THE JOURNAL OF COMPARATIVE NEUROLOGY 313~522-538 (1991)

Type I Corticosteroid Receptor-Like Immwnoreactivity in the Rat CNS: Distribution and Regulation by Corticosteroids REXFORD AHIMA, ZYGMUNT KROZOWSKI, AND RICHARD HARLAN Department of Anatomy, Tulane University School of Medicine, New Orleans, Louisiana 70112 (R.A., R.H.) and Baker Medical Institute, Prahan 3181, Australia (Z.K.)

ABSTRACT Previous maps of Type I corticosteroid receptor binding in the rat central nervous system (CNS) revealed a restricted distribution of the receptor in limbic regions, hypothalamus, and circumventricular organs. More recent studies have shown a more widespread expression of the receptor, with high levels of Type I receptor mRNA in limbic, motor, and sensory systems. We have used two antisera against peptide sequences derived from the cDNA of the human Type I corticosteroid receptor to map the regional distribution and corticosteroid regulation of the intracellular location of Type I corticosteroid receptor-like immunoreactivity (Type I-ir) in the rat CNS. Neurons showing Type I-ir were observed at all levels of the CNS. Highest densities of immunoreactive neurons were observed in limbic regions, isocortex, and some thalamic nuclei. Motor, sensory, and visceral systems often showed moderate densities of immunoreactive neurons. Type I-ir glia were observed in some fiber systems, e.g., corpus callosum, medial lemniscus, cerebral peduncles, spinal trigeminal tract, and funiculi of the spinal cord. In the majority of neurons and in glia, Type I-ir showed a diffusely nuclear and cytoplasmic location. Long-term adrenalectomy reduced immunoreactivity in most neurons and glia. Neuronal Type I-ir was localized mainly in the cytoplasm after long-term adrenalectomy. Nuclear immunoreactivity was retained in some neurons in the globus pallidus, motor trigeminal nucleus, and laminae 8 and 9 of the spinal cord. Acute treatment with corticosterone or aldosterone restored neuronal and glial Type I-ir to densities below that seen in intact rats. Key words: mineralocorticoid receptor, hippocampus, globus pallidus

Two types of corticosteroid receptors have been described in the rat CNS. The classical glucocorticoid receptor (GR) is expressed widely in CNS neurons and glia, binds selectively to dexamethasone and specific glucocorticoids, and has a low affinity and high capacity for corticosterone, the naturally occuring glucocorticoid in rats (McEwen et al., ’86; Funder and Sheppard, ’87; Gustafsson et al., ’87). GR is thought to mediate the effects of stress levels of corticosterone (DeKloet and Reul, ’87; Fuxe et al., ’87). The classical mineralocorticoid receptor (MR) binds to corticosterone and aldosterone, the principal glucocorticoid and mineralocorticoid, respectively, with a high and equivalent affinity in vitro (Beaumont and Fanestil, ’83; Krozowski and Funder, ’83). MR is largely selective for corticosterone in vivo (Arriza et al., ’88)and is thought to mediate the tonic effects of corticosterone on neuronal activity within its circadian levels, and some aspects of behavior (Bohus and De Kloet, ’81; DeKloet and Reul, ’87). In the hypothalamus and

o 1991 WILEY-LISS, INC.

circumventricular regions, MR is selective for aldosterone (McEwen et al., ’86a,b) and is thought to mediate the central effects of mineralocorticoids on salt intake (Magarinos et al., ’861, and blood pressure (Van de Berg et al., ’90). MR in the CNS is thus a glucocorticoid as well as a mineralocorticoid receptor. The term “Type I corticosteroid receptor” has been used to emphasize the dual role of MR in the CNS, and “Type I1 corticosteroid receptor” for GR (Funder and Sheppard, ’87; Arriza et al., ’88; Reul et al., ’88;Ratka et al., ’89). Circulating levels of corticosterone are 100-1,000 times higher than aldosterone. One would therefore expect Type I receptors to be fully saturated by corticosterone even under basal levels. In peripheral mineralocorticoid targets like the Accepted August 21: 1991 Address reprint requests to Richard E. Harlan, Dept. of Anatomy, Tulane University School of Medicine, 1430 Tulane Ave., New Orleans, LA 70112.

Abbreviations 1-9 3 7

10 12 Acb ACo AHP AHiPM AM AOB AOD AOL AOM AOP AOV APir APO APT Arc AV BL BLA BM BMA BST CA1-4 cc Ce CG cg CL c1 CLi CM Cp

CPU cu cu CXA DC DEn DG Dk DLG DM DPG DpMe DR ec ECIC EP EP1 FL fmi Frl Fr2 FStr G gcc Ge5 Gi G1 GP Gr Hb HDB HL IC ICj IGr InG Int IOC IOD IP Lat LaV LC LD LH LL LM

spinal cord laminae/ cerebellar lobules oculomotor nucleus facial nucleus dorsal vagal nucleus hypoglossal nucleus accumbens nucleus anterior cortical amygdaloid nucleus anterior hypothalamic area, posterior part amygdalohippocampalarea, posteromedial part anteromedial thalamic nucleus accessory olfactory bulb anterior olfactory nucleus, dorsal part anterior olfactory nucleus, lateral part anterior olfactory nucleus, medial part anterior olfactory nucleus, posterior part anterior olfactory nucleus, ventral part amygdalopiriformtransition anterior preoptic hypothalamic nucleus anterior pretectal nucleus arcuate hypothalamic nucleus anteroventral thalamic nucleus basolateral amygdaloid nucleus basolateral amygdaloid nucleus, anterior part basomedial amygdaloid nucleus, medial part basomedial amygdaloid nucleus, anterior part bed nucleus of stria terminalis fields of Ammon’s horn corpus callosurn central amygdaloid nucleus central gray cingulate cortex centrolateral thalamic nucleus claustrum caudal linear nucleus of the raphe centromedial nucleus of thalamus cerebral peduncle, basal part caudate-putamen cuneate tract cuneate nucleus cortex-amygdalatransition zone dorsal cochlear nucleus dorsal endopiriform nucleus dentate gyrus nucleus of Darkschewitsch dorsal lateral geniculate nucleus dorsomedial hypothalamic nucleus deep gray area of superior colliculus deep mesencephalic nucleus dorsal raphe nucleus external capsule external cortex of inferior colliculus endopiriform nucleus external plexiform layer forelimb area of cortex forceps minor of corpus callosum frontal cortex, area 1 frontal cortex, area 2 fundus striati gelatinous thalamic nucleus genu of corpus callosum gelatinous layer of caudal spinal trigerninal nucleus gigantocellular reticular nucleus glomerular layer globus pallidus gracile nucleus habenula nucleus of horizontal limb of the diagonal band hindlimb area of the cortex internal capsule islands of Calleja internal granular layer of the bulb intermediate gray-superior colliculus interposed cerebellar nucleus inferior olivary nucleus, central nucleus inferior olive, dorsal nucleus interpeduncular nucleus lateral cerebellar nucleus lateral amygdaloid nucleus, ventral part locus ceruleus dorsolateral thalamic nucleus lateral hypothalamic area lateral lemniscus lateral mammillary nucleus

lo LOT LP LPO LRt LSD LSI LSO LSV LVe MCPO MD MdD Med MeA MeP MG MGV Mi MiTg ML ml MM MnPO MnR Mo5 MPB MPO MS MVe ON Par1 Par2 PCRt PDTg Pe Pir PMCo Pn PnC PnO Po Pr5 PrH PT PY Re Rh RMg RN RPa RRF Rt RtTg SCh Scp

SFi

SI SNC SNR

so

Sol sp5c Sp5I Sp50 SpVe SP5 SuG TC TS TT Tu TZ VBD

vc

VL

VM VMH VP VPL VPM VTA ZI

lateral olfactory tract nucleus of the lateral olfactory tract lateral posterior thalamic nucleus lateral preoptic area lateral reticular nucleus lateral septal nucleus, dorsal part lateral septal nucleus, intermediate part lateral superior olive lateral septal nucleus, ventral part lateral vestibular nucleus magnocellular preoptic nucleus mediodorsal thalamic nucleus medullary reticular nucleus, dorsal part medial cerebellar nucleus medial amygdaloid nucleus, anterior part medial amygdaloid nucleus, posterior part medial geniculate nucleus medial geniculate nucleus, ventral part mitral cell layer of the olfactory bulb microcellular tegmental nucleus medial mammillav nucleus, lateral part medial lernniscus medial mammillary nucleus, medial part median preoptic nucleus median raphe nucleus motor trigeminal nucleus medial parabrachial nucleus medial preoptic nucleus medial septal nucleus medial vestibular nucleus olfactory nerve layer parietal cortex, area 1 parietal corticoamygdaloid nucleus parvicellular reticular nucleus posterodorsal tegmental nucleus periventricular hypothalamic nucleus piriform cortex posteromedial cortical amygdaloid nucleus pontine nuclei pontine reticular nucleus, caudal part pontine reticular nucleus, oral part posterior thalamic nuclear group principal sensory trigeminal nucleus prepositus hypoglossal nucleus paratenial thalamic nucleus pyramidal tract reuniens thalamic nucleus rhomboid thalamic nucleus raphe magnus nucleus red nucleus raphe pallidus nucleus retrorubral field reticular thalamic nucleus reticulotegmental nucleus of the pons suprachiasmatic hypothalamic nucleus superior cerebellar peduncle septofimbricalnucleus substantia innominata substantia nigra, compact part substantia nigra, reticular part supraoptic nucleus nucleus of solitary tract spinal trigeminal nucleus, dorsomedial part spinal trigeminal nucleus, pars interpolaris spinal trigeminal nucleus, oral part spinal vestibular nucleus spinal trigeminal tract superficial gray of superior colliculus tuber cinereum area triangular septal nucleus taenia tecta olfactory tubercle trapezoid nucleus nucleus of vertical limb of diagonal band ventral cochlear nucleus ventrolateral thalamic nucleus ventromedial thalamic nucleus ventromedial hypothalamic nucleus ventral pallidum ventroposterolateral thalamic nucleus ventroposteromedial thalamic nucleus ventral tegmental area zona incerta

R. AHIMA ET AL.

524

kidneys, salivary glands, and colon, selectivity for aldosterone is said to result from sequestration of corticosterone by high local levels of corticosteroid binding globulin (CBG), or inactivation by 11-beta hydroxysteroid dehydrogenase (McEwen et al., '86a; Funder and Sheppard, '87; Edwards et al., '88; Funder et al., '88; Rundle et al., '89). In the CNS, CBG levels are low (McEwen et al., '86a). Eleven-beta hydroxysteroid dehydrogenase has been localized in the brain (Krozowski et al., '90) and colocalized with Type I receptor (Sakai et al., 'go), and may confer Type I receptor selectivity for aldosterone. Previous reports based on binding of 3H aldosterone to cytosolic Type I receptors or steroid autoradiography in the rat brain suggested a restricted distribution of the receptor in the limbic brain ( R e d and De Kloet, '85; McEwen et al., '86a). Some binding was also reported in brainstem motor nuclei, reticular formation, and arachnoid (Ermisch and Ruhle, '78; Birmingham and Sar, '79), hypothalamus, and circumventricular organs (McEwen et al., '86b). The human renal Type I receptor cDNA has been cloned (Arriza et al., '87). It shares a high degree of homology with the rat brain Type I receptor cDNA (Arriza et al., '88). Contrary to results based on 3H aldosterone binding, Type I receptor mRNA (Type I mRNA) has a widespread neuronal distribution (Arriza et al., '88). Very little Type I mRNA was observed in the hypothalamus, and none in circumventricular regions. An anti-Type I receptor antiserum, MINREC2, was raised against a peptide sequence deduced from the "hinge" region of the human renal Type I receptor cDNA (Krozowski et al., '89). MINRECZ antiserum specifically stained distal tubular cells of the kidney. The pattern of immunostaining was similar to results obtained from an auto-anti idiotypic antibody (Lombes et al., '90). A second antiserum, MINREC4, was raised against a fusion protein, GTMR4, derived from an N-terminal peptide of the Type I receptor and glutathione S-transferase (GST). MINREC2 and MINREC4 antisera have the same specificity for renal Type I receptors. They have been used in limited studies on Type I receptors in the rat CNS (Bohn et al., '90; Chou et al., '90; Sakai et al., '90). The aims of the present study were to determine the distribution in the rat CNS of Type I receptor-like immunoreactive (Type I-ir) elements by using MINREC antisera and to compare the results to previous maps based on radioligand binding and Type I mRNA. Since steroid receptor immunoreactivity is dependent on the presence or absence of cognate hormones (Fuxe et al., '85; Cintra et al., '86; Gustafsson et al., '87; Blaustein and Turcotte, '89; DonCarlos et al., '89; Sar et al., 'go), we also examined the effect of adrenalectomy and corticosteroid treatment on Type I-ir.

MATERIALS AND METHODS Animals and tissue preparation Intact and 1- or 4-week adrenalectomized (a&) adult male Sprague-Dawley rats, weighing 250-300 g, were purchased from Charles River. They were kept under a 1Zh lightll2h dark cycle and allowed free access to normal rat chow. Intact rats were maintained on tap water ad libitum, and adx rats on normal saline. The adx rats were divided into three treatment groups (n = 4). One group of 1- and 4-week adx rats, respectively, received 1 mg1100 g bw corticosterone (CORT) in ZOO r*.l 0.005% ethanol-PBS in-

traperitoneally, a second group 10 pg/ 100 g bw aldosterone (ALDO)in the same vehicle, and a third group, vehicle only. All adx rats were treated 2 hours before sacrifice. The animals were anesthetized with 25 mgl100 g bw sodium pentobarbital injected intraperitoneally, perfused with phosphate-buffered saline (PBS) transcardially for 5 minutes, followed by phosphate-buffered 3% paraformaldehyde for 8 minutes. The abdominal cavities were explored to confirm adrenalectomy. Brains and spinal cords as well as parotid glands and segments of the distal colon (targets for mineralocorticoids) from all animals were dissected out, postfixed in 3% paraformaldehyde for 2 hours, cryoprotected with 30% sucrose, frozen on dry ice, and stored at - 70°C until use. Biopsies taken from the renal cortices and livers of intact rats after perfusion were postfixed in the fixative for 6 hours and processed subsequently €or embedding in paraplast as described previously (Krozowski et al., '89). The renal distal tubular epithelium and hepatocytes are classical sites of Type I and I1 corticosteroid receptors respectively (Funder and Sheppard, '87).

Immunocytochemistry Free-floating 50-pm-thick coronal sections of the brains and spinal cords were cut on a freezing microtome, washed in PBS for 1hour, and processed for immunocytochemistry with the Vectastain (ABC) peroxidase kit (Vector) according to specifications. Thirty micron thick Sections of the parotid gland and distal colon were cut on a cryostat and processed for immunocytochemistry. MINRECZ and MINREG4 antisera were used at dilutions of M O O , 1:1,000, 1:2,000,and 1:4,000. Sections were incubated with primary antisera for 48 hours at 4°C. The chromogens used were 3,3' diaminobenzidine (DAB) tetrahydrochloride or nickelDAB (Hancock, '84). Adjacent control sections in the above experiments were incubated with either normal rabbit serum, MINREC2 antiserum (1:1,000) preabsorbed with 10 FM MINRECZ peptide, MINREC4 antiserum preabsorbed with 10 KM GTMR4 fusion protein or 10 FM GST, or GST antiserum (1500, 1:1,000, 1:4,000). Preabsorption of antisera with antigen was carried out at 4°C for 24 hours. Cross-reactivity of MINREC antisera with the estrogen receptor (ER) was examined by preabsorbing 1 rnl of 1:1,000 dilutions of MINRECZ and MINREC4 with 100 nM ER obtained from partially purified MCF-7 cytosol. This quantity of ER is more than 10 times what is required to eliminate ER immunostaining in the guinea pig forebrain (DonCarlos et al., '91). For cross-reactivity of MINREC antisera with the Type I1 corticosteroid receptor, paraplast-embedded pieces of renal cortex and liver were sectioned at 7 pm, and processed for Type I-ir with MINREC antisera and preabsorption controls according to a method described by Krozowski et al. ('89). Adjacent control sections were processed for Type 11-ir with a 1:1,000 dilution of BUGR2 mouse antirat liver GR monoclonal antibody (Eisen et al., '85) or nonimmune P3 AgX- 653 myeloma cell supernatant. After immunostaining, a 1in 5 series of sections from the brains and spinal cords were counterstained lightly with 0.25% cresyl violet to facilite delineation of anatomical boundaries. Free-floating sections were mounted on chromalum-coated slides. All sections were dehydrated in 70%, 95%, and 100%ethanol, defatted with Histoclear (National Diagnostics), and coverslipped with Permount (Fisher Scientific).

CNS TYPE I RECEPTOR IMMUNOREACTIVITY

Fig. 1. MINREC4 immunostaining in the hippocampus. A. Note the high density of immunoreactive cells in the pyramidal cell layer (CA1-CM) and granule cells of the dentate gyrus (dg). The stratum oriens (so) and molecular layer (m) also show immunoreactive cells.

525

B. Preabsorption of MINREC4 antiserum with GTMR4 fusion protein abolished immunoreactivity. C: Preabsorption of MINREC4 antiserum with GST peptide did not alter immunoreactivity. Scale bar = 200 pm.

R. AHIMA ET AL.

526

Fig. 2. A. Caudate-putamen. Note the large number of medium-size immunoreactive cells. The large immunoreactive cell (arrow) shows diffuse immunoreactivity. B. Reticular thalamic nucleus. Note intensely immunoreactive cells. C. Subfornical organ. Note the large

number of immunoreactive cells. cf = column of fornix. D. Corpus callosum. Note intensely immunoreactive glia (arrows).Scale bar A and C = 100 pm; B and D = 50 pm.

Microscopy and data analysis

region from both hemispheres. Immunoreactivity and cresyl violet staining were confined to somata; therefore the distance of 150 pm between sections was considered sufficient to prevent double counting errors. The mean percentage Type I-ir cells per region was determined. Regions with 100% immunoreactive cells were rated “highest density”; greater than 70% but less than loo%, “high density”; 30-70%, “moderate density”; less than 30%,“low density.” Representative maps showing the distribution of Type I-ir cells were prepared by plotting cells at 40x onto camera lucida drawings (10 X ) matched with corresponding sections in the atlas of Paxinos and Watson (’86).

Sections were examined with a Nikon Optiphot microscope equipped with a camera lucida drawing tube, a 10 mm x 10mm ocular grid and Nikon FX35A camera. Photomicrographs were taken with Technical Pan (Kodak) film.

Distribution of Type I-ir in the CNS A 1in 5 series of cresyl violet-counterstained sections of brains and spinal cords (n = 4 per treatment group) were analyzed by two investigators. Sections were selected and matched with similar sections in the atlas of Paxinos and Watson (’86).In each hemisphere nuclear boundaries were traced out with the aid of the camera lucida tube at a magnification of x 100, and cells with immunostaining clearly above background staining within the depth of a section counted. The total number of cresyl violet stained cells within a boundary was also counted. Type I-ir cells per region were determined from pooled counts of Type I-ir and total cells from sections through the entire thickness of that

Intracellular location of Type I-ir, and regulation by corticosteroids Sections from the CNS, parotid gland, and distal colon from all treatment groups (n = 4)were analyzed blindly for intracellular location of Type I-ir by two investigators. Prior to the analysis slides were selected, matched, and

CNS TYPE I RECEPTOR IMMUNOREACTIVITY

527

Figs. 4-10. Maps of Type I-ir cells in the CNS of intact rats adapted from the atlas of Paxinos and Watson (’86). The distribution of immunoreactive cells in the isocortex and cerebellar cortex is illustrated in Figures 6B and 1OA.

coded by another investigator, who was also responsible for collating data and breaking the codes. Type I-ir in the CNS was described as “diffuse,” where nuclear staining could not be delineated from cytoplasmic staining, and somata showed little staining in dendrites; “predominantly nuclear,” where strong nuclear staining could be delineated from weaker cytoplasmic staining, and “cytoplasmic,” where immunoreactivity was confined to the cytoplasm, the nucleus was devoid of staining, and in some neurons, strong staining was present in dendrites. Type I-ir in control sections of the parotid gland and colon was also described as “difise,” “predominantly nuclear,” or “cytoplasmic.” Sections of the renal cortex and liver from intact rats were also analyzed for Type I-ir and Type 11-iraccordingto the above parameters. The effects of adrenalectomy and corticosteroid treatment on the densities of Type I-ir cells in selected regions of the CNS were analyzed semiquantitatively. The hippocampus, globus pallidus, reticular thalamic, and motor trigeminal nuclei were selected for analysis because the distribution of Type I receptor expressing cells is distinct from the closely related Type I1 receptor (Arriza et al., ’88b; Souza et al., ’89;see Table 4). Moreover, we have recently described in detail the effects of corticosteroids on the densities of

Figure 3

Fig. 3. Patterns of intracellular distribution of Type I-ir in intact rats. A. Gigantocellular reticular nucleus. Neurons that show diffuse cytoplasmic and nuclear immunoreactivity. Cytoplasmic immunoreactivity is granular or patchy. B.Lamina 9 cervical spinal cord. Motoneuron that shows predominant nuclear immunoreactivity. Note the absence of staining in the nucleolus. C . Cerebellum. Coronal section of cerebellar cortex that shows Purkinje cells (PI with intense diffuse immunoreactivity and weaker immunoreactivity in the granule cell layer (Gr). The molecular layer (Mj shows intensely immunoreactive elements, presumably stellate cells and astrocytes. Scale bar: A and B = 25 pm; C = 50 pm.

528

R. AHIMA ET AL.

Figure 5

Type 11-ir cells in the hippocampus (Ahima and Harlan, '911, thus enabling a comparison of corticosteroid regulation of the two receptors possible. Brain sections corresponding to Figure 11A-E were selected and matched from three animals in each treatment group. Cells showing cytoplasmic, diffuse, or nuclear Type I-ir were counted with the aid of ocular grids, 100 p,m x 100 pm positioned as illustrated (see Fig. 11)at x 200 magnification. In the hippocampus Type I-ir cells in the pyramidal cell layer of fields CA1 and CA3 of Ammon's horn, polymorphic layer (CA41, and dorsal and ventral blades of the granular layer of the dentate gyrus were analyzed in each hemisphere (see Fig. 11C-El. There was no significant difference in the density of immunoreactive cells per region at the three levels examined; neither did the densities of immunoreactive cells in one hemisphere differ significantly from the other in each animal. Counts of immunoreactive cells were therefore pooled per region in each animal. Immunoreactive cells in the globus pallidus and reticuIar thalamic nucleus were sampled as illustrated in Figure 11A,B. For the motor trigeminal nucleus two sections corresponding to Figures 54 and 57 of the atlas of Paxinos and Watson ('86) were selected from three animals in each treatment group. Two grids were sampled in each hemisphere and counts pooled for analysis. The density of Type I-ir cells per region were determined for each animal by dividing pooled counts by the number of grids sampled. Data from all treatment groups were analyzed using one-way analysis of variance (ANOVA) and posthoc comparisons between treatment groups carried out with Scheffe F-test.

Figure 6

RESULTS Specificity of MINREC antisera for Type I receptor In intact rats, both MINRECB and MINREC4 antisera showed the presence of Type I-ir in distal and collecting tubular cells of the kidney (micrographs not provided). Whereas most of these cells had predominantly nuclear Type I-ir, a few showed diffuse staining. Preincubation of MINRECB antiserum with MINRECB peptide and MINREC4 antiserum with GTMR4 fusion protein eliminated Type I-ir; preincubation with GST did not affect it. In contrast, there was no Type I-ir in hepatocytes that showed intense predominantly nuclear Type 11-ir with BUGRB antibody (micrographs not provided). Ductal epithelial cells of the parotid gland from intact rats, and surface columnar cells of the distal colon showed mainly apical cytoplasmic Type I-ir, even though a few cells showed nuclear immunoreactivity. Preabsorption with respective antigens eliminated immunostaining. In the CNS, the MINREC antisera produced widespread Type I-ir (Figs. l-lO).Dilution of antisera down to 1:2,000 eliminated specific signal. MINRECB antiserum produced high background signal in some regions, e.g., the striatum, making it difficult to resolve immunoreactive elements. One in 1,000 dilution of MINREC4 antiserum produced optimal signal/noise ratio. Preabsorption with GTMR4 fusion protein eliminated Type I-ir (Fig. lB), whereas

529

CNS TYPE I RECEPTOR IMMUNOREACTIVITY

CA3 and CA4 of Ammon’s horn (Figs. lA, 7A-8B), and the central amygdaloid nucleus (Fig. 7A,B). The glomerular and periglomerular layer of the olfactory bulb (Fig. 4A), medial and dorsal divisions of the anterior olfactory nucleus (Fig. 4B), endopiriform nucleus (Figs. 6A-7B), layer I11 of the piriforrn cortex (Figs. 5A-7B), islands of Calleja (Fig. 5A), amygdalopiriform transition (Fig. 8A), amygdalohippocampal transition (Fig. 8A), and the corticoamygdaloid, posterolateral, and lateral amygdaloid nuclei (Figs. 7A-8A) all showed moderate densities of immunoreactive neurons. All divisions of the bed nucleus of the stria terminalis (Fig. 6A), globus pallidus (Figs. 6B-7B), entopeduncular nucleus (Fig. 7A,B), ventral pallidum (Fig. 5B, 6A), substantia innominata (Fig. 6B), claustrum (Figs. 5A-6B), medial septum (Fig. 5B), septohippocampus, septofimbrical and triangular septal nuclei (Fig.6A,B),horizontal limb of the nucleus of the diagonal band (Fig. 6A,B), layers I11 and IV of the frontoparietal cortex (Fig. 6B), and the subiculum (Fig. 8A) also showed moderate densities of immunoreactive neur ons. Low densities of immunoreactive neurons were observed in layer I of the piriform (Figs. 5A-7B) and frontoparietal cortices (Fig. 6B), stratum oriens and molecular layers of the hippocampus (Figs. 7B, 8A), anterior amygdaloid area, basal amygdaloid nucleus (Fig. 7A,B), nucleus accumbens (Fig. 5A,B), fundus striati (Fig. 6A), and substriatal area and the basal nucleus of Meynert (Fig. 6A,B).

Diencephalon

Distribution of Type I-ir cells in the CNS

In the thalamus, the anterior, ventromedial, ventrolateral, ventroposterior, dorsolateral, paracentral, lateral and medial geniculate, reticular and subthalamic nuclei (Figs. ZB, 6B-8B) had high densities of immunoreactive neurons. The paraventricular, reuniens and gelatinous thalamic nuclei, and the habenula (Fig. 7A,B) all showed low densities of immunoreactive neurons. All other regions of the thalamus showed moderate densities of immunoreactivity. In the hypothalamus, the medial preoptic nucleus (Fig. 6B) and anterior ventromedial nucleus (Fig. 7A) showed high densities of immunoreactivity. Low densities were observed in the anterior hypothalamic area (Fig. 7A), suprachiasmatic (Fig. 6B), posterior paraventricular, central ventromedial (Fig. 7B), dorsomedial (Fig. 7B), and mammillary and supramammillary nuclei (Fig. 8A). All other hypothalamic nuclei showed moderate densities of immunoreactive neurons (Figs. 6A-8A).

Figures 1-10 illustrate the distribution of immunoreactive cells in the CNS of intact rats.

Brainstem

Telencephalon

There was a widespread distribution of Type I-ir cells in the brainstem. Moderate densities of immunoreactive neurons were observed in sensory nuclei, e.g., the superficial and intermediate gray of the superior colliculus (Fig. 8B), vestibular nuclei (Fig. lOA), principal sensory trigeminal nucleus (Fig. lOA), superior olivary nucleus (Fig. 9B), and the cuneate and gracile nuclei (Fig. 10B). Motor-related nuclei showed low to moderate densities of immunoreactive neurons. For example, the oculomotor (Fig. 8B), motor trigeminal (Fig. 9B), facial (Fig. lOA), and hypoglossal nuclei (Fig. 10B) had moderate densities of immunoreactive neurons. Nuclei involved in motor coordination such as the red nucleus (Fig. 8B) and pontine (Fig. 9A) nuclei had low to moderate densities of immunoreactive neurons, and the inferior olivary subnuclei (Fig. lOB), moderate densities. Visceral (autonomic) nuclei such as the Edinger Westphal nucleus (Fig. 8B), nucleus ambiguus, nucleus of

Figure 7

preabsorption with either GST (Fig. 1C) or MCF-7 cytosol containing 10-fold excess of ER normally required for preabsorption controls (DonCarlos et al., ’911, did not alter Type I-ir. Sections incubated in normal rabbit serum or GST antiserum did not produce specific staining.

The highest densities of Type I-ir neurons were observed in the pyramidal cells of fields CA1 and CA2 of Ammon’s horn (Fig. lA), the granule cell layer of the dentate gyrus (Figs. lA, 7B, 8A), and layer I1 of the piriform cortex (Figs. 5A-7B). All neurons in these regions were immunoreactive. High densities of immunoreactive neurons were observed in the granule cell layer of the olfactory bulb (Fig. 4A), lateral and ventral parts of the anterior olfactory nucleus (Fig. 4B), nucleus of the lateral olfactory tract (Fig. 6B), bed nucleus of the accessory olfactory tract, medial amygdaloid nucleus (Fig. 7A,B), caudate-putamen (Figs. ZA, 5B-7B1, lateral dorsal and intermediate septum (Figs. 5B, 6A), vertical limb of the diagonal band of Broca (Fig. 5B), layers 11, V, and VI of the frontoparietal (motor and sensory) and cingulate cortices (Fig. 6B), the pyramidal cell layer of fields

R. AHIMA ET AL.

530

Figure 8

solitary tract (Fig. lOB), and the dorsal vagal nucleus (Fig. lOB), showed moderate densities of immunoreactive neurons. Scattered immunoreactive neurons were observed all over the brainstem reticular formation (Figs. 8B-10B). The lateral and dorsal central gray had moderate densities of immunoreactive neurons, whereas the medial parts had very few (Figs. 8A,B, 9A).

Cerebellum and spinal cord The cerebellar cortex showed moderate densities of small Type I-ir neurons, presumably granule cells (Figs. 3C, 10A); occasional immunoreactive medium-size neurons were also observed in the granule cell layer and were presumed to be Golgi cells. There were moderate densities of immunoreactive Purkinje cells (Figs. 3C, lOA), whereas the molecular layer showed low to moderate densities of immunoreactive elements, presumably stellate cells and astrocytes (Figs. 3C, 10A).All deep cerebellar nuclei showed low to moderate densities of immunoreactive neurons (Fig. 10A). In the spinal cord, high densities of immunoreactive neurons were observed in the substantia gelatinosa and nucleus proprius, and moderate densities in the intermediolateral horn and laminae 8 and 9 of the ventral horn (Figs. 3B, 1OC). There were scattered immunoreactive neurons over the rest of the gray matter.

Glia and cireumventricularregions Low or moderate densities of small spindle or ovoid immunoreactive cells were observed in some fiber systems,

Figure 9

e.g., corpus callosum (Figs. 2D, 5A-9A), stria medullaris (Fig. 6B), medial lemniscus (Fig. 8A,B), medullary pyramids (Figs. 9B-l0B), cerebral peduncles (Fig. 8A,B), and spinal cord funiculi (Fig. 1OC). These cells were presumed to be oligodendroglia. In the circumventricular regions, immunoreactive cells were clearly observed in the subfornical organ (Fig. 2C). Immunoreactivity in other circumventricular organs was difficult to interpret, due to high levels of non-specific staining.

Regulation of Type I-ir by cortieosteroids Three patterns of intracellular distribution of Type I-ir were observed in the CNS of intact rats: (1) “diffuse” nuclear and cytoplasmic, (2) predominantly nuclear, and (3) predominantly cytoplasmic. Most immunoreactive neurons showed “diffuse” nuclear and cytoplasmic immunoreactivity (Fig. 3A,C). Some neurons, notably in the olfactory tubercle, piriform cortex, subiculum, globus pallidus, motor trigeminal nucleus, substantia gelatinosa, and laminae 8 and 9 of the spinal cord (Fig. 3B), however, showed a predominant nuclear immunoreactivity. These neurons typically showed an absence of immunoreactivity in the nucleolus (Fig. 3B). The caudate-putamen showed a large population of small and medium-sizeneurons with predominantly nuclear Type I-ir, and a few large neurons with either predominantly nuclear or diffuse immunoreactivity (Fig. 2A). Scattered neurons with cytoplasmic or diffuse immunoreactivity were observed in the brainstem reticular

531

CNS TYPE I RECEPTOR IMMUNOREACTIVITY

C

B

Figure 10

formation (Fig. 3A), magnocellular preoptic hypothalamic nucleus, and the nucleus of the horizontal limb of the diagonal band. Type Lir, where diffuse or cytoplasmic was sometimes patchy (Fig. 3A). Glial Type I-ir was often diffuse (Fig. 2D). There was no significant change in the intensity and subcellular distribution of immunoreactivity in neurons and glia after 1 week of adx in most brain regions. After 4 weeks of a&, glial immunoreactivity was barely resolvable. The intensity of immunoreactivity was markedly reduced in neurons showing diffuse or cytoplasmic Type I-ir in intact animals (compare A and B in Fig. 12). The few remaining neurons showed mainly cytoplasmic immunoreactivity. In regions like the globus pallidus, motor trigeminal nucleus, and ventral horn of the spinal cord where neurons with nuclear Type I-ir were observed in intact rats, nuclear or diffuse Type I-ir was retained in some neurons after 4 weeks of a h . Two hours treatment with CORT restored the intensity and subcellular distribution of neuronal and glial Type I-ir to patterns observed in intact rats (Fig. 12A,B,C); however, the density of labeled neurons remained below intact levels (see below). Two hours treatment with ALDO increased the intensity of Type I-ir above 4 week adx levels (Fig. 12B,D), but less than CORT-induced Type I-ir (Fig. 12C,D). In the control experiments, Type I-ir in ductal epithelial cells of the parotid gland, and surface columnar cells of the distal colon was abolished following 1 week of adx. Two hours treatment with CORT or ALDO increased immunoreactivity to near intact levels (photomicrographs not

shown). The pattern and intensity of immunoreactivity produced by treatment with the two hormones were identical.

Regulation of the density of Type I-ir cells in selected CNS regions by corticosteroids Table 4 illustrates the effects of adx and corticosteroid treatment on the density of Type I-ir cells in the hippocampus, globus pallidus, reticular thalamic, and motor trigeminal nuclei. In the hippocampus, the density of Type I-ir cells in the pyramidal layer of field CA1, the polymorphic layer (CA4), and the granular layer was not affected by 1week of a&. CA3, however, showed a significant reduction in density of immunoreactive cells. After 4 weeks of adx, all regions showed a significant decrease in densities of immunoreactive cells. Treatment with CORT increased the density of immunoreactive cells above 4-week adx but below intact levels in all regions except CA1. Treatment with ALDO produced a similar effect; however, the increase was less than that produced by CORT. In the globus pallidus there was a significant decrease in the density of immunoreactive cells 1week after adx. CORT treatment restored Type I-ir to intact levels, whereas ALDO treatment produced no effect. The response of the reticular thalamic nucleus to adx and corticosteroid treatment was similar to the globus pallidus; however, the increase in densities induced by CORT and ALDO were not different. The motor trigeminal nucleus did not show any significant response to adx or corticosteroid treatment.

R. AHIMA ET AL.

532

n Fig. 11. Drawings show positions of 100 km X 100 Fm grids used in sampling Type I-ir cells in the globus pallidus (A, B)and hippocampus (C-El. Bregma in A = - 1.30 mm, B = -2.12 mm, C = -2.56 mm, D = -3.60 mm, and E = -5.20 mm.

533

CNS TYPE I RECEPTOR IMMUNOREACTIVITY

Fig. 12. Regulation of hippocampal (CA1) Type I-ir by corticosteroids. A. Intact rat. Note the high density of cells with diffuse Type I-ir in the pyramidal cell layer. Immunoreactive elements are present in the stratum oriens (so) and molecular layer (m). B. Four week adx. Note the reduction in the density of immunoreactive cells. The few immuno-

reactive pyramidal cells show cytoplasmic Type I-ir (arrow). C. Four weekadx + 2h CORT. Type I-ir has been restored in most cells. D. Four week adx 2h ALDO. Type I-ir has been restored in most cells but the intensity is weak (compare to B and C). Scale bar = 50 pm.

DISCUSSION Specificity of MINREC antisera

receptors (Arriza et al., '87; Evans et al., '88). The distribution of Type I1 receptor (Fuxe et al., '85; Sousa et al., '89; Ahima et al., '901, ER (Pfaff et al., '73; Cintra et al., '86; DonCarlos et al., '91), AR (Sar and Stumpf, '74; Simerly et al., 'go), PR (Blaustein and Turcotte, '891, and Vitamin D receptors (Stumpf and O'Brien, '87) in the CNS differ considerably from Type I receptors (see Tables 1-3). For instance, the posterior magnocellular division of the hypothalamic paraventricular nucleus contains a high density of neurons with AR mRNA, but no neurons with Type I-ir (Table 2); the ventrolateral division of the hypothalamic ventromedial nucleus and the arcuate nucleus have much higher densities of neurons with ER mRNA than with Type I-ir (Table 2). It is unlikely that MINREC antisera recognize these steroid receptors. Moreover, we have demonstrated regulation of Type I-ir by aldosterone and corticosterone, its cognate hormones, in both the CNS and peripheral mineralocorticoid targets.

Our experimental controls support the specificity of MINREC antisera for Type I corticosteroid receptors in the CNS. MINREC immunostaining was competed out by corresponding antigens. Excess ER could not compete out MINREC staining. MINREC antisera were able to distinguish between MR (Type I receptor) and GR (Type I1 receptor) in their classical tissue sites, i.e., the distal and collecting tubular epithelium of the kidney, and hepatocytes, respectively. These two corticosteroid receptors share the greatest degree of homology among the family of steroid receptors (Arriza et al., '87). In the hinge region and N-terminal domain of the Type I receptor against which MINRECB and MINREC4 antisera were generated, there is less than 20% homology in amino acid sequence between the Type I receptor, Type I1 receptor, and progesterone receptor (PR), and even less for other steroid hormone

+

R. AHIMA ET AL.

534

TABLE 1. Distribution of Corticosteroid, Androgen, and Estrogen Receptors in Selected Regions of the Rat Telencephalon TVpe I-ir Hippocampus Pyramidal layer CA1 CA2 CA3 Dentate gyrus Granular layer Polymorphic"

++++ +++ +++ ++++ +++

Type ImRNA (Arriza et al., '88)

Type 11-ir (Ahima and Harlan, '90)

+++ (+)

+++

+++ +++ +++

?

++++ ++++ +++

ARmRNA (Simerlv et al..'90)

ERmRNA (Simerlv et al.. '90)

+++

(+I

+

+ + (+) ++ (+)

+ ?

?

+

0

I

Septum Lateral dorsal Intermediate Ventral Medial Subfornical organ Caudate-putamen Globus pallidus

+++ ++++ +++ ++ 0 +++ ++ 0 ++ i+l ++ ++ ++ +++(+I +++ ++ ++ + + + +++ 0 0 ++f ++(+I +++ ? +++ ? ? ++ ? 0 ? For Type I-ir and Type 11-ir, + + + + = all cells are immunoreactive; + + + = more than 7 0 6 but less than 100% immunoreactive cells; + + = 30-708 immunoreactive cells; + = less than 30% immunoreactive cells; 0 = no immunoreactive cells. For Type ImRNA, ARmRNA, and ERmRNA, + + + + = very strong signal; + + + = strong signal, + + = moderate signal; + = weak signal, 0 = no signal; ? = not reported; + + + (+) = between + + + + and ++ +. ~

~~

TABLE 2. Distribution of Corticosteroid, Androgen and Estrogen Receptors in Selected Regions of the Rat Diencephalon Type I-ir Thalamus Lateral dorsal Paraventricular Ventral anterior Ventral lateral Ventral posterolateral Ventral posteromedial Medial geniculate Centrolateral Reticular Hypothalamus Suprachiasmatic Supraoptic Paraventricular: Posterior magnocellular Medial parvicellular dorsal Arcuate Anterior Ventromedial: ventrolateral Tuberomammillary Lateral area

+++ +

++ +++ ++

+++ +++ ++ +++

+ + 0

+or++

++ ++

++ ++ ++

Type ImRNA (Arriza et al., '88)

Type 11-ir (Ahima and Harlan, '90)

ARmRNA (Simerly et al., '90)

+ ? + ? ++

+++ ++

+

?

+++ +

++

++

++ ++ ++ ++

+++

0 or +

0 0 0

0 0

0 0 0 0

++ +

++ 0

0

++ +++ +or++

+

ERmRNA (Simerly et al., '90) 0

+ +

+ + (+)

0

+

+ + + ++ + i+)

++ + +++ 0

+++ + +++ ++ +++

0 0 0 0 0

0

++ 0 0

++ + ++++ 0 + (+)

+++(+I

See Table 1 for legend.

Distribution of Tme I-ir in the CNS Type I-ir neurons were observed at all levels of the CNS in correspondence to the widespread expression of the receptor mRNA (Arriza et al., '881, and contrary to the limited binding sites reported earlier (Reul and DeKloet, '85; McEwen et al., '86a). Major differences between Type I-ir and Type I mRNA were, however, observed in the hypothalamus, circumventricular organs, and glia. Moderate densities of Type I-ir neurons were observed in the hypothalamus, subfornical organ, and some fiber tracts. In contrast Type I mRNA is generally low in the hypothalamus and absent in circumventricular organs and glia (Arriza et al., '88). Whereas these differences may result from the sensitivities of the methods used, it is significant that aldosterone binding has been reported in the hypothalamus, circumventricular organs (McEwen et al., '86a,b), and C6 glioma cells (Beaumont, '851, which implies that Type I receptors in these regions are functional. Low or absent Type I mRNA could result from either low transcription rates or high turnover of Type I mRNA. Alternatively, local factors in the hypothalamus, circumventricular regions, and fiber tracts could stabilize and prolong the half

life of Type I receptors, and thus account for binding and Type I-ir. The widespread distribution of Type I mRNA and Type I-ir compared to aldosterone binding, raises questions concerning the functional status of Type I receptors outside the limbic brain, the hypothalamus, and circumventricular organs. Since the Type I receptor is the high affinity glucocorticoid receptor, it is logical that to be able to mediate the diverse actions of glucocorticoids on ongoing CNS neuronal and glial functions (McEwen et al., '86a), the receptor would be expressed at all levels of the CNS, as is the case for the low affinity Type I1 receptor. However, whereas widespread binding of specific glucocorticoids to the Type I1 receptor corresponds to Type 11-ir and Type I1 mRNA to a large extent (Fuxe et al., '85; McEwen et al., '86a; Souza et al., '89; Ahima and Harlan, '901, implying a widespread role for this low affinity receptor in mediating glucocorticoid effects in neurons and glia during stress, or in response to pharmacological doses of glucocorticoids, the relatively limited distribution of Type I receptor binding sites is suggestive of a limited function for this receptor. Why would so many neurons express the Type I receptor

CNS TYPE I RECEPTOR IMMUNOREACTIVITY

535

TABLE 3. Distribution of Corticosteroid, Androgen, and Estrogen Receptors in the Rat Mes- and RhombenceDhalon Type ImRNA (Arriza et al., '88)

Type I-ir Brainstem Medial vestibular Nucleus of the solitary tract (medial) Trigeminal: motor Trigeminal: mesencephalic Nucleus ambiguus Hypoglossal Inferior olive Red nucleus Substantia nigra: Pars compacta Pars reticulata Cerebellum Deep nuclei Purkinje layer Granular layer

++

++

++ ++ ++ ++ ++ ++ ++ + ++

++

Spinal Cord Substantia gelatinosa Intermediolateral horn Ventral horn

+++ ++ ++

ARmRNA (Simerly et al., '90)

+

++ (+)

+++ +++ +++ +++ +++ +++i ++ +++

++ ( + I + (+)

?

++(+I

++

0

++(+I

+ +

0

++

++ ( + I

++(+I

0 0 0 0

++(+I + (+I

0

+ (+I

++

0 (+I 0

0

++

?

0 ?

+++ ++ +

+ (+)

++ +++

+++

+++ +++

+++ ++

++ ++

+ (+I

+++-I

ERmRNA (Simerly et al., '90)

++ +++ ++

++ (+)

++

+ (+)

Type 11-ir (A hma and Harlan, '90)

0

++ 0 0

++(+I

See Table 1for legend.

TABLE 4. Changes in the Densities of Type I-ir Cells in Selected Regions of the Rat Brain in Response to Adrenalectomy and Corticosteroid Treatment Region Hippocampus CA1 CA3 CA4 DG Globus pallidus Reticular thalamic n Motor Trigeminal n

Intact

Adxl

Adx4

65.3 2 2.0 42.9 t- 1.5 24.8 t- 1.5 46.6 It 1.3 8.5 2 0.2 10.4 t 0.8 3.9 t 0.4

56.8 t 0.5 33.8 t- 1.3* 24.8 t- 0.7 44.9 t 0.3 7.3 t 0.2* 8.6 t 0.1 3.9 t 0.2

38.5 t- 5.4* 12.8 t- 0.6* 2.1 k 0.3* 13.2 t 0.2* 6.7 t- 0.1* 4.8 t 0.2* 3.4 t 0.3

Adx4

+ CORT

45.8 t 1.9' 32.1 5 1.0'8 20.5 2 0.1*§ 41.3 t 0.5*§ 7.9 t- 0.18 9.7 t- 0.58 3.6 t- 0.2

Adx4

+ ALDO

36.0 t 0.8* 23.0 t 0.8*8t 11.7 t 0.8*81 31.6 t- 1.3*5t 6.6 2 0.3*1 8.2 t 0.55 3.2 t- 0.2

F,, 1")

D

20.6 108.8 156.7 254.1 15.6 19.5 1.1