Glutamate has been found to play an unexpectedly important role in neuroendocrine regulation in the hypothalamus, as revealed in converging experiments with ultrastructural immunocytochemistry, optical physiology with a calcium-sensitive dye, and intracellular electrical recording. There were large amounts of glutamate in boutons making synaptic contact with neuroendocrine neurons in the arcuate, paraventricular, and supraoptic nuclei. Almost all medial hypothalamic neurons responded to glutamate and to the glutamate agonists quisqualate and kainate with a consistent increase in intracellular calcium. In all magnocellular and parvocellular neurons of the paraventricular and arcuate nuclei tested, the non-NMDA (non-N-methyl-D-aspartate) glutamate antagonist CNQX (cyano-2,3-dihydroxy-7-nitroquinoxaline) reduced electrically stimulated and spontaneous excitatory postsynaptic potentials, suggesting that the endogenous neurotransmitter is an excitatory amino acid acting primarily on non-NMDA receptors. These results indicate that glutamate plays a major, widespread role in the control of neuroendocrine neurons.

GABA is the primary inhibitory transmitter of the adult hypothalamus, synthesized by many neurons and found in 50% of the presynaptic boutons. GABA causes a decrease in Ca2+ in mature hypothalamic neurons in vitro by depressing cellular activity through opening Cl- channels. Despite the early expression of GABAA receptors in the embryonic hypothalamus (E15), the cellular function of GABA in the developing hypothalamus has received little attention. In the present study the role of GABA in modulating intracellular Ca2+ in developing hypothalamic neurons was studied with fura-2 digital imaging. GABA (0.5- 500 microM) applied to embryonic hypothalamic neurons elicited a dramatic and rapid increase in intracellular Ca2+ This Ca2+ rise could be completely blocked by the GABAA antagonist bicuculline (20 microM) and persisted in the presence of tetrodotoxin (1 microM). The Ca2+ elevation induced by GABA was greater than that of equimolar concentrations of the excitatory transmitter glutamate in early development. The number of E15 neurons that responded to GABA with a Ca2+ rise increased during the first few days of culture, reaching 78% after 4 d in vitro. The Ca2+ rise was 87% blocked by cadmium (100 microM) and 85% blocked by nimodipine (1 microM), indicating that the mechanism of Ca2+ increase was primarily via L-type voltage operated Ca2+ channels. Addition of bicuculline to synaptically coupled cultures caused a significant decrease in Ca2+ 4–10 d after culturing, indicating hypothalamic neurons were secreting GABA at an early age of development, and that sufficient GABA was released to elicit an increase in Ca2+. This effect was seen even after blocking all glutamatergic activity with glutamate receptor antagonists. In contrast, GABA elicited no Ca2+ rise in older neurons (> 18 d in vitro), and the action of bicuculline reversed and caused a large increase in Ca2+ in spontaneously active neurons. Similar findings were obtained in cultures enriched in GABAergic neurons from the suprachiasmatic nucleus. To determine if the Ca2+ stimulating role of GABA on developing neurons was restricted to the hypothalamus and a few other regions, or whether it might exist throughout the brain, we examined the Ca2+ responses in cultured olfactory bulb, cortex, medulla, striatum, thalamus, hippocampus, and colliculus. The majority (75%) of developing neurons from each region showed a Ca2+ rise in response to GABA. Together these data suggest that GABA elevates Ca2+ in developing, but not mature, neurons from the hypothalamus and all other brain regions examined.

To study the distribution of glycine immunoreactive neurons in the spinal cord and brain, antisera were raised against glycine conjugated to protein carriers. High-titer rabbit glycine antiserum was purified by affinity chromatography. Testing against other amino acids and peptides with immuno dot blots and ELISA assays showed little apparent cross-reaction with glutamate, aspartate, glutamine, taurine, and 17 other amino acids and related compounds. Similarly, the antiserum showed little apparent recognition of glycine when glycine was incorporated into peptides. A slight cross-reactivity with GABA, beta- alanine, and cysteine was found. Immunocytochemical labeling of tissue sections could be blocked with glycine conjugated to a heterologous carrier protein but not by other amino acids conjugated to that protein. Immunocytochemistry at the light microscope level with immunofluorescence and silver-intensified colloidal gold revealed a wide distribution of glycine-like immunoreactivity throughout all laminae of the rat spinal cord and in all segments studied from the cervical, thoracic, lumbar, and sacral cord. Immunoreactive boutons were found terminating on both cell bodies and on dendrites. Ultrastructural analysis with postembedding colloidal gold immunocytochemistry demonstrated large numbers of immunoreactive boutons making symmetrical type synapses with neuronal perikarya, including motor neurons, and with proximal and distal dendrites. Presynaptic glycine immunoreactive boutons were found in both ventral and dorsal horn. Immunoreactivity was concentrated over regions rich in vesicles, and over mitochondria in immunoreactive boutons, but not over mitochondria in postsynaptic dendrites. Glycine-immunoreactive perikarya were identified both in the dorsal horn and in the ventral horn. Myelinated and unmyelinated glycine-immunoreactive axons were noted both in the gray and white matter of the cord. The density of immunoreactive axons varied in the white matter, with the greatest number of immunoreactive axons found in the white matter adjacent to the gray matter in lateral and ventral white. Significantly fewer immunoreactive axons were found in the white matter of the dorsal columns. Myelin sheaths around axons were unlabeled. The distribution of glycine-immunoreactive boutons correlated well with the distribution of glycine receptor immunoreactivity on postsynaptic elements of the spinal cord, tested with different monoclonal antisera against strychnine-purified glycine receptor. Glycine receptor immunoreactivity was found throughout the gray matter of both rat and primate.

Converging lines of evidence suggest that the hypothalamic suprachiasmatic nucleus (SCN) is the site of the endogenous biological clock controlling mammalian circadian rhythms. To study the calcium responses of the cellular components that make up the clock, computer-controlled digital video and confocal scanning laser microscopy were used with the Ca*+ indicator dye fluo-3 to examine dispersed SCN cells and SCN explants with repeated sampling over time. Ca2+ plays an important second messenger role in a wide variety of cellular mechanisms from gene regulation to electrical activity and neurotransmitter release, and may play a role in clock function and entrainment. SCN neurons and astrocytes showed an intracellular Ca2+ increase in response to glutamate and 5-HT, two major neurotransmitters in afferents to the SCN. Astrocytes showed a marked heterogeneity in their response to the serial perfusion of different transmitters; some responded to both 5-HT and glutamate, some to neither, and others to only one or the other. Under constant conditions, most neurons showed irregular temporal patterns of Ca2+ transients. Expression of regular neuronal oscillations could be blocked by the inhibitory transmitter GABA. Astrocytes, on the other hand, showed very regular rhythms of cytoplasmic Ca2+ concentrations with periods ranging from 7 to 20 sec. This periodic oscillation could be initiated by in vitro application of glutamate, the putative neurotransmitter conveying visual input to the SCN critical for clock entrainment. Long-distance communication between glial cells, seen as waves of fluorescence moving from cell to cell, probably through gap junctions, was induced by glutamate, 5-HT, and ATP. These waves increased the period length of cellular Ca2+ rises to 45-70 sec. Spontaneously oscillating cells were common in culture medium, serum, or rat cerebrospinal fluid, but rare in HEPES buffer. One source for cytoplasmic Ca2+ increases was an influx of extracellular Ca2+, as seen under depolarizing conditions in about 75% of the astroglia studied. All neurotransmitterinduced Ca2+ fluxes were not dependent on voltage changes, as Ca*+ oscillations could be initiated under both normal and

We thank Ms. A. Schneider for technical assistance, and Drs. F. Gage, E. Johnson and E. Shooter for nerve growth factor receptor antibody and suggestions related to it, and Drs. U. di Porzio, R. Vogt, and M. Bennett for useful discussion.

Within the hypothalamus, a large number of neuroactive substances are found, many first detected in this part of the brain. Excitatory amino acids, recognized as important transmitters in other parts of the brain, have received little attention here. To study glutamate immunoreactivity at the ultrastructural level in the hypothalamus, postembedding colloidal gold or silver-intensified gold was used. Antisera raised against glutamate conjugated with glutaraldehyde to keyhole limpet hemocyanin were specific for glutamate, tested with a battery of tests including immunodot blot, ELISA assays. Western blot, and Sepharose epoxy-conjugated amino acids. Antisera did not cross- react with other amino acids and related compounds, with proteins containing glutamate, or with polyglutamate. A population of presynaptic boutons in the suprachiasmatic, arcuate, ventromedial, supraoptic, and parvocellular and magnocellular paraventricular nuclei showed strong immunoreactivity for glutamate. Highly labeled presynaptic axons generally made asymmetrical Gray type 1 synaptic contacts with dendrites or cell bodies and had up to eight times more immunogold particles per unit area than postsynaptic dendrites. Axon terminals exhibiting strong glutamate immunoreactivity had large numbers of round, clear vesicles adjacent to the synaptic specialization together with a few larger, dense-core vesicles. The largest number of gold particles over axons were located in regions containing the small clear vesicles. Axons in general had about three times more gold particles over them than did the postsynaptic dendrites. Staining of single boutons in adjacent serial ultrathin sections with glutamate or GABA antisera showed that non-GABAergic terminals had a higher level of glutamate staining than did axons immunoreactive for GABA. In control experiments, immunostaining with glutamate antiserum could be blocked by solid-phase absorption of the antiserum with glutamate conjugated with glutaraldehyde to proteins. Aspartate was also detected with immunocytochemistry in some presynaptic boutons in the medial hypothalamus. To compare the response of neurons to aspartate and glutamate, calcium-imaging dyes were used in combination with digital video microscopy. Whereas almost all neurons showed a rise in intracellular Ca2+ in response to glutamate, many but not all of the same cells also showed a Ca2+ rise of smaller magnitude in response to aspartate. These ultrastructural immunocytochemical data, taken in conjunction with biochemical and electrophysiological experiments, suggest that glutamate, and to a lesser extent aspartate, may play an important neurotransmitter role in a wide variety of hypothalamic circuits.

To study the morphological substrate for interaction between two chemically distinct neuronal types, two double ultrastructural immunolabeling strategies were employed. In the first, two different electron-dense markers were used to examine simultaneously two different neurotransmitter-related antigens in the hypothalamic supraoptic nucleus in the same thin section. Results obtained with the first method were confirmed with a second approach based on postembedding immunostaining of alternate serial thin sections with different antisera. Antiserum against glutamate decarboxylase, the enzyme responsible for the synthesis of the inhibitory amino acid transmitter gamma-aminobutyric acid (GABA), or antisera against GABA, was used to localize immunoreactive axons in the hypothalamic supraoptic nucleus. With light microscopy, glutamate decarboxylase- and GABA-immunoreactive axon terminals immunostained with peroxidase were found arborizing throughout all areas of the nucleus; terminal boutons were found adjacent to unlabeled somata within the nucleus. Cells containing immunoreactive oxytocin, vasopressin, and neurophysin were localized with peroxidase. Glutamate decarboxylase-immunoreactive axons stained with peroxidase prior to embedding in plastic were demonstrated to contact neurons which contained vesicles immunostained with neurophysin antiserum by a post-embedding immunocytochemical procedure which used immunoglobulins or protein A adsorbed to colloidal gold as a second ultrastructural marker. Quantitative evaluation of post- embedding staining with colloidal gold using a neurophysin primary antiserum indicated a specific antigen localization in neurosecretory vesicles. A critical factor in this double-labeling paradigm was that immunological reagents used in the second series did not cross-react with those used in the first series, regardless of the species of origin of antisera. To provide further verification of GABAergic synapses on neurophysin-containing neurons, alternate serial ultrathin sections were stained with colloidal gold using antisera against either neurophysin or GABA; boutons immunoreactive for GABA made synaptic contact with supraoptic neurons containing neurophysin immunoreactivity. Converging results obtained with these two procedures indicate that GABAergic axons synapse directly on neurons containing oxytocin or vasopressin in the rat hypothalamic supraoptic nucleus.(ABSTRACT TRUNCATED AT 400 WORDS)

The hypothalamus is the crucial part of the brain that regulates homeostasis throughout the body. It governs the endocrine and autonomic nervous systems, temperature, heart rate, emotional and motivational states, reproduction, energy and water balances, and circadian rhythms. In contrast to the prevailing belief that hypothalamic neurons use peptides, neuromodulators, or other slow-acting agents as their principal neuroactive substances, we present data indicating that the primary excitatory transmitter released by medial hypothalamic neurons is glutamate. This surprising new evidence is based on three converging approaches: Immunogold cytochemistry revealed that some hypothalamic neurons and their processes in vitro contained high amounts of immunoreactive glutamate. Ca2+ digital video imaging showed that cytoplasmic Ca2+ levels of cultured neurons, elevated because of spontaneous presynaptic release of a hypothalamic transmitter, were reduced by perfusion with the selective glutamate receptor antagonists cyano-2,3-dihydroxy-7-nitroquinoxaline and 2-aminophosphonovaleric acid. Electrophysiological analysis of whole-cell patch-clamp recordings from single and pairs of monosynaptically coupled hypothalamic neurons in culture showed that virtually all spontaneous and evoked EPSPs appear to be mediated by synaptic secretion of glutamate.