K. Kosaka

Chemically defined neuron groups and their subpopulations in the glomerular layer of the rat main olfactory bulb: prominent differences in the intraglomerular dendritic arborization and their relationship to olfactory nerve terminals

In the glomerular layer of the rat main olfactory bulb, we previously reported three chemically defined interneuron groups: GABA-like immunoreactive, calretinin-immunoreactive and Calbindin-D28k-immunoreactive groups [Kosaka K. et al. (1995) Neurosci. Res. 23, 73-88]. In the present study, we analysed the structural features of these three neuron groups using confocal laser scanning light microscopy, focusing on their dendritic arborization pattern, especially on their close apposition to olfactory receptor terminals labeled by olfactory marker protein. Each glomerulus consisted of two zones, the olfactory nerve zone and the non-olfactory nerve zone. The former was mainly occupied by olfactory nerve preterminals and terminals as well as their targets, postsynaptic fine dendritic portions of intrinsic neurons. The latter non-olfactory nerve zone was occupied mainly by olfactory marker protein-negative profiles. Processes of GABAergic neurons and those of one of their subpopulations, tyrosine hydroxylase-immunoreactive neurons, were numerous both in the olfactory nerve and non-olfactory nerve zones, resulting in their frequent close apposition to olfactory marker protein-immunoreactive elements. Combined confocal laser scanning light microscopic electron microscopic examination revealed synaptic contacts from olfactory nerve terminals on tyrosine hydroxylase-immunoreactive processes at these sites of close apposition. In contrast, calretinin-immunoreactive and Calbindin-D28k-immunoreactive processes, particularly Calbindin-D28k-immunoreactive ones, were distributed almost exclusively in the non-olfactory nerve zone, as if they avoided the olfactory nerve zone, showing a net or honeycomb pattern. Thus, calretinin-immunoreactive and Calbindin-D28k-immunoreactive processes were not or very rarely closely apposed to olfactory nerve terminals. These findings suggested that there might be some differences among chemically defined interneuronal groups in their synaptic contacts from olfactory nerves. Further quantitative image analysis clearly exhibited the prominent differences among these neuron groups in their intraglomerular dendritic arborization in relation with the olfactory nerve zone, i.e. the percentages of the area in the olfactory nerve zone occupied by GABAergic and tyrosine hydroxylase-immunoreactive processes were about 10%, respectively, whereas those of calretinin-immunoreactive and Calbindin-D28k-immunoreactive processes were only about 1% and 0.3%, respectively. These findings suggested that so-called periglomerular cells in glomeruli might be heterogeneous not only in their chemical nature, but also in their dendritic arborization pattern and synaptic contacts from olfactory nerve terminals.

Chemically defined neuron groups and their subpopulations in the glomerular layer of the rat main olfactory bulb—IV. Intraglomerular synapses of tyrosine hydroxylase-immunoreactive neurons

Synapses of intraglomerular processes of tyrosine hydroxylase-immunoreactive neurons in the rat main olfactory bulb were examined by electron microscopic immunocytochemistry. Prominent characteristics of intraglomerular synapses of tyrosine hydroxylase-immunoreactive elements were that the vast majority (about 80%) of their synaptic inputs were asymmetrical synapses from olfactory nerve terminals and, though far smaller in proportion, one half of the remaining were asymmetrical synapses from mitral/tufted cell dendrites and the other half were symmetrical synapses from gamma-aminobutyric acid-like immunoreactive elements. So far, we have observed no typical reciprocal synapses between tyrosine hydroxylase-immunoreactive processes and mitral/tufted dendrites; however, we have often identified serial synapses; that is, asymmetrical synapses from olfactory nerve terminals or mitral/tufted cell dendrites to tyrosine hydroxylase-immunoreactive processes, and then symmetrical synapses from the latter to different mitral/tufted cell dendrites. These synaptic connections of tyrosine hydroxylase-immunoreactive neurons were very different from those of Calbindin-D(28k)-immunoreactive neurons, which received no synaptic contact directly from olfactory nerve terminals but formed reciprocal synapses with mitral/tufted cells as we analysed previously.Thus, our present and previous electron microscopic studies combined with confocal laser scanning light microscopy clearly indicated for the first time the heterogeneity of periglomerular neurons, not only in their chemical and morphological features, but also in their synaptic organization in the olfactory glomerulus.

Quantitative analysis of GABA-like-immunoreactive and parvalbumin-containing neurons in the CA1 region of the rat hippocampus using a stereological method, the disector

The numerical density of neurons in the CA1 region of the rat dorsal hippocampus has been estimated by a stereological method, the disector, using pairs of video images of toluidine blue-stained, plastic-embedded, 0.5-μm-thick sections, 3 μm distant from each other. The chemical properties of those disector-counted cells were further analyzed by postembedding immunocytochemical methods on adjacent, semithin sections using antibodies against gamma-aminobutyric acid (GABA) and a specific calcium-binding protein, parvalbumin (PV). The density of neurons in the CA1 region was 35.2 × 103/mm3; numerical densities in the stratum oriens (SO), stratum pyramidale (SP), and strata radiatum-lacunosum-moleculare (SRLM) were 11.3 × 103/mm3, 272.4 × 103/mm3, and 1.9 × 103/mm3, respectively. The numerical densities of GABA-like immunoreactive (GABA-LIR) and PV-immunoreactive (PV-IR) neurons were 2.1 × 103/mm3 and 1.1 × 103/mm3, respectively, which were 5.8% and 3.2% of all neurons, respectively. In the CA1 region only about 60% of PV-positive neurons were GABA-LIR. However, taking the previous observation into consideration that almost all hippocampal PV-positive neurons were immunoreactive for the GABA-synthesizing enzyme glutamic acid decarboxylase (GAD), neurons that were immunoreactive to either GABA or PV or both (GABA+ and/or PV + neurons) were regarded as a better representative of GABAergic neurons in this region; thus, the numerical density of these GABA + and/or PV + neurons was 2.5 × 103/mm3 and they were 7.0% of all neurons in the CA1 region. Lamellar analysis showed that the numerical densities of GABA+ and/or PV+, GABA-LIR, and PV-IR neurons were highest in the SP, where they were 8.2 × 103/mm3, 6.2 × 103/mm3, and 5.4 × 103/mm3, respectively. The results of the present study indicate that the proportions of GABAergic neurons and a subpopulation of them, PV-containing GABAergic neurons, to other presumable non-GABAergic neurons are far smaller in the CA1 region of the hippocampus than in several neocortical regions previously reported.

Electron microscopic serial-sectioning/reconstruction study of parvalbumin-containing neurons in the external plexiform layer of the rat olfactory bulb

Neurons containing a calcium-binding protein parvalbumin in the external plexiform layer of the rat olfactory bulb were identified light microscopically with the pre-embedding immunocytochemistry and were subsequently analysed with the electron microscopic serial-sectioning and three-dimensional reconstructions. In the present study we chose several different types of parvalbumin-immunoreactive neurons identified light microscopically as Van Gehuchten cell type, superficial short-axon cell type and multipolar cell type. Parvalbumin-immunoreactive somata were similar to one another in their ultrastructural characteristics, showing nuclear indentations, moderately developed Golgi apparatus and abundant mitochondria; these structural features appeared to resemble those of the short axon cells around the glomeruli and in the granule cell layer reported in previous electron microscopic studies. All neurons analysed in the present study made symmetrical synapses on to dendrites and somata of presumed mitral/tufted cells and received asymmetrical synapses from them, and occasionally formed reciprocal synapses with them. On the parvalbumin-immunoreactive processes, the asymmetrical synapses nearly equalled the symmetrical ones in number and about 30-50% of them were identified as reciprocal pairs. In contrast, no presynaptic sites were observed on parvalbumin-immunoreactive somata, and thick portions (more than approximately 2 microns in diameter) of the proximal dendrites, where they were occasionally postsynaptic in some asymmetrical and symmetrical synapses from parvalbumin-immunonegative profiles. Characteristically, parvalbumin-immunoreactive process frequently make direct contacts with one another; processes regarded light microscopically as arising from a soma or a dendrite or parvalbumin-immunoreactive neurons were sometimes revealed to be separate but directly contacting processes with electron microscopic examinations. Although puncta adherentia were occasionally observed between these contact sites, so far neither gap junctions nor chemical synapses were observed. Until now, it has been believed that in the external plexiform layer only granule cells form reciprocal synapses with mitral/tufted cells. However, the present study clearly demonstrates that interneurons different from granule cells, namely GABAergic neurons containing a calcium-binding protein parvalbumin, also make reciprocal synapses with mitral/tufted cells in the external plexiform layer. Therefore, neuronal processes making reciprocal synapses with mitral/tufted cells in the external plexiform layer cannot be determined a priori as granule cell processes.

Axons and axon terminals of cerebellar Purkinje cells and basket cells have higher levels of parvalbumin immunoreactivity than somata and dendrites: quantitative analysis by immunogold labeling

The immunointensities of calcium-binding proteins parvalbumin (PV) and calbindin D28K were quantified in different parts of Purkinje cells and interneurons (basket cells and stellate cells) of the rat cerebellum. An electron microscopic, postembedding immunogold procedure on Lowicryl K4M-embedded thin sections was applied. Neuronal profiles were identified by double-labeling immunocytochemistry using the combination of the two primary antibodies, mouse monoclonal anti-rat calbindin D28K and rabbit polyclonal anti-rat PV. The secondary antibodies were conjugated with colloidal gold of different sizes (10 and 15 nm diameter). In the cerebellar cortex, double-labeled profiles were identified as Purkinje cells and profiles labeled only with anti-PV were identified as inteneurons. The densities of gold particles were used for statistical comparison of the relative levels of PV and calbindin D28K in somata, dendrites, dendritic spines, axons and axon terminals of Purkinje cells, and interneurons. The axons and axon terminals of Purkinje cells and basket cells had significantly higher levels of PV immunoreactivity than Purkinje cell somata, primary, secondary, and tertiary dendrites, and dendritic spines, as well as interneuron somata. On the other hand, the present study could not determine conclusively whether calbindin D28K was distributed homogeneously throughout soma, dendrites, and axons of Purkinje cells or was also concentrated in Purkinje cell axons. To estimate absolute PV concentrations, we made a series of artificial standard samples which were aldehyde-fixed 10% bovine serum albumin containing given concentrations of PV (0, 12.5, 25, 50, 100, 200, and 400 μM, 1 and 2 mM), and calibration curves were deduced from quantitative immunogold analyses of these standard samples. We also analyzed a fast twitch muscle, the superficial part of the gastrocnemius muscle (GCM), whose PV content was previously reported in a biochemical study; the comparison between gold particle densities of GCM and standard samples indicated that these artificial standard samples could be used to estimate the approximate intracellular concentrations of PV. Based on these analyses PV concentrations were estimated as 50-100 μM in Purkinje cell somata and dendrites as well as interneuron somata, and as 1 mM or more in axons and axon terminals of Purkinje cells and basket cells.

Calcium-binding protein parvalbumin-immunoreactive neurons in the rat olfactory bulb

The laminar distribution and morphological features of parvalbumin-immunoreactive [PV(+l)] neurons, one of the subpopulations of GABAergic neurons, were studied in the rat olfactory bulb at a light microscopic level. In the main olfactory bulb of adult rats, PV(+) neurons were mainly located in the external plexiform layer (EPL), and a few were scattered in the glomerular layer (GL), mitral cell layer (ML), and granule cell layer (GRL); whereas PV(+) neurons were rarely seen in the accessory olfactory bulb. The inner and outer sublayers of the EPL (ISL and OSL) appeared to be somewhat different in the distribution of PV(+) somata and features of PV(+) processes. PV(+) somata were located throughout the OSL, and PV(+) processes intermingled with one another, making a dense meshwork in the OSL; whereas, in the ISL, PV(+) somata were mainly located near the inner border of the EPL, and PV(+) processes made a sparser meshwork than that in the OSL. PV(+) neurons in the EPL were apparently heterogeneous in their structural features and appeared to be classifiable into several groups. Among them there appeared five distinctive types of PV(+) neurons. The most prominent group of PV(+) neurons in the OSL were superficial short-axon cells, located in the superficial portion of this sublayer and giving rise to relatively thick processes, in horizontal or oblique directions, which usually bore spines and varicosities. Another prominent group of PV(+) neurons extended several short, branched dendrites with spines and varicosities, which appeared to intermingle with one another, making a relatively small, spherical or ovoid dendritic field around the cell bodies; most of them resembled Van Gehuchten cells reported in previous Golgi studies. A third distinctive and most numerous group of PV(+) neurons were of the multipolar type; their somata and processes were located throughout the EPL. Their relatively smooth processes with frequent varicosities and a few spines were extended horizontally or diagonally throughout the EPL. A fourth group, which could be a subtype of the multipolar type, were located in or just above th ML and extended several thin, smooth dendrites in the EPL, some of which appeared to reach the border between the GL and EPL. Occasionally, axonlike processes arose from their cell bodies and extended into the ML. This fourth type of PV(+) neuron was named inner short-axon cells. A fifth group of neuron was located in the ML; processes of these neurons were extended horizontally, so they were named inner horizontal cells. PV(+) processes from the fourth and the fifth group of cells appeared to make contacts on mitral cell somata. In the GL some presumably periglomerular cells were also PV(+). In the GRL, PV(+) neurons were small in number, but they were also heterogeneous in their structural features; Some were identified as Golgi cells. This study shows a tremendous heterogeneity in morphological features of a chemically defined subpopulation of GABAergic interneurons in the olfactory bulb.

Possible coexistence of amino acid (?-aminobutyric acid), amine (dopamine) and peptide (substance P); neurons containing immunoreactivities for glutamic acid decarboxylase, tyrosine hydroxylase and substance P in the hamster main olfactory bulb

SummaryThe coexistence of immunoreactivities for glutamic acid decarboxylase (GAD), tyrosine hydroxylase (TH) and substance P (SP) was revealed in the hamster main olfactory bulb, using the peroxidase-antiperoxidase immunohistochemical method. Adjacent 40 μm thick Vibratome sections were incubated in different antisera and those cells which were bisected by the plane of sectioning were identified at the paired surfaces of two consecutive sections. The coexistence of the immunoreactivities for 1) TH and GAD, 2) TH and SP and 3) GAD and SP in the same cells could thus be determined by observing the immunoreactivity of the two halves of the cell incubated in two different antisera. About 70% of TH-like immunoreactive (TH-LI) neurons in the periglomerular region also contained GAD-like immunoreactivity, whereas about 45% of GAD-LI ones were also TH-like immunoreactive. Furthermore, almost all (more than 95%) of SP-LI neurons contained both GAD-like and TH-like immunoreactivities. These observations indicate that in the periglomerular region of the hamster main olfactory bulb, some neurons (about 9% of all neurons containing TH-like and/or GAD-like immunoreactivities) may contain three different categories of neuroactive substances, that is, amino acid (GABA), amine (dopamine) and peptide (SP).

Tyrosine hydroxylase-like immunoreactive neurons in the olfactory bulb of the snake, Elaphe quadrivirgata, with special reference to the colocalization of tyrosine hydroxylase- and GABA-like immunoreactivities.

SummaryThe distribution and structural features of tyrosine hydroxylase-like immunoreactive (TH-LI) neurons were studied in the olfactory bulb of a snake, Elaphe quadrivirgata, by using pre-and post-embedding immunocytochemistry at the light microscopic level. In contrast to rodent olfactory bulbs previously reported, many TH-LI neurons were seen not only in the main olfactory bulb (MOB) but also in the accessory olfactory bulb (AOB). With regard to the TH-like immunoreactivity, there appeared no appreciable differences between MOB and AOB. As in mammalian MOB, the majority of TH-LI neurons were clustered in the periglomerular region and appeared to send their dendritic branches into glomeruli, which as a whole make an intense TH-LI band in the glomerular layer (GML). In the external plexiform/mitral cell layer (EPL/ML) of MOB and AOB as well as in the outer sublamina of the internal plexiform layer (OSL) of AOB, an appreciable number of TH-LI neurons were scattered, extending dendritic processes which appeared to make a loose meshwork. TH-LI neurons in EPL/ML (including OSL) appeared to consist of at least two morphologically different types. The first had a small perikaryon and one or two smooth dendrites which usually extended to GML and were frequently confirmed to enter into glomeruli. The second had a larger perikaryon and 2–3 dendrites which branched into several varicose processes extending in EPL/ML/OSL but appeared not to enter into glomeruli. The TH-like immunoreactivity was rarely seen in the internal plexiform layer and internal granule cell layer. The colocalization of GABA-like and TH-like immunoreactivities was further studied. Almost all TH-LI neurons in both EPL/ ML/OSL and GML contained GABA-like immunoreactivity irrespectively of the type of TH-LI cells.

Postnatal X-ray irradiation effects on glomerular layer of rat olfactory bulb: quantitative and immunocytochemical analysis

SummaryIn the rat olfactory bulb, the majority of interneurons in the glomerular layer (GL) are supposed to be generated during first postnatal week. Low and repeated doses of X-rays (200 rad x 4 and 200 rad x 6) were used during this period to impair the development of interneurons. The resulting effects on olfactory bulb neurons were examined stereologically and immunocytochemically in animals of 4 and 12 weeks of age. Quantitative analysis showed that, 1) the volume of the GL decreased to 55% (1200 rad) – 70% (800 rad) of control, 2) numerical cell densities in GL decreased to 40% (1200 rad) – 60% (800 rad) of control, thus resulting in 3) a decrease of the total cell number in GL to 20% (1200 rad) – 40% (800 rad) of control in irradiated olfactory bulbs of animals 4 weeks old. In comparison, mitral cells, which are generated prenatally, were much less affected (total cell number: 70–80% of control), indicating a selective loss of cells generated during the first postnatal week in GL. Effects on somata and processes immunoreactive for GABA, tyrosine hydroxylase (TH), calbindin D-28K and parvalbumin (PV) were examined in irradiated bulbs of both 4 and 12 week-old rats. All of these immunoreactive elements showed a drastic decrease in all layers. Semiquantitative analysis showed that in the GL, calbindin D-28K immunoreactive (calbindin D-28K(+)) neurons decreased more extensively than TH immunoreactive (TH(+)) and GABA-like immunoreactive (GABA(+)) neurons; that is, TH(+) and GABA(+) neurons decreased to 20% (1200 rad) – 40% (800 rad) of control, whereas calbindin D-28K(+) neurons decreased to 10% (1200 rad) – 30% (800 rad) of control in the GL of irradiated bulbs. These findings indicated that larger proportions of calbindin D-28K(+) neurons might be generated during the first postnatal week than those of GABA(+) and TH(+) neurons. Furthermore, in irradiated bulbs the proportion of GABA(-)TH(+) cells in TH(+) cells increased to about twice of control, and the estimated total numbers of GABA(-)TH(+) cells in irradiated rats were 95% (800 rad) and 40% (1200 rad) of control. These observations suggest that the majority of GABA(-)TH(+) neurons were less affected by X-ray irradiation during the first postnatal week and thus that they might be generated in the prenatal period. Since during the first 2 postnatal weeks, neurons showing GABA(-)TH(+) were not seen in GL (Kosaka et al. 1987a), the majority of GABA(-)TH(+) neurons in adult olfactory bulb were assumed to change their phenotype at some postnatal developmental period.