Rapid corticosterone-induced changes in gene expression in rat hippocampus display type II glucocorticoid receptor specificity.

Corticosteroids influence a wide range of neuronal activities by binding to either of two different glucocorticoid receptors found in rat brain. To investigate genomic responses in brain to stress levels of circulating corticosterone (CORT), we isolated hippocampal total RNA and poly(A)-containing RNA from rats treated with 10 mg/day CORT or vehicle. RNA translation products were resolved by 2-dimensional gel electrophoresis and fluorography. Select changes in four translation products after acute CORT treatment were inferred from up to 100-fold increases in three polypeptides and a 2-fold decrease in another. While adrenalectomy decreased levels of the inducible RNA sequences (adrenalectomized vs. intact controls), CORT increased the inducible sequences above their levels in intact controls. Rapid increases within 2 h of CORT treatment were seen for RNAs coding for 35, 33, and 20 kilodalton polypeptides. However, RNA coding for a 50 kilodalton polypeptide had a delayed decrease, first seen after 32 h CORT. The CORT increases displayed type II glucocorticoid receptor-specificity: RU 28362 greater than or equal to CORT greater than aldosterone greater than dihydrotestosterone = control. Since type II receptors are only substantially occupied by stress levels of CORT, these changes in gene expression are candidates for molecular stress responses in the brain.

Steroid effects on protein synthesis in cultured smooth muscle cells from rat aorta.

To examine the direct effects of steroids on vascular smooth muscle, we have incubated rat aortic vascular smooth muscle cells in culture either steroid-free, or with natural and synthetic corticosteroids (RU26988, dexamethasone, corticosterone, 9 alpha-fluorocortisol, aldosterone, deoxycorticosterone) or sex steroids (estradiol, 5 alpha-dihydrotestosterone). At the end of 24 h, cultures were pulsed with [35S]methionine for 2 h, the cells lysed, and patterns of incorporation of isotope into protein determined by two-dimensional gel electrophoresis and autoradiography. Neither estradiol nor 5 alpha-dihydrotestosterone altered protein synthetic profiles compared with control (steroid-free) incubations. In contrast, cultures exposed to the six corticosteroids at 10(-7) M showed an identical pattern of response (6 proteins increased, 6 proteins decreased). This response appears to be glucocorticoid specific, since the mineralocorticoids (9 alpha-fluorocortisol, aldosterone, and deoxycorticosterone) did not have any effects over and above those seen with the pure glucocorticoid RU26988. We interpret these data as evidence for a putative glucocorticoid domain of at least 12 proteins in rat vascular smooth muscle cells. In contrast, there appear to be no comparable estrogen-, androgen-, or mineralocorticoid-specific changes in these cells.

Transcriptional control of glial fibrillary acidic protein and glutamine synthetase in vivo shows opposite responses to corticosterone in the hippocampus.

Transcriptional regulation of two astrocyte genes, glial fibrillary acidic protein (GFAP) and glutamine synthase (GS), by glucocorticoids was determined by nuclear run-on assay with hippocampal tissues from adult male F344 rats. Transcriptional responses of GFAP to corticosterone were slower than those observed for GS, but were more sensitive to changes in plasma corticosterone. The transcription of GFAP did not change 2 h after the injection of 10 mg corticosterone, but was reduced by 50% at 6 and 24 h. In contrast, corticosterone increased GS transcription at 2 and 6 h. Seven days after adrenalectomy, GFAP, but not GS, transcription was increased. Corticosterone replacement (200 micrograms/ml in the drinking water) suppressed GFAP, but did not increase GS transcription in adrenalectomized rats. Therefore, GFAP transcription is more sensitive to low physiological levels of corticosterone than transcription of GS. The slower response of GFAP than GS to corticosterone suggests that glucocorticoids may have indirect effects on GFAP expression that require additional transcriptional regulators besides the glucocorticoid receptor.

Glial fibrillary acidic protein: regulation by hormones, cytokines, and growth factors.

Levels of glial fibrillary acidic protein (GFAP), an astrocyte-specific intermediate filament protein, are altered during development and aging, GFAP also responds dynamically to neurodegenerative lesions. Changes in GFAP expression can occur at both transcriptional and translational levels. Modulators of GFAP expression include steroids, cytokines, and growth factors. GFAP expression also shows brain region-specific responses to sex steroids and of astrocyte-neuronal interactions. The 5'-upstream sequences of rat, mouse, and human are compared for the presence of response elements that are candidates for transcriptional regulation of GFAP. We propose that the regulation of the GFAP gene has evolved a system of controls that allow integrated responses to neuroendocrine and inflammatory modulators.

Aldosterone binding sites in aortic cell cultures from spontaneously hypertensive rats.

Spontaneously hypertensive rats and rats made hypertensive by deoxycorticosterone-salt treatment have in common increased Na+ and K+ permeability and transport in their aortic cells. These changes may be important factors in the development of the hypertensive state and may be mediated by mineralocorticoid binding to intracellular sites in the aorta. Therefore, we examined 3H-aldosterone binding in aortic cell cultures from spontaneously hypertensive rats and normotensive Wistar-Kyoto rats. Vascular corticoid binding sites in the two strains were compared by Scatchard analysis of Kd and Bmax, pH and temperature stability, and subcellular binding. By all of these criteria normotensive rats. These results indicate that the underlying genetic defect in spontaneous hypertension is not an intrinsic cellular defect which alters mineralocorticoid binding in the aorta.

Food restriction delays the age-related increase in GFAP mRNA in rat hypothalamus.

Astrogliosis with advancing age is correlated with increased expression of glial fibrillary acidic protein (GFAP). Hypothalamic GFAP mRNA prevalence was determined in male F344 rats of different ages that were fed ad lib (AL) and compared with that of rats that were food-restricted (FR) to 60% of AL levels. Hypothalamic GFAP mRNA increased 3-fold at 24 to 25 months in AL rats compared with 3 and 6 month groups. There were no differences in GFAP mRNA levels between AL and FR rats from 3 to 18 months. However, GFAP mRNA was significantly lower in FR than in AL rats at 24 to 25 months; FR rats reached the level of GFAP mRNA in 24 to 25 months AL rats by 33 months. Hypothalamic glutamine synthetase mRNA also increased with age in both dietary groups but did not differ between dietary groups at any age. The observation that FR delays the increased expression of GFAP in the hypothalamus during aging lends support to the hypothesis that upregulation of GFAP mRNA is a biomarker of brain aging.

GFAP mRNA increases with age in rat and human brain.

Glial fibrillary acidic protein (GFAP) mRNA was examined by RNA blot hybridization in three age groups of two cohorts of male F-344 rats and in 47 human postmortem brain samples. GFAP mRNA increased in the hippocampus and striatum of 24 versus 6- to 7-month-old rats. Another astrocytic molecular marker, glutamine synthetase mRNA, did not change with age in rat brain. Rat GFAP mRNA prevalence was inversely correlated with serum testosterone but not correlated with serum corticosterone. In human hippocampus, frontal and temporal cortex, GFAP mRNA also increased in older (60-79 years) compared with middle-aged (25-59 years) individuals. In contrast, mitochondrial cytochrome oxidase subunit 1 mRNA did not change between age groups in any region. By combining the three regions for further analysis, GFAP mRNA increased with age irregardless of gender, alcoholism in the middle-aged group, or whether brains were classified as normal or neuropathologic (excluding Alzheimer's disease pathology). These data indicate that increased GFAP protein or GFAP-immunoreactive astrocytes in rats and humans may result from transcriptional or post-transcriptional regulation and extend the number to three species (including mouse) showing an increase in GFAP mRNA with age. Factors that are known to regulate GFAP mRNA expression in young brains are considered as possible causes of age-related increases.

Overlapping but not identical protein synthetic domains in cardiovascular cells in response to glucocorticoid hormones.

We have compared, by two-dimensional gel electrophoresis, the effects of glucocorticoids on newly synthesized proteins of cultured cardiovascular cells (rat synthetic- and contractile-state vascular smooth muscle, rat cardiac muscle and non-muscle, and bovine endothelium) and rat hepatoma (H4) cells, as a non-cardiovascular target. In each cell type, glucocorticoids consistently affected a particular set of proteins termed the glucocorticoid domain for that target tissue. Cardiovascular cells exhibited overlapping but non-identical glucocorticoid domains; these domains varied in number of proteins, proportion of induced and repressed proteins, and overlap (extent of shared protein responses) between the various cardiovascular cell types studied. On physico-chemical and co-electrophoresis data the glucocorticoid domain of cardiovascular cells also overlapped with that of hepatoma cells, but to a lesser extent. The glucocorticoid domain in cardiac muscle cells included five proteins not found in cardiac non-muscle cells. Contractile-state vascular smooth muscle cells showed some, but not all, of the proteins affected by glucocorticoids in synthetic-state cells; the only response seen in the contractile but not vascular smooth muscle cells was a protein of Mr-52K and pI-5.3, which was also induced in spontaneously contractile cardiac muscle. Further characterization of contractile and synthetic vascular smooth muscle and cardiac muscle proteins affected by glucocorticoids is necessary to ascertain their role(s) in specialized cardiovascular functions.

Decreased aldosterone receptor affinity in aortic cells from the Fischer 344 rat.

The Fischer 344 rat strain represents a uniform population that is immune to salt induced hypertension and resistant to mineralocorticoid hypertension. We have compared aldosterone binding in aortic cells cultured from salt-resistant Fischer 344 rats to that from salt-sensitive Wistar-Kyoto controls for aldosterone binding. Aortic smooth muscle cells of both strains contain two classes of aldosterone binding sites: corticoid receptor I with high affinity and low capacity and corticoid receptor II with low affinity and high capacity. The corticoid receptor I of Fischer 344 rats has a significantly (P less than 0.001) lower affinity than that of age and sex-matched Wistar-Kyoto controls, but the binding capacity was the same. There was no difference between the strains in the affinity or binding capacity of corticoid receptor II. These results indicate that mineralocorticoid binding may be important in susceptibility and resistance to hypertension and support the contention that mineralocorticoids regulate blood pressure in part by direct action on vascular smooth muscle cells.

Resistance of the dentate gyrus to induced apoptosis during ageing is associated with increases in transforming growth factor-beta1 messenger RNA.

Up-regulation of endogenous neurotrophic factors may protect against apoptosis during ageing. Recent studies showed that the expression of several neurotrophic factors increased with age in specific regions of the rat brain. Previously, we showed that removal of trophic adrenal steroids by adrenalectomy induced apoptosis in the dentate gyrus of adult rats, which was accompanied by increased expression of transforming growth factor-beta1 (TGF-beta1) messenger RNA. In this study, we compared the relative densities of apoptotic cells in the dentate gyrus with TGF-beta1 messenger RNA expression in virgin male Fischer 344 rats ranging from 2 to 26 months of age across three treatment groups: adrenalectomy, adrenalectomy with corticosterone replacement, or sham operation. Seven days after adrenalectomy an increase in the density of apoptotic cells was observed in rats of all age groups compared with sham-operated and corticosterone-treated groups. By in situ hybridisation, the glial messenger RNAs, TGF-beta1 and glial fibrillary acidic protein as a marker of ageing, increased following adrenalectomy in the dentate gyrus in rats of all ages compared with control groups. Interestingly, within adrenalectomy groups, both the number and density of apoptotic cells decreased significantly by 6-8 months with a further decrease at 24-26 months of age. Furthermore, the amount of apoptosis corresponded to changes in TGF-beta1 expression, notably in: (i) adrenalectomised rats showing a significant inverse correlation and (ii) 24-26-month-old rats with the lowest induced apoptosis showing increased expression at the time of adrenalectomy. These studies show that resistance to adrenalectomy-induced apoptosis in the dentate gyrus is associated with increases in TGF-beta1 messenger RNA expression. Furthermore, the endogenous up-regulation of neurotrophic factors, such as the increase in TGF-beta1 expression in the oldest rats, suggests that the aged brain may have compensatory mechanisms, which protect against apoptosis.

Transforming growth factor beta 1 and fibronectin messenger RNA in rat brain: responses to injury and cell-type localization.

Transforming growth factor-beta 1 rapidly increases in adult rat brain in response to experimental lesions. This study characterized the schedule of changes, regional distribution, and cellular localization of striatal transforming growth factor-beta 1 messenger RNA and fibronectin messenger RNA following partial striatal deafferentation by frontal cortex ablation. Frontal cortex ablation induced striatal transforming growth factor-beta 1 messenger RNA elevations that coincided temporally and overlapped anatomically with the course of degeneration of cortico-striatal afferent fibers. Within three days post-lesioning, transforming growth factor-beta 1 messenger RNA was localized at the cortical wound. By 10 days, the anatomical site of transforming growth factor-beta 1 messenger RNA expression shifted to the dorsal half of the deafferented striatum and co-localized with OX-42+ immunostained microglia-macrophage at the site of degenerating afferent terminals. Similarly, fibronectin messenger RNA also shifted from the cortical wound to the deafferented striatum by 10 days post-lesioning. Fibronectin messenger RNA was localized to glial fibrillary acidic protein+ immunostained astrocytes surrounding degenerating corticostriatal afferents. Infusion of transforming growth factor-beta 1 peptide elevated striatal and cortical fibronectin messenger RNA. These findings suggest that microglia-macrophage associated with degenerating afferent fibres can upregulate transforming growth factor-beta 1 messenger RNA and may influence fibronectin messenger RNA synthesis in reactive astrocytes. This study suggests that transforming growth factor-beta 1 has a role in controlling extracellular matrix synthesis following brain injury, which is analogous to that in peripheral wound healing.

Transforming growth factor-beta 1 induces neuronal and astrocyte genes: tubulin alpha 1, glial fibrillary acidic protein and clusterin.

Transforming growth factor-beta 1 was studied as a possible regulator of messenger RNAs in astrocytes and neurons that increase after hippocampal deafferentation by perforant path transection: tubulin alpha 1, clusterin and glial fibrillary acidic protein messenger RNA. Because transforming growth factor-beta 1 messenger RNA is increased after this lesion, we examined which messenger RNA lesion responses could be induced by transforming growth factor-beta 1 alone. Porcine transforming growth factor-beta 1 infused into the lateral ventricle elevated the messenger RNAs for tubulin alpha 1, clusterin and glial fibrillary acidic protein 24 h after infusion in the ipsilateral hippocampus. As assayed by nuclear run-on, the transcription of glial fibrillary acidic protein RNA was increased in the ipsilateral hippocampus after perforant path transection and in primary rat astrocyte cultures by transforming growth factor-beta 1. In contrast, transforming growth factor-beta 1 did not change apolipoprotein-E messenger RNA or transcription, or growth associated protein-43 messenger RNA levels. We conclude that transforming growth factor-beta 1 increases subsets of neuronal and astrocyte messenger RNAs coding for cytoskeletal proteins that are also elevated in response to experimental lesions and Alzheimer's disease. This suggests that transforming growth factor-beta 1 might be a local organizing factor of neuronal and astrocyte responses to brain injury.

Transforming growth factor-beta1 induces transforming growth factor-beta1 and transforming growth factor-beta receptor messenger RNAs and reduces complement C1qB messenger RNA in rat brain microglia.

Transforming growth factor-beta1 is a multifunctional peptide with increased expression during Alzheimer's disease and other neurodegenerative conditions which involve inflammatory mechanisms. We examined the autoregulation of transforming growth factor-beta1 and transforming growth factor-beta receptors and the effects of transforming growth factor-beta1 on complement C1q in brains of adult Fischer 344 male rats and in primary glial cultures. Perforant path transection by entorhinal cortex lesioning was used as a model for the hippocampal deafferentation of Alzheimer's disease. In the hippocampus ipsilateral to the lesion, transforming growth factor-beta1 peptide was increased >100-fold; the messenger RNAs encoding transforming growth factor-beta1, transforming growth factor-beta type I and type II receptors were also increased, but to a smaller degree. In this acute lesion paradigm, microglia are the main cell type containing transforming growth factor-beta1, transforming growth factor-beta type I and II receptor messenger RNAs, shown by immunocytochemistry in combination with in situ hybridization. Autoregulation of the transforming growth factor-beta1 system was examined by intraventricular infusion of transforming growth factor-beta1 peptide, which increased hippocampal transforming growth factor-beta1 messenger RNA levels in a dose-dependent fashion. Similarly, transforming growth factor-beta1 increased levels of transforming growth factor-beta1 messenger RNA and transforming growth factor-beta type II receptor messenger RNA (IC(50), 5pM) and increased release of transforming growth factor-beta1 peptide from primary microglia cultures. Interactions of transforming growth factor-beta1 with complement system gene expression are also indicated, because transforming growth factor-beta1 decreased C1qB messenger RNA in the cortex and hippocampus, after intraventricular infusion, and in cultured glia. These indications of autocrine regulation of transforming growth factor-beta1 in the rodent brain support a major role of microglia in neural activities of transforming growth factor-beta1 and give a new link between transforming growth factor-beta1 and the complement system. The auto-induction of the transforming growth factor-beta1 system has implications for transgenic mice that overexpress transforming growth factor-beta1 in brain cells and for its potential role in amyloidogenesis.

A common 43 K protein induced by glucocorticoids in a variety of cells and tissues.

We have investigated the domain of adrenal steroid action in a variety of mammalian cells and tissue by high resolution two-dimensional gel electrophoresis and autoradiography. Following in vivo or in vitro treatment with steroids, minced tissue or cells were pulsed with [35S]methionine, and they newly synthesized polypeptides compared with controls by visual inspection of partial protein maps. We have found a similar protein (Mr approximately 41-44K, pI approximately 6.3-6.5) to be consistently increased by dexamethasone in 3 rat tissues and 9 human, bovine and rat cell lines. The protein was constitutively synthesized in all targets; quantitative measurements on the magnitude of response show a 2-fold induction by dexamethasone in all systems, in vivo and in vitro. Glucocorticoid-specific hormones but not sex steroids increase the rate of synthesis; deoxycorticosterone has agonist or antagonist effects on 43K synthesis in different systems studied. Co-electrophoresis experiments indicated that 43K proteins co-migrate in at least 3 rat cell lines from tissues with diverse functions. The functional significance of this common glucocorticoid-induced protein is unknown; its ubiquity and electrophoretic properties suggest a highly conserved, common indicator of glucocorticoid action in diverse target cells and tissues, rather than a tissue-specific response.

Glucocorticoid-induced proteins in bovine endothelial cells.

High circulating levels of corticosteroids are associated with elevated blood pressure and an increased incidence of vascular damage. We have used high resolution two-dimensional gel electrophoresis and autoradiography to analyse direct steroid effects on the newly synthesized proteins in bovine aortic endothelial cells. At medium concentrations of 10(-7) M, corticosteroids but not sex steroids increased the synthesis of two endothelial proteins: el (M.W. approximately 43K, pI approximately 6.3) and e2 (M.W. approximately 28K, pI approximately 5.9). The responses were consistent for different endothelial cell cultures, different gel runs, and for natural and synthetic glucocorticoids, including the 'pure' synthetic glucocorticoid, RU 26988. Mineralocorticoids (aldosterone, deoxycorticosterone) at 10(-7) M were less potent in their effect on the synthesis of e2, and had no discernible effect on the synthesis of e1. These specific effects on cellular protein synthesis demonstrate that glucocorticoids can act directly on endothelial cells. These direct actions may thus mediate altered endothelial cell function in clinical and experimental cardiovascular conditions characterized by corticosteroid excess.

Messenger RNA for glial fibrillary acidic protein is decreased in rat brain following acute and chronic corticosterone treatment.

RNA coding for a 50 kDa polypeptide decreased by 50% in 5 brain regions after corticosterone (CORT) treatment (40 mg/kg for 3 days). By hybrid selection and in vitro translation, the 50 kDa polypeptide is identified as glial fibrillary acidic protein (GFAP). Hippocampal GFAP mRNA (2.9 kb) decreases in a dose-dependent manner in response to CORT by RNA blot hybridization using a mouse GFAP cRNA probe; a similar decrease in response to the glucocorticoid agonist, RU 28362, is consistent with a type II glucocorticoid receptor-mediated effect. GFAP mRNA is decreased in both hippocampus and cortex following acute (1-3 days) and chronic (3 days to 3 months) CORT treatment. GFAP gene expression is disinhibited in the rat hippocampus by 7 days post adrenalectomy but not by 3 days. Finally, two clones (CR46 and CR59) that were isolated from a rat hippocampal cDNA library by differential hybridization, show decreased RNA abundance in CORT-treated rats compared to controls. A partial DNA sequence derived from the two clones exhibits 94% nucleotide identity and 96% derived amino acid identity with mouse GFAP mRNA. These results indicate that GFAP mRNA is under negative regulation by glucocorticoids and suggests that glucocorticoids may be used to inhibit GFAP gene expression in vivo in order to assess the role of GFAP in temporal aspects of central nervous system damage.

Corticosterone differentially regulates the bilateral response of astrocyte mRNAs in the hippocampus to entorhinal cortex lesions in male rats.

This study examined the effect of adrenalectomy (ADX) and corticosterone (CORT) replacement on the levels of two astrocyte mRNAs during responses to unilateral entorhinal cortex lesions (ECL) to identify molecular mechanisms involved in glucocorticoid modulation of astrocyte activation following deafferentation. Both glial fibrillary acidic protein (GFAP) and sulfated glycoprotein-2 (SGP-2) mRNA were increased in the ipsilateral hippocampus 4 days following unilateral ECL. In unlesioned ADX rats CORT replacement decreased both messages in the hippocampus. CORT replacement suppressed the ECL-induced increase of GFAP mRNA in the contralateral, but not ipsilateral hippocampus of ADX rats. In contrast, CORT decreased SGP-2 mRNA both ipsi- and contralaterally. It is clear that several regulatory mechanisms are responsible for maintaining a physiological balance of astrocyte activity in the adult brain, and that changes in circuit integrity and the endocrine milieu can alter this balance.

Gene products of corticosteroid action in hippocampus.

We used two methods to examine altered patterns of gene expression in rat hippocampus in response to administered glucocorticoids: analysis of RNA in vitro translation products on 2-d gels and cloning of cDNAs from a rat hippocampal library by differential hybridization (+/- CORT). We determined that two of the CORT-responsive cDNA clones encoded the 35- and 50-kd RNA translation products and identified them as GPDH and GFAP, respectively, by sequence analysis. Cloned mRNAs that increased and decreased in response to CORT were determined to be under positive and negative regulation by glucocorticoids in intact rats. Despite their similarities in glucocorticoid response characteristics, we found three subsets of hippocampal mRNA responses to CORT and shaking stress which differ in temporal and level-dependent aspects of CORT regulation. In addition, GPDH gene expression represents a glucocorticoid-dependent stress response which is rapidly increased in a dose- and stressor-dependent manner. It is a candidate for a sensitive indicator of stress responsiveness in the brain as a function of neuroendocrine activity. Mechanisms of adaptation to stress in the brain are likely to involve responses that are both mediated by glucocorticoids and opposed by them. GFAP and TGF-beta 1 mRNA responses may be examples of the latter, since they are decreased in response to glucocorticoids, are under negative regulation by glucocorticoids in intact rats, and are increased in response to brain injury and disease and during aging. If these astrocytic and microglial responses are involved in cellular defense mechanisms in the brain, then their regulation by glucocorticoids would be important in maintaining and restoring cellular homeostasis in physiological and pathophysiological states. Future studies using these sensitive probes for glucocorticoid-regulated gene expression may identify new mechanisms by which the brain coordinates acute and chronic responses to stress and disease.

Adrenalectomy-induced apoptosis and glial responsiveness during ageing.

Cell death by apoptosis occurs in mature neurones of adult rat dentate gyrus following adrenalectomy. In these studies, apoptosis in the dentate gyrus was induced 3 days after adrenalectomy in male Fischer 344 rats ranging in age from 2-3 to 24-26 months. Glial fibrillary acidic protein and transforming growth factor-beta 1 mRNAs were increased in dentate gyrus of all age groups after adrenalectomy, but there was no effect of age on these responses. Furthermore, corticosterone treatment in the drinking water of adrenalectomized rats prevented both cell death and increases in glial mRNAs in all age groups that were tested (up to 16-18 months). Therefore, adrenalectomy-induced apoptosis can be used to study mechanisms of cell death and survival in aged brain.

NT-4/5 protects against adrenalectomy-induced apoptosis of rat hippocampal granule cells.

Adrenalectomy (ADX) in rats has been shown to induce apoptosis of hippocampal granule cells. We tested whether neurotrophins are able to protect hippocampal neurons in this neurodegeneration model. Acid fucshsin stain was used to identify pyknotic cells in ADX rats treated for 4 days with NT-3, NT-4-5 or cytochrome-C, as a control protein. Cytochrome-C injections slightly decreased cell death on the ipsilateral side. NT-3 did not further promote this effect. Significantly less cell death was observed bilaterally in hippocampus treated with NT-4/5. TUNEL end labeling also confirmed the results. Our results demonstrated that NT-4/5, but not NT-3, promotes hippocampal neuron survival in adrenalectomized rats. They further show that injections of a control solution can induce a local protective effect.