Action of Glucocorticoids on Survival of Nerve Cells: Promoting Neurodegeneration or Neuroprotection?1

Authors

  • I. M. Ábrahám,

    1. Laboratory of Molecular Neuroendocrinology, Institute of Experimental Medicine of the Hungarian Academy of Sciences, Budapest, Hungary.
    2. Department of Molecular Neurobiology, Graduate School of Behavioural and Cognitive Neurosciences, University of Groningen, The Netherlands.
    Search for more papers by this author
  • T. Harkany,

    1. Department of Molecular Neurobiology, Graduate School of Behavioural and Cognitive Neurosciences, University of Groningen, The Netherlands.
    Search for more papers by this author
  • K. M. Horvath,

    1. Department of Molecular Neurobiology, Graduate School of Behavioural and Cognitive Neurosciences, University of Groningen, The Netherlands.
    Search for more papers by this author
  • P. G. M. Luiten

    1. Department of Molecular Neurobiology, Graduate School of Behavioural and Cognitive Neurosciences, University of Groningen, The Netherlands.
    Search for more papers by this author

  • 1

    The authors dedicate this review to the memory of Professor Béla G. J. Bohus.

Dr Paul G. M. Luiten, Department of Molecular Neurobiology, University of Groningen, Kerklaan 30, PO Box 14, NL-9750 AA Haren, The Netherlands (e-mail: p.g.m.luiten@biol.rugnl).

Abstract

Extensive studies during the past decades provided compelling evidence that glucocorticoids (GCs) have the potential to affect the development, survival and death of neurones. These observations, however, reflect paradoxical features of GCs, as they may be critically involved in both neurodegenerative and neuroprotective processes. Hence, we first address different aspects of the complex role of GCs in neurodegeneration and neuroprotection, such as concentration dependent actions of GCs on neuronal viability, anatomical diversity of GC-mediated mechanisms in the brain and species and strain differences in GC-induced neurodegeneration. Second, the modulatory action of GCs during development and ageing of the central nervous system, as well as the contribution of altered GC balance to the pathogenesis of neurodegenerative disorders is considered. In addition, we survey recent data as to the possible mechanisms underlying the neurodegenerative and neuroprotective actions of GCs. As such, two major aspects will be discerned: (i) GC-dependent offensive events, such as GC-induced inhibition of glucose uptake, increased extracellular glutamate concentration and concomitant elevation of intracellular Ca2+, decrease in GABAergic signalling and regulation of local GC concentrations by 11β-hydroxysteroid dehydrogenases; and (ii) GC-related cellular defence mechanisms, such as decrease in after-hyperpolarization, increased synthesis and release of neurotrophic factors and lipocortin-1, feedback regulation of Ca2+ currents and induction of antioxidant enzymes. The particular relevance of these mechanisms to the neurodegenerative and neuroprotective effects of GCs in the brain is discussed.

Survival of the mammalian organism in a constantly changing environment demands a continuous adaptation to external and internal challenges. In case of threatening environmental stimuli, overt defensive responses, also termed stress responses, enable the organism to cope with novel conditions via the activation of complex neuroendocrine, metabolic and behavioural processes. Stress-related sensory information are conveyed to corticotropin-releasing hormone (CRH)-secreting neurones in the paraventricular nucleus (PVN) of the hypothalamus to initiate the neuroendocrine stress cascade. CRH then stimulates the release of the adrenocorticotropic hormone (ACTH) from the pituitary, which in turn leads to the secretion of glucocorticoids (GCs; cortisol in human, corticosterone in rat) from the adrenal cortex as an end-point of the activation of the hypothalamic-pituitary-adrenal (HPA) axis. Besides many peripheral sites of action, the brain is the major neural target for the action of GCs. Due to their lipophilic nature GCs readily pass the blood–brain barrier and bind to intracellular mineralocorticoid (MR) and glucocorticoid (GR) receptors in neurones and glia (1–6). MRs prevail in limbic brain areas, in particular in the hippocampus, while GRs are widely expressed throughout the brain (4–6). Basal corticosterone concentrations predominantly occupy MR, whereas GR can additionally be activated to MR only when corticosterone levels are high (i.e. at the circadian peak and during stress) (7, 8). From a pharmacological point of view, MRs bind corticosterone with an affinity 10-fold higher than that of GR, while the latter receptor type can be challenged by the selective agonist dexamethasone (7). Major actions of GCs in the brain are multifold, as they facilitate the ability of the organism to cope with, adapt to, and recover after stressful stimuli. Primarily, the GC feedback maintains the basal activity of the HPA system and facilitates the termination of stress-induced HPA activation (9–12). Importantly, GCs also influence emotional (13, 14) and learning and memory processes (15, 16), and are involved in the coordination of circadian events, such as the sleep–awake cycle and food intake (17, 18).

Acutely or chronically unbalanced (significantly reduced or enhanced) GC concentrations may directly threaten physiological neuronal functions, and enhance neuronal vulnerability under pathological (toxic) conditions (19, 20). Additionally, clinical observations indicate that elevated GC concentrations and/or changes in the daily profile of GC release may be of particular relevance to the pathogenesis of psychiatric disorders, such as depression and post-traumatic stress disorder (PTSD), as well as chronic neurodegenerative diseases like Alzheimer's disease (19–22). On the other hand, a fine-tuned action of GCs is essential for neural development, particularly by their control of cellular differentiation, and for maintenance of neural integrity and function in adulthood as these hormones exert neuroprotective roles in adult brain in a narrow concentration-window (23–26).

In the present review, some aspects of recently acquired insight in the role of GC in promoting neuronal death and survival, including concentration dependence, regional specificity in the brain, influence of development and ageing, and species and strain differences, are discussed. Moreover, the contribution of GCs to the pathogenesis of neurodegenerative disorders is addressed. Finally, possible distinct mechanisms of GC-induced neurodegenerative and neuroprotective cellular processes are summarized, providing a background for their complex effect on the viability of nerve cells.

GCs and their role in neurodegeneration and neuroprotection

GCs are routinely used as effective anti-inflammatory protective agents in a broad variety of clinically defined pathological conditions. In contrast, GCs appear to have a cardinal role in the pathophysiology of stress-related disorders, and age-related dysregulation of GC metabolism may be a risk factor that contributes to neuronal death in neurodegenerative processes (Fig. 1). An obvious conclusion of current literature is that the rather controversial pharmacological effects of GCs in the central nervous system (CNS) are based on condition-dependent actions of GCs profoundly affecting neuronal viability.

Figure 1.

Outline of the neurodegenerative and neurotrophic effects of different corticosterone concentrations in the brain. Note that glucocorticoid actions exhibit a strict dose–response relationship. bFGF, basic fibroblast growth factor; GABA, γ-aminobutyric acid; GCs, glucocorticoids; NGF, nerve growth factor; PLA2, phospholipase A2.

Concentration and duration dependence

Absence of GCs for 3 days due to adrenalectomy (ADX), where circulatory levels of plasma corticosterone were eliminated, results in pyknosis, fragmentation of DNA and formation of chalescin chromatin and of multiple spherical nuclear bodies in granule cells of the dentate gyrus (DG) in adult rats, resembling cellular hallmarks of neuronal apoptosis (23, 27), as was confirmed by electron microscopy (27). Additionally, the apoptotic nature of ADX-induced neuronal death is further supported by evidence that DNA fragmentation in the DG is associated with increased expression of key regulator proteins of the cell cycle (e.g. p53) which are also involved in apoptosis (28–30). Moreover, ADX-induced neuronal apoptosis is generally accompanied by compensatory hypertrophy of astroglial cells, likely indicating activation of intrinsic neuroprotective mechanisms, such as the release of neurotrophic substances (e.g. nerve growth factor and S-100β) in the close vicinity of the injured neurone (31). Besides direct effects of ADX on neuronal survival, the lack of corticosterone also enhances the vulnerability of neurones to excitotoxic insults. In this regard, recent studies in the authors' laboratories demonstrated that ADX significantly potentiated β-amyloid (Aβ)- and N-methyl-d-aspartate (NMDA)-induced excitotoxicity on cholinergic neurones of the rat magnocellular nucleus basalis (26) (Fig. 2).

Figure 2.

Effect of different plasma corticosterone concentrations on the loss of cholinergic projection fibres in the somatosensory cortex following unilateral injections of NMDA (60 nmol/1 µl) in the cholinergic magnocellular nucleus basalis. Acetylcholinesterase (AChE)-positive fibre loss was expressed as the percentage difference between AChE fibre density in the lesioned and contralateral sides of the brain. (a) Histogram of AChE-positive fibre loss in the different experimental groups expressed as means ± SEM. (b) Scatter-plot of the relationship of corticosterone concentrations and percentage of fibre loss. The horizontal dashed line indicates the level of NMDA-induced neuronal injury in adrenally intact animals. Reprinted with permission (26).

Overexposure to GCs may also cause obvious neuroanatomical changes in neurones. In case the GC concentration is of inefficient magnitude or duration to exert direct neurotoxicity, GCs may endanger neurones by their ability to enhance the vulnerability of nerve cells to neurotoxic insults (32, 33). In this regard, stress levels of corticosterone (approximately 300 nM or higher) were demonstrated to increase neuronal damage in the hippocampus following acute hypoxia/ischaemia in both rats and gerbils, as well as to augment the effect of hypoglycemia in hippocampal tissue explants (34–37). Moreover, stress levels of corticosterone contribute to the augmentation of neuronal loss induced by a variety of neurotoxins, such as NMDA (Fig. 2), Aβ (26), kainic acid (38), 3-acetylpiridine (34) and ethylcholine aziridinium (39, 40). Further increment of these noxious stimuli in the presence of selective GR agonists suggests that the deleterious effects of corticosterone are predominantly mediated by GRs.

Long-lasting exposure to high doses of corticosterone exerts deleterious effects in the brain, as it may induce (ir)reversible damage to neurones. Repeated episodes of restraint stress for 3 weeks may elicit reversible neuronal atrophy in the CA3 region of the hippocampus indicated by the temporary disappearance, or shortening, of apical dendritic branches of CA3 pyramidal cells (41, 42). Noxious effects of restraint stress may be mediated by GCs, as they can be mimicked by 3 weeks of treatment with high doses of corticosterone (43) and eliminated when corticosterone secretion is continuously blocked by cyano-acetone treatment (43, 44). Interestingly, combining chronic restraint stress and oral corticosterone administration failed to show neuronal atrophy in hippocampus (45). It was also recognized that prolonged application of GCs might lead to irreversible neurodegeneration of hippocampal neurones in vivo. In fact, studies of different laboratories indicated that exposure to high concentrations of corticosterone for 12 h per day for 3 months resulted in an approximate 20% loss of hippocampal neurones (46–50). In contrast, however, recent investigations challenged these pioneering data, as they failed to show significant loss of neurones following prolonged exposure to stress or supra-physiological doses of corticosterone in adult rats and tree shrews (51, 52).

In contrast, slightly elevated plasma corticosterone concentrations may exert neuroprotective effects against excitotoxic insults (25, 26). GC-induced neuroprotection is likely to be mediated by MR action, as treatment with MR ligands (23, 53) but not with dexamethasone prevents ADX-induced apoptosis (54, 55). Recent work by ourselves supports these findings since slightly elevated corticosterone concentrations in blood plasma (ranging from 20 to 270 nM) profoundly attenuate both NMDA and Aβ toxicity on cholinergic neurones of the rat magnocellular nucleus basalis (26).

Taking the above data together, a bidirectional effect of corticosterone as a function of concentration can be suggested that regulates neuronal survival. Our recent studies in vivo further confirm this hypothesis, since we clearly showed that corticosterone exerts a bidirectional effect on excitotoxic damage, which became apparent in the U-shaped profile of a dose–response relationship between plasma corticosterone concentration and the extent of excitotoxic injury to cholinergic basal forebrain neurones. Whereas ADX (eventual loss of serum corticosterone) and highly elevated corticosterone concentrations (310–650 nM) potentiate both NMDA- and Aβ-triggered excitotoxicity, moderate levels of plasma corticosterone in a narrow concentration window significantly protect against excitotoxic neuronal damage (Fig. 2). This dual type of glucocorticoid effect is relatively common since, for example, corticosterone-induced ionic conductances and changes in amino acid-mediated neuronal inputs also exhibit U-shaped dose–response relationships (56).

Diverse effect of GCs on neuronal functions in different brain structures

Compelling experimental evidence indicates that lack of GCs or exposure to elevated concentrations of GCs is damaging to hippocampal neurones, whereas other brain regions remain largely intact (46–49). In fact, the hippocampus is the main extra-hypothalamic target of corticosterone and several distinct populations of hippocampal neurones abundantly express GRs and MRs (2, 4, 6). In general terms, hippocampal subfields respond differently to varying concentrations of corticosterone. Lack of corticosterone (as a consequence of ADX) deteriorates the integrity of granule cells in the DG with the concomitant degeneration of granule cell axons (also referred to as mossy fibres) invading the CA3 region, while their postsynaptic targets (CA3 pyramidal cells) and presynaptic terminals localized to granule cells remain largely intact (27). On the other hand, nerve cells of the DG apparently withstand the damaging effects of high GC levels. Recent pharmacological data indicate that high concentrations of GCs are predominantly toxic to neurones in the CA3 region of the hippocampus (46, 57, 58), while damage to CA1 pyramidal cells was less frequently demonstrated (59). Nevertheless, a consistent finding is that GCs in high concentrations may cause atrophy of apical dendrites of CA3 neurones, accompanied by the reduction of their dendritic branching and of dendritic length, and rearrangement of neurotransmitter-containing vesicles in the active zones of their synapses (42–44, 46, 60). Interestingly, GC-induced dendritic atrophy is restricted to the apical surface of CA3 pyramidal cells, whereas their basal dendrites remain largely unaffected (60, 61). Additionally, chronic stress-induced dendritic involution shows sex differences, as female rats did not develop significant dendritic atrophy in the apical field of CA3 pyramidal cells in response to repeated restraint stress (62). Taking the above data together, it may be assumed that effects of GCs on the morphology of apical dendrites may occur secondarily to changes in their presynaptic input. Because neurones of the CA3 subfield receive dense innervation from mossy fibres, it is tempting to speculate that their degeneration is at least partly due to GC-induced changes in the function of granule cells (61).

Whereas GC-induced neurodegeneration is regarded as specific for the hippocampus, GCs may enhance the vulnerability of a broad variety of neurone populations to secondary stimuli in the brain. In this regard, GCs are known to augment excitotoxic damage in rat striatum and cholinergic basal forebrain following infusion of numerous neurotoxins including NMDA (26, 63), quinolinic acid (64) and Aβ (26). Moreover, GC may significantly increase acute ischaemic damage to the neocortex in adult rats (35, 65).

Neuroprotective action of GCs in a narrow concentration window can be observed throughout the brain. As mentioned above, lack of corticosterone results in widespread apoptosis in the DG of the hippocampus. Based on these findings, Sloviter et al. postulated that GCs are protective ‘obligatory growth factors’ for hippocampal granule cells (27). Moreover, corticosterone was shown to prevent NMDA excitotoxicity on cholinergic projection neurones of the basal forebrain (26), while continuous stimulation of GRs by dexamethasone decreased the extent of focal lesions following traumatic brain injury in the neocortex (66), and attenuated hypoxic-ischaemic brain damage in neonatal rats (67).

Species and strain differences and susceptibility to GC-induced neurodegeneration

Comparison of the presence and/or severity of GC-induced neurotoxic mechanisms in different animal species or in distinct strains of rats reveals some peculiar phenomena. Data from several studies indicate that ADX-induced loss of granule cells in the DG occurs uniformly irrespective of the rat strains employed (27, 31, 68). It is worth noting, however, that considerable variability may be present in the severity and temporal profile of neuronal death in the DG, with some animals within a population being devoid of any significant granule cell pathology (68–70). Such an inconsistency might be explained by the existence of extra-adrenal ectopic tissue capable of converting and secreting considerable amounts of GCs, or to incomplete ADX. Interestingly, the above phenomenon may not only be observed under experimental conditions, but may also be of clinical importance, as a case report indicated selective loss of granule cells in a female patient with adrenocortical deficiency syndrome (71).

Several lines of evidence indicate that different rat strains exhibit significant variations in their susceptibility to GC-induced neurodegeneration. Whereas chronic exposure to GCs in Fisher-344 rats results in a mortality rate of approximately 50% and loss of hippocampal CA3 pyramidal neurones (46), such a fatal reaction and cell loss cannot be observed either in Long-Evans (72) or Wistar rats (50) during or after chronic corticosterone treatment. Furthermore, GC-induced reversible dendritic atrophy was demonstrated in Sprague-Dawley rats (42). Interestingly, development of corticosterone-induced neurodegeneration is also highly dependent on particular primate strains investigated. Dendritic atrophy, shrinkage of nerve cells and condensation of nucleoplasm in CA3 of the hippocampus, was only demonstrated in vervet monkeys after 1 year of intrahippocampal corticosteroid pellet implantation (73), while elevated cortisol concentrations in the absence of stress did not produce significant neuronal loss in the hippocampal formation of pigtailed macaques (74). The reason of these strain differences is not clear but the role of some genetic components that compromises GC-induced neurodegeneration can be proposed. In contrast, GC-induced enhancement of neurotoxic insults may be uniformly observed among rat strains, as was described for Sprague-Dawley, Long-Evans, Wistar, Holtzman and spontaneously hypertensive rats (39, 46, 75–77).

Development and ageing of the brain

During the perinatal period, GCs are essential for normal development of the CNS (21, 78). Their other important role during development is to protect the developing CNS from deleterious impacts, which is underscored by experimental studies reporting protective effects of dexamethasone in a neonatal model of hypoxic-ischaemic brain injury (79–81). The contribution of developmentally regulated activity of the HPA axis to neuronal vulnerability during the entire life span is indicated by recent studies showing long-lasting effects, e.g. the ACTH4−9 analogue ORG 2766 and of postnatal handling on experimentally induced excitotoxic brain damage and age-related neurodegeneration. In this regard, Nyakas et al. (82) reported that early postnatal administration of ORG 2766 induced permanent upregulation of MR in rat hippocampus. Long-term upregulation of MR may be of significance in mediating the ‘delayed’ neuroprotective effect of ORG 2766 on cholinergic neurones of the rat magnocellular nucleus basalis following exposure to toxic concentrations of NMDA in adulthood (83). Furthermore, early postnatal handling (daily short-term separation of the pups from the mother) was demonstrated to attenuate the age-related decline of cognitive function (84) and to affect excitotoxic brain damage in adult animals (85). Postnatal handling as a crucial early life experience is the postnatal part of environmental enrichment possibly due to improved maternal care (86). Neuroendocrinological and neuroanatomical data indicate that blunted stress reactivity of the HPA axis, upregulation of GR expression and attenuated reactivity of stress-related neuronal circuits, as indicated by c-fos labelling studies (87), may underlie the lifelong effects of postnatal handling (88–90). It is worth noting that in adult and ageing rodents environmental enrichment might decrease spontaneous apoptotic cell death in the hippocampus and attenuate kainate-induced seizures and excitotoxic neuronal injury (91).

On the other hand, ageing of the brain is associated with reduced ability of neurones to maintain their ion (e.g. Ca2+) homeostasis (92), breakdown of nutrient supply due to pathological changes in brain microvasculature (93) and enhanced ACTH and corticosterone responses to stress, which ultimately lead to the enhancement of GC-induced neurodegeneration of nerve cells. The pioneering studies of Landfield et al. (94) provided the first indication on the role of GCs in age-related dysfunction of the hippocampus, and demonstrated that adrenal hypertrophy and subsequent prolonged hypersecretion of corticosterone positively correlate with increased neuronal damage and accumulation of reactive astroglia in the hippocampus during ageing. Additionally, elevated basal corticosterone levels accelerated hippocampal neuronal damage (95), whereas resupplementation of low doses of corticosterone in middle-aged ADX rats for 9 months significantly reduced neuronal loss and glial reactivity in the hippocampus, as compared to age-matched controls (96). Pharmacological studies demonstrated that antagonism of GR with RU486 significantly attenuated the age-related hippocampal damage (97), while ageing exacerbates dexamethasone-induced apoptosis in DG (98). Moreover, exposure to repeated stress delivered by means of foot-shocks for 6 months significantly increased morphological indices of hippocampal ageing (58). Taking the above data together, the ‘GC cascade hypothesis of aging’ was postulated (99), which refers to a vicious cycle of events during ageing. First, sustained elevation of corticosterone concentrations results in progressive injury of hippocampal neurones. Dysfunction of the hippocampus, in turn, downregulates the negative feedback of adrenal GC secretion, which further enhances corticosteroid secretion and escalates the neurodegenerative process. Developmental suppression of stress reactivity by postnatal handling provides further support for the above hypothesis because aged postnatally handled rats did not show increased basal plasma corticosterone levels, and exhibited a reduced rise in plasma corticosterone levels following acute stress concomitant with a significant decrease in age-related loss of hippocampal neurones (81, 100). It is noteworthy, however, that the observation of age-associated impairment of cognitive performance without a reduction in the number of hippocampal neurones may challenge the primary role of GC-induced hippocampal dysfunction during ageing (72, 101).

Neurodegenerative disorders

In general terms, exogenous administration of GCs is an effective strategy to suppress inflammatory reactions in several neurological disorders, such as multiple sclerosis, traumatic brain and spinal cord injury. However, pharmacological evidence indicates that chronic administration of the GR agonist dexamethasone (as part of the anti-inflammatory therapy for rheumatoid arthritis) leads to widespread neuronal atrophy in the brain (102). Concordantly, the results of several clinical studies lend support to the notion that sustained endogenous elevation of plasma corticosterone levels may be regarded as a major predisposing factor for neurodegenerative disorders in the human and mammalian brain (19, 20). Such a hypothesis is further supported by the observation that individuals with Alzheimer's and Parkinson's disease exhibit significantly higher total plasma cortisol concentrations whereas the relative diurnal variation of cortisol secretion did not differ from healthy controls (21, 22). Moreover, a significant correlation between increased plasma cortisol levels and the degree of mental deterioration as well as decreased volume of the hippocampus was demonstrated in Alzheimer's disease patients (103). Additionally, evidence has recently accumulated on the role of high GCs concentration as a ‘risk factor’ also in the pathogenesis of several psychiatric disorders, such as depression and PTSD. Alterations in the hippocampal volume of individuals with major depression exhibited marked negative correlation with plasma concentration of cortisol (104). It should be noted, however, that the rather limited neuronal death might only contribute to a minor extent to this reduction of hippocampal volume (105). Interestingly, patients with PTSD also exhibit a reduced volume of the hippocampus even without cortisol hypersecretion (106, 107), indicating episodic stress and/or temporary hypersecretion of cortisol, which might elicit irreversible changes in susceptible brain structures and potentially lead to the subsequent manifestation of PTSD. Moreover, in endocrinological disorders, such as Cushing's syndrome, the high plasma cortisol concentration may deteriorate the function and structure of the CNS. In fact, a reduced volume of the hippocampus and impaired cognitive function (e.g. verbal recall, declarative memory) were shown to correlate with an average increase in plasma cortisol levels in patients with Cushing's syndrome (108, 109). Interestingly, hippocampal atrophy can be suppressed by normalization of cortisol levels in Cushing's disease (110).

Possible mechanisms of glucocorticoid action on neurones

The remainder of this review is aimed at analysing the cellular mechanisms underlying the bidirectional action of GCs on neuronal survival, namely promoting neurodegeneration or rescuing nerve cells therefrom. To begin with, however, the critical role of distinct classes of corticosteroid receptors in determining the cellular outcome of GC action has to be defined. Deleterious effects of GCs are mediated by low-affinity GR (97), which is also evidenced by the pharmacological observation that synthetic GR agonists [e.g. methylprednisolone (64, 111) and dexamethasone (35, 50)] share the ability to enhance the vulnerability of neuronal cells, whereas non-GR ligands do not exert toxic properties in the rodent brain (112, 113). Recently, Almeida et al. (114) provided further indication on the critical role of GR in neuronal survival in rats. Stimulation of GR was shown to lead to neuronal apoptosis by significantly increased expression of the pro-apoptotic molecule Bax, relative to that of the anti-apoptotic molecules Bcl-2 or Bcl-X, whereas opposite, neuroprotective effects were observed following stimulation of MR. Recent anatomical data in primates, however, do not support these findings because only very low GR levels were demonstrated in the hippocampus, concomitant with high GR density in the PVN and cerebral cortex of rhesus monkeys (115). It appears therefore likely that the effects of GCs are predominantly mediated by MRs present in the hippocampal formation of primates (115).

GC-induced neurodegeneration: offensive action of GCs

Major mechanisms by which GCs may hamper neuronal functions involve inhibition of the uptake of key nutrients, such as glucose, from the general circulation (116), modulation of both excitatory (glutamatergic) and inhibitory (GABAergic) neurotransmission (117, 118), direct enhancement of intracellular Ca2+ signalling (119) that may lead to or potentiate excitotoxic neuronal death and GC metabolization by 11β-hydroxysteroid dehydrogenases (11β-HSD) (120).

Because bioavailability of glucose is critical for the maintenance of the energetic state of neurones, endogenous compounds that are capable of blocking the uptake and/or metabolism of glucose may potentially threaten physiological neuronal functions. GCs are catabolic hormones because they increase the plasma glucose level by stimulation of glycolysis, concomitant with the arrest of cellular glucose uptake by the inhibition of its trans-endothelial transport both to the brain and to peripheral organs (121). Additionally, studies on the hippocampus revealed an approximate 20–30% inhibition of glucose transport as well as blockade of glucose uptake of both hippocampal neurones and glia (19, 116). Whereas such an extent of metabolic compromise on hippocampal neurones may acutely be insufficient to directly kill neurones, it may effectively contribute to GC-induced enhancement of neuronal injury following excitotoxic attacks (122).

During the past decade, ample evidence emerged on a critical role of GCs in modulating glutamate metabolism and glutamatergic neurotransmission in the brain (117, 123). Glutamate and glutamate-related excitatory amino acid (EAA) neurotransmitters (e.g. aspartate) bind to a broad variety of metabotropic and ionotropic glutamate receptors in the brain, some of which (primarily the class of NMDA receptors) are key mediators of enhancing intracellular Ca2+ signalling as well as mobilizing cytosolic free Ca2+. Under physiological circumstances, Ca2+ mobilization via NMDA receptors leads to the enhancement of synaptic communication in both pre- and postsynaptic elements and to the formation of highly active perforated synaptic contacts (124). Under pathological conditions, however, overt glutamate release and uncontrolled rise in the intracellular Ca2+ concentration in neurones are central in the pathogenesis of several disorders, such as ischaemia (125), hypoxia (126) and epileptic seizures (126). Excessive amounts of cytosolic free Ca2+ may induce mitochondrial dysfunction, subsequent increases in the formation of reactive oxygen intermediates and promiscuous activation of Ca2+-dependent proteases, lipases and nucleases, leading potentially to damage to eventually all biomolecules (127). Results of both in vitro and in vivo studies indicated that GCs might trigger mechanisms that render glutamate transmission to excitotoxic dimensions in the brain. In this regard, by means of microdialysis, high GC concentrations were shown to increase the extracellular glutamate level in the hippocampus of freely moving conscious animals (25, 118, 123, 128, 129). It is noteworthy that GCs elicit an abrupt rise in the extracellular glutamate concentration in vivo, which cannot be inhibited by applying either specific antagonist for corticosteroid receptors, or by inhibitors of protein synthesis (130). In vitro studies with cultured cells suggest that direct GC-induced inhibition of glutamate uptake by glial cells might play a predominant role in the above process (131).

Chronic absence as well as high concentrations of corticosterone may directly modulate Ca2+ currents in nerve cells (7, 119, 132, 133). In fact, ADX significantly increases the amplitude of voltage dependent Ca2+ currents (134), concomitant with a relatively small Ca2+-dependent K+ conductance in hippocampal neurones (132). Moreover, electrophysiological studies on slice preparations of ADX rats demonstrated that GR agonists and high concentrations of GCs augment the voltage-gated Ca2+ conductance, Ca2+-dependent K+ currents and prolong the duration of Ca2+ spikes (135, 136). Chronic exposure to high corticosterone concentrations affects the Ca2+ conductance of neuronal membranes in a similar fashion (134), which becomes even more deleterious if the age-related elevation of basal GC concentration is considered (132, 137). Interestingly, molecular pharmacological data substantiate the direct effects of GCs on Ca2+ currents because high GC concentrations elevated the cellular mRNAs content for Ca2+ channel subunits in ADX rats, whereas chronic supply with low levels of corticosterone led to a reduced Ca2+ channel subunit mRNA expression (138). Additionally, high concentrations of GCs repress the expression of plasma membrane Ca2+ pump isoform 1 mRNA in several brain regions, thereby providing a unique way to control Ca2+ extrusion from nerve cells (139). Landfield et al. (140) postulated that excessive activation of GR and dysregulation of the intracellular Ca2+ homeostasis may be two distinct phases of a single process that significantly increases the susceptibility of hippocampal neurones to neurodegeneration during ageing and in Alzheimer's disease. A limitation of the above hypothesis is that its underlying mechanisms were predominantly identified in CA1 neurones, whereas mechanisms in other hippocampal subfields, or other brain regions remained largely elusive. This, in turn, does not allow ready extrapolation of the pharmacological data on ADX-and GC-induced neurodegeneration to other brain regions (i.e. basal forebrain cholinergic nuclei, or the cerebral cortex) affected in Alzheimer's disease.

In general terms, GABA exerts inhibitory activity in the CNS and an ‘intrinsic neuroprotective potential’ against excitotoxic brain damage via distinct classes of GABA receptors that are involved in the generation of inhibitory postsynaptic currents. High concentrations of GCs also deteriorate GABAergic neurotransmission in the brain, as they significantly reduce the efficacy of GABAergic signalling by decreasing the binding of GABA, neurosteroids and benzodiazepines to GABA receptors (141, 142). Electrophysiological studies performed on brain slices support the above findings, as the presence of high GC concentrations blocked the generation of GABAA receptor-mediated inhibitory postsynaptic potentials (143). It is worth noting that GC-induced derangement of GABAergic neurotransmission may be likely due to changes in GABA receptor sensitivity, as microdialysis studies in freely moving rats failed to show significant effects of intracerebral GC administration on extracellular GABA concentrations (118, 130).

Recent studies revealed an additional enzymatic mechanism critical in controlling prereceptor metabolism of GCs. In particular, the key enzyme 11β-HSD has two isoforms, 11β-HSD1 and 11β-HSD2, that exert opposite actions. Whereas 11β-HSD1 is the predominant 11β-reductase that in many tissues increases local GC concentrations in vivo, 11β-HSD2 is a 11β-dehydrogenase that inactivates cortisol (120). Because 11β-HSD1 is abundantly expressed in the hippocampus (144), it may augment GC-induced neurodegeneration of hippocampal neurones (120, 145). The pivotal role of 11β-HSD1 in mediating deleterious effects of chronic excess GC concentrations on cognitive function is further substantiated by recent observations in 11β-HSD1 knockout mice indicating that the lack of 11β-HSD1 prevents GC-induced cognitive decline (146).

Neuroprotective (defensive) mechanisms

Whereas eventual lack and overt exposure to GCs elicits or enhances neurodegeneration in the CNS, fine-tuned plasma corticosterone concentrations in the physiological or slightly elevated range (20–270 nM) may enhance the neuroprotective potential of nerve cells. The neuroprotective effects of GCs may be mediated by distinct mechanisms involving modulation of Ca2+ currents (119), enhanced synthesis of lipocortin-1 (147) and neurotrophic factors (148), and their ability to attenuate lipid peroxidation (149).

GCs may dose-dependently regulate ionic conductances in neuronal membranes. Because some of the electrophysiological aspects of GC action have already been reviewed in detail above, here additional electrophysiological consequences of GC-induced high intracellular Ca2+ levels are emphasized with regard to neuroprotective mechanisms. Elevation of corticosterone levels initiates large Ca2+ currents and leads to prolonged after-hyperpolarization of neuronal membranes (119, 132, 135). Ca2+ influx itself inhibits subsequent Ca2+ currents through various mechanisms, including feedback inhibition of voltage gated Ca2+ channels, and of NMDA channels via Ca2+-dependent phosphatases (150, 151). Although these processes are likely only consequences of high intracellular Ca2+ levels, and they are possibly not regulated directly by GCs, they may contribute to GC-induced enhancement of the latency between consecutive membrane depolarizations, delay the generation of action potentials and counteract excitotoxicity-related over-activation of nerve cells.

Controlled activation of GR also initiates changes in the expression of a broad variety of substances with neurotrophic potential, such as lipocortin-1 (147), basic fibroblast growth factor (bFGF) and nerve growth factor (NGF) (148, 152). In this regard, lipocortin-1 is known to inhibit the synthesis of prostaglandins and leukotrienes by inhibiting phospholipase A2, the key enzyme of the arachidonic acid cascade, and thereby precludes the subsequent production of potentially cytotoxic oxygen radicals (147). In fact, lipocortin-1 was demonstrated to act as a neuroprotective agent against ischaemic insults and to inhibit neuronal damage induced by infusion of NMDA receptor agonists in the brain (153). Taking the above observations together, GC-induced synthesis of lipocortin-1 may be regarded as a potential means to rescue nerve cells from excitotoxic injury.

GCs also stimulate the production of neurotrophic factors that are vital for the development and survival of certain populations of nerve cells in the brain. In situ hybridization studies revealed that systemic corticosterone administration elicits the temporal induction of mRNA species coding for bFGF and NGF in the cerebral cortex (148), where GCs are not particularly known to be deleterious. Conversely, ADX leads to a decreased expression of neutrophins, which may be directly associated with increased neuronal damage following ischaemic conditions or hypoglycemic stress (152, 154). Because NGF administration prevents ischaemia- and NMDA receptor stimulation-induced elevation of the intracellular Ca2+ concentration in a dose-dependent manner, enhanced NGF synthesis may potentially enhance neuronal survival after excitotoxic insults (155). It is worth noting that immobilization stress with concomitant exposure to GCs was reported to block NGF mRNA expression in the hippocampus (156), where the vast majority of GC-induced neuronal damage was identified. Furthermore, this result indicates that GCs, in a narrow concentration window and in a complex interaction with other potential factors, may (dys)regulate neutrophin expression and neutrophin-linked signal transduction pathways.

Due to their lipophilic nature, GCs may accumulate in biological membranes, where they intercalate lipid molecules and protect them from oxygen radical-induced peroxidation (19). It is likely that this phenomenon is purely dependent on the concentration of available free GCs, as supra-physiological GC levels are required, and synthetic non-GC steroid compounds also exhibit antioxidant activity (157). Importantly, the neuroprotective efficacy of GCs (as antioxidants) was demonstrated both in animal models of spinal cord trauma (149) and in clinical trials (158). It should be noted that high GC concentrations can effectively reduce the activity of the detoxifying enzyme glutathione peroxidase in rat hippocampus, striatum and cerebral cortex (159), which may significantly affect neuronal survival when cells are exposed to agents (e.g. Aβ) capable of inducing oxidative neuronal injury.

Conclusions

In the present review, recent knowledge on the role of GCs in both neurodegenerative and neuroprotective processes is summarized, along with the possible cellular mechanisms executing either offensive or defensive effects of GCs on neurones. The complexity and variability of GC-induced cellular events is predominantly due to the concentration-dependent action of GCs, and their interaction with other mechanisms (e.g. glutamate excitotoxicity) substantially involved in the induction of neuronal injury and cell death. Lack, or high, GC concentrations may enhance the impact of neurotoxic challenges in the CNS, whereas a narrow concentration window (ranging from physiological to slightly elevated GC levels) may protect nerve cells from excitotoxic hazards (Figs 1 and 2). It should be noted, however, that rather controversial data exist in the literature regarding the role of corticosterone as a neurotoxic substance, whereas GC-induced enhancement of neuronal vulnerability is a widely accepted phenomenon. GCs-induced neurodegeneration is largely dependent on the age and strain of the experimental subjects and on the neuroanatomical area investigated. In contrast, GC-induced enhancement of neuronal vulnerability is a general principle, which was demonstrated in several animal species, strains, age groups and brain regions. Taking the above considerations together, it seems likely that GCs do not pose a direct neurotoxic threat, but increase or boost the noxious effect of coinciding neurotoxic stimuli. In this regard, the presence of chronically elevated high GC levels and subsequent activation of stress-related intracellular signalling, together with a pathological increase in glutamatergic neurotransmission, may effectively contribute to the pathogenesis of neurodegenerative disorders.

Bidirectional effects of GCs on neuronal functions are supported by GC-induced distinct offensive and defensive molecular events. Hypothetically, these mechanisms are in a well-controlled balance at physiological GC concentrations. Slightly increased GC levels may induce compensatory neuroprotective processes, whereas lack or high concentrations of GCs shift this sensitive balance into the neurotoxic range. The set-point of GC balance may significantly vary in different brain structures, under distinct pathological conditions, and be affected by the age of individuals. Although particular cellular mechanisms of GC action are as yet well characterized, further studies must reveal how, and when, a transition from the neuroprotective (physiological) to the neurodegeneration-promoting range of GC balance occurs.

Acknowledgements

The authors thank the reviewers of this manuscript for their helpful comments and Dr K. J. Kovács for her critical appraisal of an earlier version of this review. This work was supported by grants of the Hungarian National Science Foundation (OTKA, #F030681 to I.M.Á., #F023865 and #F035254 to T.H.) and by a joint grant of the Netherlands Science Foundation (NWO) and OTKA (#048-011-006 to P.G.M.L.).

Accepted 20 June 2001

Ancillary