GABAergic innervation of somatostatin-containing neurosecretory cells of the anterior periventricular hypothalamic area: a light and electron microscopy double immunolabelling study

Kisspeptin Stimulates Growth Hormone Release by Utilizing Neuropeptide Y Pathways and Is Dependent on the Presence of Ghrelin in the Ewe

Although kisspeptin is the primary stimulator of gonadotropin-releasing hormone secretion and therefore the hypothalamic-pituitary-gonadal axis, recent findings suggest kisspeptin can also regulate additional neuroendocrine processes including release of growth hormone (GH). Here we show that central delivery of kisspeptin causes a robust rise in plasma GH in fasted but not fed sheep. Kisspeptin-induced GH secretion was similar in animals fasted for 24 hours and those fasted for 72 hours, suggesting that the factors involved in kisspeptin-induced GH secretion are responsive to loss of food availability and not the result of severe negative energy balance. Pretreatment with the neuropeptide Y (NPY) Y1 receptor antagonist, BIBO 3304, blocked the effects of kisspeptin-induced GH release, implicating NPY as an intermediary. Kisspeptin treatment induced c-Fos in NPY and GH-releasing hormone (GHRH) cells of the arcuate nucleus. The same kisspeptin treatment resulted in a reduction in c-Fos in somatostatin (SS) cells in the periventricular nucleus. Finally, blockade of systemic ghrelin release or antagonism of the ghrelin receptor eliminated or reduced the ability of kisspeptin to induce GH release, suggesting the presence of ghrelin is required for kisspeptin-induced GH release in fasted animals. Our findings support the hypothesis that during short-term fasting, systemic ghrelin concentrations and NPY expression in the arcuate nucleus rise. This permits kisspeptin activation of NPY cells. In turn, NPY stimulates GHRH cells and inhibits SS cells, resulting in GH release. We propose a mechanism by which kisspeptin conveys reproductive and hormone status onto the somatotropic axis, resulting in alterations in GH release.

Glutamatergic innervation of growth hormone-releasing hormone-containing neurons in the hypothalamic arcuate nucleus and somatostatin-containing neurons in the anterior periventricular nucleus of the rat

Growth hormone-releasing hormone (GHRH) and somatostatin are the two main hypothalamic neurohormones, which stimulate or inhibit directly hypophysial growth hormone (GH) release. Majority of the GHRH neurons projecting to the median eminence is situated in the arcuate nucleus and the somatostatin neurons in the anterior periventricular nucleus. Data suggest that the excitatory amino acid glutamate may play an important role in the control of hypothalamic neuroendocrine neurons and processes including the control of GH. There is a dense plexus of glutamatergic fibres in the hypothalamic arcuate and anterior periventricular nucleus. The aim of the present studies was to examine the relationship of these fibres to the GHRH neurons in the arcuate nucleus and to somatostatin neurons in the anterior periventricular nucleus. Double-labelling immuno-electron microscopy was used. Glutamatergic structures were identified by the presence of vesicular glutamate transporter 2 (VGluT2) (a selective marker of glutamatergic elements) immunoreactivity. A significant number of VGluT2-immunoreactive boutons was observed to make asymmetric type of synapses with GHRH-immunostained nerve cells in the arcuate and with somatostatin neurons in the anterior periventricular nucleus. A subpopulation of somatostatin-immunoreactive neurons displayed also VGluT2 immunoreactivity. Our findings provide direct neuromorphological evidence for the view that the action of glutamate on GH release is exerted, at least partly, directly on GHRH and somatostatin neurons releasing these neurohormones into the hypophysial portal blood.

Modulation Of [35S]-tert-butylbicyclophosphorothionate binding by somatostatin in rat hypothalamus

1. The present study was designed to assess the effect of the tetradecapeptide somatostatin on the GABA(A) receptor complex in the rat hypothalamus. 2. GABA(A) receptors were labelled with [35S]-tert-butylbicyclophosphorothionate (TBPS), which binds in or near the chloride channel, and binding as assessed by in vitro quantitative autoradiography using a computer-assisted image analysis system. 3. Somatostatin inhibited the binding of [35S]-TBPS to the convulsant site of the hypothalamic GABA(A) receptor complex of rat slide-mounted hypothalamic structures in a concentration-dependent manner with an affinity in the micromolar range (10(-6) to 3 x 10(-6) mol/L). Somatostatin appeared to mimic the effects of the neurosteroid 5alpha-pregnane-3alpha ol-one (5alpha3alphaP), GABA and picrotoxin on [35S]-TBPS binding in the rat hypothalamus in all structures examined. Furthermore, GABA or muscimol (a GABA(A) receptor agonist), when added to the incubation medium, enhanced the capacity of somatostatin to inhibit [35S]-TBPS binding, with an IC50 of 10(-7) mol/L. However, incubation with bicuculline (a GABA(A) receptor antagonist) led to the abolition of the inhibitory effect of somatostatin on [35S]-TBPS specific binding in rat hypothalamus. 4. The present results demonstrate the presence of a modulatory effect of somatostatin on the GABA(A) receptor complex in rat hypothalamic structures. Furthermore, the data suggest that somatostatin allosterically modifies [35S]-TBPS binding through a mechanism similar to that of GABA. Taken together, these results provide evidence for the presence of somatostatin- GABA interactions in rat hypothalamus.

Prolactin: structure, function, and regulation of secretion

Prolactin is a protein hormone of the anterior pituitary gland that was originally named for its ability to promote lactation in response to the suckling stimulus of hungry young mammals. We now know that prolactin is not as simple as originally described. Indeed, chemically, prolactin appears in a multiplicity of posttranslational forms ranging from size variants to chemical modifications such as phosphorylation or glycosylation. It is not only synthesized in the pituitary gland, as originally described, but also within the central nervous system, the immune system, the uterus and its associated tissues of conception, and even the mammary gland itself. Moreover, its biological actions are not limited solely to reproduction because it has been shown to control a variety of behaviors and even play a role in homeostasis. Prolactin-releasing stimuli not only include the nursing stimulus, but light, audition, olfaction, and stress can serve a stimulatory role. Finally, although it is well known that dopamine of hypothalamic origin provides inhibitory control over the secretion of prolactin, other factors within the brain, pituitary gland, and peripheral organs have been shown to inhibit or stimulate prolactin secretion as well. It is the purpose of this review to provide a comprehensive survey of our current understanding of prolactin's function and its regulation and to expose some of the controversies still existing.

Neuropeptide Y and somatostatin in the anterior piriform cortex alter intake of amino acid-deficient diets

Neuropeptides affect food intake via peripheral and brainstem mechanisms, but their roles in mediating feeding via the cerebral cortex have received little attention. The anterior piriform cortex (APC) appears to play a critical role in neuroperception of deficiencies of essential amino acids (AA) and the anorectic response to such deficiencies. The neural circuitry underlying the role of this paleocortex in these events is not understood. We have shown that neurons containing neuropeptide Y (NPY) and somatostatin (SOM) are cytoarchitecturally in positions to relate synaptically to the neurons of the APC which may mediate responses to AA. Thus, we hypothesized that NPY and SOM administered intracortically to the APC would directly affect food intake in a threonine-imbalanced model. We determined that NPY at 1-1.5 nmol decreased intake of the AA-deficient diet for 3 h, with a cumulative effect that extended through 6 h. SOM had a dual effect; at 1 pmol it increased intake of the AA-deficient diet for 3 h; at 2 nmol, SOM decreased intake of the AA-deficient diet for over 9 h, with a cumulative effect that persisted through 12 h. In the first 3 h, intake of animals receiving 1 pmol of SOM differed significantly from those receiving 2 nmol. These results suggest that NPY and SOM affect the cortical circuitry responsible for recognition of deficiencies in nutritionally essential AA, and that the timing of the cortical responses to the peptides may be related to the time course of the anorectic responses.

Organisation of the cerebellar nucleus of the dogfish, Scyliorhinus canicula L.: a light microscopic, immunocytochemical, and ultrastructural study

Elasmobranchs possess a well-developed cerebellum with an associated cerebellar nucleus. To determine whether the organization of this nucleus is comparable with that of the deep cerebellar nuclei of mammals, we studied the dogfish cerebellar nucleus with light microscopic methods (Nissl stain, Golgi method, reduced silver stain, NADPH-diaphorase histochemistry and immunocytochemistry) and with electron microscopy. We found the dogfish cerebellar nucleus to consist of about 1,050 large neurons, the ratio of Purkinje cells to cerebellar nucleus neurons being about 17:1. Immunocytochemistry showed large glutamatergic neurons in the main portions of the nucleus and small glutamate- and/or alpha-aminobutyric acid (GABA)-immunoreactive cells in the subventricular region of the nucleus. Large glutamatergic neurons corresponded to bipolar or triangular cells revealed by Golgi methods. Application of horseradish peroxidase to the cerebellar cortex produced the labelling of beaded fibres of Purkinje cells in the cerebellar nucleus. Unlike in mammals, GABAergic innervation of the cerebellar nucleus was scare: Purkinje cell axon terminals in the cerebellar nucleus did not appear to be GABA-immunoreactive, most GABAergic fibres being found in the subventricular neuropile. Some fibres immunoreactive to serotonin and somatostatin were also observed in the subventricular neuropile of the cerebellar nucleus. Three neuron types were distinguished with electron microscopy (types A to C). Type A cells were abundant and smooth-surfaced, and appeared to correspond to Golgi-impregnated neurons and large glutamate-immunoreactive cells. Type B neurons were scarce and possessed dendrites covered by sessile or stalked spines. Type C neurons were small cells located mainly in the medialmost region of the nucleus and corresponded to subventricular glutamate- and GABA-immunoreactive cells. Six types of synaptic bouton were observed (types I to VI). The most abundant (type I boutons) made symmetrical contacts and appeared to correspond to Purkinje cell axons. Type I boutons were the only type observed on perikarya and initial axon segments of type A cells. Type IV and type V boutons made complex glomerular-like asymmetrical contacts with spines of type B cells. Type VI boutons appeared to correspond to peptidergic and/or monoaminergic axons. The functional significance of these results is discussed.

Neuroendocrine regulation of growth hormone secretion in sheep. VII. Effects of GABA

The effects of intravenous (i.v.) or intracerebroventricular (i.c.v.) administration of gamma aminobutyric acid (GABA) on plasma growth hormone (GH) concentrations have been examined in sheep. Intravenous administration of GABA resulted in a rapid, significant (P < 0.001) increase in plasma GH. Administration of 10 mg of GABA i.c.v. produced a significant (P < 0.001) increase in GH release. By contrast, 100 mg given i.c.v. was inhibitory and resulted in a decrease (P < 0.05) in plasma GH levels. Concurrent administration of somatostatin (0.5 microgram/min i.v. over 1 h) did not alter the plasma GH response to 10 mg GABA given i.c.v. These data are consistent with the concept of dual sites of action for GABA in regulating GH release in sheep, but the exact mechanism through which this effect is mediated remains unclear.

Expression of GABAA receptor alpha 2 sub-unit mRNA by periventricular somatostatin neurones in the rat hypothalamus

Pharmacological evidence suggests that GABA may play an important role in regulating the secretory and synthetic activity of the hypothalamic periventricular somatostatin (SOM) neurones controlling growth hormone secretion. In this study, we have utilized a dual labelling in situ hybridization technique to examine whether the alpha 2 sub-unit of the GABAA receptor, which is abundant in this region, is expressed by periventricular SOM neurones. Neurones expressing SOM were detected using an alkaline phosphatase-labelled antisense oligonucleotide and the alpha 2 sub-unit with an 35S-labelled antisense oligonucleotide. Hybridization experiments with the alpha 2 sub-unit probe alone confirmed the high level of mRNA expression for this sub-unit in the rat periventricular region and simultaneous hybridization experiments with both probes revealed that > 90% (93 +/- 2%) of periventricular SOM neurones express the alpha 2 sub-unit of the GABAA receptor. These results provide the first direct evidence that periventricular SOM cells possess GABAA receptors and suggest that the great majority of these neurones synthesize a GABAA receptor containing the alpha 2 sub-unit.

Local origins of some GABAergic projections to the paraventricular and supraoptic nuclei of the hypothalamus in the rat

Axonal transport and immunohistochemical methods were used to characterize the organization of glutamic acid decarboxylase-immunoreactive (GAD-ir) projections to the paraventricular (PVH) and supraoptic (SO) nuclei of the hypothalamus in the rat. In line with prior reports, GAD-ir varicosities were found to be densely and quite uniformly distributed throughout the hypothalamus, including the PVH and the SO. Nonetheless, the periventricular part of the PVH was consistently found to contain a disproportionately high density of GAD-ir elements. Small crystalline implants of the retrograde tracer, true blue, into the PVH labeled GAD-ir cells in the anterior perifornical region, portions of the anterior hypothalamic area immediately ventral to the PVH, a region just dorsal to the rostral SO and extending caudomedially over the optic chiasm and tract, and within the anterior one-third of the PVH itself. Because possible uptake of retrograde tracer by local dendritic processes might have yielded false positive filling of nearby GAD-ir cells, anterograde transport, Phaseolus vulgaris-leucoagglutinin, and combined anterograde transport-immunohistochemical methods were used to attempt to confirm these four putative local sources of GAD-ir inputs. Tracer injections in each of the above mentioned regions labeled sparse to moderate axonal projections to the PVH, which ramified preferentially in the parvicellular division of the nucleus. Projections to the magnocellular division of the PVH and the SO were generally sparse and inconsistently observed in this material. A variable, and generally small, proportion of anterogradely labeled axons and terminals in the PVH also displayed GAD-ir. These results suggest that GABAergic projections to visceromotor cell types in the PVH and SO arise, at least in part, from several diffusely distributed local sources. The fact that these afferents were found to terminate preferentially in the parvicellular division of the PVH makes it likely that additional sources of GABAergic projections to the magnocellular neurosecretory system remain to be identified. Peri- and intranuclear GABAergic neurons could provide an intermediary by which documented (and generally inhibitory) limbic system influences on neuroendocrine function are exerted.

Regulation of somatostatin synthesis by GABAA receptor stimulation in mouse brain

Neuroanatomical data have documented the existence of synaptic contacts between gamma-aminobutyric acid (GABA) terminals and periventricular hypothalamic somatostatin (SRIF) neurons. In other brain regions, like the cortex or hippocampus, GABA and SRIF are colocalized in short interneurons. These observations suggest that GABA modulates SRIF neuronal activity. In order to test this hypothesis, we studied the effects of the in vivo stimulation of the GABAA receptor (muscimol, 0.75 mg/kg + diazepam, 2.5 mg/kg) on SRIF content and preproSRIF mRNA levels, in mouse brain. Chronic (7 days), but not acute, treatment induced a 38% decrease in hypothalamic SRIF content (as estimated by RIA), a 20% decrease in cortex and no effect in the striatum. The decrease in hypothalamic and cortical SRIF levels lasted until 24 h after cessation of the treatment. In the hypothalamus, prosomatostatin mRNA levels were estimated by Northern blot analysis using a 32P-labeled 45-mere oligoprobe. ProSR1F mRNA hypothalamic levels were equally (48%) decreased by the acute and chronic treatments and remained lower than controls 48 h after the last injection. Quantitative in situ hybridization was used to examine the regional distribution of GABA-induced acute inhibition of proSR1F mRNA densities, using the same oligomere labeled with 35S. ProSR1F mRNA levels were decreased by 35% in the periventricular hypothalamic nucleus. In contrast, no significant modification was observed in cortex, striatum and hilus of the dentate gyrus of the dorsal hippocampus. The present data demonstrate a regionally selective inhibitory action of GABA, mediated by GABAA receptors stimulation, on the biosynthetic mechanisms of the long projecting neuroendocrine SRIF neurons of the anterior periventricular nucleus of the hypothalamus.

Dual immunocytochemistry using 125I-labeled protein A: a new electron microscopic technique applied to the investigation of chemical connectivity and axonal transmitter co-localization in the brain

We have developed a double labeling immunocytochemical method utilizing peroxidase conjugated Fab fragments and 125I-labeled protein A to localize two neuronal markers on the same light or electron microscopic section with primary antibodies raised in the same animal species. The technique is applicable to the study of chemical connectivity in the brain, as illustrated by data obtained in the hypothalamus using rabbit polyclonal antisera against tyrosine hydroxylase (TH), phenylethanolamine-N-methyltransferase (PNMT), neuropeptide Y (NPY), and vasoactive intestinal peptide (VIP). Moreover, due to a high level of sensitivity and resolution, the technique offers considerable advantages over many previously developed dual labeling immunocytochemical methods for the demonstration of transmitter axonal co-localizations. Utilizing the peroxidase Fab/[125I]protein A method, we present here the first direct evidence that PNMT is present in many endings also containing NPY in the thalamic and hypothalamic paraventricular nuclei and in the arcuate nucleus. The method also may be combined as required with other labeling methods for localizing more than two neurochemical markers on one and the same electron microscopic section.

Electrophysiological properties of neurones in the region of the paraventricular nucleus in slices of rat hypothalamus

1. Neurones in the region of the hypothalamic paraventricular nucleus (PVN) of the rat were studied with intracellular recording in the coronal slice preparation. Three types of hypothalamic neurones were distinguished according to their membrane properties and anatomical positions. Lucifer Yellow or ethidium bromide was injected intracellularly to determine the morphology of some recorded cells. 2. The most distinctive electrophysiological characteristic was the low-threshold depolarizing potentials which were totally absent in type I neurones, present but variable in type II neurones and very conspicuous in type III neurones. Type II neurones generally showed relatively small low-threshold depolarizations (26.5 +/- 2.2 mV) which generated at most one to two action potentials. Type III neurones, on the other hand, generated large low-threshold potentials (40.3 +/- 2.8 mV) which gave rise to bursts of three to six fast action potentials. Deinactivation of the low-threshold conductance in both type II and type III neurones was voltage- and time-dependent. Low-threshold potentials persisted in TTX (1-3 microM) but were blocked by solutions containing low Ca2+ (0.2 mM) and Cd2+ (0.5 mM), suggesting they were Ca(2+)-dependent. 3. Type I neurones had a significantly shorter membrane time constant (14.5 +/- 1.7 ms) than those of type II (21.6 +/- 1.7 ms) and type III neurones (33.8 +/- 4.4 ms). Input resistance and resting membrane potential did not differ significantly among the cell groups. 4. Current-voltage (I-V) relations were significantly different among the three cell types. Type I neurones had linear I-V relations to -120 mV, while type III neurones all showed marked anomalous rectification. I-V relations among type II neurones were more heterogeneous, although most (75%) had linear I-V curves to about -90 to -100 mV, inward rectification appearing at more negative potentials. 5. Type I neurones generated fast action potentials of relatively large amplitude (64.2 +/- 1.1 mV, threshold to peak) and long duration (1.1 +/- 0.1 ms, measured at half-amplitude); the longer duration was due to a shoulder on the falling phase of the spike. Type II neurones had large spikes (66.5 +/- 1.6 mV) of shorter duration (0.9 +/- 0.1 ms) with no shoulder. Type III neurones had relatively small spikes (56.1 +/- 2.2 mV) of short duration (0.8 +/- 0.1 ms) with no shoulder. 6. The three cell populations showed different patterns of repetitive firing in response to depolarizing current pulses. Type I neurones often generated spike trains with a delayed onset and little spike-frequency adaptation.(ABSTRACT TRUNCATED AT 400 WORDS)

Immunocytochemical and autoradiographic studies of the endocrine cells interacting with GABA in the rat stomach

There are now increasing evidences suggesting that GABA is able of direct interaction with certain endocrine cells. In the present study, highly specific anti-GABA-glutaraldehyde antibodies and 3H-GABA uptake were used at the light and electron microscope levels to investigate the occurrence of cells containing endogenous GABA or taking up exogenous GABA in the mucosal antrum and corpus of the rat stomach. Only certain endocrine cell types of both regions were immunostained or grain-labelled. However, the morphology of their secretory granules did not allow to identify the nature of their hormone with certainty but suggested that somatostatin-like cells could interact with GABA. The combination of gastrin and somatostatin immunodetection with 3H-GABA uptake autoradiography at the light microscope level, revealed that a subpopulation of somatostatin-like cells and other still unidentified endocrine cells are able to take up GABA, while the gastrin-like cells are not. These results reinforce the hypothesis that certain endocrine cell types of the diffuse endocrine system of the digestive tract are able to directly interact with GABA.

Peripheral administration of picrotoxin and bicuculline stimulates in vivo somatostatin release from rat median eminence

The gamma-Aminobutyric acid-A (GABAA) antagonist picrotoxin and bicuculline were administered to male rats to determine their effects on somatostatin (SRIF) release, measured in unanesthetized animals stereotaxically implanted with push-pull cannula in the median eminence (ME). I.p. injection (3 mg/kg) of picrotoxin (n = 5) or bicuculline (n = 6) significantly increased (35.4 +/- 10.8 vs 13.7 +/- 4.3 pg/15 min; P less than 0.03 and 38 +/- 3.5 vs 14 +/- 1.8 pg/15 min; P less than 0.001, respectively) SRIF release from the ME compared to baseline levels measured in the same animals. In contrast, with local perfusion of picrotoxin, (10(-4) to 10(-6) M) SRIF release from the ME was not affected. These data suggest a physiological endogenous inhibitory tone of GABA on SRIF release.

Neurons containing messenger RNA encoding glutamate decarboxylase in rat hypothalamus demonstrated by in situ hybridization, with special emphasis on cell groups in medial preoptic area, anterior hypothalamic area and dorsomedial hypothalamic nucleus

Previous deafferentation studies have suggested that most hypothalamic GABAergic innervation originates from neurons within the hypothalamus. We have investigated the distribution of GABAergic cell groups in the rat hypothalamus by means of the in situ hybridization technique, using a cDNA probe for messenger RNA encoding glutamate decarboxylase. Several major GABAergic cell groups were demonstrated, including cells of the tuberomammillary nucleus, arcuate nucleus, suprachiasmatic nucleus, medial preoptic area, anterior hypothalamic area, the dorsomedial hypothalamic nucleus, perifornical area, and lateral hypothalamic area. The most prominent glutamate decarboxylase mRNA-containing cell groups were located in the medial preoptic area, anterior hypothalamic area and dorsomedial hypothalamic nucleus, and were composed of small- to medium-sized neurons. Compared to previously well-characterized GABAergic cell groups in the tuberomammillary nucleus, reticular thalamic nucleus, and non-pyramidal cells of cerebral cortex, the cells of these GABAergic groups demonstrated only weak cDNA labelling, indicating that they contain lower levels of glutamate decarboxylase mRNA. Several types of control experiments supported the specificity of this cDNA labelling, and the GABAergic nature of these cell populations was further supported by detection of glutamate decarboxylase and GABA immunoreactivity. Abundance of GABAergic cells in many hypothalamic nuclei indicates that GABA represents quantitatively the most important transmitter of hypothalamic neurons, and may be involved in neuroendocrine and autonomic regulatory functions.

Ultrastructural localization of somatostatin-like immunoreactivity in the rat dentate gyrus

Neurons containing somatostatin (SOM) are enriched in the dentate gyrus. We sought to establish the ultrastructural localization of this peptide in the dentate gyrus of the rat brain with a double-bridged peroxidase-antiperoxidase (PAP) method localizing antisera directed against somatostatin (SOM)-28 and SOM-28. Initial light microscopic observations confirmed that the majority of perikarya and thick varicose processes with intense SOM-like immunoreactivity (SOM-LI) were observed in the hilus. Fine varicose processes with SOM-LI were found throughout all layers of the dentate gyrus but were most intense in the outer third of the molecular layer (ML), where an occasional perikaryon with SOM-LI was seen. By electron microscopy, SOM-LI was found in neuronal perikarya, dendrites, axons, and axon terminals. Two types of SOM-containing perikarya were observed. The first type was small (6-10 microns), round or avoid, and had a labeled cytoplasma with abundant Golgi complexes and a dense accumulation of PAP-reaction product. The second type of perikarya was larger (11-16 microns) and had a more abundant cytoplasm than the first type, but the Golgi complexes did not appear labeled. Most (96% of 374) of the synapses on the SOM-labeled perikarya and dendrites were from terminals without SOM-LI which formed nearly equal proportions of asymmetric and symmetric junctions. The remainder of the presynaptic terminals contained SOM-LI and made primarily symmetric synapses. Synaptic junctions from both unlabeled and labeled terminals were primarily on the shafts of the small (0.5-1.5 microns) SOM-immunoreactive dendrites. The terminals with SOM-LI (0.25-1.3 microns) contained many small, clear vesicles and from zero to four large dense-core vesicles. Terminals with SOM-LI were associated 1) with one unlabeled perikaryon or dendrite (49% of 215 in the hilus; 76% of 326 in the ML); 2) with two unlabeled perikarya or dendrites simultaneously (5% hilus; 4% ML); and 3) with one SOM-containing perikaryon or dendrite (6% hilus; 3% ML). In all three types of associations, synaptic contacts on perikarya were few while the majority were with small (distal) dendrites. Moreover, most of the terminals with SOM-LI formed symmetric junctions or lacked membrane specializations but were without any apparent glial intervention in the plane of section analyzed. The remaining SOM-labeled terminals (40% hilus; 17% ML) were without any apparent synaptic relations. However, a few of these terminals were in direct apposition to other terminals, some of which were also SOM-immunoreactive.(ABSTRACT TRUNCATED AT 400 WORDS)