Using double-immunolabeling, we attempted to localize the glucocorticoid receptor (GR) and methionine-enkephalin-octapeptide (ENK-8) in the rat spinal cord. The 2nd to 8th cervical segments were analyzed in intact and in colchicine-injected animals. GR immunoreactivity was detected in the nuclei of neurons in all segments, and was most intense in small neurons in the superficial layers (laminae 1–3) of the dorsal horn. Medium-sized neurons in laminae 4–6 and the reticular formation, as well as the large ventral horn neurons, all showed moderate immunoreactivity. Double-immunolabeling revealed that, in all regions examined except the ventral horn, some of the GR-positive neurons were also ENK-8-positive. These findings suggest that enkephalinergic neurons in the spinal cord may be regulated by glucocorticoids, each of which is presumably related to the neuronal antinociceptive mechanism in the dorsal horn.
The receptors of adrenal steroids are heterogeneous and show a region-specific localization in the central nervous system (CNS) (Fuxe et al., 1985a; McEwen et al., 1986; Ahima and Harlan, 1990). Two receptor systems for glucocorticoids have been verified through the ligand binding property. Type I receptors have a high affinity to corticosterone; in contrast, type II receptors (the classical glucocorticoid receptors, GR) are characterized by a high affinity to dexamethasone. It has been demonstrated by immunohistochemistry that type I receptor immunoreactivity (IR) is localized in hippocampal pyramidal neurons and septal neurons (Ahima et al., 1991). GR-IR is distributed more widely in neurons of the hippocampus, hypothalamus, brainstem, and spinal cord (Fuxe et al., 1985a, b; Hisano et al., 1988). Immunohistochemical analysis of the spinal cord has shown that GR-IR is localized in motor and dorsal horn neurons (Ahima and Harlan, 1990). These findings suggest that adrenal steroids influence both the sensory and motor functions of the spinal cord (Hall, 1982). Physiological studies have also demonstrated steroidal action on spinal neurons (De Nicola et al., 1989).
Neuronal peptides in the spinal cord have been documented as intrinsic mediators of primary sensation and its modulation. Chao and McEwen (1990) revealed that preproenkephalin mRNA synthesis was regulated by glucocorticoid in rat strial neurons. Furthermore, Yin and Howells (1992) showed that dexamethasone has stimulatory effects on proenkephalin gene expression in cultured rat glioma cells. Of the peptides, opioids which include enkephalin are known to participate in antinociception. Immunohistochemical analysis of the opioid system has demonstrated that enkephalin-IR exists in dorsal horn neurons (Finley et al., 1981; Katoh et al., 1988). Therefore, the dorsal horn may be one of the sites for steroid-opioid interaction in the spinal cord. In the present study, we used double-immunolabeling to examine the neuro-anatomical relationship between GR, which is described as having a high affinity to dexamethasone-positive neurons, and enkephalinergic neurons in the rat spinal cord.
MATERIALS AND METHODS
Animals and fixation
Twenty adult male Wistar rats (250–350 g body weight) were used. Under chloral hydrate anesthesia (45 mg/100 g body weight), 10 animals received colchicine (100 μg/10 μl/rat) (Nakarai, Kyoto, Japan) by injection into the cisterna cerebellomedullaris 24 hr before sacrifice to facilitate the detection of enkephalinergic neurons. The intact and the colchicine-injected animals were fixed by transaortic perfusion with 0.1 M sodium cacodylate buffer (pH 7.4) followed by a fixative (4% paraformaldehyde, 0.15 % glutaraldehyde, and 0.2% picric acid in the same buffer). The 2nd to 8th cervical segments of the spinal cord were dissected out and immersed in fresh fixative overnight at 4°C. The tissues were cut into 30 μm-thick vibratome sections and stored in cold fixative (4% paraformaldehyde, and 0.2% picric acid in the same buffer) until immunostaining.
To immunostain GR, two anti-GR antibodies were used. The first was a mouse monoclonal antibody (BuGR-2) that recognizes the 17 amino acid sequence (407–423) near the DNA-binding portion of the rat liver GR (Gametchu and Harrison, 1984). The second was a polyclonal antibody (PolyGR) purchased from Affinity BioReagents (Golden, CO, USA). This antibody, raised in rabbits against a 22 amino acid synthetic peptide (amino acids 346–367 of the human GR), is reported to recognize, by Western Blot, a single 97 kDa band corresponding to the rat GR (Cidlowski et al., 1990). The immunohistochemical specificity of both antibodies has also been evaluated by other studies (Hisano et al., 1988; Ahima and Harlan, 1990; Cidlowski et al., 1990). For the present immunolabeling, the antibodies were diluted (1:500 for BuGR-2; 1:1,000 for PolyGR) with 0.02 M phosphate buffered saline (PBS, pH 7.4) containing 10% normal horse or goat serum (Cappel, West Chester, PA, USA) and 0.1% sodium azide. To immunostain enkephalinergic neurons, anti-methionine-enkephalin-octapeptide (ENK-8) antiserum (Peninsula Laboratories, Inc. San Carlos, CA, USA) was used. The specificity of the antiserum was characterized according to the same absorption test as previously described (Katoh et al., 1988). The antiserum was diluted (1:5,000) with PBS containing liver extract and a carrier protein for immunostaining.
Sections were first washed with PBS and immersed in 10% normal horse or goat serum in PBS for 60 min at 25°C to suppress background antibody binding onto the tissues. For single-immunolabeling of GR-IR, the sections were processed as follows: (a) BuGR-2 (60 hr, 4°C) or PolyGR (15 hr, 4°C); (b) biotinylated horse anti-mouse IgG (1:100, 2 hr, 32°C, Vectastain ABC Kit, Vector, Burlingame, CA, USA) for BuGR-2 or biotinylated goat anti-rabbit IgG (1:200, 2 hr, 32°C, Vector) for PolyGR; and (c) avidin-biotin-peroxidase complex (2 hr, 32°C, Vector). After each incubation with the immunoreagents, the sections were thoroughly washed with PBS. The immunoreaction was revealed with either diaminobenzidine tetrahydrochloride (DAB) (Hisano et al., 1988) or DAB-nickel ion solution (Ni-DAB, Hancock, 1984). For double-immunolabeling, only the sections from colchicine-injected animals were used.
After revealing GR-IR with Ni-DAB, the sections were further processed for immunolabeling of ENK-8 as follows: (a) anti-ENK-8 serum (36 hr, 4°C); (b) biotinylated goat anti-rabbit IgG (1:200, 2 hr, 32°C, Vector); and (c) avidin-biotin-peroxidase complex (2 hr, 32°C, Vectastain ABC Elite Kit, Vector). The immunoreaction for ENK-8 was revealed with DAB, so that GR-IR was labeled with Ni-DAB (dark blue) and ENK-8-IR with DAB (brown). The vibratome sections were not treated with hydrogen peroxide (which suppresses the reaction of endogenous peroxidase), to prevent interference with the immunore-activity of BuGR-2 and PolyGR. To serve as negative controls to assess the immunoreaction with BuGR-2 or PolyGR, sections were incubated with the diluents alone. These preparations showed no immunoreaction. In this study, the absorption control experiment was not performed because of the difficulty of obtaining the immunogens. The staining specificity was previously examined (Hisano et al., 1988; Ahima and Harlan, 1990; Cidlowski et al., 1990).
From each of 10 colchicine-injected rats, we sampled 10 sections of cervical spinal cord at random. The labeled neurons per section were counted. The mean and standard deviation (SD) were calculated for the number of GR-positive neurons and ENK-8-positive neurons for each section.
Localization of GR-immunoreactivity
Both antibodies (BuGR-2 and PolyGR) brought about almost the same results in the demonstration of GR-IR (Fig. 1a, b). At any location within the spinal cord, GR-IR was confined to the nuclei of positive neurons and glia-like cells. The immunoreaction was consistently slightly more intense in sections stained with PolyGR than in those stained with BuGR-2. By the Ni-intensification stronger signals were obtained than those with DAB alone. Positive signals detected by Ni-DAB were found as small particles in the cell nuclei, while such particles were unclear by DAB alone (Figs. 1c, 2, 3, 4, and 5).
In both intact and colchicine-injected animals, many GR-positive neurons were revealed throughout the cervical spinal cord. The distribution and intensity of the immunoreaction did not differ between the two groups. In the dorsal horn, a large population of small GR-positive neurons was revealed in laminae 1–3, chiefly in lamina 2 (Fig. 1). In laminae 4–6 and the reticular formation, medium-sized neurons were moderately, sometimes strongly, immunoreactive for GR. In lamina 9 of the ventral horn, nuclei of almost all large neurons (probably motor neurons) expressed moderate to strong GR-IR. In addition, the nuclei of glia-like cells were usually immuno-stained throughout the white matter and in the pia mater.
Double-immunolabeling for GR and ENK-8
We examined only colchicine-injected rats because it was easier to detect ENK-8-positive neurons in them. In the dorsal horn of all cervical segments, double-labeled neurons were frequently encountered in the dorsal horn (5.0±0.9 neurons, mean±SD per section), chiefly in lamina 2 (Fig. 2). ENK-8-positive fibers were also revealed. GR-positive neurons were 25.0±2.7 neurons, and ENK-8-positive neurons were 5.5±1.1 neurons in the dorsal horn per section.
Double-labeled neurons accounted for 20. 0% of GR-positive neurons and 91.0% of ENK-8-positive neurons. In laminae 4–6 (Fig. 3) and in the reticular formation (Fig. 4), a few medium-sized GR-positive neurons were immunoreactive for ENK-8. About a half of ENK-8-positive neurons in these regions were also positive for GR. In the ventral horn, all the GR-positive neurons (2.6±0.8 neurons per section) were large-sized and did not contain ENK-8 (Fig. 5).
The present study shows the co-localization of GR-IR and ENK-8-IR in neurons of the cervical spinal cord in the rat. Previous immunohistochemical studies have described GR-positive neurons in the superficial layers of the dorsal horn in the spinal cord (Fuxe et al., 1985b; Ahima and Harlan, 1990). Using BuGR-2, Ahima and Harlan (1990) demonstrated that GR-positive neurons are located in lamina 2 of the dorsal horn and in laminae 8 and 9 of the ventral horn at cervical, thoracic, and lumbar levels. Anti-GR antibody (PolyGR) and BuGR-2, each of which recognizes a different domain of the receptor protein, were used in the present study. The distribution of GR-positive neurons was identical for the two antibodies and consistent with previously reported data (Fuxe et al., 1985b; Ahima and Harlan, 1990). Therefore, the present study not only reconfirms the existence of glucocorticoid receptors in lamina 2, 8, and 9, but also documents their existence in lamina 1, 3, 4–8, the reticular formation, and the glia-like cells of the spinal cord.
Coexistence of GR and enkephalin has been shown in the basal ganglia and paraventricular nucleus of the hypothalamus (Cintra et al., 1991; Ceccatelli et al., 1989), and it was suggested that glucocorticoid may influence pain and the mood of the animals through the neurons in the hypothalamus. However, previous studies have not identified the neuroactive substances that GR-positive neurons produce in the spinal cord. The most salient result of the present study is that some of the GR-positive neurons were enkephalinergic at least in the cervical segments of the spinal cord.
The co-localization of GR-IR and ENK-8-IR was most frequently detected in neurons of the dorsal horn and sometimes in laminae 4–8 of the cervical spinal cord. This suggests that glucocorticoids regulate a certain function of the enkephalinergic neurons, probably at the genomic level. By a molecular biological approach in cultured rat glioma cells, Yin and Howells have recently reported that dexamethasone stimulates proenkephalin expression (1992). Furthermore, the rat proenkephalin gene has been reported to contain a nucleotide sequence highly homologous to a glucocorticoid response element (Beato, 1989; Muller and Renkawitz, 1991). Using the in situ hybridization technique, Chao and McEwen (1990) demonstrated that glucocorticoids regulate preproenkephalin mRNA synthesis in rat striatal neurons. In the sympathoadrenal system of the rat, dexamethasone produces effects at the level of enkephalin (Stachowiak et al., 1988). Increased glucocorticoid levels in response to stress lead to increased sympathetic enkephalin levels (Henion et al., 1993). Thus, glucocorticoids may also influence peptide production in enkephalinergic neurons of the spinal cord.
Enkephalinergic neurons in the dorsal horn are thought to mediate antinociception to noxious stimuli from peripheral regions (Hökfelt et al., 1977; Glazer and Basbaum, 1983). Therefore, glucocorticoids may influence this function of enkephalinergic neurons. Analgesia is known to be evoked under some stressful conditions. In stress-induced analgesia, glucocorticoids are considered critical factors in a neuronal analgesic pathway within the brain following activation of the hypothalamic-pituitary-adrenal axis (MacLennan et al., 1982; Ratka et al., 1988). At present, there is no neurophysiological evidence for a positive regulation by glucocorticoids of the release of enkephalins from dorsal horn neurons under stressful conditions. However, by taking into consideration the molecular biological evidence obtained by Yin and Howells (1992), the present data may suggest that the opioid interacts with the steroid, and the interaction relates to stress-induced analgesia in the spinal cord.
Recent immunohistochemical studies have described the co-localization of ENK-8 and substance P (Katoh et al., 1988) or gamma-aminobutyric acid (Todd et al., 1992) in dorsal horn neurons of the rat. Therefore, glucocorticoids may be involved in the synthesis of such neuroactive substances in enkephalinergic neurons. The results of the present double-immunolabeling analysis suggest the presence of GR-positive “non-enkephalinergic” neurons in the dorsal horn, though the technical limitation of immunohistochemistry should be kept in mind. This may indicate that glucocorticoids regulate neurons other than the enkephalinergic kind. After peripheral nerve stimulation, c-Fos-related proteins are expressed in dorsal horn neurons (Hunt et al., 1987), some of which are positive for dynorphin (Noguchi et al., 1991). Molecular biological studies on cultured cells have suggested a glucocorticoid-mediated down-regulation of c-Fos mRNA (Yin and Howells, 1992) and the negative influence of glucocorticoids on the expression of collagenase through an interaction of GR and the AP-1 complex (c-Fos/c-Jun) (Jonat et al., 1990). Thus, it may be possible that adrenal steroids affect the dorsal horn through various neuronal functions including those of ENK-8.
The authors sincerely thank Professor Yoshihiro Fukui, Department of Anatomy, School of Medicine, The University of Tokushima, and Professor Setsuji Hisano, Department of Anatomy, School of Medicine, The University of Tsukuba, for technical advice and helpful discussions. Thanks are also owed to Dr. R.W. Harrison (University of Arkansas for Medical Sciences) for his kindly providing BuGR-2.