Immunocytochemical distribution of ionotropic glutamate receptor subunits in the spinal cord of the rabbit

ELSEVIER

Journal of Chemical Neuroanatomy 11 (1996) 267-278

Immunocytochemical distribution of ionotropic glutamate receptor subunits in the spinal cord of the rabbit Agn6s Bonnot a'*, Marc Corio b, G6rard Tramu b, Denise Viala a aLaboratoire des Neurosciences de la Motricitd, Universitd Bordeaux I, CNRS URA339, Avenue des Facultks, 33405 Talence Cedex, France bLaboratoire de Neurocytochimie Fonctionnelle, Universitd Bordeaux I, CNRS URA339, Avenue des Facultks, 33405 Talence Cedex, France

Received 12 February 1996; revised 5 June 1996; accepted 24 July 1996

Abstract

Several histochemical and physiological studies in the literature suggest that ionotropic glutamate receptors are involved in various sensory and motor control mechanisms at the spinal level. The present immunocytochemical study used three specific antibodies to GluR2,4, GluR5,6,7 and to NMDAR1 to differentiate between the regional distribution of ct-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA), kainate and N-methyl-D-aspartate (NMDA) subtypes of glutamate receptors throughout the rabbit spinal cord. All of these immunoreactivities were prominent in the superficial dorsal horn and motor column. Each antibody gave rise to regionally specific immunostaining patterns but which were similar at all spinal levels. Numerous small neurons in superficial laminae were immunostained with GluR2,4 antibody while only neuropilar elements were immunostained with the two other antibodies. Cell bodies of the intermediate zone and fibres in the motor column were particularly densely immunostained with GluR5-7. Such an immunostaining pattern, which was particularly abundant with the GIuR5-7 antibody, suggests the presence, at the spinal level, of an extensive population of neurons exhibiting a high density of kainate receptors. Immunostaining with NMDAR1 antibody was less dense in comparison with the two others and especially in the motoneuron area. The present results provide the first immunohistochemical comparison between the respective regional distibutions of the three types of ionotropic glutamate receptors in the spinal cord. Their parallel distributions throughout the spinal cord support the concept of a tight functional cooperation between NMDA and non-NMDA receptors which has been extensively described for spinal events. Keywords: AMPA receptors; Kainate receptors; NMDA receptors; Immunocytochemistry; Spinal cord; Rabbit

1. Introduction

Many of the actions of the principal excitatory neurotransmitter L-glutamate (Glu) in the m a m m a l i a n central nervous system are mediated by ionotropic receptors. The three classes of ionotropic glutamate receptors (GluRs) termed, after their principal agonist, N-methyl-o-aspartate (NMDA), ~-amino-3-hydroxy-5methyl-4-isoxazole propionic acid (AMPA) and kainate (KA) receptors, are traditionally divided into two major sub-classes: N M D A and n o n - N M D A receptors.

*Corresponding author. Tel.: +33 56 848921; fax: +33 56 848901; e-mail: [email protected]. 0891-0618/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved P 11 SO891-0618(96)00173-1

N M D A and n o n - N M D A receptors are involved in mediating synaptic transmission at different neuronal levels in the spinal cord. Several lines of evidence suggest that Glu is a neurotransmitter in primary afferent fibers conveying somesthetic information to the spinal cord: (1) the behavioural effects of intrathecal injection of excitatory amino acids (EAA) agonists (Raigorodsky and Urca, 1990) or antagonists (Nasstr6m et al., 1992), (2) their electrophysiological effect on excitatory responses of lumbar neurons to mechanical stimuli in rat dorsal horn (DH) in vitro (King and Lopezgarcia, 1994), and (3) the neuroanatomical expression of Glu-like immunoreactivity in primary afferent terminals throughout the D H (A1ghoul et al., 1993; Broman et al., 1993). In addition, the

268

A. Bommt et al. ' Jm~rnal ~/ Chemical Neuroanatomy 11 (1996) 267 278

efficiency of EAA antagonists to attenuate excitotoxic cell death following various spinal cord injuries (ischemia, epilepsy, trauma) suggests that glutamatergic afferent inputs would contribute significantly to the neuronal death via GluRs located on these neurons (Sanner et al., 1994). The crucial role of EAAs in the activation of the locomotor generators of vertebrates (Cazalets et al., 1992; Douglas et al., 1993) suggests that several types of GluRs could participate in the genesis of locomotor activity by acting on the spinal oscillatory network. Finally, electrophysiological studies of the muscle relaxant properties of EAA antagonists revealed that GluRs were involved not only in polysynaptic reflexes but also in monosynaptic ones, suggesting that GluRs would be functionally involved at the motoneuronal level (Turski et al., 1990). Molecular cloning of GluRs has revealed, so far, the existence of 16 channel subunits (Sommer and Seeburg, 1992) whose association in various homomeric or heteromeric configurations accounts for a significant functional diversity within each channel class. For the AMPA receptor channel, four subunits termed GluRI 4 have been characterized. High-affinity KA receptors can be generated in vitro from subunits GluR5 7 and KA1 or KA2. NMDA receptors can be reconstituted as heteromeric structures from two subunit types, the NR1 subunit and one of four subunits (NR2A NR2D). The growing knowledge of the GluR molecular structure and of drug specificity has enabled many studies related to the neuroanatomical determination of the expression pattern of the GluR channel subunits to be performed in brain structures. To date, only a few studies have been carried out on vertebrate spinal cord. Concerning drugs specific for each GluR subtype, all autoradiographic studies, to date, have found the highest density of binding sites in the superficial DH (Jansen et al., 1990; Chinnery et al., 1993; Henley et al., 1993). A synthesis of in situ hybridization studies with autoradiographic data has given more detail about the respective distribution of the various subunits (Furuyama et al., 1993; Henley et al., 1993; T611e et al.. 1993, 1995). An extensive immunocytochemical study on the anatomical localization of AMPA receptor subunits performed in the rat spinal cord has provided an accurate assessment of subunit distribution due to the good spatial resolution of the technique (Petralia et al., 1994; Tachibana et al., 1994). The aim of the present immunohistochemical study is to compare the respective distribution, throughout the rabbit spinal cord, of three kinds of immunoreactivity, each of them being specific for one of the three GluRs: AMPA, KA or NMDA receptors. The data are discussed in the context of the findings of previous ligandbinding autoradiographic, in situ hybridization and electrophysiological studies. Preliminary results of this work have been published in abstract form (Bonnot et al., 1995).

2. Materials and methods

Five adult rabbits (2.5 3 kg) were anaesthetized with pentobarbital (35 mg/kg i.v.) and the rib cage was opened for subsequent transcardial perfusions, first with NaCI 0.9% to flush out the blood, followed by 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). For each rabbit, cervical, thoracic, lumbar and sacral spinal segments were dissected and postfixed in the same fixative for about 2 h at 4°C. The various segments were then placed overnight in a cryoprotectant solution of 20% sucrose in phosphate buffer and then cut with a cryostat in the transverse plane. In order to identify the spinal laminae and nuclei in the gray substance of rabbit spinal cord, serial sections were collected on gelatined slides and stained with cresyl violet. For immunocytochemical studies, coronal free-floating sections (40 jam) were rinsed in 0.1 M phosphate buffer saline (PBS) without detergent and then preincubated 30 rain in the same buffer containing 0.2% bovine serum albumin (BSA) and 0.3% casein. Sections were incubated for 3 days at 4°C in mouse monoclonal anti-NMDARI IgG, anti-GluR2,4 IgG (recognizing both GIuR2 and GluR4 subunits) or anti-GluR5-7 lgM (recognizing a common epitope in GluR5,6,7), respectively, diluted 1:250, 1:500 and 1:2500 in the protein-complemented buffer containing 0.01% sodium azide. The antibodies were purchased from Pharmingen (USA) and their immunological properties were previously described (Huntley et al., 1993; Siegel et al., 1994, 1995). After rinsing in PBS for l h, the sections were subsequently incubated for 2 h in biotinylated goat anti-mouse immunoglobulins (Jackson Laboratories, USA), diluted 1:2000 and in peroxidase-conjugated streptavidin (Jackson Laboratories, USA), diluted 1:1000. The sections were washed in PBS and the peroxidase reaction product was visualized by the glucose oxidase-nickel-DAB method (Shu et al., 1988). Some sections were treated for the silver intensification technique of the nickel-DAB end product of the peroxidase reaction (Merchenthaler et al., 1989). All sections were mounted on gelatin-coated slides, dehydrated and coverslipped with Eukitt. Immunostaining controls included deletion of the primary antibodies during the procedure and showed no immunoreaction in accordance with previous specificity studies.

3. Results

For the same monoclonal antibody, the immunoreactivity patterns were quite similar whatever the cervical, thoracic, lumbar or sacral spinal cord level of the rabbit. After staining with cresyl violet, we could identify all the laminae and nuclei in accordance with those

A. Bonnot et al. /Journal of Chemical Neuroanatomy 11 (1996) 267-278

A

B GluR2,4

~~. C GluR5-7

D

269

NMDAR 1

i" / ,,';':""

1

..... .-f

"'" --- -i:.'."

o ",'~, •

Fig. 1. Schematic drawings of the cytoarchitectonic subdivisions of segments C6, LI, L7 and S1 of rabbit spinal cord (A) and schematic mapping representing the distribution and density of GIuR2,4 (B), GIuR5-7 (C) and NMDAR1 (D) immunostaining at the same spinal levels (low-magnification: × 20). Dots represent fibers; full circles, small neurons; squares, medium-sized and large neurons; open stars, elongated neurons; full stars, large multipolar neurons.

270

A. Bonnot et al. ,' Journal q f Chemical Neuroanatomy 11 (1996) 267 278

described by Rexed (1954) in the cat. Silver intensification of the immunoreaction was useful to increase the efficiency of detection of various neuronal elements. In Fig. 1, immunostaining has been schematized for C6, L1, L7 and S1 representative medullar levels from the observation of both intensified and non-intensified sections. Unless specified, the following descriptions are valid for all the levels and the overall results relative to the comparative densities between immunoreactive structures for each antibody are semi-quantitatively summarized in Table 1.

3.1. Distribution of GluR2,4 immunostaining In the dorsal horn (DH) the vast majority of GluR2,4 immunoreactivity was observed in the superficial laminae. Numerous small cell bodies, elongated or round in shape were densely immunostained in laminae I III (Fig. 1B and Fig. 2A,B). They were scarce in other parts of the D H except in the medial part of lamina VI where only a few cells were usually observed. Neuropilar immunostaining, with a fine graining aspect, displayed a distribution similar to that described above and densely stained in lamina I, again mostly in the dorsal part of lamina III and in medial part of lamina VI. At variance with the small neurons, homogeneously widespread, immunostained fibers were less dense in lamina II (Fig. 2B). In addition, at thoracic and high lumbar levels (Fig. 1B), the immunostaining displayed a continuous localisation at the medial edge of the DH. A few medium-sized bi- or multipolar perikarya (about Table l Summary of the distribution of the GIuR2,4, GIuR5 7 and NMDAR1 immunoreactivities in rabbit spinal cord ]mmunoreactivity densities

Dorsal horn Perikarya Neuropil Dorsal horn Perikarya Neuropil Central grey Perikarya Neuropil Intermediate Perikarya Neuropil Ventral horn Perikarya Neuropil Ventral horn Perikarya Neuropil

GIuR2,4

GtuR5 7

NMDARI

+ + ++

0 ++

0 ? +

(laminae l - l l I )

(laminae IV VI) +t

+

(+)

(+)

+5 (+)

+ +

0

0

+ 0

++ (+)

0

+0

4- + +.+

+ 0

++ (+)

+±+ + ++

0

(lamina X)

zone (lamina VII)

(lamina VIII)

(lamina IX) L

0, none: ( + ) , occasional: + , little abundant: + + , quite abundant: + + + , very abundant.

2 per section), lightly to moderately stained, were found sporadically in laminae IV, V and VI (Fig. 2C). In the intermediate zone (IZ) small perikarya with the same characteristics as above were densely immunostained in lamina X around the central canal and some immunopositive fibers were generally observed in the same area. A few medium-sized or large round neurons, moderately immunostained (about 3 per section), were always found in lamina VII (Fig. 2D). In the ventral horn (VH), at lumbosacral levels, one or two large immunostained perikarya similar to those of lamina VII appeared to be easily detectable in lamina VIII (Fig. 2E). A few elongated multipolar cell bodies heterogeneous in size and moderately immunostained were found sporadically in this lamina at lumbosacral levels, and more commonly in cervical sections (Fig. 2F). A similar distribution of neuronal elements was not found at thoracic and high lumbar levels. Large multipolar neurons, moderately to highly stained, were widespread in the various somatic motor nuclei of lamina IX (Fig. 2G). Silver intensification of the immunoreaction resulted in a strong staining of these large cell bodies (Fig. 2H). In addition, occasional fibers were weakly detectable in lamina IX.

3.2. Distribution of GluR5- 7 immunostaining In the DH, GluR5 7 immunostaining appeared quite abundant overall in the superficial laminae (Fig. 1C and Fig. 3A). However, at variance with GIuR2,4 immunostaining, no cell body staining could be observed in this area; only neuropilar elements, containing distinct puncta moderately concentrated, were found in laminae I III, with a higher density in laminae I and II. In the inferior part of the DH where only occasional GIuR5 7 immunoreactive fibers were observed, a few mediumsized neurons moderately immunostained were detected in laminae IV, V (about 1 per lamina) and VI, with a higher density in the latter (about 2 per lamina). In the IZ, medium-sized or large bi- or multipolar round neurons, moderately to highly stained, were particularly concentrated in lamina VII (about 7 per section), overall in the medial part near lamina X (Fig. 3B). Generally, one large neuron could be observed in each sacral section, while some medium-sized neurons were immunostained in each lumbar section. In the VH, a quite abundant (about 7 per section) GluR5 7 immunostaining of both elongated and large round cell bodies were clearly observed in lamina VIII at the cervical, lumbar and sacral levels (Fig. 3C). In addition, the numerous large multipolar neurons of lamina IX were strongly immunostained (Fig. 3D). Both neuronal elements were surrounded by a dense network of thick stained fibers (Fig. 3E and F).

A. Bonnot et al. /Journal of Chemical Neuroanatomy 11 (1996) 267-278

271

,v~,

¸

Fig. 2. Light micrograph of transverse sections of the spinal cord immunostained with antibodies to GluRs. Dorsal is up and calibration bar is 100 ~na. GluR2,4 immunostaiaing. Superficial DH at S1 (A) and L1 levels (B) exhibits small round neurons and fibers. (C) In the top of the Sl level micrograph, the lower part of the superficial DH immunostaining is visible, and in deep DH an arrowhead points to a medium-sized neuron. In lamina VII at C6 level (D) and in lamina VIII at Sl level (E), an arrowhead points to a large neuron. In (D), central canal (cc) is underlined by artefactual reaction product (intensified edge effect). (F) Immunoreactive profile of an elongated neuron (arrowhead) is found in lamina VIII at C6 level. VH at L6 (G) and L7 levels (H) contains immunoreactive motoneurons. All sections are silver intensified except B, E and G.

272

A. Bonnot et al. ,: Journal ~/' Chemical Neuroanatomy I I (1996) 267 278

Fig. 3. Light micrograph of transverse sections of the spinal cord immunostained with antibodies to GluRs. Dorsal is up and calibration bar is 100 rim. GluR5-7 immunostaining. L6 level sections exhibit in superficial DH (A) immunoreactive fibers and in IZ (B), numerous immunoreactive neurons. In lamina VIII, arrowheads point at C6 level (C), to immunoreactive profiles of elongated neurons isolated or in cluster, surrounded by fibers and at L7 level (D) to a large isolated one. VH at L7 (E) and L6 levels (F) shows immunostained motoneurons and thick fibers.

3.3. Distribution of NMDAR1 immunostaining A dense accumulation of fibers highly immunosrained with N M D A R l antibody was localized in laminae I III of the DH while a few fibers were sporadically observed in other parts of the D H (Fig. 1D; Fig. 4A). A few medium-sized perikarya lightly immunostained could be detected in laminae IV,V and VI (about 3 per section) after silver intensification of the immunoreaction (Fig. 4B). In the IZ, whatever the level, at least one mediumsized or large neuron was lightly immunostained in each section in lamina VII (Fig, 4C). They were more numerous (about 3 per section) in thoracic and high

lumbar sections. After silver intensification, some fibers were seen to be immunostained in lamina X, especially at lumbar and sacral levels (Fig. 4D). In the VH, at lumbar and sacral levels, occasional (less than 1 per section) elongated cell bodies were lightly immunostained without intensification in lamina VIII. In contrast, they were more abundant (about 4 cells per section) at the cervical level. In addition, a few large round neurons, moderately immunostained, were encountered at lumbosacral levels in lamina VIII (Fig. 4D). At all spinal levels, large multipolar neurons were hardly detected in lamina IX without intensification of the Shu immunoreaction (Fig. 4E and F).

.4. Bonnot et al. / Journal of Chemical Neuroanatomy 11 (1996) 267-278

273

B

Fig. 4. Light micrograph of transverse sections of the spinal cord immunostained with antibodies to GluRs. Dorsal is up and calibration bar is 100 ~tm. NMDAR1 immunostaining. Superficial DH at L7 level (A) exhibits immunoreactive fibers. (B), In DH at S1 level, an arrowhead points to a medium-sized neuron. In lamina VII (C) and in lamina VIII (D), at S1 level, an arrowhead indicates a large neuron. (E, F) VH at L7 level presents large immunoreactive motoneurons. All sections are silver intensified except C and E.

4. Discussion The present results obtained by immunocytochemistry using antibodies that recognize GIuR2,4, GluR57 and NMDAR1 (subunits specific for AMPA, KA and NMDA receptors, respectively) demonstrated the existence of each of these GluR subunits with the

most prominent localization in the superficial dorsal horn (DH) and motor column (Table 1). The distribution pattern was continuous and similar throughout the spinal cord. In addition, immunoreactive structures were found in other regions such as the deepest part of the DH and the intermediate zone (IZ).

274

A. Bonnot et al. / Journal of Chemical Neuroanatomy 11 (1996) 267-278

Antibody to GIuR5-7 gave the most abundant immunostaining and the greatest contrast of immunostaining between highly and lightly stained structures. Consequently, it was of little use, in this case, to silver-intensify the immunoreaction. Even if comparison of the various immunoreactivity distributions suggested that the three GIuR channels, AMPA, KA and NMDA may colocalize in many stuctures throughout the spinal cord, in particular superficial neuropil and motoneurons, all of these sub-types displayed a specific immunostaining pattern that fitted well with their functional profiles. However, such a light microscopic study does not enable a precise description of the subcellular localization of the immunoreactivity and thus, only reflects the potentiality for the stained neurons to express a subclass of GIuR. In addition, electron microscopic immunocytochemical studies showed cytoplasmic labelling in both perikaria and neuropil that could be associated with either synthesis, degradation or transit of receptor (Petralia et al., 1994; Tachibana et al., 1994; Jaarsma et al., 1995). Our immunohistochemical results correlate well with certain electrophysiological results revealing a functional diversity of GluRs. Moreover, as compared to autoradiography and hybridization histochemistry, immunocytochemistry gives better spatial resolution. Therefore, this kind of approach is particularly useful to observe neuropilar labelling, this being all the more necessary since a previous study revealed that the receptor density of all EAA receptors are generally higher on the dendrites than on the soma (Arancio et al., 1993). 4. I. A comparison of immunoreactivities in the D H

In the present study, the neuropil was shown to be immunostained by the three types of antibodies in the superficial DH, while neuronal immunostaining was exclusively observed with anti-GluR2,4 antibody. To date, about three studies have dealt with the immunocytochemical localization of the AMPA receptors subunits using antibodies directed against GluR1, GluR2/3 (or GluR2/3/4c) and GIuR4 subunits in the cervical (Furuyama et al., 1993; Martin et al., 1993) or entire (Tachibana et al., 1994) rat spinal cord. The data from latter studies were in agreement with our results demonstrating that overall neuronal immunostaining was concentrated in the three most dorsal layers. In all cases, they found a much more prominent GluR2/3 immunoreactivity as compared to either GIuRI or GIuR4. While no GIuR4-IR cells and fibers were observed by Furuyama et al. (1993), the other two studies described a dense staining of neuropil with GIuR4 and a light cellular staining. Consequently, the GluR2/4 immunostaining pattern that we obtained must be due mainly to the immunostaining of GluR2 subunits localized on the small neurons and to the immunostaining of both

GluR2 and GIuR4 subunits on fibers. The former assumption coincides well with data from in situ hybridization studies, performed respectively at cervical and lumbar levels of the rat spinal cord, that revealed a preferential labelling of GIuR2 mRNA cells evenly expressed in superficial laminae of the DH (Furuyama et al., 1993; T611e et al., 1993). In addition, in the Martin et al. (1993) study, layer 3 was much more enriched in GluR4 immunoreactivity than layers 1 and 2. Therefore, the dense plexus of fibers we describe in the superficial part of lamina III could correspond to a GluR4-immunoreactivity-enriched region. According to T611e et al. (1993), superficial neuropilar immunostaining could correspond, to some extent, to dorsal dendrites of deeper neurons. The latter study also indicated the expression of GIuR3/4 heteromers in these deeper neurons, suggesting that GIuR2,4 antibody would recognize preferentially GIuR4 subunit on cell bodies we have observed in laminae IV-VI. To date, spinal cord has not been studied for its GluR5-7 immunostaining. For KA receptors, an in situ hybridization study showed a GluR5 mRNA labelling only in cells disseminated throughout the DH (Furuyama et al., 1993). In a more detailed study, which analyzed the distribution of transcripts of all GluR subunits, it appeared, in line with our data, that the levels of KA receptor labelling were much lower than those of the AMPA receptors (T611e et al., 1993). In addition, using in situ hybridization KA2 subunit was shown to be the main member of the KA receptor subclass present on the perikarya of the DH. Consequently, in the present study, where the KA2 subunit was not immunostained, the prominent neuropilar immunostaining of superficial laminae could mask a lighter GIuR5-7 cellular immunostaining. In contrast, in the deepest part of the DH where few immunoreactive fibers were observed, it was possible to detect occasional medium-sized neurons. NMDAR1 being the subunit most likely to be contained in all functional NMDA receptors (Nakanishi et al., 1992), the observed distribution of NMDARI immunostaining must be representative of the possible sites of action of endogenous agonists on the NMDA receptor subtype. Previous immunocytochemical resuits, obtained on rat cervical cord, described a dense neuropilar staining in superficial laminae and lamina X that coincided well with our results (Petralia et al., 1994). However, in the latter study, moderately stained neurons were found in all laminae while, in our experiments, occasional cells were observed only in the deepest part of the DH after silver intensification. Such a discrepancy also appeared with in situ hybridization results that demonstrated a predominant expression of NMDAR 1 mRNAs either in the deep layers of the DH (Furuyama et al., 1993) or in the substantia gelatinosa (T611e et al., 1995). In addition, a study dealing with the

A. Bonnot et aL /Journal of Chemical Neuroanatomy 11 (1996) 267-278

regional distribution in the rat lumbar spinal cord of the various NMDAR1 splice m R N A variants described large neurons in laminae IV and V containing virtually all the variants (T611e et al., 1995) that could fit with our cellular immunostaining. These neuronatomical localizations of GluR subunits can account for some of the known electrophysiological properties of spinal cord neurons under physiological and pathophysiological conditions. There is abundant electrophysiological evidence that N M D A and nonN M D A receptors tightly cooperate in mediating synaptic transmission during dorsal root stimulation (Evans and Long, 1989). Non-NMDA receptors substantially contribute to fast neurotransmission of noxious (Neugebauer et al., 1993) and innocuous (King and Lopezgarcia, 1994) cutaneous stimuli, while the role of N M D A receptors is restricted to longer latency synaptic components. Small GluR2,4-immunoreactive neurons observed in laminae I - I I I could correspond to deutoneurons of the superficial laminae which are able to mediate mainly innocuous stimuli from A0cl3 cutaneous fibers (Ishida and Shinozaki, 1991). GIuR5-7 neuropilar immunostaining observed in the same laminae could partly correspond to dendritic postsynaptic receptors receiving noxious stimuli from C cutaneous fibers. Non-NMDA presynaptic autoreceptors on group A and C endings could account, in addition to postsynaptic receptors of superficial or deeper deutoneurons, for both GIuR2,4 and GIuR5-7 neuropilar staining. In the same way, our NMDAR1 neuropilar immunostaining could result from both autoreceptors of primary afferent endings and dendritic postsynaptic receptors of C fibers. At variance with KA receptors which exhibit somatic subtypes containing principally the KA2 subunit, our study suggests an exclusive dendritic localization of N M D A receptors on superficial deutoneurons. Finally, the finding in the deep DH of immunoreactive neurons with the three used antibodies supports the suggestion that GIuR subtypes would be located on Wide Dynamic Range (WDR) neurons of laminae IV-VI which receive nociceptive information: non-NMDA receptors would preferentially mediate muscular information coming from direct primary afferents, while N M D A receptors would be involved in polysynaptic processing of cutaneous information (Song and Zhao, 1993). 4.2. A comparison of immunoreactivities in the VH

Motoneuronal intensity in the rabbit ventral horn (VH), as visualized by non intensified immunostaining, differed widely depending on the antibody used. Such results may be explained, to a large extent, by the relative density of somatic receptors. In immunocytochemical studies performed on rat spinal cord, motoneurons stained intensely with antibodies to GluR2/3

275

and GIuR4 (Furuyama et al., 1993; Martin et al., 1993; Tachibana et al., 1994). In parallel, both in situ hybridization studies described an abundant expression of GIuR2, GIuR3 and GIuR4 genes (Furuyama et al., 1993; Tflle et al., 1993). In rat cervical spinal cord, motoneurons were found to be intensely immunostained with NMDAR1 antibody (Petralia et al., 1994), and they also exhibited strong signals for the transcript (Furuyama et al., 1993; Tflle et al., 1993, 1995). In addition, neuropilar staining was described with GluR2/3 and 4 (Tachibana et al., 1994). In the present study, a moderate to light motoneuronal immunostaining was obtained with GluR2,4 and N M D A R 1, respectively, and a few puncta were observed with GIuR2,4 after silver intensification. The weakness of AMPA receptor immunostaining obtained with this antibody might be explained by the lack of immunoreaction with the GIuR3 subunit of the receptor. However, such an explanation is excluded for NMDAR1 immunostaining since NMDAR1 subunit is known to be present in all functional receptors (Nakanishi et al., 1992). In addition, our results reporting a weak motoneuronal immunostaining for both AMPA and N M D A receptors, are also supported by autoradiographic studies (Jansen et al., 1990; Chinnery et al., 1993) where no high binding activity was found in relation to motoneurons. In the present study, the strongest staining on perikarya, dendrites and endings was obtained with the GIuR5-7 antibody, contrasting with previous hybridization works which report a moderate signal on motoneurons for GluR5 and KA1 probes (Furuyama et al., 1993; Trlle et al., 1993). This discrepancy does not exist in the cerebellar cortex, since Jaarsma et al. (1995) using the same GIuR5-7 antibody obtained a high immunolabelling in good agreement with hybridization results (Wisden and Seeburg, 1993). The strength of the immunoreaction can hardly be explained by the enhanced avidity of this antibody (IgM) since, on the contrary, its large molecular size would reduce its penetration into the tissue. Thus, our present results suggest that KA receptors would be an important subtype of GIuR involved in synaptic transmission at the somatic motoneuronal level through both postsynaptic receptors and presynaptic autoreceptors. Such an assumption is supported by previous electrophysiological findings: KA was found to be the most efficient agonist to increase the firing of VH cells in slices of rat spinal cord and quisqualate, an agonist of AMPA receptors induced also a facilitatory effect which was greater than that of N M D A itself (Hiriyama et al., 1990). Secondly, in vivo studies of spinal reflex activity revealed that only nonN M D A antagonists were able to completely abolish the monosynaptic reflexes and the excitability of motoneurons, while N M D A receptors contribute little to this (Turski et al., 1990; Fenaux et al., 1991; Farkas and

276

A. Bonnot et al, / Journal o f Chemical Neuroanatomy 11 (1996) 267-278

Ono, 1995). Finally, the considerable spatial spread of the monosynaptic contacts involving non-NMDA receptors along the dendrites of motoneurons probably explains the blockade or reduction by a non-NMDA antagonist of the EPSPs evoked by monosynaptic afferent stimulation (Walmsley and Bolton, 1994). This supports that the strong GluR5-7 neuropilar immunostaining in VH would derive, to some extent, from postsynaptic dendritic receptors. 4.3. A comparison of immunoreactivities in the IZ We have described, at all spinal levels, a substantial neuronal immunostaining in lamina VII whatever the antibody used. In previous neuroanatomical studies, few data were given about this area, except for AMPA subunits which were localized with antibody to GluR2/ 3 and 4 in lateral regions of lamina VII known to contain Renshaw cells and Ia inhibitory cells (Tachibana et al., 1994). In the present study, very few medium-sized neurons were immunostained with GluR2,4 and GluR5-7 antibodies in similar regions. In return, a few medium-sized or large neurons which were quite strongly immunostained with all antibodies occurred in the medial part of lamina VII. At thoracic, high lumbar and low sacral levels, it cannot be excluded that some of the stained neurons described in lamina VII would be autonomic preganglionic neurons. In lamina VIII, at the lumbosacral level, such neurons are scattered with smaller fusiform perikarya, like those described in the Tachibana et al. (1994) study. In lamina X, only small GIuR2,4 immunoreactive neurons are observed. Accordingly, AMPA-selective subunit mRNAs and GluR1 neuronal immunostaining were found in this area (T611e et al., 1993; Tachibana et al., 1994). Some electrophysiological findings provide suggestions about the functional role of such GluR immunostaining in IZ. The key role of EAA receptors in spinally mediated rhythmic motor activity had been demonstrated in various in vitro (Kudo and Yamada, 1987: Smith and Feldman, 1987; Cazalets et al., 1990) and in vivo (Fenaux et al., 1991; Douglas et al., 1993) mammalian preparations. In addition, it has been shown recently, on lumbosacral slices of rat spinal cord, that a population of neurons ventrolateral to the central canal could generate spontaneous or NMDA-evoked rhythmic low-frequency membrane voltage oscillations (Hochman et al., 1994). Similarly located neurons were found to be activated by stimulation of the mesencephalic locomotor region (Jordan and Noga, 1991) and to receive reticulospinal input (Kuypers et al., 1962; Martin et al., 1982; Light, 1983; Nahin et al., 1983). In the same context, in the rabbit spinal cord, the most dorsally located neurons firing at the locomotor rate have been recorded in the IZ (Viala et al., 1991;

Bonnot et al., 1996). On the other hand, the various neuroanatomical attempts to localize potential spinal locomotor networks using activity-dependent markers (Viala et al., 1988; Dai et al., 1990; Kjaerulff et al., 1994; Carr et al., 1994) have found consistent labelling in the central and intermediate grey matter. Thus, our findings strengthen the hypothesis that EAA-sensitive neurons located in the intermediate grey could participate in locomotor genesis and they will provide a framework for further electrophysiological localization of GIuR agonist-evoked rhythmically active interneurons in the rabbit spinal cord. In conclusion, this study provides several lines of evidence for the cooperation of the various types of GluRs at all the rostro-caudal levels of the rabbit spinal cord. These receptors are differentially localized on many neuronal structures, that may constitute not only afferent pathways, various local spinal neurons (deutoneurons, interneurons, premotoneurons, motoneurons) and associated neuropil, but also cells of origin of ascending tracts. They also constitute potential targets for endogenous EAA action in both physiological and pathological spinal events. The differential dorso-ventral density between the three types of GIuR is in accordance with many functional suggestions derived from electrophysiological data or recent studies using activity-dependent markers. In particular, the finding that KA, in young rat spinal cord, induced cobalt uptake in most neuronal perikarya while the NMDAinduced uptake was the lowest (Nagy et al., 1994) also suggests a prominent role for KA receptors at the spinal level. However, the real physiological significance of the variations in subunit composition of each type of GluR remains unknown, and highlights the possible large variety of functional effects related to glutamate and aspartate release at the spinal level.

Acknowledgements The authors would like to express their gratitude to Dr. T. Durkin for correcting the English version of the manuscript and to M. Chaigniau for assistance in the preparation of photomicrographs.

References Al-ghoul, W.M., Volsi, G.L., Weinberg, R.J. and Rustioni, A. (1993) Glutamate immunocytochemistry in the dorsal horn after injury or stimulation of the sciatic nerve of rats. Brain Res. Bull. 30, 453 459. Arancio, O., Yoshimura, M., Murase, K. and MacDermott, A.B. (1993) The distribution of excitatory amino acid receptors on acutely dissociated dorsal horn neurons from postnatal rats. Neuroscience 52, 159 167.

A. Bonnot et al./ Journal of Chemical Neuroanatomy 11 (1996)267-278 Bonnot, A., Bouc, A,M., Viala, D. and Viala, G. (1996) Effect of posture on the onset of fictive locomotion in the decerebrate rabbit preparation. Neurosci. Lett. 208, 1-4. Bonnot, A., Corio, M., Tramu, G. and Viala, D. (1995) Immunocytochemical localization of different glutamate receptors in the lumbar spinal cord of the rabbit: relationships with locomotor control. Behav. Pharmacol. 6, 616. Broman, J., Anderson, S. and Ottersen, O.P. (1993) Enrichment of glutamate-like immunoreactivity in primary afferent terminals throughout the spinal cord dorsal horn. Eur. J. Neurosci. 5, 1050-1061. Carr, P.A., Noga, B.R., Nance, D.M. and Jordan, L.M. (1994) Intracellular labelling of cat spinal neurons using a tetramethylrhodamine-dextran amine conjugate. Brain Res. Bull. 34, 447451. Cazalets, J.R., Grillner, P., Menard, I., Cr6mieux, J. and Clarac, F. (1990) Two types of motor rhythm induced by NMDA and amines in an in vitro spinal cord preparation of neonatal rat. Neurosci. Lett. I l l , 116-122. Cazalets, J.R., Sqalli-Houssaini, Y. and Clarac, F. (1992) Activation of the central pattern generators for locomotion by serotonin and excitatory amino acids in neonatal rat. J. Physiol. (Lond.) 455, 187- 204. Chinnery, R.M., Shaw, P.J., Ince, P.G. and Johnson, M. (1993) Autoradiographic distribution of binding sites for the nonNMDA receptor antagonist [3H]CNQX in human motor cortex, brainstem and spinal cord. Brain Res. 630, 75-81. Dai, X., Douglas, J.R., Nagy, J.I., Noga, B.R. and Jordan, L.M. (1990) Localization of spinal neurons activated during treadmill locomotion using the c-fos immunohistochemical method. Soc. Neurosci. Abstr. 16, 889. Douglas, J.R., Noga, B.R., Dai, X. and Jordan, L.M. (1993) The effects of intrathecal administration of excitatory amino acid agonists and antagonists on the initiation of locomotion in the adult cat. J. Neurosci. 13, 990-1000. Evans, R.H. and Long, S.K. (1989) Primary afferent depolarization in the rat spinal cord is mediated by pathways utilising NMDA and non-NMDA receptors. Neurosci. Lett. 100, 231-236. Farkas, S. and Ono, H. (1995) Participation of NMDA and nonNMDA excitatory amino acid receptors in the mediation of spinal reflex potentials in rats: an in vivo study. Br. J. Pharmacol. 114, 1193-1205. Fenaux, F., Corio, M., Palisses, R. and Viala, D. (1991) Effects of an NMDA-receptor antagonist, MK801, on central locomotor programming in the rabbit. Exp. Brain Res. 86, 393-401. Furuyama, T., Kiyama, H., Sato, K., Park, H.T., Maeno, H., Takagi, H. and Tohyama, M. (1993) Regio-specific expression of subunits of ionotropic glutamate receptors (AMPA-type, KA-type and NMDA receptors) in the rat spinal cord with special reference to nociception. Mol. Brain Res. 18, 141-151. Henley, J.M., Jenkins, R. and Hunt, S.P. (1993) Localization of glutamate receptor binding sites and mRNAs to the dorsal horn of the rat spinal cord. Neuropharmacology 32, 37-41. Hiriyama, T., Ono, H. and Fukuda, H. (1990) Effects of excitatory and inhibitory amino acid agonists and antagonists on ventral horn ceils in slices of spinal cord isolated from adult rats. Neuropharmacology 29, 1117-1122. Huntley, G.W., Rogers, S.W., Moran, T., Janssen, W., Archin, N., Vickers, J.C., Cauley, K., Heinemann, S.F. and Morrison, J.H. (1993) Selective distribution of kainate receptor subunit immunoreactivity in monkey neocortex revealed by a monoclonal antibody that recognizes glutamate receptor subunits GIuR5/6/7. J. Neurosci. 13, 2965-2981. Hochman, S., Jordan, L.M. and MacDonald, J.F. (1994) N-methylD-aspartate receptor-mediated voltage oscillations in neurons surrounding the central canal in slices of rat spinal cord. J. Neurophysiol. 72, 565-577.

277

Ishida, M. and Shinozaki, H. (1991) Novel kainate derivatives: potent depolarizing actions on spinal motoneurones and dorsal root fibres in newborn rats. Br. J. Pharmacol. 104, 873-878. Jaarsma, D., Wenthold, R.J. and Mugnaini, E. (1995) Glutamate receptor subunits at mossy fiber-unipolar brush cell synapses: light and electron microscopic immunocytochemical study in cerebellar cortex of rat and cat. J. Comp. Neurol. 357, 145-160. Jansen, K.L.R., Faull, R.L.M., Dragunow, M. and Waldvogel, H. (1990) Autoradiographic localization of NMDA, quisqualate and kainic acid receptors in human spinal cord. Neurosci. Lett. 108, 53-57. Jordan, L.M. and Noga, B.R. (1991) Spinal interneurons synaptically activated by stimuli applied to the mesencephalic locomotor region. Soc. Neurosci. Abstr. 17, 1223. King, A.E. and Lopezgarcia, J.A. (1994) Excitatory amino acid receptor-mediated neurotransmission from cutaneous afferents in rat dorsal horn in vitro. J. Physiol. (Lond.) 472, 443-457. Kjaerulff, O., Barajon, I. and Kiehn, O. (1994) Sulphorodamine-labelled cells in the neonatal rat spinal cord following chemically induced locomotor activity in vitro. J. Physiol. (Lond.) 478, 265-273. Kudo, N. and Yamada, T. (1987) N-methyl-D,L-aspartate-induced locomotor activity in a spinal cord-hindlimb muscles preparation of the newborn rat studied in vitro. Neurosci. Lett. 75, 43-48. Kuypers, H.G.J.M., Fleming, W.R. and Farinholt, J.W. (1962) Subcorticospinal projections in the rhesus monkey. J. Comp. Neurol. 118, 107-137. Light, A.R. (1983) Spinal projections of physiologically identified axons from the rostral medulla of cats. Adv. Pain Res. Ther. 5, 373-379. Martin, L.J., Blackstone, C.D., Levey, A.I., Huganir R.L. and Price, D.L. (1993) AMPA glutamate receptor subunits are differentially distributed in rat brain. Neuroscience 53, 327-358. Martin, G.F., Cabana, T., Ditirro, F.J., Ho, R.H. and Hunbertson, A.O. (1982) Raphespinal projections in the North American opossum: evidence for connectionnal heterogeneity. J. Comp. Neurol. 208, 67-84. Merchenthaler, I., Stankovics, J. and Gallyas, F. (1989) A highly sensitive one-step method for silver intensification of the nickel-diaminobenzidine end product of peroxidase reaction. J. Histochem. Cytochem. 37, 1563-1565. Nagy, I., Woolf, C.J., Dray, A. and Urban, L. (1994) Cobalt accumulation in neurons expressing ionotropic excitatory amino acid receptors in young rat spinal cord: morphology and distribution. J. Comp. Neurol. 344, 321-335. Nahin, R.L., Madsen, A.M. and Giesler, G.J. (1983) Anatomical and physiological studies of the gray matter surrounding the spinal cord central canal. J. Comp. Neurol. 220, 321-335. Nakanishi, N., Axel., R. and Schneider, N.A. (1992) Alternative splicing generates functionally distinct N-methyl-D-aspartate receptors. Proc. Natl. Acad. Sci. USA 89, 8552-8556. Nasstr6m, J., Karlsson, U. and Post, C (1992) Antinociceptive actions of different classes of excitatory amino acid receptor antagonists in mice. Eur. J. Pharmacol. 212, 21-29. Neugebauer, V., Liicke, T. and Schaible, H.G. (1993) Differential effects of N-methyl-D-aspartate (NMDA) and non-NMDA receptor antagonists on the responses of rat spinal neurons with joint input. Neurosci. Lett. 155, 29-32. Petralia, R.S., Yokotani, N. and Wenthold, R.J. (1994) Light and electron microscope distribution of the NMDA receptor subunit NMDAR1 in the rat nervous system using a selective anti-peptide antibody. J. Neurosci. 14, 667-696. Raigorodsky, G. and Urca, G. (1990) Involvement of N-methyl-D-aspartate receptors in nociception and motor control in the spinal cord of the mouse: behavioral, pharmacological and electrophysiological evidence. Neuroscience 36, 601-610.

278

A. Bonnot et al./ Journal of Chemical Neuroanatomy I I (1996) 267-278

Rexed, B. (1954) A cytoarchitectonic atlas of the spinal cord in the cat. J. Comp. Neurol. 100, 297-380. Sanner, C.A,, Cunningham, T.J. and Goldberger, M.E. (1994) NMDA receptor blockade rescues Clarke's and red nucleus neurons after spinal hemisection. J. Neurosci. 14, 6472 6480. Shu, S., Ju, G. and Fan, L. (1988) The glucose oxidase-DAB-nickel method in peroxidase histochemistry of the nervous system. Neurosci. Lett. 85, 163-171. Siegel, S.J., Brose, N., Janssen, W., Gassic, G.P., Jahn, R., Heinemann, S.F. and Morrison, J.H. (1994) Regional, cellular and ultrastuctural distribution of N-methyl-D-aspartate receptor subunit 1 in monkey hippocampus. Proc. Natl. Acad. Sci. USA 91, 564 568. Siegel, S.J., Janssen, W.G., Tullai, J.W., Rogers, S.W., Moran, T., Heinemann, S.F. and Morrison, J.H. (1995) Distribution of the excitatory amino acid receptor subunits GIuR2(4) in monkey hippocampus and colocalization with subunits GluR5 7 and NMDARI. J. Neurosci. 15, 2707-2719. Smith, J.C. and Feldman, J.L. (1987) In vitro brainstem-spinal cord preparation for study of motor systems for mammalian respiration and locomotion. J. Neurosci. Methods 21, 321 333. Sommer, B. and Seeburg, P.H. (1992) Glutamate receptor channels: novel properties and new clones. Trends Pharmacol. Sci. 13, 291 - 296. Song, X.J. and Zhao, Z.Q. (1993) Differential effects of NMDA and non-NMDA receptor antagonists on spinal cutaneous vs. muscular nociception in the cat. Neuroreport 4, 17-20. Tachibana, M., Wenthold, R.J., Morioka, H. and Petralia, R.S.

(1994) Light and electron microscopic immunocytochemical localization of AMPA-selective glutamate receptors in the rat spinal cord. J. Comp. Neurol. 344, 431-454. T611e, T.R., Berhele, A., Ziegl/insberger, W., Seeburg, P.H. and Wisden, W. (1993) The differential expression of 16 NMDA and non-NMDA receptor subunits in the rat spinal cord and in periaqueductal gray. J. Neurosci. 13, 5009-5028. Trlle, T.R., Berthele, A., Laurie, D.J., Seeburg, P.H. and Zieglg/insberger, W. (1995) Cellular and subcellular distribution of NMDAR1 splice variant mRNA in the rat lumbar spinal cord. Eur. J. Neurosci. 7, 1235 1244. Turski, L., Bressler, K., Klockgether, T. and Stephens, D.N. (1990) Differential effects of the excitatory amino acid antagonists, 6cyano-7-nitroquinoxaline-2,3-dione (CNQX) and 3-(( _+)-2-carboxypiperazon-4-yl)propyl-l-phosphonic acid (CPP), on spinal reflex activity in mice. Neurosci. Lett. 113, 66-71. Viala, D., Buisseret-Delmas, C. and Portal, J.J. (1988) An attempt to localize the lumbar locomotor generator in the rabbit using 2-deoxy-[14C]glucose autoradiography. Neurosci. Lett. 86, 139 143. Viala, D., Viala, G. and Jordan, M, (1991) lnterneurons of the lumbar cord related to spontaneous locomotor activity in the rabbit. Exp. Brain Res. 84, 117--186. Walmsley, B. and Bolton, P.S. (1994) An in vivo pharmacological study of single group Ia fibre contacts with motoneurones in the cat spinal cord. J. Physiol. (Lond.) 481, 731 741. Wisden, W. and Seeburg, P.H. (1993) A complex mosaic of highaffinity kainate receptors in rat brain. J. Neurosci. 13, 3582-3597.