THE JOURNAL OF COMPARATIVE NEUROLOGY 29546'7484 (1990)
Calcium-Binding Proteins, Parvalbuminand Calbindin-D 28k-Immunoreactive Neurons in the Rat Spinal Cord and Dorsal Root Ganglia:A Light and Electron Microscopic Study MIKL6S ANTAL, TAA'rMAsF. FREUND, AND ERIKA POLGAR Department of Anatomy, University Medical School, Debrecen, H-4012, Hungary (M.A., E.P.); MRC Anatomical Neuropharmacology Unit, Department of Pharmacology, Oxford, OX1 3QT, England (M.A., T.F.F.); 1st Department of Anatomy, Semmelweis University Medical School, Budapest, H-1540, Hungary (T.F.F.)
ABSTRACT The distribution of two calcium-binding proteins, parvalbumin (PV) and calbindin-D 28K (CaBP), was studied by the peroxidase-anti-peroxidaseimmunohistochemical method at the light and electron microscopic level in the rat spinal cord and dorsal root ganglia. The possible coexistence of these two proteins was also investigated. PV-positive neurons were revealed in all layers of the spinal cord, except lamina I, which was devoid of labelling. Most of the PV-positive cells were found in the inner layer of lamina 11, lamina 111, internal basilar nucleus, central gray region, and at the dorsomedial and ventromedial aspects of the lateral motor column in the ventral horn. Neuronal processes intensely stained for PV sharply delineated inner lamina 11. With the electron microscope most of them appeared to be dendrites, but vesicle containing profiles were also found in a smaller number. CaBP-positive neurons appeared to be dispersed all over the spinal gray matter. The great majority of them were found in laminae I, 11, IV; the central gray region; the intermediolateral nucleus; and in the ventral horn just medial to the lateral motor column. Laminae I and I1 were densely packed with CaBP-positive punctate profiles that proved to be dendrites and axons in the electron microscope. A portion of labelled neurons in lamina IV and on the ventromedial aspect of the lateral motor column in the ventral horn disclosed both PV- and CaBPimmunoreactivity. All of the funiculi of the spinal white matter contained a large number of fibres immunopositive for both PV and CaBP. The highest density of CaBP-positive fibres was found in the dorsolateral funiculus, which was also densely packed with PV-positive fibres. PV-positive fibres were even more numerous in the dorsal part of the dorsal funiculus. The territory of the gracile funiculus in the brachial cord and that of the pyramidal tract in its whole extent were devoid of labelled fibres. In the thoracic cord, the dorsal nucleus of Clarke received a large number of PV-positive fibres. Dorsal root ganglia displayed both PV- and CaBP-immunopositivity. The cell diameter distribution histogram of PV-positive neurons disclosed two peaks-one a t 35 pm and the other at 50 pm. CaBP-positive cells in the dorsal root ganglia corresponded to subgroups of small and large neurons with mean diameters of 25 pm and 45 pm, respectively. On the basis of the location of perikarya and dendritic arborization patterns, PV- and CaBP-positive cells are correlated with previously described and neurochemically or physiologically characterized neurons of the spinal cord and dorsal root ganglia. K e y words: rat, spinal cord, immunohistochemistry
Accepted December 14,1989.
o 1990 WILEY-LISS, INC.
M. ANTAL ET AL.
468
Figure 1
PARVALBUMIN AND CALBINDIN-D 28K IN RAT SPINAL CORD
469
Fig. 2. Electron micrographs of PV-positive profiles in inner lamina 11. a: PV-positive dendrite in a glomerular-like synaptic arrangement. The central axon, containing mitochondria and loosely arranged spheroid vesicles, established asymmetric synaptic contacts with three dendrites, one of which is PV-positive. d,, d,: unlabelled dendrites. b:
PV-positive dendrite receives a synapse from a small profile densely packed wit,h synaptic vesicles. c,d: Vesicle containing PV-positive profiles establishing symmetric synaptic contacts with dendrites. Arrows point at synaptic contacts. Bars: 0.5 pm.
The presence of specific calcium-binding proteins has long been known in the central nervous system (Taylor, '74; Baron et al., '75; Lin et al., '80; Baimbridge and Parkes, '81; Celio and Heizmann, '81; Heizmann, '84). Calmodulin, a multifunctional calcium-binding protein, is present in all eukaryotic cells (Means et al., '82), but two other calciumbinding proteins that have been isolated from the brain,
parvalbumin (PV) and calbindin-D 28k (CaBP), appear to be markers for specific subpopulations of neurons (Celio and Heizmann, '81; Baimbridge and Miller, '82; Baimbridge et al., '82; Celio, '86; Stichel et al., '87). PV was found almost exclusively in GABA-ergic cells in the rat cerebral cortex (Celio, '86) and hippocampus (Kosaka e t al., '87), in septohippocampal neurons (Freund, '89), and in the lateral geniculate nucleus of the cat (Stichel et al., '88). Furthermore, only a subpopulation of GABA-ergic neurons appears to contain PV in these areas. In the rat cerebral cortex, P V was demonstrated in approximately 70 76 of GABA-ergic cells (Celio, '86), but in the hippocampus only 2070 of GAD-immunoreactive neurons proved to be PV positive (Kosaka et al., '87). Data available in this field strongly suggest that P V is a marker of GABA-ergic neurons that may be characterized by their high frequency firing rate in mammals (Heizmann, '84; Schwartzkroin and Kunkel, '85; Celio, '86; Kawaguchi et al., '87; Kosaka et al., '87; Stichel et al., '88). Although PV and CaBP are members of a homologous group of calcium-binding proteins (Heizmann and Berchtold, '871, the two proteins are located in largely different
Fig. 1. Transverse sections of the dorsal part of the brachial spinal cord processed for P V (a) and CaBP (b). a: PV-immunopositive punctate profiles delineate inner lamina 11. Stained neurons are in laminae 11-IV and the internal basilar nucleus. The dorsolateral and cuneate funiculus contain stained fibres in high density, while the gracile funiculus and the pyramidal tract is devoid of labelling. b: Laminae I and I1 are densely filled with CaBP-positive punctate profiles. Laminae I-IV contain numerous labelled neurons. In the white matter most of the labelled fibres can be seen in the dorsolateral funiculus, while the gracile funiculus and the pyramidal tract is nearly devoid of labelling. Borders of Rexed laminae I-IV are drawn with dashed lines; the laminae are indicated by Roman numerals. PT, pyramidal tract; GF, gracile funiculus; CF, cuneate funiculus; DLF, dorsolateral funiculus; IBN, internal basilar nucleus. Bars: 100 pn.
470
M. ANTAL ET AL.
Fig. 3. PV-positive neurons in laminae I-IV. a: Fusiform neuron with rostrocaudally oriented dendritic arbor in inner lamina 11. Sagittal section. b Same type of neurons shown in a in frontal section. c: Stained neurons in outer lamina I1 and at the border of laminae I1 and 111. Sagittal section. d: The same type of neuron shown in c at the border of
laminae I1 and 111 in frontal section. e: “Antenna-like” neuron in lamina IV. Sagittal section. Two other neurons are located at the border of laminae I1 and 111. f, g: “Pyramidal” neurons in lamina I[I. Sagittal section. Bars: 50 pm.
neuron populations in the cerebral cortex and hippocampus of the rat (Bairnbridge and Miller, ’82; Kosaka et al., ’87; Stichel et al., ’87). The segregation of neurons containing PV and CaBP is not complete, since both proteins have been demonstrated in rat cerebellar Purkinje cells and in periglomerular cells of the olfactory bulb (Celio and Heizmann, ’81; Bairnbridge and Miller, ’82). The presence of calcium-binding proteins has received very little attention in the spinal cord and dorsal root ganglia. Recently, it has been Ieported that about 20% of neurons are immunopositive for CaBP in dorsal root ganglia of the chick (Philippe and Droz, ’88); furthermore, only two of the six ultrastructurally well-defined subclasses of primary sensory neurons (Rambourg et al., ’83) have been found to be labelled. Although calcium-binding proteins have not been demonstrated in the spinal cord, we postulated that localization of these proteins could be used as marker of distinct populations of spinal neurons. If the presence of PV and CaBP could be related to physiological properties of neurons, then the immunohistochemical localization of these proteins would provide an excellent tool to study the functional anatomical organization of interneuronal circuits in the spinal cord.
In the experiments presented here, we have investigated the distribution of PV- and CaBP-immunopositive neurons and the coexistence of these two proteins in the spinal cord and dorsal root ganglia of the rat. In addition, we have analyzed the distribution of these proteins at the ultrastructural level and investigated the synaptic relations of the immunoreactive profiles in the superficial dorsal horn.
MATERIALS AND METHODS Preparation of tissue sections Six female Sprague-Dawley albino rats were deeply anaesthetized with chloral hydrate (35 mg/kg i.p.j and perfused through the heart first with Tyrode’s solution (oxygenated with a mixture of 95% 02,5% CO,), followed by a fixative containing 1% glutaraldehyde, 2 %o paraformaldehyde, and 0.2% picric acid in 0.1 M phosphate buffer (PB, pH: 7.4). The brain and spinal ganglia were removed and fixed in the same fixative for 1-3 hours. Blocks of brachial, thoracic, and lumbar segments of the spinal cord as well as the spinal ganglia were dissected and immersed in 10% and 20% sucrose dissolved in 0.1 M PB until they sank. Tissue blocks
PARVALBUMIN AND CALBINDIN-D 28K I N RAT SPINAL CORD
47 1
Fig. 4. Electron micrographs of CaBP-positive profiles in laminae I and 11. a: Labelled dendrites in glomerular-like synaptic arrangement. The central axon is densely packed with spheroid synaptic vesicles and surrounded by a number of dendrites. One of the Cal3P-positive profiles receives a synapse from the central axon (arrow). b Labelled dendrite establishes asymmetric synaptic contact (arrow) with a large axon terminal filled with numerous spheroid and a few dense-core synaptic vesicles and mitochondria. e: Stained dendrite receives synaptic contact
(arrow) from a large axon. The axon contains spheroid synaptic vesicles in a relatively small number. d CaBP-positive dendrite in synaptic contact (arrow) with a small calibre axon that contains pleomorphic synaptic vesicles. e: Small labelled axon establishes asymmetric synaptic contact (arrows) with a large dendrite. f: Stained axon containing spheroid and dense-core synaptic vesicles makes an asymmetric synaptic contact (arrow) with a large dendrite. Bars: 0.5 pm.
and ganglia were freeze-thawed in liquid nitrogen and sectioned at 60 fim on a vibratome. Following several washes in 0.1 M PB sections were treated with 1% sodium borohydride for 30 minutes, then washed again extensively in t,he same buffer.
Immunohistochemistry The free-floating sections were first incubated with antiparvalbumin (diluted 1:400) or anticalbindin (diluted 1: 1000) for 2 days at 4°C. The sections were then transferred
472
M. ANTAL ET AL.
Fig. 5. CaBP-positiveneurons in laminae I-IV. a: Fusiform cell with rostrocaudally oriented dendrites in lamina I. Frontal section. b: Multipolar cell with radially arborizing dendrites in lamina I. Frontal section. c: Small neurons in lamina 11. Arrows point at dendrites arising from the perikarya. Frontal section. d Fusiform neuron with longitudi-
nally running dendrites at the border of laminae I1 and 111. Sagittal section. e-g: Neurons in lamina 111. Sagittal (e,f) and frontal (g) sections. h, i: “Antenna-like” neurons in lamina IV. Sagittal sections. Bars:50 pm.
to goat antirabbit IgG (ICN, diluted 1:40) for 6-8 hours a t room temperature, followed by an overnight incubation at 4 O C in rabbit-PAP (DAKO, diluted 1:lOO). Prior to the antibody treatments sections were kept in 20% normal goat serum (ICN) for 40-50 minutes. All incubations were performed under continuous gentle agitation, and all of the antibodies were diluted in 50 mM TBS (pH: 7.4) to which 0.5% Triton X-100 and 1%normal goat serum (ICN) were added. Between incubations in the antibody solutions, sections were rinsed three times for 30 minutes in the same buffer. Immunoreactions were visualized by treating the sections with 0.05% diaminobenzidine (SIGMA) and 0.01 % H,O, in 50 mM Tris buffer (pH: 7.4). Immunostained sections were mounted on chrome alum-gelatin coated slides, air dried, dehydrated in ethanol, cleared in xylene, and mounted with XAM neutral medium. Alternate vibratome sections were processed for electron microscopy in the same way except that Triton X-100 was not included in any of the incubating solutions. After the chromogen reaction, sections were treated with 1%090,for
45 minutes, then dehydrated and embedded into Durcupan ACM (Fluka). Ultrathin sections were collected onto formvar-coated grids and counterstained with lead citrate. Primary antisera. Antisera to calbindin and parvalbumin were kindly donated by K.G. Baimbridge. Antiserum against monkey cerebellar calbindin was raised in rabbit, and it was shown to detect a single band of 28 kD protein in both one and two dimensional gel electrophoresis immunoblots. Antiserum to rat muscle parvalbumin was also raised in rabbit. The immunological and immunocytochemical characteristics of these antisera have been published earlier (Baimbridge et al., ’82; Sloviter, ’89). Controls. To test the specificity of the incubation procedure some sections were incubated in normal rabbit serum (diluted 1:200) instead of the primary antiserum. No specific staining was observed in these sections, but a homogeneous weak reaction endproduct covered the surface of the sections because of nonspecific binding of the antibodies. Some other sections that were processed through the full sequence of the immunohistochemical procedure were incubated in
PARVALBUMIN AND CALBINDIN-D 2 8 K IN RAT SPINAL CORD
473
Fig. 6. PV- and CaBP-positive neurons in the basal part of the dorsal horn. a: PV-positive cells in the internal basilar nucleus. b, c: CaBP-positive cells in the internal basilar nucleus. d: CaBP-positive
cells in the lateral part of laminae V-VI. e: PV-positive cell in the lateral part of laminae V-VI. Transverse sections. CF, cuneate funiculus; PT, pyramidal tract. Bars:50 pm.
the chromogen but without adding H,O,, to control the binding of diaminobenzidine to the tissue. In these sections there was no reaction endproduct associated with neuronal structures.
limited number at their origin from the cell body, fibres in the white matter were always strongly labelled.
Morphometry The size distribution of PV- and CaBP-immunopositive cells in the spinal ganglion was analyzed. Contours of profiles of 198 PV- and 226 CaBP-labelled cells found on the upper and lower surfaces of randomly selected sections were drawn with a camera lucida. Contours of the profiles, as they appeared on the camera lucida drawings with a final magnification of 550 times, were entered into an IBM PC with the aid of a graphics tablet. Perimeters of the contours were measured, the mean diameter of profiles was calculated (profile size), and the frequency distribution of profiles of various sizes was computed. From the profile size distribution histogram, cell size (diameter of labelled cells) distribution was derived according to the method of Weibel(’79).
RESULTS Cell bodies and dendrites were intensely labelled both for PV and CaBP. Although axons could be identified only in a
Superficial dorsal horn (laminae I, 11) PV-immunopositive elements. In the superficial dorsal horn the inner layer of Rexed lamina I1 was densely packed with immunopositive neuronal elements which appeared mostly as punctate profiles in a transverse section (Figs. la, 11). With the electron microscope most of them showed the ultrastructural characteristics of dendrites (Fig. 2a, b), whereas a fraction of labelled profiles contained synaptic vesicles (Fig. 2c,d). The peroxidase reaction end product was chiefly associated with microtubules and postsynaptic membranes, but mitochondria and synaptic vesicles were also frequently outlined with the label (Fig. 2a-d). Presynaptic unlabelled profiles establishing synapses with PV-positive dendrites could be classified into two morphological types. The first type of presynaptic axons was large and scalloped, filled with mitochondria and spheroid synaptic vesicles, and established asymmetric synaptic contacts with PV-positive dendrites (Fig. 2a). In addition to the PV-positive postsynaptic dendrite, the axon was usually surrounded by a number of unlabelled dendrites in a
474
M. ANTAL ET AL.
Fig. 7. a: PV-immunopositive elements in the dorsal funiculus and dorsal nucleus of Clarke a t the level of the thoracic spinal cord. The midline is indicated by a dashed line. GF, gracile funiculus; CF, cuneate funiculus; PT, pyramidal tract; dnC, dorsal nucleus of Clarke. Arrows
point at fibres running from the dorsal funiculus toward the dorsal nucleus of Clarke. b, c: CaSP-positive neurons in the intermediolateral nucleus in frontal (b) and transverse (c) sections. Bars: 50 pm.
glomerular-like synaptic arrangement (Fig. 2a). The second type of axon presynaptic to labelled dendrites was significantly smaller. The synaptic vesicles were generally aggregated in smaller or larger groups (Fig. 2b), but sometimes they were more loosely arranged.
PV-positive presynaptic profiles contained spheroid vesicles which were usually accumulated adjacent to the synaptic articulation (Fig. 2c,d). Besides vesicles, these profiles contained mitochondria and microtubules (Fig. 2c,d). They established symmetric synaptic contacts with dendrites.
PARVALBUMIN AND CALBINDIN-D 2 8 K IN RAT SPINAL CORD
476
Fig. 8. PV- and CaBP-positive neurons in the central gray region in transverse (a,b) and frontal (c-f) sections cut at the level of the central canal (c,d) and subjacent t o it ( e , f). Arrowhead labels a stained neuron
in a, arrows point at dendrites of CaBP-positive neurons in b. The midline is at the left margin in c and d, and indicated by dashed lines in e and f. Bars: 50 pm.
Most of these vesicle-containing profiles appeared to be axons, but the possibility can not be excluded that some of them may be presynaptic dendrites. PV-positive cell bodies were also found in the superficial dorsal horn, but lamina I was completely devoid of labelled neurons (Figs. la, 11). Stained neurons were primarily present in the inner layer of lamina 11, where punctate profiles formed a well defined immunopositive zone described above (Fig. la). Sections cut in the frontal and sagittal planes revealed that all labelled cells in the inner lamina I1 presented elongated cell bodies oriented in the rostrocaudal direction (Fig. 3a,b). Stem dendrites originated from the rostra1 and caudal poles of the soma, and extended in the rostrocaudal direction. The medio-lateral and dorsoventral extension of the dendritic tree was restricted (Fig. 3a,h), however the dendrites spanned a distance of 300-500 p m rostrocaudally (Fig. 3a).
Stained neurons of similar form were in smaller number in the outer layer of lamina I1 and a t the border between laminae I1 and 111 (Figs. la, 3c-e). The initial course of their dendrites was directed toward inner lamina I1 (Fig. 3c,e), and after reaching this layer the dendrites stretched along a distance of 300-500 pm in the rostrocaudal direction (Fig. 3c). The mediolateral and dorsoventral extensions of the dendritic tree were significantly smaller (Fig. 3c-e). CaBP-immunopositive elements. CaBP immunoreactivity was stronger than the labelling obtained for PV in the superficial dorsal horn. Laminae I and I1 were darkly stained with punctate profiles as they appeared in a transverse section (Figs. l b , 11).Electron microscopic analysis revealed that the punctate profiles represented both dendrites and axons. The ultrastructural distribution of the reaction endproduct showed a similar, if not identical, pattern to that which was produced by the reaction for PV.
476
M. ANTAL ET AL.
Fig. 9. PV- and CaBP-positive neurons in the ventral horn. a, b Labelled neurons are shown at the dorsomedial and ventromedial aspects of the lateral motor column, and in lamina VIII. Transverse sections. Borders of Rexed laminae VIII and IX are drawn by dashed
lines, the laminae are indicated by Roman numerals. c, d Micrographs of frontal sections cut at the level of labelled neurons at the ventromedial aspect of the lateral motor column. The midline is at the right margin in all micrographs. Bars: 100 pm.
Electron-dense deposits were principally associated with microtubules and postsynaptic membrane specializations, but peroxidase reaction endproduct was also frequently precipitated on the outer membrane of mitochondria and synaptic vesicles (Fig. 4a-f). CaBP-positive dendrites and axons were found in synaptic contacts with unlabelled profiles. Four morphologically distinct types of unlabelled axonal profiles established synapses with CaBP-positive dendrites. One of them was relatively large, scalloped, densely packed with spheroid synaptic vesicles and formed asymmetric synaptic contacts with labelled dendrites (Fig. 4a). This type of axonal varicosity was generally surrounded by a number of labelled and unlabelled dendrites in glomerular-like synaptic arrangement. The second type of presynaptic profile was domeshaped, densely packed with large spheroid synaptic vesicles and mitochondria. A few dense-core vesicles were always dispersed among the spheroid ones (Fig. 4b). The third type of axons contained also spheroid vesicles, but in a significantly smaller number than the previous ones (Fig. 4c). The synaptic vesicles in these profiles were usually aggregated in small areas, and the relatively small amount of cytoplasmic organelles (mitochondria and neurofilaments) endowed these profiles with a light appearance (Fig. 4c). Small-calibre axons containing pleomorphic vesicles constituted the fourth group of axons (Fig. 4d). CaBP-positive axons could be classified into two morphological types. Both of them established asymmetric synaptic contacts with unlabelled dendrites. Axonal profiles in the first group were much smaller than in the second, and
profiles of the second type contained also dense-core vesicles (Fig. 4e,f). CaBP-positive perikarya were also revealed in the superficial dorsal horn (Figs. l b , 11).On the basis of the shape of cell bodies and dendritic arborization pattern, labelled neurons in lamina I could be classified into two distinct morphological types. One of them presented a fusiform cell body elongated rostrocaudally. The two dendrites originating from the rostra1 and caudal extremities of the cell body arborized in a narrow, longitudinal strip of lamina I (Fig. 5a). The second type of neurons had a multipolar perikaryon which emitted three or four dendritic trunks (Fig. 5b). The dendrites gave rise only to a few side branches along their course and extended exclusively in lamina I. A large number of small neurons expressed positive immunoreaction for CaBP in lamina I1 (Figs. Ib, 5c). They presented round cell bodies and thin, slender dendrites which were difficult to recognize as they usually left the section within a few micrometers from the cell body independent from the orientation of the section (Figs. Ib, 2c). In the most inner part of lamina 11, near the border between laminae I1 and 111, a few neurons with spindle-shaped bodies and dendrites arborizing primarily in the rostrocaudal direction were also revealed in sagittal sections (Fig. 5d).
Nucleus proprius (laminae 111, IV) PV-immunopositive neurons. PV-positive cells in lamina I11 appeared to be pyramidal in shape (Figs. la, 3f, 9). The apical dendrite arising from the dorsal aspect of the cell body ran a t first dorsally, and then, before reaching
477
PARVALBUMIN A N D CALBINDIN-D 2 8 K IN RAT SPINAL CORD lamina 11, branched once or twice. Daughter branches, with a few exceptions, entered inner lamina I1 where they turned into the rostrocaudal direction and could be traced for a distance of 100-200 pm (Fig. 3f, g). The basal dendrites arborized in lamina 111. A few labelled cells were also found in lamina IV (Figs. l a , 3e). Their perikarya gave rise to two or three stem dendrites, which took a course toward more superficial layers (Fig. 3e). Almost all of the dendrites reached lamina 111, and a portion of them could be traced even into lamina 11. The long axis of the cell bodies and the dendritic arbors were oriented in the sagittal plane, and the dendritic tree spanned across a few hundred micrometers in this direction. CaBP-immunopositive neurons. As they appeared in horizontal and sagittal sections, CaBP-positive neurons in lamina I11 were oriented also rostrocaudally (Fig. 5e-g). One or two slender, poorly arborizing dendrites originating from both the rostra1 and caudal poles of their fusiform cell bodies endowed these cells with a bipolar, or bitufted appearance. Their dendritic arbors were never found to exceed 50 pm in the rostrocaudal direction, and were much narrower in other directions. The morphological characteristics of CaBP-positive neurons in lamina IV strikingly resembled PV-positive cells found in the same layer, but they outnumbered the PVpositive ones two or three times (Figs. lb, 5h, i). The poorly arborizing dendrites of these neurons were oriented in the rostrocaudal plane and extended a few hundred micrometers in this direction. The dendrites were directed toward the more superficial laminae of the dorsal horn, and a portion of them could be traced as far as lamina I1 (Fig. 5h, i).
Basal part of the dorsal horn PV-immunopositive neurons. A considerable number of PV-positive, round, or slightly multipolar neurons were found in the medial part of the basis of the dorsal horn at all levels of the spinal cord (Figs. la, 6a, 11).The area included the medial part of laminae V and VI and corresponded to the internal basilar nucleus. In addition, a few stained cells were scattered lateral to this area (Figs. la, 6e, 11).Three or four dendritic trunks originated from the cell body endowing these neurons with a triangular or quandrangular appearance. Dendrites gave rise to two or three orders of branches along the 200-300 pm course over which they could be traced (Fig. 6e). CaBP-immunopositive neurons. The location of neurons labelled for CaBP in the basal dorsal horn was the same as that of PV-positive ones (Figs. l b , 6b-d, 11).A number of them revealed multipolar perikarya giving rise to 4-5 dendrites (Fig. 6b); others presented bipolar cell bodies elongated in the dorsoventral axis in the internal basilar nucleus (Fig. 6c). Neurons located in the lateral part of laminae V and VI tended to form cellular aggregates, in which three or four multipolar neurons were in close vicinity (Fig. 6d).
Dorsal nucleus of Clarke An area involving the dorsomedial region of lamina VII and the ventromedial part of lamina VI was intensely stained for PV at the level of thoracic segments (Figs. 'ia, 11). It was conspicuously demarcated from the unstained surroundings and corresponded to the dorsal nucleus of Clarke. A few PV-positive fibres were seen to be separated from the axonal mass of the dorsal funiculus and to run ventrally toward this PV-positive area (Fig. 7a). In several
cases, these fibres could be traced as far as their entrance to the nucleus. It appeared as if PV-immunopositive elements in the dorsal nucleus of Clarke might be the terminals, or collaterals of PV-positive fibres emerging from the dorsal funiculus.
Intermediolateral nucleus A t the thoracic level of' the spinal cord CaBP-positive neurons were revealed in the intermediolateral nucleus (Figs. 7b, c, 11).Frontal sections demonstrated that these cells were arranged in small groups each consisting of 3-4 multipolar neurons. The cell nests were separated by 70-100 pm intervals from each other and interconnected with bundles of neuronal processes (Fig. 7b). Besides these rostrocaudally oriented fibres, the cells gave rise also to medially and ventrally oriented processes (Fig. 7c). PV-immunoreactive neurons were never found in the intermediolateral nucleus.
Central gray region (laminaX) PV-immunopositive neurons. Strong immunopositivity was revealed in the central gray region. The labelling was most pronounced on the lateral and ventral aspects of the central canal, while it was light in the dorsal part of the region (Fig. 8a). PV-positive multipolar neurons were scattered around the central canal (Fig. 8c), and their dendrites formed a dendritic network which appeared as a dense aggregate of punctate profiles in transverse sections (Fig. 8a). Dendrites of neurons subjacent to the central canal crossed the midline, even cell bodies themselves, were located at the midline in a number of cases (Fig. 8e). CaBP-immunopositive neurons. The distribution of CaBP immunoreactivity in the central gray region was similar to the staining obtained for PV. The labelling was most prominent at the lateral and ventral aspects of the central canal (Fig. 8b). Dendrites of neurons positive for CaBP extended in all directions in frontal sections (Fig. 8d). The presence of shorter or longer portions of labelled processes corroborated the impression about their diffuse dendritic arhorization pattern (Fig. 8b). CaBP-positive neurons subjacent to the central canal were more numerous than PV-positive ones in the same location. Some of their dendrites crossed the midline and formed a dendritic plexus interconnecting the two sides of the spinal cord (Fig. 8f).
Ventral horn PV-immunopositive neurons. PV-positive neurons were found in three locations in the ventral horn. Multipolar neurons were revealed in a large number in an area corresponding to the ventro-lateral part of lamina VII. The most ventral neurons of this cell group were seen in the dorsal part of lamina IX (Fig. 9a). The second group of PV-positive neurons was located in the ventromedial region of the ventral horn, which could be defined as lamina VIII (Fig. 9a). These neurons disclosed spindle-shaped, mediolaterally oriented cell bodies with two dendritic stems originating from the medial and lateral extremities of the somata. The third population of PV-positive cells appeared in the most ventral part of the ventral horn, medial to the lateral motor column, and they disclosed round, slightly multipolar cell bodies (Fig. 9a). Frontal sections indicated that these neurons, as well as the others found in other locations, formed a cell column along the ventral horn (Fig. 9c). PV-positive neurons of the ventral horn were revealed in the largest number at the brachial level of the spinal cord. They were
478
M. ANTAL ET AL.
PARVALBUMIN AND CALBINDIN-D 28K IN RAT SPINAL CORD much more sparsely distributed in the lumbar cord, and only a few labelled cells were found in the thoracic region (Fig. 11). CaBP-immunopositive neurons. CaBP-positive neurons were detected in similar regions where PV-positive cells were found. In the lateral part of lamina VII, only a few solitary neurons were stained (Fig. 9b). Their multipolar cell bodies gave rise to long, poorly arborizing dendrites (Fig. 9b). Most CaBP-positive neurons of lamina VIII lay very near the border between the gray and white matter (Fig. 9b). They presented fusiform cell bodies with two dendrites originating from the medial and lateral extremities of the perikarya. Both the location and the morphology of CaBPpositive neurons were similar to that of PV-positive cells in the ventral part of the ventral horn (Fig. 9b,d). Labelled neurons were packed in circumscribed cell groups in the brachial spinal cord; they were more diffusely arranged at the level of lumbar segments; and only a few stained cells were found in the thoracic part of the cord (Fig. 11).
White matter P V-immunopositiuity. Labelled fibres were revealed in a large number in all funiculi at all levels of the spinal cord (Figs. la, 6a, 7a, 9a, c), but they were distributed heterogeneously in the different regions of the white matter. The dorsolateral funiculus was more densely filled than the other parts of the lateral funiculus (Fig. la), while the ventralmost part of the dorsal funiculus, where the pyramidal tract is located, was almost completely devoid of immunoreaction (Figs. la, 6a, 7a, 8a, 9a). The strongest labelling was revealed in the dorsal part of the dorsal funiculus (Figs. l a , 6a, 7a). In this region stained fibres were homogeneously distributed at lumbar levels, but the medial part of the dorsal funiculus contained fewer labelled fibres than the lateral part in the thoracic spinal cord (Fig. 7a). Furthermore, the medial part corresponding to the gracile funiculus did not reveal any immunopositivity at the level of brachial segments (Fig. la). CaBP immunopositiuity. An intense labelling was obtained for CaBP all over the white matter, with the most dense staining in the dorsolateral funiculus (Figs. l b , 9b, d). In the dorsal funiculus the labelling was less intense and showed a pattern similar to that obtained for PV (Fig. lb). The most ventral part of the dorsal funiculus, the site of the pyramidal tract was similarly devoid of labelled fibres (Fig. lb). The density of stained fibres was remarkably low in the medial part of the dorsal funiculus a t the brachial level (Fig. lb), whereas the density of stained fibres was very similar medially and laterally in the lumbar cord.
Coexistence of PV- and CaBP-positivity Consecutive sections reacted for PV and CaBP revealed that some neurons in lamina IV and at the ventromedial aspect of the lateral motor column contained both PV and
Fig. 10. Micrographs from consequtive sections processed for either P V or CaBP. Identical areas of lamina I\' (a-d) and of the most ventral part of the ventral horn (e, f) are shown. Dendrites of PV-positive cells (b,d,f) can be traced into the consecutive section processed for CaBP (a,c,e). Arrowheads indicate the joining points of dendrites. White asterisks show labelled cell bodies. Black asterisks indicate a dendrite in c and d, which can be traced from one section to the other but does not belong to the labelled cell in d. c, capillaries serving as landmarks in fitting the micrographs.Bars: 50 Hm.
479
CaBP (Fig. 10). Dendrites of labelled cells in one section reacted for PV could be traced into the next section reacted for CaBP (Fig. 10). However, this method enabled us to identify only a fraction of neurons that contain both PV and CaBP, since tracing dendrites from one section to the next was very difficult in many cases.
Dorsal root ganglia PV-irnmunopositiue neurons. PV-positive cells were found in a considerable number in dorsal root ganglia (DRG); nearly one-third of the neurons proved to be immunoreactive for PV (Fig. 12a). On the basis of the cell diameter, labelled cells represented large- and mediumsized neurons of the DRG; the frequency distribution histogram of cell body diameter had two peaks, one at 35 pm and the other at 50 pm (Fig. 13a). CaBP-immunopositive neurons. CaBP-positive neurons were found nearly in the same number as the PVpositive ones (Fig. 12b). CaBP-positive cells represented subpopulations of small and large neurons; their cell diameter distribution histogram showed peaks at 25 pm and 45 pm (Fig. 13b).
DISCUSSION Lamina I Only CaBP-positive neurons are present in this lamina. The shape of cell bodies and orientation of the dendritic arbor of CaBP-positive cells resemble Golgi-impregnated neurons of Lima and Coimbra ('86), who have classified lamina I neurons into six morphologically distinct groups. One type of CaBP-positive neurons with fusiform cell body and rostrocaudally oriented dendritic arbor is reminiscent of their fusiform cells with longitudinal dendrites, while the flattened and pyramidal neurons of Lima and Coimbra ('86) resemble the second population of CaBP-positive cells with triangular or quadrangular cell bodies and poorly arborizing dendrites. Flattened and pyramidal neurons give rise to long projecting axons (Lima and Coimbra, '86), and the staining of such perikarya after mesencephalic injections of H R P (Menbtrey et al., '82) suggests that this type of CaBPpositive cells represents spinomesencephalic neurons. Fusiform neurons also project to the mesencephalon. Intracellular labelling of physiologically identified spinomesencephalic tract neurons, and extracellular injections of neuronal tracers into the mesencephalon have revealed a number of fusiform rostrocaudally oriented neurons in lamina I (Hylden et al., '86, '89; Leah et al., '88). The fact that the dorsolateral funiculus is densely packed with CaBP-positive fibres supports this notion, since the axons of the spinomesencephalic tract neurons ascend here (Hylden e t al., '89).
Lamina I1 Neurons immunopositive for calcium-binding proteins are present in the highest density in lamina 11. PV-positive cells are situated almost exclusively in the inner part of lamina 11, and they constitute a morphologically homogeneous neuron population. Their morphological features suggest that they correspond to islet cells which represent one of the principal cell types in lamina I1 (Gobel, '75, '78; Gobel et al., '80; Todd, '88). Neurons with similar morphology and location show glutamic acid decarboxylase (GAD) immunopositivity (Hunt et al., %l), which suggest that PV-immunoreactivity coexists with GABA in the islet cells.
M. ANTAL ET AL.
480
Parvalbumin
Calbindin-D 28k
Br . ..: ..
Th
L
Figure 11
,
Calcium-Binding Proteins, Parvalbuminand Calbindin-D 28k-Immunoreactive Neurons in the Rat Spinal Cord and Dorsal Root Ganglia:A Light and Electron Microscopic Study MIKL6S ANTAL, TAA'rMAsF. FREUND, AND ERIKA POLGAR Department of Anatomy, University Medical School, Debrecen, H-4012, Hungary (M.A., E.P.); MRC Anatomical Neuropharmacology Unit, Department of Pharmacology, Oxford, OX1 3QT, England (M.A., T.F.F.); 1st Department of Anatomy, Semmelweis University Medical School, Budapest, H-1540, Hungary (T.F.F.)
ABSTRACT The distribution of two calcium-binding proteins, parvalbumin (PV) and calbindin-D 28K (CaBP), was studied by the peroxidase-anti-peroxidaseimmunohistochemical method at the light and electron microscopic level in the rat spinal cord and dorsal root ganglia. The possible coexistence of these two proteins was also investigated. PV-positive neurons were revealed in all layers of the spinal cord, except lamina I, which was devoid of labelling. Most of the PV-positive cells were found in the inner layer of lamina 11, lamina 111, internal basilar nucleus, central gray region, and at the dorsomedial and ventromedial aspects of the lateral motor column in the ventral horn. Neuronal processes intensely stained for PV sharply delineated inner lamina 11. With the electron microscope most of them appeared to be dendrites, but vesicle containing profiles were also found in a smaller number. CaBP-positive neurons appeared to be dispersed all over the spinal gray matter. The great majority of them were found in laminae I, 11, IV; the central gray region; the intermediolateral nucleus; and in the ventral horn just medial to the lateral motor column. Laminae I and I1 were densely packed with CaBP-positive punctate profiles that proved to be dendrites and axons in the electron microscope. A portion of labelled neurons in lamina IV and on the ventromedial aspect of the lateral motor column in the ventral horn disclosed both PV- and CaBPimmunoreactivity. All of the funiculi of the spinal white matter contained a large number of fibres immunopositive for both PV and CaBP. The highest density of CaBP-positive fibres was found in the dorsolateral funiculus, which was also densely packed with PV-positive fibres. PV-positive fibres were even more numerous in the dorsal part of the dorsal funiculus. The territory of the gracile funiculus in the brachial cord and that of the pyramidal tract in its whole extent were devoid of labelled fibres. In the thoracic cord, the dorsal nucleus of Clarke received a large number of PV-positive fibres. Dorsal root ganglia displayed both PV- and CaBP-immunopositivity. The cell diameter distribution histogram of PV-positive neurons disclosed two peaks-one a t 35 pm and the other at 50 pm. CaBP-positive cells in the dorsal root ganglia corresponded to subgroups of small and large neurons with mean diameters of 25 pm and 45 pm, respectively. On the basis of the location of perikarya and dendritic arborization patterns, PV- and CaBP-positive cells are correlated with previously described and neurochemically or physiologically characterized neurons of the spinal cord and dorsal root ganglia. K e y words: rat, spinal cord, immunohistochemistry
Accepted December 14,1989.
o 1990 WILEY-LISS, INC.
M. ANTAL ET AL.
468
Figure 1
PARVALBUMIN AND CALBINDIN-D 28K IN RAT SPINAL CORD
469
Fig. 2. Electron micrographs of PV-positive profiles in inner lamina 11. a: PV-positive dendrite in a glomerular-like synaptic arrangement. The central axon, containing mitochondria and loosely arranged spheroid vesicles, established asymmetric synaptic contacts with three dendrites, one of which is PV-positive. d,, d,: unlabelled dendrites. b:
PV-positive dendrite receives a synapse from a small profile densely packed wit,h synaptic vesicles. c,d: Vesicle containing PV-positive profiles establishing symmetric synaptic contacts with dendrites. Arrows point at synaptic contacts. Bars: 0.5 pm.
The presence of specific calcium-binding proteins has long been known in the central nervous system (Taylor, '74; Baron et al., '75; Lin et al., '80; Baimbridge and Parkes, '81; Celio and Heizmann, '81; Heizmann, '84). Calmodulin, a multifunctional calcium-binding protein, is present in all eukaryotic cells (Means et al., '82), but two other calciumbinding proteins that have been isolated from the brain,
parvalbumin (PV) and calbindin-D 28k (CaBP), appear to be markers for specific subpopulations of neurons (Celio and Heizmann, '81; Baimbridge and Miller, '82; Baimbridge et al., '82; Celio, '86; Stichel et al., '87). PV was found almost exclusively in GABA-ergic cells in the rat cerebral cortex (Celio, '86) and hippocampus (Kosaka e t al., '87), in septohippocampal neurons (Freund, '89), and in the lateral geniculate nucleus of the cat (Stichel et al., '88). Furthermore, only a subpopulation of GABA-ergic neurons appears to contain PV in these areas. In the rat cerebral cortex, P V was demonstrated in approximately 70 76 of GABA-ergic cells (Celio, '86), but in the hippocampus only 2070 of GAD-immunoreactive neurons proved to be PV positive (Kosaka et al., '87). Data available in this field strongly suggest that P V is a marker of GABA-ergic neurons that may be characterized by their high frequency firing rate in mammals (Heizmann, '84; Schwartzkroin and Kunkel, '85; Celio, '86; Kawaguchi et al., '87; Kosaka et al., '87; Stichel et al., '88). Although PV and CaBP are members of a homologous group of calcium-binding proteins (Heizmann and Berchtold, '871, the two proteins are located in largely different
Fig. 1. Transverse sections of the dorsal part of the brachial spinal cord processed for P V (a) and CaBP (b). a: PV-immunopositive punctate profiles delineate inner lamina 11. Stained neurons are in laminae 11-IV and the internal basilar nucleus. The dorsolateral and cuneate funiculus contain stained fibres in high density, while the gracile funiculus and the pyramidal tract is devoid of labelling. b: Laminae I and I1 are densely filled with CaBP-positive punctate profiles. Laminae I-IV contain numerous labelled neurons. In the white matter most of the labelled fibres can be seen in the dorsolateral funiculus, while the gracile funiculus and the pyramidal tract is nearly devoid of labelling. Borders of Rexed laminae I-IV are drawn with dashed lines; the laminae are indicated by Roman numerals. PT, pyramidal tract; GF, gracile funiculus; CF, cuneate funiculus; DLF, dorsolateral funiculus; IBN, internal basilar nucleus. Bars: 100 pn.
470
M. ANTAL ET AL.
Fig. 3. PV-positive neurons in laminae I-IV. a: Fusiform neuron with rostrocaudally oriented dendritic arbor in inner lamina 11. Sagittal section. b Same type of neurons shown in a in frontal section. c: Stained neurons in outer lamina I1 and at the border of laminae I1 and 111. Sagittal section. d: The same type of neuron shown in c at the border of
laminae I1 and 111 in frontal section. e: “Antenna-like” neuron in lamina IV. Sagittal section. Two other neurons are located at the border of laminae I1 and 111. f, g: “Pyramidal” neurons in lamina I[I. Sagittal section. Bars: 50 pm.
neuron populations in the cerebral cortex and hippocampus of the rat (Bairnbridge and Miller, ’82; Kosaka et al., ’87; Stichel et al., ’87). The segregation of neurons containing PV and CaBP is not complete, since both proteins have been demonstrated in rat cerebellar Purkinje cells and in periglomerular cells of the olfactory bulb (Celio and Heizmann, ’81; Bairnbridge and Miller, ’82). The presence of calcium-binding proteins has received very little attention in the spinal cord and dorsal root ganglia. Recently, it has been Ieported that about 20% of neurons are immunopositive for CaBP in dorsal root ganglia of the chick (Philippe and Droz, ’88); furthermore, only two of the six ultrastructurally well-defined subclasses of primary sensory neurons (Rambourg et al., ’83) have been found to be labelled. Although calcium-binding proteins have not been demonstrated in the spinal cord, we postulated that localization of these proteins could be used as marker of distinct populations of spinal neurons. If the presence of PV and CaBP could be related to physiological properties of neurons, then the immunohistochemical localization of these proteins would provide an excellent tool to study the functional anatomical organization of interneuronal circuits in the spinal cord.
In the experiments presented here, we have investigated the distribution of PV- and CaBP-immunopositive neurons and the coexistence of these two proteins in the spinal cord and dorsal root ganglia of the rat. In addition, we have analyzed the distribution of these proteins at the ultrastructural level and investigated the synaptic relations of the immunoreactive profiles in the superficial dorsal horn.
MATERIALS AND METHODS Preparation of tissue sections Six female Sprague-Dawley albino rats were deeply anaesthetized with chloral hydrate (35 mg/kg i.p.j and perfused through the heart first with Tyrode’s solution (oxygenated with a mixture of 95% 02,5% CO,), followed by a fixative containing 1% glutaraldehyde, 2 %o paraformaldehyde, and 0.2% picric acid in 0.1 M phosphate buffer (PB, pH: 7.4). The brain and spinal ganglia were removed and fixed in the same fixative for 1-3 hours. Blocks of brachial, thoracic, and lumbar segments of the spinal cord as well as the spinal ganglia were dissected and immersed in 10% and 20% sucrose dissolved in 0.1 M PB until they sank. Tissue blocks
PARVALBUMIN AND CALBINDIN-D 28K I N RAT SPINAL CORD
47 1
Fig. 4. Electron micrographs of CaBP-positive profiles in laminae I and 11. a: Labelled dendrites in glomerular-like synaptic arrangement. The central axon is densely packed with spheroid synaptic vesicles and surrounded by a number of dendrites. One of the Cal3P-positive profiles receives a synapse from the central axon (arrow). b Labelled dendrite establishes asymmetric synaptic contact (arrow) with a large axon terminal filled with numerous spheroid and a few dense-core synaptic vesicles and mitochondria. e: Stained dendrite receives synaptic contact
(arrow) from a large axon. The axon contains spheroid synaptic vesicles in a relatively small number. d CaBP-positive dendrite in synaptic contact (arrow) with a small calibre axon that contains pleomorphic synaptic vesicles. e: Small labelled axon establishes asymmetric synaptic contact (arrows) with a large dendrite. f: Stained axon containing spheroid and dense-core synaptic vesicles makes an asymmetric synaptic contact (arrow) with a large dendrite. Bars: 0.5 pm.
and ganglia were freeze-thawed in liquid nitrogen and sectioned at 60 fim on a vibratome. Following several washes in 0.1 M PB sections were treated with 1% sodium borohydride for 30 minutes, then washed again extensively in t,he same buffer.
Immunohistochemistry The free-floating sections were first incubated with antiparvalbumin (diluted 1:400) or anticalbindin (diluted 1: 1000) for 2 days at 4°C. The sections were then transferred
472
M. ANTAL ET AL.
Fig. 5. CaBP-positiveneurons in laminae I-IV. a: Fusiform cell with rostrocaudally oriented dendrites in lamina I. Frontal section. b: Multipolar cell with radially arborizing dendrites in lamina I. Frontal section. c: Small neurons in lamina 11. Arrows point at dendrites arising from the perikarya. Frontal section. d Fusiform neuron with longitudi-
nally running dendrites at the border of laminae I1 and 111. Sagittal section. e-g: Neurons in lamina 111. Sagittal (e,f) and frontal (g) sections. h, i: “Antenna-like” neurons in lamina IV. Sagittal sections. Bars:50 pm.
to goat antirabbit IgG (ICN, diluted 1:40) for 6-8 hours a t room temperature, followed by an overnight incubation at 4 O C in rabbit-PAP (DAKO, diluted 1:lOO). Prior to the antibody treatments sections were kept in 20% normal goat serum (ICN) for 40-50 minutes. All incubations were performed under continuous gentle agitation, and all of the antibodies were diluted in 50 mM TBS (pH: 7.4) to which 0.5% Triton X-100 and 1%normal goat serum (ICN) were added. Between incubations in the antibody solutions, sections were rinsed three times for 30 minutes in the same buffer. Immunoreactions were visualized by treating the sections with 0.05% diaminobenzidine (SIGMA) and 0.01 % H,O, in 50 mM Tris buffer (pH: 7.4). Immunostained sections were mounted on chrome alum-gelatin coated slides, air dried, dehydrated in ethanol, cleared in xylene, and mounted with XAM neutral medium. Alternate vibratome sections were processed for electron microscopy in the same way except that Triton X-100 was not included in any of the incubating solutions. After the chromogen reaction, sections were treated with 1%090,for
45 minutes, then dehydrated and embedded into Durcupan ACM (Fluka). Ultrathin sections were collected onto formvar-coated grids and counterstained with lead citrate. Primary antisera. Antisera to calbindin and parvalbumin were kindly donated by K.G. Baimbridge. Antiserum against monkey cerebellar calbindin was raised in rabbit, and it was shown to detect a single band of 28 kD protein in both one and two dimensional gel electrophoresis immunoblots. Antiserum to rat muscle parvalbumin was also raised in rabbit. The immunological and immunocytochemical characteristics of these antisera have been published earlier (Baimbridge et al., ’82; Sloviter, ’89). Controls. To test the specificity of the incubation procedure some sections were incubated in normal rabbit serum (diluted 1:200) instead of the primary antiserum. No specific staining was observed in these sections, but a homogeneous weak reaction endproduct covered the surface of the sections because of nonspecific binding of the antibodies. Some other sections that were processed through the full sequence of the immunohistochemical procedure were incubated in
PARVALBUMIN AND CALBINDIN-D 2 8 K IN RAT SPINAL CORD
473
Fig. 6. PV- and CaBP-positive neurons in the basal part of the dorsal horn. a: PV-positive cells in the internal basilar nucleus. b, c: CaBP-positive cells in the internal basilar nucleus. d: CaBP-positive
cells in the lateral part of laminae V-VI. e: PV-positive cell in the lateral part of laminae V-VI. Transverse sections. CF, cuneate funiculus; PT, pyramidal tract. Bars:50 pm.
the chromogen but without adding H,O,, to control the binding of diaminobenzidine to the tissue. In these sections there was no reaction endproduct associated with neuronal structures.
limited number at their origin from the cell body, fibres in the white matter were always strongly labelled.
Morphometry The size distribution of PV- and CaBP-immunopositive cells in the spinal ganglion was analyzed. Contours of profiles of 198 PV- and 226 CaBP-labelled cells found on the upper and lower surfaces of randomly selected sections were drawn with a camera lucida. Contours of the profiles, as they appeared on the camera lucida drawings with a final magnification of 550 times, were entered into an IBM PC with the aid of a graphics tablet. Perimeters of the contours were measured, the mean diameter of profiles was calculated (profile size), and the frequency distribution of profiles of various sizes was computed. From the profile size distribution histogram, cell size (diameter of labelled cells) distribution was derived according to the method of Weibel(’79).
RESULTS Cell bodies and dendrites were intensely labelled both for PV and CaBP. Although axons could be identified only in a
Superficial dorsal horn (laminae I, 11) PV-immunopositive elements. In the superficial dorsal horn the inner layer of Rexed lamina I1 was densely packed with immunopositive neuronal elements which appeared mostly as punctate profiles in a transverse section (Figs. la, 11). With the electron microscope most of them showed the ultrastructural characteristics of dendrites (Fig. 2a, b), whereas a fraction of labelled profiles contained synaptic vesicles (Fig. 2c,d). The peroxidase reaction end product was chiefly associated with microtubules and postsynaptic membranes, but mitochondria and synaptic vesicles were also frequently outlined with the label (Fig. 2a-d). Presynaptic unlabelled profiles establishing synapses with PV-positive dendrites could be classified into two morphological types. The first type of presynaptic axons was large and scalloped, filled with mitochondria and spheroid synaptic vesicles, and established asymmetric synaptic contacts with PV-positive dendrites (Fig. 2a). In addition to the PV-positive postsynaptic dendrite, the axon was usually surrounded by a number of unlabelled dendrites in a
474
M. ANTAL ET AL.
Fig. 7. a: PV-immunopositive elements in the dorsal funiculus and dorsal nucleus of Clarke a t the level of the thoracic spinal cord. The midline is indicated by a dashed line. GF, gracile funiculus; CF, cuneate funiculus; PT, pyramidal tract; dnC, dorsal nucleus of Clarke. Arrows
point at fibres running from the dorsal funiculus toward the dorsal nucleus of Clarke. b, c: CaSP-positive neurons in the intermediolateral nucleus in frontal (b) and transverse (c) sections. Bars: 50 pm.
glomerular-like synaptic arrangement (Fig. 2a). The second type of axon presynaptic to labelled dendrites was significantly smaller. The synaptic vesicles were generally aggregated in smaller or larger groups (Fig. 2b), but sometimes they were more loosely arranged.
PV-positive presynaptic profiles contained spheroid vesicles which were usually accumulated adjacent to the synaptic articulation (Fig. 2c,d). Besides vesicles, these profiles contained mitochondria and microtubules (Fig. 2c,d). They established symmetric synaptic contacts with dendrites.
PARVALBUMIN AND CALBINDIN-D 2 8 K IN RAT SPINAL CORD
476
Fig. 8. PV- and CaBP-positive neurons in the central gray region in transverse (a,b) and frontal (c-f) sections cut at the level of the central canal (c,d) and subjacent t o it ( e , f). Arrowhead labels a stained neuron
in a, arrows point at dendrites of CaBP-positive neurons in b. The midline is at the left margin in c and d, and indicated by dashed lines in e and f. Bars: 50 pm.
Most of these vesicle-containing profiles appeared to be axons, but the possibility can not be excluded that some of them may be presynaptic dendrites. PV-positive cell bodies were also found in the superficial dorsal horn, but lamina I was completely devoid of labelled neurons (Figs. la, 11). Stained neurons were primarily present in the inner layer of lamina 11, where punctate profiles formed a well defined immunopositive zone described above (Fig. la). Sections cut in the frontal and sagittal planes revealed that all labelled cells in the inner lamina I1 presented elongated cell bodies oriented in the rostrocaudal direction (Fig. 3a,b). Stem dendrites originated from the rostra1 and caudal poles of the soma, and extended in the rostrocaudal direction. The medio-lateral and dorsoventral extension of the dendritic tree was restricted (Fig. 3a,h), however the dendrites spanned a distance of 300-500 p m rostrocaudally (Fig. 3a).
Stained neurons of similar form were in smaller number in the outer layer of lamina I1 and a t the border between laminae I1 and 111 (Figs. la, 3c-e). The initial course of their dendrites was directed toward inner lamina I1 (Fig. 3c,e), and after reaching this layer the dendrites stretched along a distance of 300-500 pm in the rostrocaudal direction (Fig. 3c). The mediolateral and dorsoventral extensions of the dendritic tree were significantly smaller (Fig. 3c-e). CaBP-immunopositive elements. CaBP immunoreactivity was stronger than the labelling obtained for PV in the superficial dorsal horn. Laminae I and I1 were darkly stained with punctate profiles as they appeared in a transverse section (Figs. l b , 11).Electron microscopic analysis revealed that the punctate profiles represented both dendrites and axons. The ultrastructural distribution of the reaction endproduct showed a similar, if not identical, pattern to that which was produced by the reaction for PV.
476
M. ANTAL ET AL.
Fig. 9. PV- and CaBP-positive neurons in the ventral horn. a, b Labelled neurons are shown at the dorsomedial and ventromedial aspects of the lateral motor column, and in lamina VIII. Transverse sections. Borders of Rexed laminae VIII and IX are drawn by dashed
lines, the laminae are indicated by Roman numerals. c, d Micrographs of frontal sections cut at the level of labelled neurons at the ventromedial aspect of the lateral motor column. The midline is at the right margin in all micrographs. Bars: 100 pm.
Electron-dense deposits were principally associated with microtubules and postsynaptic membrane specializations, but peroxidase reaction endproduct was also frequently precipitated on the outer membrane of mitochondria and synaptic vesicles (Fig. 4a-f). CaBP-positive dendrites and axons were found in synaptic contacts with unlabelled profiles. Four morphologically distinct types of unlabelled axonal profiles established synapses with CaBP-positive dendrites. One of them was relatively large, scalloped, densely packed with spheroid synaptic vesicles and formed asymmetric synaptic contacts with labelled dendrites (Fig. 4a). This type of axonal varicosity was generally surrounded by a number of labelled and unlabelled dendrites in glomerular-like synaptic arrangement. The second type of presynaptic profile was domeshaped, densely packed with large spheroid synaptic vesicles and mitochondria. A few dense-core vesicles were always dispersed among the spheroid ones (Fig. 4b). The third type of axons contained also spheroid vesicles, but in a significantly smaller number than the previous ones (Fig. 4c). The synaptic vesicles in these profiles were usually aggregated in small areas, and the relatively small amount of cytoplasmic organelles (mitochondria and neurofilaments) endowed these profiles with a light appearance (Fig. 4c). Small-calibre axons containing pleomorphic vesicles constituted the fourth group of axons (Fig. 4d). CaBP-positive axons could be classified into two morphological types. Both of them established asymmetric synaptic contacts with unlabelled dendrites. Axonal profiles in the first group were much smaller than in the second, and
profiles of the second type contained also dense-core vesicles (Fig. 4e,f). CaBP-positive perikarya were also revealed in the superficial dorsal horn (Figs. l b , 11).On the basis of the shape of cell bodies and dendritic arborization pattern, labelled neurons in lamina I could be classified into two distinct morphological types. One of them presented a fusiform cell body elongated rostrocaudally. The two dendrites originating from the rostra1 and caudal extremities of the cell body arborized in a narrow, longitudinal strip of lamina I (Fig. 5a). The second type of neurons had a multipolar perikaryon which emitted three or four dendritic trunks (Fig. 5b). The dendrites gave rise only to a few side branches along their course and extended exclusively in lamina I. A large number of small neurons expressed positive immunoreaction for CaBP in lamina I1 (Figs. Ib, 5c). They presented round cell bodies and thin, slender dendrites which were difficult to recognize as they usually left the section within a few micrometers from the cell body independent from the orientation of the section (Figs. Ib, 2c). In the most inner part of lamina 11, near the border between laminae I1 and 111, a few neurons with spindle-shaped bodies and dendrites arborizing primarily in the rostrocaudal direction were also revealed in sagittal sections (Fig. 5d).
Nucleus proprius (laminae 111, IV) PV-immunopositive neurons. PV-positive cells in lamina I11 appeared to be pyramidal in shape (Figs. la, 3f, 9). The apical dendrite arising from the dorsal aspect of the cell body ran a t first dorsally, and then, before reaching
477
PARVALBUMIN A N D CALBINDIN-D 2 8 K IN RAT SPINAL CORD lamina 11, branched once or twice. Daughter branches, with a few exceptions, entered inner lamina I1 where they turned into the rostrocaudal direction and could be traced for a distance of 100-200 pm (Fig. 3f, g). The basal dendrites arborized in lamina 111. A few labelled cells were also found in lamina IV (Figs. l a , 3e). Their perikarya gave rise to two or three stem dendrites, which took a course toward more superficial layers (Fig. 3e). Almost all of the dendrites reached lamina 111, and a portion of them could be traced even into lamina 11. The long axis of the cell bodies and the dendritic arbors were oriented in the sagittal plane, and the dendritic tree spanned across a few hundred micrometers in this direction. CaBP-immunopositive neurons. As they appeared in horizontal and sagittal sections, CaBP-positive neurons in lamina I11 were oriented also rostrocaudally (Fig. 5e-g). One or two slender, poorly arborizing dendrites originating from both the rostra1 and caudal poles of their fusiform cell bodies endowed these cells with a bipolar, or bitufted appearance. Their dendritic arbors were never found to exceed 50 pm in the rostrocaudal direction, and were much narrower in other directions. The morphological characteristics of CaBP-positive neurons in lamina IV strikingly resembled PV-positive cells found in the same layer, but they outnumbered the PVpositive ones two or three times (Figs. lb, 5h, i). The poorly arborizing dendrites of these neurons were oriented in the rostrocaudal plane and extended a few hundred micrometers in this direction. The dendrites were directed toward the more superficial laminae of the dorsal horn, and a portion of them could be traced as far as lamina I1 (Fig. 5h, i).
Basal part of the dorsal horn PV-immunopositive neurons. A considerable number of PV-positive, round, or slightly multipolar neurons were found in the medial part of the basis of the dorsal horn at all levels of the spinal cord (Figs. la, 6a, 11).The area included the medial part of laminae V and VI and corresponded to the internal basilar nucleus. In addition, a few stained cells were scattered lateral to this area (Figs. la, 6e, 11).Three or four dendritic trunks originated from the cell body endowing these neurons with a triangular or quandrangular appearance. Dendrites gave rise to two or three orders of branches along the 200-300 pm course over which they could be traced (Fig. 6e). CaBP-immunopositive neurons. The location of neurons labelled for CaBP in the basal dorsal horn was the same as that of PV-positive ones (Figs. l b , 6b-d, 11).A number of them revealed multipolar perikarya giving rise to 4-5 dendrites (Fig. 6b); others presented bipolar cell bodies elongated in the dorsoventral axis in the internal basilar nucleus (Fig. 6c). Neurons located in the lateral part of laminae V and VI tended to form cellular aggregates, in which three or four multipolar neurons were in close vicinity (Fig. 6d).
Dorsal nucleus of Clarke An area involving the dorsomedial region of lamina VII and the ventromedial part of lamina VI was intensely stained for PV at the level of thoracic segments (Figs. 'ia, 11). It was conspicuously demarcated from the unstained surroundings and corresponded to the dorsal nucleus of Clarke. A few PV-positive fibres were seen to be separated from the axonal mass of the dorsal funiculus and to run ventrally toward this PV-positive area (Fig. 7a). In several
cases, these fibres could be traced as far as their entrance to the nucleus. It appeared as if PV-immunopositive elements in the dorsal nucleus of Clarke might be the terminals, or collaterals of PV-positive fibres emerging from the dorsal funiculus.
Intermediolateral nucleus A t the thoracic level of' the spinal cord CaBP-positive neurons were revealed in the intermediolateral nucleus (Figs. 7b, c, 11).Frontal sections demonstrated that these cells were arranged in small groups each consisting of 3-4 multipolar neurons. The cell nests were separated by 70-100 pm intervals from each other and interconnected with bundles of neuronal processes (Fig. 7b). Besides these rostrocaudally oriented fibres, the cells gave rise also to medially and ventrally oriented processes (Fig. 7c). PV-immunoreactive neurons were never found in the intermediolateral nucleus.
Central gray region (laminaX) PV-immunopositive neurons. Strong immunopositivity was revealed in the central gray region. The labelling was most pronounced on the lateral and ventral aspects of the central canal, while it was light in the dorsal part of the region (Fig. 8a). PV-positive multipolar neurons were scattered around the central canal (Fig. 8c), and their dendrites formed a dendritic network which appeared as a dense aggregate of punctate profiles in transverse sections (Fig. 8a). Dendrites of neurons subjacent to the central canal crossed the midline, even cell bodies themselves, were located at the midline in a number of cases (Fig. 8e). CaBP-immunopositive neurons. The distribution of CaBP immunoreactivity in the central gray region was similar to the staining obtained for PV. The labelling was most prominent at the lateral and ventral aspects of the central canal (Fig. 8b). Dendrites of neurons positive for CaBP extended in all directions in frontal sections (Fig. 8d). The presence of shorter or longer portions of labelled processes corroborated the impression about their diffuse dendritic arhorization pattern (Fig. 8b). CaBP-positive neurons subjacent to the central canal were more numerous than PV-positive ones in the same location. Some of their dendrites crossed the midline and formed a dendritic plexus interconnecting the two sides of the spinal cord (Fig. 8f).
Ventral horn PV-immunopositive neurons. PV-positive neurons were found in three locations in the ventral horn. Multipolar neurons were revealed in a large number in an area corresponding to the ventro-lateral part of lamina VII. The most ventral neurons of this cell group were seen in the dorsal part of lamina IX (Fig. 9a). The second group of PV-positive neurons was located in the ventromedial region of the ventral horn, which could be defined as lamina VIII (Fig. 9a). These neurons disclosed spindle-shaped, mediolaterally oriented cell bodies with two dendritic stems originating from the medial and lateral extremities of the somata. The third population of PV-positive cells appeared in the most ventral part of the ventral horn, medial to the lateral motor column, and they disclosed round, slightly multipolar cell bodies (Fig. 9a). Frontal sections indicated that these neurons, as well as the others found in other locations, formed a cell column along the ventral horn (Fig. 9c). PV-positive neurons of the ventral horn were revealed in the largest number at the brachial level of the spinal cord. They were
478
M. ANTAL ET AL.
PARVALBUMIN AND CALBINDIN-D 28K IN RAT SPINAL CORD much more sparsely distributed in the lumbar cord, and only a few labelled cells were found in the thoracic region (Fig. 11). CaBP-immunopositive neurons. CaBP-positive neurons were detected in similar regions where PV-positive cells were found. In the lateral part of lamina VII, only a few solitary neurons were stained (Fig. 9b). Their multipolar cell bodies gave rise to long, poorly arborizing dendrites (Fig. 9b). Most CaBP-positive neurons of lamina VIII lay very near the border between the gray and white matter (Fig. 9b). They presented fusiform cell bodies with two dendrites originating from the medial and lateral extremities of the perikarya. Both the location and the morphology of CaBPpositive neurons were similar to that of PV-positive cells in the ventral part of the ventral horn (Fig. 9b,d). Labelled neurons were packed in circumscribed cell groups in the brachial spinal cord; they were more diffusely arranged at the level of lumbar segments; and only a few stained cells were found in the thoracic part of the cord (Fig. 11).
White matter P V-immunopositiuity. Labelled fibres were revealed in a large number in all funiculi at all levels of the spinal cord (Figs. la, 6a, 7a, 9a, c), but they were distributed heterogeneously in the different regions of the white matter. The dorsolateral funiculus was more densely filled than the other parts of the lateral funiculus (Fig. la), while the ventralmost part of the dorsal funiculus, where the pyramidal tract is located, was almost completely devoid of immunoreaction (Figs. la, 6a, 7a, 8a, 9a). The strongest labelling was revealed in the dorsal part of the dorsal funiculus (Figs. l a , 6a, 7a). In this region stained fibres were homogeneously distributed at lumbar levels, but the medial part of the dorsal funiculus contained fewer labelled fibres than the lateral part in the thoracic spinal cord (Fig. 7a). Furthermore, the medial part corresponding to the gracile funiculus did not reveal any immunopositivity at the level of brachial segments (Fig. la). CaBP immunopositiuity. An intense labelling was obtained for CaBP all over the white matter, with the most dense staining in the dorsolateral funiculus (Figs. l b , 9b, d). In the dorsal funiculus the labelling was less intense and showed a pattern similar to that obtained for PV (Fig. lb). The most ventral part of the dorsal funiculus, the site of the pyramidal tract was similarly devoid of labelled fibres (Fig. lb). The density of stained fibres was remarkably low in the medial part of the dorsal funiculus a t the brachial level (Fig. lb), whereas the density of stained fibres was very similar medially and laterally in the lumbar cord.
Coexistence of PV- and CaBP-positivity Consecutive sections reacted for PV and CaBP revealed that some neurons in lamina IV and at the ventromedial aspect of the lateral motor column contained both PV and
Fig. 10. Micrographs from consequtive sections processed for either P V or CaBP. Identical areas of lamina I\' (a-d) and of the most ventral part of the ventral horn (e, f) are shown. Dendrites of PV-positive cells (b,d,f) can be traced into the consecutive section processed for CaBP (a,c,e). Arrowheads indicate the joining points of dendrites. White asterisks show labelled cell bodies. Black asterisks indicate a dendrite in c and d, which can be traced from one section to the other but does not belong to the labelled cell in d. c, capillaries serving as landmarks in fitting the micrographs.Bars: 50 Hm.
479
CaBP (Fig. 10). Dendrites of labelled cells in one section reacted for PV could be traced into the next section reacted for CaBP (Fig. 10). However, this method enabled us to identify only a fraction of neurons that contain both PV and CaBP, since tracing dendrites from one section to the next was very difficult in many cases.
Dorsal root ganglia PV-irnmunopositiue neurons. PV-positive cells were found in a considerable number in dorsal root ganglia (DRG); nearly one-third of the neurons proved to be immunoreactive for PV (Fig. 12a). On the basis of the cell diameter, labelled cells represented large- and mediumsized neurons of the DRG; the frequency distribution histogram of cell body diameter had two peaks, one at 35 pm and the other at 50 pm (Fig. 13a). CaBP-immunopositive neurons. CaBP-positive neurons were found nearly in the same number as the PVpositive ones (Fig. 12b). CaBP-positive cells represented subpopulations of small and large neurons; their cell diameter distribution histogram showed peaks at 25 pm and 45 pm (Fig. 13b).
DISCUSSION Lamina I Only CaBP-positive neurons are present in this lamina. The shape of cell bodies and orientation of the dendritic arbor of CaBP-positive cells resemble Golgi-impregnated neurons of Lima and Coimbra ('86), who have classified lamina I neurons into six morphologically distinct groups. One type of CaBP-positive neurons with fusiform cell body and rostrocaudally oriented dendritic arbor is reminiscent of their fusiform cells with longitudinal dendrites, while the flattened and pyramidal neurons of Lima and Coimbra ('86) resemble the second population of CaBP-positive cells with triangular or quadrangular cell bodies and poorly arborizing dendrites. Flattened and pyramidal neurons give rise to long projecting axons (Lima and Coimbra, '86), and the staining of such perikarya after mesencephalic injections of H R P (Menbtrey et al., '82) suggests that this type of CaBPpositive cells represents spinomesencephalic neurons. Fusiform neurons also project to the mesencephalon. Intracellular labelling of physiologically identified spinomesencephalic tract neurons, and extracellular injections of neuronal tracers into the mesencephalon have revealed a number of fusiform rostrocaudally oriented neurons in lamina I (Hylden et al., '86, '89; Leah et al., '88). The fact that the dorsolateral funiculus is densely packed with CaBP-positive fibres supports this notion, since the axons of the spinomesencephalic tract neurons ascend here (Hylden e t al., '89).
Lamina I1 Neurons immunopositive for calcium-binding proteins are present in the highest density in lamina 11. PV-positive cells are situated almost exclusively in the inner part of lamina 11, and they constitute a morphologically homogeneous neuron population. Their morphological features suggest that they correspond to islet cells which represent one of the principal cell types in lamina I1 (Gobel, '75, '78; Gobel et al., '80; Todd, '88). Neurons with similar morphology and location show glutamic acid decarboxylase (GAD) immunopositivity (Hunt et al., %l), which suggest that PV-immunoreactivity coexists with GABA in the islet cells.
M. ANTAL ET AL.
480
Parvalbumin
Calbindin-D 28k
Br . ..: ..
Th
L
Figure 11
,
