Disruptive effects of glucocorticoids on glutathione peroxidase biochemistry in hippocampal cultures

Authors


Address correspondence and reprint requests to Robert Sapolsky, Department of Biological Sciences, Stanford University, Gilbert Laboratory, MC 5020, Stanford, CA 94305–5020, USA. E-mail: sapolsky@stanford.edu

Abstract

Glucocorticoids (GCs), the adrenal steroids secreted during stress, compromise the ability of hippocampal neurons to survive various necrotic insults. We have previously observed that GCs enhance the hippocampal neurotoxicity of reactive oxygen species and, as a potential contributor to this, decrease the activity of the antioxidant enzyme, glutathione peroxidase (GSPx). In this report, we have studied the possible mechanisms underlying this GC effect upon GSPx in primary hippocampal cultures and have observed several results. (i) Corticosterone (the GC of rats) decreased glutathione levels; this was predominately a result of a decrease in levels of reduced glutathione (GSH), the form of glutathione which facilitates GSPx activity. (ii) Corticosterone also decreased levels of NADPH; this may help explain the effect on GSH as NADPH is required for regeneration of GSH from oxidized glutathione. (iii) However, the corticosterone effect on total glutathione levels could not just be caused by the NADPH effect, as there were also reduced levels of oxidized glutathione. (iv) Corticosterone caused a small but significant decrease in GSPx activity over a range of glucose concentrations; this occurred under circumstances of an excess of glutathione as a substrate, suggesting a direct effect of corticosterone on GSPx activity. (v) This corticosterone effect was likely to have functional implications, in that enhancement of GSPx activity (to the same magnitude as activity was inhibited by corticosterone) by GSPx overexpression protected against an excitotoxin. Thus, GCs have various effects, both energetic and non-energetic in nature, upon steps in GSPx biochemistry that, collectively, may impair hippocampal antioxidant capacity.

1These authors

contributed equally to this manuscript.

Abbreviations used

CPA, cis-parinaric acid

DTNB

5,5′-dithiobis-2-nitrobenzoic acid

GC

glucocorticoid

GSH

glutathione

GSPx

glutathione peroxidase

GSSG

oxidized glutathione

HSV

KA

MEM

minimum essential medium

MOI

multiplicities of infection

NEM

N-ethyl maleimide

PBS

phosphate-buffered saline

ROS

reactive oxygen species.

Glucocorticoids (GCs), the adrenal steroids secreted during stress, can have adverse effects in the nervous system and particularly in the hippocampus, which is a primary GC target. An excess of GCs or stress can atrophy dendritic processes, inhibit neurogenesis, and, at an extreme, be neurotoxic (reviewed in Reagan and McEwen 1997). GCs can also compromise the ability of hippocampal neurons to survive necrotic insults, including hypoxia-ischemia, seizure, hypoglycemia, anti-metabolites, β-amyloid, and the gp120 protein of HIV (Sapolsky 1996). A fair amount is known about the mechanisms underlying these endangering effects; GCs worsen both the glutamate accumulation and the calcium mobilization that are central to necrotic neuron death. These damaging actions reflect both direct GC effects [e.g. altering profiles of subunits of calcium channels (Nair et al. 1998) or inhibiting a calcium-ATPase (Bhargava et al. 2000)], as well as indirect ones, secondary to disrupting hippocampal energetics.

The ability of GCs to worsen this glutamate/calcium pathway generates the prediction that the hormone should also worsen degenerative events downstream of the calcium excess. Supporting this, GCs worsen the cytoskeletal proteolysis, the microtubule abnormalities, and the accumulation of reactive oxygen species (ROS) during necrotic insults to hippocampal neurons (Sapolsky 1996; Reagan and McEwen 1997). This final endpoint is particularly interesting, given the increasing emphasis on the role of ROS in necrotic neuronal injury.

ROS are important to neuronal function at many levels, from simple modulation of the function of various receptors to degradation of mitochondrial function and triggering of complex cell-death pathways (Coyle and Puttfarcken 1993). The high metabolic rate of the brain results in considerable ROS accumulation, making detoxification of ROS a critical function. Previous work indicates that GCs increase the oxidative load on hippocampal neurons following insults, secondary to the effects of the hormone on glutamate and calcium trafficking (reviewed in Sapolsky 1996). In addition, the hormone has a direct effect upon oxygen radical pathways, in that GCs decrease the activity of the key antioxidant enzyme glutathione peroxidase (GSPx) in the hippocampus, in vivo, following excitotoxic seizures (McIntosh et al. 1998a,b). This is important because the glutathione system provides both protection against and repair of ROS damage, and may be linked to oxidative stress in a number of neurological disorders (Dringen 2000). Because of the ramifications of this observation, we have explored the mechanisms underlying this GC effect on GSPx activity.

GSPx function requires an adequate supply of reduced glutathione (GSH) to serve as an electron donor. GSH is regenerated by NADPH from oxidized glutathione (GSSG), and thus GSPx is ultimately sensitive to the energy state of the neuron (via the pentose phosphate pathway) (Dringen 2000). Our observation that GCs decrease the activity of GSPx in vivo following a necrotic insult (McIntosh et al. 1998a) suggested two possible mechanisms to explain the GC action. First, the steroid could be decreasing the intrinsic activity and/or levels of the enzyme. Alternatively or additionally, GSPx activity could be decreased secondarily to GCs decreasing the availability of GSH as a substrate. In the present study, we tested these possibilities.

Materials and methods

Materials

Materials included glutathione reductase, glutathione, NADPH, 5,5′-dithiobis-2-nitrobenzoic acid (DTNB), kainic acid (KA), N-ethyl maleimide (NEM), t-butyl hydroperoxide, anti-MAP2 antibody, corticosterone (all from Sigma, St Louis, MO, USA), C18 Sep-Pak cartridge (Waters Associates, Farmingham, MA, USA), cis-parinaric acid (Molecular Probes, Eugene, OR, USA), biotinylated anti-mouse IgG, ABC reagent and ABTS reagent (all from Vector, Burlingame, CA, USA).

Culture preparation

Mixed hippocampal cultures were grown from day-18 fetal rats (Brooke et al. 1995). Briefly the procedure was as follows: after dissection the tissue was treated with papain (Worthington Biochemical, Freehold, NJ, USA) according to the manufacturer's instructions. The cells were dissociated, filtered through an 80-µm cell strainer and resuspended in a modified minimum essential medium (MEM) media (UCSF Tissue Culture Facility, San Francisco, CA, USA) and supplemented with 10% horse serum (Hyclone, Logan, UT, USA). Cells were plated at a density of 20 000/cm2 on 96-, 24- or 6-well plates coated with poly-d-lysine. The cells were used on days 10–14 in culture. Under these conditions, cultures are typically approximately 30% neuronal.

Glutathione assay

The standard GSH/GSSG recycling assay was used to measure both total glutathione levels and oxidized glutathione (GSSG) levels (Vandeputte et al. 1994). Mixed neuronal cultures on 6-well plates were lysed by physical scraping into 100 md potassium phosphate buffer with 5 md ethylenediaminetetraacetic acid (EDTA). To determine total glutathione, 10 md DTNB was mixed with an equal volume of lysate. This DTNB/lysate mix (30 µL) was then added to a cuvette with 0.5 units of glutathione reductase, and 220 nmol NADPH. The change in absorbance was measured at 412 nm for 60 s. To measure glutathione disulfide, or GSSG, 10 md NEM, which reacts with reduced glutathione, was mixed with an equal volume of lysate. The mixture was run through a C18 Sep-Pak cartridge (Waters Associates), which binds the NEM/reduced glutathione complex thereby allowing GSSG to flow through. Oxidized glutathione was measured in the same manner as total glutathione. In these experiments, corticosterone (the species-typical GC of rats), dissolved in ethanol, was added to the cells 24 h prior to assay, producing a final concentration of 1 µd; control cultures received ethanol alone. One hour prior to assay, KA was added to the cells to produce a final concentration of 50 µd. A Bradford protein assay was run to normalize for protein.

Lipid peroxidation assay

Lipid peroxidation measurements utilized cis-parinaric acid (CPA; Molecular Probes, P1901), a molecule that is incorporated into the lipid bilayer of live neurons (Kuypers et al. 1987; Hedley and Chow 1992). CPA is converted into the trans-isomer after peroxidation and causes a loss of fluorescence that can be detected using a fluorimeter with an excitation wavelength of 312 nm and an emission wavelength of 414 nm. Neuronal cultures on 24-well plates were treated with 50 µd CPA for 4 h prior to the assay to allow time for CPA to be incorporated into lipid bilayers. One hour prior to assay, KA at a final concentration of 50 µd was added. Corticosterone was added 24 h prior to assay.

Glutathione peroxidase assay

GSPx was detected according to Buckman et al. (1993). Cell lysate was combined with 110 µd NADPH, 1 md glutathione, 0.6 U/mL glutathione reductase in 100 md potassium phosphate buffer to a volume of 500 µL and equilibrated for 8 min. The reaction was initiated with 10 µL t-butyl hydroperoxide, mixed rapidly, and measured for 3 min at 340 nm absorbency. A critical point in the assay is that all the substrates are in excess. Therefore, GSPx activity is an indicator of either the ability of GSPx to function or the number of enzyme molecules. It does not directly correlate with the GSPx activity that would be observed in an intact cell, where substrates may be a limiting factor.

Construction of amplicon plasmids

Amplicon plasmid pα4GSPxα22βgal was constructed as follows (Fig. 1): human GSPx cDNA was isolated from the pbsGPXp120/23 plasmid (graciously provided by Michael Kelner) as a Cla1/Xho1 fragment. The fragment was then blunt-end ligated into the BamH1 site of pG310. This expression vector contains the polyadenylation (polyA) signal (from nucleotides +3270 to +3430) of the human cytomegalovirus ie1 gene in a pGEM-2 plasmid background. The GSPx coding sequence together with the polyA tail was then isolated as a BamH1/BglII fragment and blunt-end ligated into the HindIII site of pα22βgal, downstream of the α4 promoter. The control vectors pα22βgal and pα4sβgal have been described previously (Ho 1994; Lawrence et al. 1995).

Figure 1.

Schematic diagram showing the structure of the amplicon pα4GSPxα22βgal. The GSPx and lacZ genes are controlled by the α4 and α22 promoters, respectively. The oris, together with the “a” sequence provide the necessary HSV replication signals. The polyA signal for the GSPx gene was derived from the human cytomegalovirus (HCMV) ie1 gene and that for lacZ was derived from the simian virus 40 early gene. The ampicillin resistance gene of the parent plasmid pGEM2 is also indicated.

Generation of herpes simplex virus-1 (HSV) vectors

Our protocols for generating viral vectors have been previously described in detail (Ho 1994). Briefly, the amplicon plasmids, either pα4GSPxα22βgal or pα4sβgal, were transfected into E5 cells with lipofectamine in T150 cm2 flasks. Following transfection, after 16–20 h, cells were infected with helper virus strain d120 at either 0.03 or 0.1 multiplicities of infection (MOI). The cells were harvested at the point of 100% cytopathicity. The resulting vector stocks were further prepared by sonication and centrifugation at 3000 g for 10 min. The supernatants were purified by a spin at 70 000 g for 16–18 h through a 25% sucrose cushion in phosphate-buffered saline (PBS). The resulting pellets were resuspended in PBS and titers of amplicon were determined by infecting Vero cells and then counting the number of βgal-expressing cells. The titers of helper virus were determined by infecting E5 cells and using a plaque counting assay.

Infection of cultures with HSV vectors

The titer for the pα4sGSPxα22βgal viral vector used in the experiments was 3.3 × 107 infectious viral particles for amplicon and 7.5 × 107 for helper virus. For the pα4sβgal control, the titer was 2.7 × 107 for the amplicon and 4.7 × 107 for the helper virus. The quantity of amplicon was matched for the experiments. The vector was used to infect mixed neuronal cultures at various MOI ranging from 0.1 to 1. This calculation was based on 1.5 million cells/well of a 6-well plate. The GSPx activity assay, as described above, was run 16 h later. In addition, a Bradford protein assay was run to correct for plate to plate variation. Studies were carried out in 20 md glucose in glucose-free MEM media, supplemented with indicated amounts of glucose.

Neuroprotection experiments

Neurotoxicity was determined using a previously published method (Brooke et al. 1999). Cultures in 96-well plates at 20 md glucose were treated with either the α4sGSPxα22βgal or α4sβgal vectors. Cells were then fixed and neurons stained immunohistochemically with MAP2 monoclonal antibody (Sigma), followed by secondary antibody Vectastain and ABTS kit (all from Vector). ABTS produces a water-soluble green colored product that can be read in an ELISA reader at 405 nm to give a measure of the number of neurons remaining in the culture. In each experiment, control wells consisted of treatment with vehicle solutions only. The values obtained in these wells were set to 0% neuronal loss and all experimental wells were taken as a percentage of these wells.

Redox microphysiometry

For redox microphysiometry (Rabinowitz et al. 1998), approximately 3.0 × 105 cells of mixed primary neuronal culture were plated onto the membrane of a poly-d-lysine coated Cytosensor cell capsule. The cell capsule is assembled and loaded into a microphysiometer sensor chamber with a gold electrode. Prior to each experiment, the chamber and electrode are washed with distilled water and equilibrated with redox medium consisting of PBS (145 md Na+, 4 md K+, 1 md Mg2+, 1 md Ca2+, 143 md Cl, 10 md phosphate, pH 7.4) supplemented with 20 md HEPES, 1 mg/mL of endotoxin-free bovine serum albumin, glucose levels from 0.5 md to 20 md, 300 µd ferricyanide, 300 µd ferrocyanide, and 10 µd menadione made in dimethylsulfoxide (DMSO). The chamber is maintained at 37°C and perfused at 1 μL s−1 with redox medium. The experiments are performed as quickly as possible to mitigate any toxic effects of the menadione. The chamber is perfused for 58-s intervals, followed by a 32-s interval with perfusion. The change in redox potential is measured during seconds 20–30 of the no-perfusion period and defines the reduction rate. In the presence of menadione, the observed reduction rate is mainly a result of the NADPH concentration in the cell.

Data analysis

Data analysis was performed using a combination of Edcel (Microsoft, Belginham, WA, USA) and Sdcel Sdcel (Jandel Scientific, Berkeley, CA, USA). Unless otherwise indicated, data were analyzed by dcela followed by post-hoc test. Statistical significance was determined at the p < 0.05 level. Plots were drawn using SdcelaPdcela software.

Results

We examined the effects of 1 µdcela corticosterone (the species-typical GC of rats) upon GSPx activity in primary hippocampal cultures under conditions of a non-limiting excess of GSH as substrate (Fig. 2). Corticosterone caused a small but significant decrease in activity. This effect occurred in glucose concentrations ranging from hypo- to hyperglycemic, and showed no trend towards an energy dependency.

Figure 2.

Effect of corticosterone on GSPx activity, expressed relative to corticosterone-free controls treated with the same dose of glucose. The x-axis indicates glucose concentration in the media. Corticosterone decreased GSPx activity (F = 4.165; d.f. = 1/75; p < 0.05 by two-way dcela; n ranged from 5 to 14/group). Absolute levels of activity (units of GSPx activity/mg protein) were: 0.0843 ± 0.03 at 25 mdcela glucose; 0.1 ± 0.025 at 5 mdcela; 0.05 ± 0.01 at 1 mdcela; 0.06 ± 0.01 at 0.5 mdcela; 0.06 ± 0.005 at 0.1 mdcela. Mean ± SEM.

We then examined the effects of 1 µdcela corticosterone on levels of total glutathione, GSH and GSSG (Fig. 3). The hormone had no effect on these endpoints in cultures incubated with 5 mdcela glucose. Under 20 mdcela glucose conditions, corticosterone caused a significant decrease in total glutathione concentrations, attributable to roughly equal and significant decreases in both GSH and GSSG levels. Although the glutamatergic excitotoxin KA alone had no effect on these endpoints (relative to KA- and corticosterone-free cultures; data not shown), KA in combination with corticosterone significantly altered glutathione profiles (Fig. 4), causing significant declines in total glutathione levels under both 5 and 20 mdcela glucose conditions. Given that the vast majority of total glutathione is comprised of GSH, the variation in total glutathione is substantially a result of variations in GSH. Thus, it was not surprising that there were also significant declines in GSH. Roughly similar declines occurred with GSSG levels; because of the large variance, however, these were not statistically significant.

Figure 3.

Effect of corticosterone on total glutathione, GSH and GSSG levels, expressed as percentage decline from corticosterone-free controls. *p < 0.05, ***p < 0.001; Newman-Keuls post-hoc test, in comparison to corticosterone-free control under the same conditions, following one-way dcela. The number of data points used, n, ranged from 6 to 26. Absolute numbers in this and Fig. 4 were in arbitrary units. Mean ± SEM.

Figure 4.

Effect of corticosterone on total glutathione, GSH and GSSG levels in kainic acid-treated cultures, expressed as percentage decline from corticosterone-free controls. *p < 0.05, ***p < 0.001; Newman–Keuls post-hoc test, in comparison to corticosterone- and kainic acid-free control under the same conditions, following one-way dcela. The number of data points used, n, ranged from 7 to 14. Mean ± SEM. As noted in the text, kainic acid alone had no effect on these endpoints.

As noted, regeneration of GSH from GSSG requires NADPH. Because of the demonstration of a corticosterone-induced decrease in GSH levels, along with prior demonstrations that GCs decrease ATP levels in hippocampal neurons and glia during insults, we next examined the effects of 1 µdcela corticosterone on NADPH levels. The hormone caused a 35 ± 11% decline in NADPH levels, relative to corticosterone-free controls, under 20 mdcela glucose conditions ( p < 0.05, post-hoc test following two-way dcela; n = 4), and a 20 ± 2% decline under 5 mdcela glucose conditions (p < 0.002; n = 4).

Having demonstrated some corticosterone effects upon GSPx biochemistry, we then wished to determine whether such actions were likely to be of physiological relevance. We first tested whether corticosterone worsened some of the sequelae of ROS accumulation. Corticosterone (1 µdcela), when administered alone, caused a significant degree of lipid peroxidation, over a range of hyper- to hypoglycemic glucose concentrations (Fig. 5). Moreover, the hormone also exacerbated KA-induced lipid peroxidation over a wide range of glucose availability (of note, the converse did not hold, in that KA did not exacerbate GC-induced peroxidation). Corticosterone also worsened KA-induced neurotoxicity (Table 1).

Figure 5.

Effects of corticosterone and/or kainic acid (50 µdcela) on lipid peroxidation. *p < 0.05, **p < 0.01, ***0.010.001; Newman–Keuls post-hoc test following one-way dcela, compared with insult-free control under equivalent glucose conditions. n = 22, 15, 10, 10 and 6 for 25, 5, 2.5, 1, 0.1 mdcela glucose, respectively. Absolute numbers were in arbitrary fluorescence units. Mean ± SEM.

Table 1.  Effects of corticosterone and of kainic acid on neuronal survival in hippocampal cultures
TreatmentPercentage neuron death
  1. n = 17/group; *p < 0.05, ***p < 0.001; Newman–Keuls post-hoc test following one-way dcela in comparison to control. ###p < 0.001 by post-hoc test when compared with kainic acid alone. Media contained 5 mdcela glucose.

Control 0 ± 3
Kainic acid (50 µdcela)19 ± 3*
Corticosterone (1 µdcela) 5 ± 4
Kainic acid + corticosterone39 ± 2***,###

These data indicate that corticosterone exacts an oxidative price upon these hippocampal neurons. However, they do not indicate whether corticosterone augments KA-induced neurotoxicity at least in part because of its effects on GSPx activity. We next tested this possibility. This could be carried out in a number of ways. As a first approach, one could selectively decrease GSPx activity to the same extent as does corticosterone, but in the absence of the hormone, and determine whether that manipulation augments KA toxicity. No such manipulation was available, however. Instead, we used the converse approach of overexpressing GSPx; we did so by constructing pα4GSPxα22βgal, a bipromoter herpes simplex virus-1 amplicon vector overexpressing GSPx. In what would be superficially the most appropriate version of this experiment, one would test whether overexpressing GSPx to the point of reversing the corticosterone effect on GSPx activity would also block the corticosterone effect on neurotoxicity. However, as demonstrated, corticosterone not only decreases GSPx activity but also decreases substrate availability to GSPx. This would confound the interpretation of any effects of GSPx overexpression.

Thus, we studied this issue in the absence of corticosterone, determining whether increasing GSPx activity (with this vector) to roughly the same extent as corticosterone reduces such activity would enhance survival in the face of kainic acid. We first documented the functioning of this vector: infection with it caused a significant increase in GSPx activity, relative to infection with pα4βgal, a control vector expressing reporter gene alone (11 ± 0.3% increase at an MOI of 0.1, p < 0.02; 18 ± 1% increase at an MOI of 0.2, p < 0.05. n = 3). We then found that infection with pα4GSPxα22βgal at an MOI of 0.2 was significantly protective against KA neurotoxicity (Fig. 6).

Figure 6.

Protective effects of GSPx overexpression on kainic acid-induced neurotoxicity. Cultures were infected at a multiplicity of infection of 0.2 with either pα4GSPxα22βgal (overexpressing GSPx), or control vector expressing βgal alone. *p < 0.05, **p < 0.01; Newman–Keuls post-hoc comparison following one-way dcela, comparison with control vector at the same kainic acid concentration. n = 35/group. Mean ± SEM.

Discussion

As reviewed, GCs can compromise the ability of hippocampal neurons to survive various necrotic insults. One component of this endangerment involves the steroid increasing the glutamate and calcium burdens on these neurons. Such an increase is likely to lead to ROS accumulation through calcium-dependent activation of nitric oxide synthase, of phospholipases and of xanthine oxidase. In addition to increasing the ROS load on hippocampal neurons, GCs also compromise antioxidant defenses, decreasing the activity of GSPx (McIntosh et al. 1998a). In the present report, we have uncovered some mechanisms in primary hippocampal cultures by which this inhibition may arise.

Corticosterone effects on GSPx activity

As a first observation, we found that corticosterone decreased GSPx activity. Because this assay was carried out under circumstances of substrate excess, this implied a direct corticosterone effect, rather than one secondary to substrate limitation. Potentially, corticosterone could be decreasing the activity and/or the number of GSPx molecules; if the latter was occurring, this could involve effects on GSPx transcription, translation or degradation. The question of which one is occurring is currently under study. In support for direct transcriptional effects, 800 bp upstream of the human GSPx gene are two near consensus hexameric half-site sequences for the glucocorticoid response element (GRE) (Moscow et al. 1992). The corticosterone inhibition of GSPx activity was not augmented under low glucose conditions. Furthermore, while the inhibition of GSPx activity was significant, it was small, ranging up to approximately 14%. Whether this is of physiological significance is considered below.

Corticosterone effects on availability of substrate for GSPx

We also observed that corticosterone decreased total glutathione levels both in the presence and absence of KA. This effect was matched (both statistically, and on the level of percentage decline) by an inhibition of GSH levels. This was not surprising, given that the vast majority of total glutathione is comprised of GSH (Dringen 2000), and agrees with a prior report of GCs decreasing GSH levels in blood and muscle (Orzechowski et al. 2000). The steroid also inhibited GSSG levels; however, these effects were either somewhat smaller than the inhibition of glutathione and GSH levels or more variable, resulting in fewer instances of statistical significance. As a precedent for this pattern, various types of ischemic insults to the brain can reduce total glutathione without altering GSSG (Cooper et al. 1980).

The mechanisms underlying these corticosterone effects are likely to be complex. As noted, statistically the glutathione depletion was preferentially attributable to a depletion of GSH. NADPH is required for the regeneration of GSH from GSSG, and the 20–35% depletion of NADPH by corticosterone under the same glycemic conditions could theoretically explain the decline in GSH. As a possible explanation for the decline in NADPH levels, GCs decrease glucose transport in hippocampal neurons and glia (Kadekaro et al. 1988; Horner et al. 1990; Virgin et al. 1991), and glucose availability can regulate neuronal NADPH levels and GSH/GSSG ratios (Delgado-Esteban et al. 2000).

Initially, this seems unlikely to be an explanation for the data observed. Should corticosterone solely be constraining the regeneration of GSH, there should be a complementary accumulation of GSSG, and no change in total glutathione levels. Instead, total levels declined as did GSSG in some instances. However, this could have been caused by the release of GSSG into the medium, for which there is some precedent (Dringen 2000).

Corticosterone might also have been directly decreasing the total glutathione pool. As possible mechanisms, glutathione is both taken up through the blood–brain barrier as well as synthesized within the brain (Kannan et al. 1990; Favilli et al. 1997), and corticosterone may potentially be altering either of these effects. As an additional possible mechanism, excitotoxic insults deplete intracellular glutathione in the nervous system, as a result of its release into the extracellular space (Orwar et al. 1994; Yang et al. 1994; Oyama et al. 1997; Almeida et al. 1998). Although some authors have interpreted this as reflecting the release of glutathione from glia for transfer to neurons (i.e. a beneficial compensation) (Yudkoff et al. 1990; Dringen et al. 1997), at least some of the efflux appears to reflect loss of neuronal glutathione (Zangerle et al. 1992). We are presently studying whether corticosterone alters such glutathione efflux, and from which cell type.

As noted, excitatory amino acids can deplete total glutathione, something that we did not observe with KA. However, the sole report of this finding (Oyama et al. 1997) involved generally higher concentrations and a longer exposure period to KA than in the present study.

Physiological consequences of these corticosterone actions

There are likely to be a number of ramifications of these corticosterone effects on glutathione profiles. The decrease in GSH levels will have obvious implications, insofar as it is a substrate for the peroxidase activity of GSPx. When coupled with the small, directly inhibitory effects of corticosterone on GSPx itself, this may mean that the steroid can limit the antioxidant efficacy of GSPx during excitotoxic insults. As discussed, we have observed this in vivo (McIntosh et al. 1998a). Moreover, we find that corticosterone blocks the increase in GSPx activity seen after exposure of hippocampal cultures to gp120, the neurotoxic coat protein of HIV (Brooke et al. 2002).

Independent of its role as a substrate for GSPx activity, GSH itself protects against oxidative insults (Saez et al. 1990; Winterbourn and Metodiewa 1994), facilitating the glutathione S-transferase-mediated detoxification of lipid peroxides (Lawrence and Burk 1976; Prohaska and Ganther 1976) and the non-enzymatic detoxification of peroxides (Spector et al. 1997). The importance of these direct glutathione actions is demonstrated by the fact that glutathione depletion enhances loss of dopamine neurons from GSPx-deficient animals (Nakamura et al. 2000).

Thus, corticosterone has various potentially adverse effects on GSPx-related biochemistry. Supporting this, we observe that corticosterone worsened KA-induced neurotoxicity and lipid peroxidation (the latter most likely being predominately in neurons). Moreover, corticosterone worsens the neurotoxicity of ROS generators (Sapolsky et al. 1988; McIntosh and Sapolsky 1996). However, these actions could arise from effects on other aspects of ROS biochemistry, rather than upon GSPx and related compounds (e.g. the up-regulation of 5-lipoxygenase in the hippocampus by GCs; Uz et al. 1999). We examined this more directly. When we overexpressed GSPx in order to increase GSPx activity by approximately 15% (the magnitude of the corticosterone inhibition of GSPx activity), we saw significant protection from KA neurotoxicity (in agreement with the finding that GSPx overexpression protects cultured cortical neurons from amyloid β-peptide; Barkats et al. 2000). The magnitude of protection per neuron is likely to be even greater, in that this titer of vector infects only approximately 50% of neurons under these culturing conditions (McLaughlin et al. in preparation). (Moreover, under these conditions, approximately 15% of glia are infected – whether this may indirectly aid neuronal survival is currently under study.) These findings suggest that the corticosterone effect on GSPx could well have implications for neuronal vulnerability. That such a small corticosterone effect might be consequential is commensurate with the extremely tight control of oxidative biochemistry. For example, Sohal and Dubey (1994) showed that a 10% increase in hydrogen peroxide was associated with a 20% increase in protein carbonyl formation in young adult house flies, and that carbonyl content was inversely correlated with their life expectancy.

In conclusion, corticosterone has a variety of disruptive effects upon GSPx and glutathione antioxidant defenses, and this could well help explain how the steroid increases the neurotoxicity of a variety of necrotic insults. One fairly puzzling feature of these findings concerns energy availability. The endangerment of neurons by GCs has been attributed to their adverse effects upon cerebral metabolism (Sapolsky 1996); the hormone decreases glucose uptake in hippocampal neurons and glia (Kadekaro et al. 1988; Horner et al. 1990; Virgin et al. 1991), and accelerates the decline of ATP concentrations during necrotic insults (Tombaugh and Sapolsky 1992; Lawrence and Sapolsky 1994). In this model, the energy disruption compromises the ability of neurons to carry out the costly tasks of containing the excesses of glutamate, cytosolic calcium and ROS. Superficially, the present data fit that model, insofar as corticosterone decreased NADPH levels, with likely influences upon GSH levels and, ultimately, GSPx activity. However, it is more striking the extent to which the present findings are independent of energy status (i.e. a decrease in both GSH and GSSG, rather than a shift from the former to the latter; a decrease in GSPx activity, independent of substrate availability). Moreover, in one instance (Fig. 3), corticosterone decreased glutathione levels to a greater extent at 20 mdcela glucose than at 5 mdcela. Thus, it appears as if the disruptive effects of corticosterone on GSPx-related biochemistry are not merely secondary to adverse effects upon hippocampal energetics. The nature of these non-energetic mechanisms is currently under study.

Acknowledgements

Technical support was provided by Harden McConnell, William Ogle and Judith Vacchino. Manuscript assistance was provided by Angela Lee, Lisa Pereira and Mani Roy. Funding was provided by NIH grants MH53814 and NS37520, and State of California TRDRP Grant 8RT-0073, to RS, and an Undergraduate Research Opportunity grant to RP.

Ancillary