapses, and occasional boutons with symmetric contacts.(ABSTRACT TRUNCATED AT 400 WORDS)

Localization of GAD- and GABA-like immunoreactivity in ground squirrel retina: retrograde labeling demonstrates GAD-positive ganglion cells

Glutamic acid decarboxylase (GAD)- and gamma-aminobutyric acid (GABA)-like immunoreactivity was examined in the retina of the 13-lined ground squirrel (Spermophilus tridecemlineatus). Labeling was observed in the inner nuclear layer (INL), inner plexiform layer (IPL) and ganglion cell layer (GCL). The immunoreactive cell bodies in the inner third of the INL were 6-13 microns in diameter and, because of their size and location it was considered that these were amacrine cells. Labeling in the IPL was concentrated in 5 bands corresponding to laminae 1a, 1c, 2, 4 and 5. In the GCL a heterogeneous population of neurons exhibited GAD- and GABA-like immunoreactivity. The soma diameters of the GCL cells ranged from 5 to 17 microns. These may represent displaced amacrines and/or ganglion cells. To determine if any of the immunoreactive cells in the GCL were ganglion cells, double labeling experiments were performed using rhodamine latex microspheres ('beads') as retrograde neuronal tracers. Rhodamine beads were injected into the superior colliculus, and retinas with retrogradely labeled ganglion cells were subsequently incubated with the anti-GAD antiserum. These experiments revealed a small population of GAD-positive ganglion cells, setting a lower limit for the total number of GABAergic ganglion cells.

GAD and GABA in an enriched population of cultured GABAergic neurons from rat cerebral cortex

To study various aspects of GABAergic metabolism in an easily accessible system, dissociated cells from postnatal rat cerebral cortex were cultured in a serum-based medium and characterized morphologically and biochemically. The majority (70-96%) of the neurons were GABAergic as determined by three double-labeling procedures. The specific activity of glutamine synthetase in the cultures was 4-5% of the levels in rat astrocyte cultures and intact rat brain, indicating that glia were a minor component. The developmental increase of GABA levels preceded the increase of GAD activity in both immunocytochemical and biochemical experiments. GABA turnover rates also increased with culture age and were 20-30% of GAD activity. Four anti-GAD antibodies, which recognize GAD subunits with differing molecular masses to varying degrees, were used to stain cultured neurons and make immunoblots. Immunoblots showed that the neurons contained two major subunits of GAD which differed in mass by 2 kDa. All four antibodies immunostained both neuronal perikarya and neurites but one antibody, which on the immunoblots predominantly labeled the GAD protein with the lower molecular weight, showed a somewhat more pronounced punctate staining, possibly indicating a principal localization to neurites.

Immunocytochemistry of the taurine biosynthesis enzyme, cysteine sulfinate decarboxylase, in the cerebellum: evidence for a glial localization

Immunocytochemistry of cysteine sulfinate decarboxylase was developed in the cerebellum with an antiserum that we have recently produced using as immunogen a homogeneous fraction purified about 2000-fold from liver. In the rat, this antiserum immunoprecipitated the enzymatic activity from brain, labeled one band (molecular weight = 51,000) on immunoblots of an enriched fraction from brain but none with a brain crude extract and thus appeared to be specific. In the cerebellum numerous immunolabeled cells were found in the white matter that were typically arranged in rows like oligodendrocytes. A few immunolabeled cells were scattered in the granular layer. Around the Purkinje cells numerous small satellite cells were immunostained that sent faintly labeled radial fibers through the molecular layer. These cells were thus identified as Golgi epithelial cells with their Bergmann fibers. Purkinje cells were not labeled either at the cell body or at the nerve ending levels. In the molecular layer no cells were found consistently and convincingly immunolabeled that could correspond to the stellate cells. No immunopositive punctae typical of the interneuron nerve endings were ever observed. A significant cysteine sulfinate decarboxylase activity was also measured in glial cell fractions enriched in oligodendrocytes and astrocytes respectively. The glial localization of taurine biosynthesis enzyme in the cerebellum does not support the hypothesis that it could be involved in neurotransmission, but rather suggests that taurine is manufactured by glial cells and accordingly may play a physiological role in relation to glial function.

Identification of the 64K autoantigen in insulin-dependent diabetes as the GABA-synthesizing enzyme glutamic acid decarboxylase

The pancreatic islet beta-cell autoantigen of relative molecular mass 64,000 (64K), which is a major target of autoantibodies associated with the development of insulin-dependent diabetes mellitus (IDDM) has been identified as glutamic acid decarboxylase, the biosynthesizing enzyme of the inhibitory neurotransmitter GABA (gamma-aminobutyric acid). Pancreatic beta cells and a subpopulation of central nervous system neurons express high levels of this enzyme. Autoantibodies against glutamic acid decarboxylase with a higher titre and increased epitope recognition compared with those usually associated with IDDM are found in stiff-man syndrome, a rare neurological disorder characterized by a high coincidence with IDDM.

An animal model of prophylactic cranial irradiation: histologic effects at acute, early and delayed stages

Wistar rats (body wt. 200 g) were subjected to a fractionated course of radiation similar to that used in prophylactic brain irradiation for small cell carcinoma of the lung (2000 cGy in 5 fractions over 5 days with 60Co). Effects of this regimen were assessed by histologic examination of brain sections at 1 week, 1 month and 6 months post-irradiation. With conventional stains there were no apparent differences between control and irradiated brains at any of the post-irradiation intervals. Immunohistochemistry for neurotransmitter synthetic enzymes tyrosine hydroxylase and glutamate decarboxylase, as well as histochemistry for acetylcholinesterase, failed to uncover any changes in the irradiated animals. Immunohistochemistry for glial fibrillary acidic protein, an astrocyte marker, also showed no differences in the irradiated groups. However, an antibody against a major histocompatibility complex, class II antigen (OX-6) revealed a microglial response in grey and white matter beginning at 1 month and increasing up to the 6 month post-irradiation interval. The neuroanatomical basis for this microglial response was suggested by the results of silver stains for nerve axons, which revealed axonal loss in striatal white matter bundles in a pattern implicating vascular insufficiency.

Fine structure of the magnocellular subdivision of the ventral anterior thalamic nucleus (VAmc) of Macaca mulatta: I. Cell types and synaptology

The nucleus ventralis anterior pars magnocellularis (VAmc) is recognized only in primates and is the major recipient of the nigrothalamic projections. The neuronal and synaptic composition of this nucleus in the rhesus monkey was studied with the use of a variety of neuroanatomical techniques that included quantitative morphometry, anterograde and retrograde labeling with WGA-HRP from the prefrontal cortex, and immunocytochemistry for glutamic acid decarboxylase (GAD). Two major cell types were identified in the nucleus: thalamocortical projection neurons (PN) that were multipolar cells of various sizes, and small GAD-immunoreactive cells, apparently local circuit neurons (LCN). The approximate ratio of the two types of cells was 10:1. The major type of bouton encountered in the neuropil was of medium to large-sized (areas from 1.5 to 12 microns 2) and mostly of en passant type. These terminals formed symmetric contacts, contained moderate amounts of pleomorphic or mostly flat synaptic vesicles and large numbers of mitochondria, and displayed numerous puncta adhaerentia. All of these boutons exhibited positive GAD immunoreactivity. These boutons constituted the only synaptic population on somata and primary dendrites of PN and formed an overwhelming majority on the secondary PN dendrites. There were fewer of these axon terminals on tertiary PN and LCN dendrites. Additionally, boutons with similar features formed synapses on axon hillocks or initial axonal segments of PN, and somata or very proximal parts of primary dendrites of LCN. With the exception of the boutons in the last two locations, all of the other boutons in this group were shown to be terminals of the nigrothalamic afferents in the parallel EM autoradiographic study (Kultas-Ilinsky and Ilinsky: J. Comp Neurol. 294:479-489, '90). The second major bouton population in the VAmc was represented by small to medium-sized terminals (areas range from 0.2 to 2.0 microns 2) that formed distinct asymmetric contacts and contained large numbers of round vesicles and few or no mitochondria. These boutons were labeled anterogradely from the cortex and dominated on distal PN and LCN dendrites. Some of them were found on secondary PN dendrites where they formed synapses either directly or indirectly via LCN dendrites and dendro-dendritic contacts. The latter arrangements, i.e., serial synapses, were also formed between the cortical boutons and PN somata or tertiary dendrites.(ABSTRACT TRUNCATED AT 400 WORDS)

Specific antiserum and monoclonal antibodies against the taurine biosynthesis enzyme cysteine sulfinate decarboxylase: identity of brain and liver enzyme

Cysteine sulfinate decarboxylase (CSD), the putative biosynthetic enzyme for taurine, was purified 1,800-fold with a 1% yield from rat liver, where it was found to be 20-fold enriched compared with brain. The final fraction was homogeneous, as ascertained through sodium dodecyl sulfate-polyacrylamide gel electrophoresis and reverse-phase HPLC. An antiserum was raised in the rabbit that (a) quantitatively immunoprecipitated CSD activity and (b) immunolabeled only one band (MW = 51,000) on an immunoblot from liver homogenate. Monoclonal antibodies were also raised that recognized the CSD protein and immunolabeled the same 51-kilodalton protein on an immunoblot from liver homogenate. In a brain extract, two CSD activities had been previously found and named CSDI and CSDII, according to their chromatographic elution patterns. We have compared the properties of CSDI from brain--the most likely enzyme involved in the biosynthesis of taurine in the brain, according to previous investigations-and CSD from liver: Both activities (a) were similarly eluted on ion-exchange and hydroxyapatite chromatographies, (b) showed the same elution pattern on gel filtration with an apparent native molecular weight of approximately 63,000, and (c) were immunoprecipitated in a strictly identical manner by the antiserum against liver CSD. Moreover, this antiserum as well as the monoclonal antibodies immunolabeled a single band (51 kilodaltons) on an immunoblot from brain CSD-enriched fraction or liver fraction. All these data show that CSDI from brain and liver CSD are the same monomeric enzyme. They also indicate that a specific antiserum against rat liver CSD has been raised that can be used for immunocytochemical visualization of CSD-containing cells in the brain.

Displaced cholinergic, GABAergic amacrine cells in the rabbit retina also contain adenosine

It is generally accepted that the purine nucleoside, adenosine, plays a neuromodulatory role in the central nervous system (CNS) (Daly et al., 1981; Phillis & Wu, 1983; Williams, 1986; Williams, 1987; Snyder, 1985). Adenosine is thought to exert its primary effects presynaptically, by inhibiting the release of neurotransmitters including gamma-aminobutyric acid (GABA) and acetylcholine (ACh) (Phillis & Barraco, 1985; Proctor & Dunwiddie, 1987). In mammalian retina, cell bodies that are strongly labeled for adenosine-like immunoreactivity (ALIR) have been localized to the ganglion cell layer (GCL) (Braas et al., 1987; Blazynski et al., 1989). Rabbit retinal cells that are labeled by markers for both ACh and GABA are located in the GCL and inner nuclear layer (INL) (Tauchi & Masland, 1984; Vaney & Young, 1988 b; Brecha et al., 1988). It is now demonstrated in the rabbit retina that approximately 50% of the cells labeled for ALIR within the GCL represent true ganglion cells, with the remainder presumed to be displaced cholinergic amacrine cells (DAPI accumulating). In addition, some of these same cells also demonstrate immunoreactivity to glutamate decarboxylase (GAD), involved in the biosynthesis of the neurotransmitter GABA. Thus, in a particular class of retinal neurons, two fast-acting neurotransmitters as well as a putative neuromodulator have been co-localized.

Localization of GABA- and GAD-like immunoreactivity in the turtle retina

gamma-aminobutyric acid (GABA) has been reported to be an important neurotransmitter in the retinas of many species. This immunocytochemical study detailed the localization of antigens resembling GABA and glutamic acid decarboxylase (GAD, an enzyme involved in the synthesis of GABA), in retinal neurons in the turtle, Pseudemys scripta elegans. GABA-like immunoreactivity was present within somata in the inner and outer regions of the inner nuclear layer, within somata in the ganglion cell layer, and in processes in the outer plexiform layer, inner plexiform layer, and ganglion cell axon layer. GAD-like immunoreactivity was found in somata in the inner and outer regions of the inner nuclear layer and in processes in the inner and outer plexiform layers. Cell counts indicated more somata with GABA-like than GAD-like immunoreactivity in the inner nuclear layer. Double-label studies showed that every somata in the inner nuclear layer which had GAD-like immunoreactivity also had GABA-like immunoreactivity, but that many somata had only GABA-like immunoreactivity. The stratification of immunoreactivity within the inner plexiform layer was analyzed using a scanning densitometer. We described the strate within the inner plexiform layer such that S0 represented the inner nuclear layer/inner plexiform layer border and S100 represented the inner plexiform layer/ganglion cell layer border. Analysis of GAD-like labeling yielded seven distinct strata with peak densities at positions S8, S19, S28, S42, S59, S75, and S93. GABA-like labeling provided five distinct strata with peak densities at positions S17, S28, S67, S84, and S95. The strata with peaks of GABA-like immunoreactivity at S17 and S28 were in statistically identical locations to corresponding strata with GAD-like immunoreactivity. The strata with GABA-like immunoreactivity at S67, S84, and S95 did not have statistically identical peaks of correlated GAD-like immunoreactivity, although there were corresponding strata with GAD-like immunoreactivity nearby. Antiserum directed against GABA failed to produce labeled strata at positions corresponding to the strata with GAD-like immunoreactivity at S8 and S42. In summary, our results indicated that the antisera we used, which were directed against GABA and GAD, produced significantly different labeling in the inner nuclear layer, inner plexiform layer, and the ganglion cell body and axon layers of the turtle retina. Until the physiological significance of these differences is resolved, studies employing these markers to investigate the function of GABA in the turtle retina should be interpreted with caution.

Localization of putative GABAergic neurons in the larval tiger salamander retina by immunocytochemical and autoradiographic methods

Putative GABAergic neurons in the larval tiger salamander retina were localized by a comparative analysis of glutamate decarboxylase immunoreactivity (GAD-IR), GABA-like immunoreactivity (GABA-IR), and high-affinity 3H-GABA uptake at the light microscopical level. Preliminary data showed that all GAD-IR neurons were double labeled for GABA-IR. However, because the weak somatic labeling with GAD-IR, we could not determine if the converse were true. Neurons commonly labeled with GABA-IR and 3H-GABA uptake include horizontal cells, type I (outer) and type II (inner) bipolar cells, type I (inner) and type II (outer) amacrine cells, and cell bodies in the ganglion cell layer (GCL). In addition, interplexiform cells were identified with GABA-IR. The presence of GABA-IR ganglion cells was indicated by GABA-IR fibers in the optic fiber layer and optic nerve as well as by a GABA-IR cell in the GCL that included a labeled axon. The percentage of labeled somas in the inner nuclear layer (INL) compared to all cells in each layer was similar for the two methods: 30% in INL 1 (outer layer of somas), 15% in INL 2 (middle layer), 43-52% in INL 3 (inner layer), and about 21-26% in the GCL. Labeled processes were found in three bands in the inner plexiform layer, with the densest band located in the most proximal part. Postembedding labeling of 1-micron Durcupan resin sections for GABA-IR showed the same general pattern as obtained with 10-microns cryostat sections, with additional staining, however, of type II (inner) bipolar cell Landolt's clubs. Extensive colocalization of labeling was indicated, and we conclude that GABA-IR can serve as a valid and reliable marker for GABA-containing neurons in this retina and suggest that GABA serves as a transmitter for horizontal cells, several types of amacrine cell, a type of interplexiform cell, and perhaps a small percentage of type I and type II bipolar cells and ganglion cells.

Distribution of glutamic acid decarboxylase and gamma-aminobutyric acid in the hypoglossal nucleus in the rat

Immunocytochemistry was used to investigate the distribution of glutamic acid decarboxylase (GAD) and gamma-aminobutyric acid (GABA) in the hypoglossal nucleus (XII) of the adult rat. The distribution of GAD and GABA was found to be co-extensive throughout XII. Although immunoreactivity was moderately dense in all regions, the intensity of staining was greatest in the ventral district of XII particularly ventromedially in the caudal half of the nucleus. Immunoreactive terminal-like profiles were observed around motoneuron somata and dendrites. There also was evidence of sparse mediolaterally-oriented densities of immunoreactivity at the junction of XII and the dorsal vagal nucleus and between dorsal and ventral districts of XII. In addition, neurons staining positive for GABA were found scattered within XII laterally and immediately outside of XII in and around the Nucleus of Roller. These observations suggest a complex, differential distribution of GABA in XII and are discussed in relation to tongue motor behavior.

GABA nerve endings in the rat red nucleus combined detection with serotonin terminals using dual immunocytochemistry

Immunocytochemical methods were used to examine the ultrastructural features and cellular interrelationships of GABA and serotonin afferent fibers to the rat red nucleus. GABAergic nerve endings were identified in two ways, either using a pre-embedding immunoperoxidase procedure with an antibody against glutamate decarboxylase, the GABA-synthesizing enzyme, or after post-embedding immunogold labelling with an anti-GABA antibody. With the latter approach, it was possible to simultaneously visualize the GABAergic and serotoninergic innervation of the red nucleus (magnocellular part) in electron microscope preparations. This procedure involved GABA labelling of ultrathin sections obtained from specimens previously immunostained for serotonin using the pre-embedding peroxidase-antiperoxidase technique. The doubly stained sections showed gold and peroxidase markers to be present in two distinct populations of axonal varicosities. Unlike the GABAergic nerve endings, which were found to be profusely distributed throughout the nucleus, the serotonin nerve endings were relatively scarce. They contacted dendrites of large-sized neurons usually endowed with several GABA-gold labelled terminals. Not uncommonly, direct appositions between serotonin and GABA-positive terminals were also encountered. These data provide morphological evidence that red nucleus outputs may be dually regulated by GABAergic and serotoninergic afferents, while suggesting that presynaptic GABA/serotonin interactions might also play a significant part in red nucleus functions.

Perfusion-fixation of the human brain for immunohistochemistry: comparison with immersion-fixation

A method of perfusion-fixation of the human brain is described and compared with immersion-fixation by immunoperoxidase staining for several substances (tyrosine hydroxylase, substance P, choline acetyltransferase, glutamate decarboxylase, Met-enkephalin, and neuron-specific enolase) in human striatum. Results from 1-cm slices fixed by immersion for 1, 2, 4 and 8 days were compared with results from slices of perfused brain postfixed for the same time periods. The fixative used in all steps was 4% paraformaldehyde at 4 degrees C. In the immersion-fixed brains, optimal immunoreaction for tyrosine hydroxylase and glutamate decarboxylase was limited to a depth of 1-2 mm from the surface of the brain slice. In contrast, staining density in perfusion-fixed brains was relatively homogeneous and of high quality. The other antigens studied displayed more uniform staining throughout the section with both perfused and immersed brains. Investigators intending to study human brain immunohistochemistry using immersion-fixation should be aware of the possibility of depth-related variations in staining intensity and would be wise to determine whether this effect is significant for the antigens they choose to study.

Taurine biosynthesis in rat brain: a new specific and sensitive microassay of cysteine sulfinate decarboxylase (CSDI) activity through selective immunotrapping and its use for distribution studies

Cysteine sulfinate decarboxylase (CSD), the putative biosynthetic enzyme for taurine, has been shown to exist in two forms in rat brain, respectively CSDI and CSDII, one of which (CSDII) is considered to be in fact glutamate decarboxylase (GAD). CSDI assay after immunotrapping was made possible by using an anti-CSD antiserum raised in sheep immunized with a partially purified CSD fraction from liver. This antiserum immunoprecipitated both liver CSD and brain CSDI activities with the same affinity but did not inhibit their enzymatic activities. The immunotrapping of CSDI was selective without any contamination by GAD/CSDII activity. The immunotrapped CSD activity, which corresponded exactly to the amount of CSD not precipitated by a GAD/CSDII antiserum, was not inhibited by a specific irreversible GAD inhibitor. A quantitative, selective and sensitive assay was thus developed by measuring CSD activity on the solid phase after immunotrapping. Kinetic parameters of the immunotrapped enzyme remained unchanged. CSDI activity represented only a fraction, around 20% with saturating concentration of substrate, of the total CSD activity in rat brain homogenate. This indicates that most studies on total CSD activity dealt essentially with CSDII activity that is indeed GAD. Regional and subcellular distributions of CSDI have been determined. CSDI activity was about threefold higher in the richest (cerebellum) compared to the poorest (striatum) region without any correlation with GAD/CSDII distribution. Subcellular distribution showed a fourfold enrichment of CSDI activity in the synaptosomal fraction. The precise role of CSDI and CSDII in the biosynthesis of taurine in vivo remains to be elucidated.

Taurine biosynthesis in rat brain in vivo: lack of relationship with cysteine sulfinate decarboxylase glutamate decarboxylase-associated activity (GAD/CSDII)

Two distinct forms of cysteine sulfinate decarboxylase (CSD), respectively, CSDI and CSDII, have already been separated in rat brain. One of them, CSDII, appeared to be closely associated with glutamate decarboxylase (GAD). We have investigated whether the taurine concentration in brain was dependent on CSDII activity in vivo. CSDI and CSDII activities were specifically measured in crude brain extracts after selective immunotrapping. After 4 days of chronic treatment of mice with gamma-acetylenic gamma-aminobutyric acid, a drastic and identical decrease in CSDII and GAD activities was observed in the brain. Taurine concentration and CSDI activities were not significantly altered. Following striato-nigral pathway lesioning in the rat brain, GAD and CSDII show an identical 80% decrease in the substantia nigra. In contrast, CSDI activity and taurine concentration in the substantia nigra were similarly but only slightly affected with an about 30% decrease. Our results provide further evidence that GAD and CSDII are indeed the same enzyme. They show that CSDII does not play any role in the biosynthesis of taurine in vivo. Our findings suggest that CSDI might be the biosynthetic enzyme for taurine in vivo and that there might be some endings projecting into the substantia nigra that contain CSDI and taurine.

Double-label analysis of GAD- and GABA-like immunoreactivity in the rabbit retina

Rabbits were injected intravitreally with colchicine and the retinas were subsequently processed for simultaneous visualization of GAD-like and GABA-like immunoreactivity (LIR) at the light microscopical level. The use of specific antisera from unrelated species minimized the possibility of crossreactivity between the two markers. Colocalization of GABA-LIR and GAD-LIR was observed in approximately 32% of somata on the inner border of the inner nuclear layer (INL) and roughly 15% of the somata in the ganglion cell layer (GCL). All GAD-positive cells were double labeled whereas 15% of the GABA-LIR cells in the INL and 52% of those in the GCL were negative for GAD-LIR. Both markers were seen throughout the inner plexiform layer. In addition, processes coursing through the OPL, perhaps from interplexiform cells, and horizontal cell bodies were double-labeled with these GABAergic probes.

Immunocytochemical analysis of the GABAergic innervation of oxytocin- and vasopressin-secreting neurons in the rat supraoptic nucleus

Antisera specific for gamma-aminobutyric acid (GABA) or its biosynthetic enzyme, glutamate decarboxylase, were used in pre- and postembedding immunocytochemical techniques at the light and electron microscopic levels, to visualize the GABAergic innervation of the hypothalamic supraoptic nucleus. Immunostaining for glutamate decarboxylase or gamma-aminobutyric acid were also combined with oxytocin and vasopressin immunolocalization, thereby permitting evaluation of the contribution of the innervation onto each type of neuron in this nucleus. Light microscopy of semithin plastic sections or vibratome slices stained for glutamate decarboxylase or gamma-aminobutyric acid, with peroxidase-antiperoxidase as immunolabel, revealed an extensive punctate labeling in the supraoptic nucleus and its immediate surroundings. Quantitative analysis of glutamate decarboxylase immunostaining in semithin sections indicated a comparable density of immunopositive punctae at the anterior and posterior levels of the nucleus (14-27 X 10(6) per mm3 tissue). Glutamate decarboxylase- or gamma-aminobutyric acid-immunoreactive cell bodies were never observed within the nucleus although they were detected in the hypothalamus immediately dorsolateral to the nucleus. Electron microscopy of vibratome slices treated with antiglutamate decarboxylase or antigamma-aminobutyric acid and peroxidase-antiperoxidase, or of ultrathin sections stained directly with antigamma-aminobutyric acid and immunoglobulin-coupled colloidal gold, showed that the immuno-reactive punctae represented, in the main, axonal terminals. They invariably contained small, rounded clear vesicles and, at times, one or two larger, dense cored vesicles; they all formed symmetrical synapses onto magnocellular cell bodies and dendrites. Oxytocin and vasopressin neurons were contacted in a similar fashion by glutamate decarboxylase- or gamma-aminobutyric acid-positive boutons in semithin sections of the nucleus stained simultaneously for glutamate decarboxylase and oxytocin and in ultrathin sections stained for glutamate decarboxylase or gamma-aminobutyric acid and oxytocin or vasopressin. Glutamate decarboxylase- or gamma-aminobutyric acid-positive terminals often formed synapses onto two postsynaptic elements in the same plane of section ("double" synapses), a synaptic configuration usually encountered in supraoptic nuclei of lactating animals. In such cases, the postsynaptic somata were oxytocinergic.(ABSTRACT TRUNCATED AT 400 WORDS)

Phylogenesis of brain glutamic acid decarboxylase from vertebrates: immunochemical studies

Brain high-speed supernatants from various lower and higher vertebrates were subjected to sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis, electroblot on nitrocellulose membranes, and immunolabelling using an anti-glutamic acid decarboxylase (anti-GAD) antiserum prepared from rat antigen. Rat brain extracts showed two distinct immunolabelled bands (MW 59,000 and 62,000 daltons). The molecular weight of the native enzyme was 120,000 daltons. The immunoblot pattern was not affected by a 3-h incubation of the homogenate. In the substantia nigra, the decrease in the immunolabelling of both bands corresponded very closely to the decrease of GAD activity following lesioning of the striato-nigral pathway. Moreover, experiments with preadsorbed antiserum showed that both subunits have common antigenic determinants. The immunolabelling was consistently more intense over the lightest band. The autoradiography of immunoprecipitated rat brain GAD, iodinated prior to electrophoresis, revealed two radiolabelled bands corresponding to the two immunolabelled ones. Their radioactivity was found in a one-to-five ratio which closely paralleled their respective immunolabelling intensity. Thus, the two subunits recognized by the antiserum are not present in stoichiometric proportions in the rat brain high-speed supernatant. These findings suggest the existence of two homodimeric GAD with common antigenic determinants which are present in different amounts. Immunoprecipitation curves of brain GAD from rat, mouse, rabbit, monkey, human, quail, frog, and trout were similar, with a less than 10-fold maximum shift in affinity for GAD. GAD immunoblots from the various higher vertebrates showed a pattern similar to that obtained in rat.(ABSTRACT TRUNCATED AT 250 WORDS)

Parvalbumin in most gamma-aminobutyric acid-containing neurons of the rat cerebral cortex

gamma-Aminobutyric acid (GABA) is one of the major inhibitory neurotransmitters in the central nervous system. In the cerebral cortex, GABA-containing cells represent a subpopulation of interneurons. With semithin frozen sections, it is possible to demonstrate that most GABA neurons in the rat somatosensory cortex contain the calcium-binding protein parvalbumin and that parvalbumin is found virtually only in GABA neurons. Parvalbumin seems to influence the electrical properties and enzymatic machinery to modulate neuronal excitability and activity. The specific role of parvalbumin in GABA-containing cortical cells may be related to controlling the effectiveness of their inhibitory action.

Immunocytochemical demonstration of topographic ordering of Purkinje cell axon terminals in the fastigial nuclei of the rat

We have used a monoclonal antibody, mabQ113, which selectively stains a subset of cerebellar Purkinje cells, to study the topography of the corticonuclear projection of the median vermis to the fastigial nuclei in the rat. The fastigial projection zone contains both mabQ113+ and mabQ113- Purkinje cells grouped into parasagittal bands. The immunoreactivity extends throughout the Purkinje cell including the axon terminals and thus it is possible to investigate the topographic distribution of mabQ113+ terminals. In the fastigial nuclei the target cells receiving mabQ113+ axon terminals are concentrated in the caudal pole. In the rostral pole, cells receive anti-GAD+ terminals but not mabQ113+ terminals. There is no gradient in anti-GAD staining. Double-labelling experiments with mabQ113, anti-GAD, and an antisynaptic antibody mabQ155 suggest that there is little or no mixing of mabQ113+ and mabQ113- Purkinje cell terminals on the same target neuron.

Colocalization of GAD-like immunoreactivity and 3H-GABA uptake in amacrine cells of rabbit retina

Rabbit retinas were double labeled to determine the degree of colocalization of glutamic-acid-decarboxylase-like immunoreactivity (GAD-like IR) and 3H-GABA uptake using light (LM) and electron microscopic (EM) autoradiography. Both GAD-like IR and 3H-GABA uptake were found in amacrine cell bodies in the inner nuclear layer (INL) as well as in cell bodies in the ganglion cell layer (GCL), and throughout the inner plexiform layer. GAD-like IR was found in 32% of the amacrine cells in the INL, 86% of which also showed 3H-GABA uptake; 3H-GABA uptake was observed in 38% of the amacrine cells. However, only 72% of these cells showed GAD-like IR. Labeled cells in the GCL were only 10-15% as common as similarly labeled cells in the INL. As in the INL, all GAD-positive cells in the GCL were double labeled, but only 53% of the cells taking up 3H-GABA were double labeled. We suggest that labeled cells in the GCL were ganglion cells rather than displaced amacrine cells. Cells, in both the INL and GCL, that showed 3H-GABA uptake but no GAD-like IR had a higher average grain density than double-labeled cells, indicating that uptake by these cells was specific. The relevance to GABAergic function of 3H-GABA uptake without an indication of GAD-like IR is yet to be determined. Statistical analysis at the EM level showed that one-third of the GAD-positive synaptic terminals of amacrine cells were double labeled after a 4-month exposure. Longer exposures at the EM level should reveal a higher percentage of GAD-positive terminals because at the LM level, one-half of the double-labeled cell bodies were "lightly" labeled with grains. The high degree of colocalization of GAD-like IR and 3H-GABA uptake suggests that both markers may be useful for labeling GABAergic neurons in the rabbit retina.

Glutamate decarboxylase-immunoreactive boutons in synaptic contacts with hypothalamic dopaminergic cells: a light and electron microscopy study combining immunocytochemistry and radioautography

Double post-embedding immunolabeling of both tyrosine hydroxylase and glutamate decarboxylase on 1-micron semi-thin sections allowed the visualization of numerous endings that use gamma-aminobutyrate as a transmitter apposed to dopaminergic cell bodies in the periventricular-arcuate hypothalamic complex. Up to fifteen glutamate decarboxylase-positive contacts per tyrosine hydroxylase-positive cell profile could be observed. In some favourable planes of section glutamate decarboxylase-positive endings were also seen in close apposition to proximal dopaminergic dendrites. About 250 tyrosine hydroxylase-positive cell profiles, whose diameter approached the maximum diameter of the dopaminergic cells, were surveyed. An average of 7.4 glutamate decarboxylase-positive contacts were counted on these profiles. From these figures it was estimated that a dopaminergic cell body was contacted on average by 75-175 terminals that use gamma-aminobutyrate as a transmitter. At the electron-microscopic level, the nature of these contacts was investigated by a method combining radioautographic detection of cell bodies having taken up tritiated dopamine and pre-embedding immunostaining of glutamate decarboxylase containing endings. Glutamate decarboxylase-positive axon terminals were seen apposed to somatic and dendritic elements. On some favorable planes of section, they were found to be engaged in morphologically defined synaptic complexes of the symmetrical or asymmetrical type. A number of the postsynaptic perikarya were labelled by tritiated dopamine and, in agreement with the light microscopic observations, they were frequently seen in contact with more than one immunopositive ending. The present findings provide a morphological substratum for a direct gamma-aminobutyrate control of the tuberoinfundibular dopaminergic neurons. Such a control could account more particularly for the central, stimulatory effects of gamma-aminobutyrate on prolactin secretion.

Light- and electron-microscopic immunocytochemistry of glutamic acid decarboxylase (GAD) in the basal hypothalamus: morphological evidence for neuroendocrine gamma-aminobutyrate (GABA)

GABAergic cells and axon terminals were localized in the basal hypothalamus of differnet species (rat, mouse and cat), by means of an immunocytochemical approach using a specific and well-characterized antiserum to the GABA biosynthetic enzyme, glutamate decarboxylase. Light-microscopic visualization was performed with an indirect immunofluorescence method and electron-microscopic observations were made on material with pre-embedding staining and use of the peroxidase-antiperoxidase procedure. At the light-microscopic level, a dense immunofluorescent plexus was observed over both the medial and lateral parts of the external layer of the median eminence. The labelling extended from the rostal part of the median eminence up to the pituitary stalk. Over the subependymal and internal layers only a few immunoreactive dots were visible, except around the blood vessels where they appeared more concentrated. Immunoreactive varicosities could be found following the outlines of the capillary loops and lining tanycyte processes, especially in the median eminance midportion. At the electron-microscopic level, the immunolabelling was exclusively found over neuronal profiles in the median eminence. The latter represented a small fraction of the total number of varicosities visible on the same section. Labelled profiles typically contained numerous small clear synaptic vesicles and only a few or no dense-core vesicles. In the subependymal and internal layers, rare labelled endings were found close to ependymal cells or among transversally cut fibers, respectively. In the palisadic zone, elongated positive boutons were visible intermingled with bundles of unlabelled axons and glial or ependymal processes. In the neurohemal contact zone, immunoreactive endings were observed among unlabelled neurosecretory endings in close vicinity to fenestrated capillary perivascular space. Small moderately intense immunofluorescent varicosities were observed all over the hypothalamus. The density of the glutamate decarboxylase-positive network was higher than in most diencephalic regions. Intraventricular or topical injection of colchicine allowed the visualization of small lightly immunoreactive cells in the diffusion area of colchicine. In the arcuate nucleus labelled axonal endings containing small pleomorphic synaptic vesicles and sometimes a few dense-core vesicles were observed at the electron-microscopic level. Typical synaptic junctions were commonly found between positive endings and unlabelled perikarya, or more frequently, unlabelled dendrites. These findings show that glutamate decarboxylase-containing endings are localized ed in several strategic sites for potential GABAergic neuroendocrine regulations. The GABAergic endings found among neurosecretory endings in the neurohemal contact zone may provide the morphological support for the release of gamma-aminobutyrate into the portal blood flow as an hypothalamic hypophysiotropic hormone.