Differential effect of nitric oxide on glutathione metabolism and mitochondrial function in astrocytes and neurones: implications for neuroprotection/neurodegeneration?

The tripeptide glutathione (GSH; γ-glutamylcysteinylglycine) is an important antioxidant within the brain (Dringen et al. 2000). GSH is synthesized by the ATP-dependent enzymes glutamate-cysteine ligase (GCL; EC 6.3.2.3 formerly known as γ-glutamylcysteine synthetase; Yip and Rudolph 1976; Meister and Anderson 1983) and GSH synthetase (EC 6.3.2.3). GCL is heterodimeric with a large catalytic subunit (GCL_h) and a small modifier subunit (GCL_l: Huang et al. 1993; Tu and Anders 1998). The modifier subunit modulates enzyme activity by affecting the affinity of the catalytic subunit for both substrates and inhibitors (Huang et al. 1993; Tu and Anders 1998). GCL is thought to be the rate-limiting enzyme of GSH synthesis (Lu 2000). Astrocytes in culture have been shown to release GSH (Yudkoff et al. 1990; Sagara et al. 1996; Stone et al. 1999). The ectoenzyme γ-glutamyltranspeptidase (γ-GT, γ-glutamyltransferase) metabolizes GSH to cysteinylglycine and a γ-glutamyl moiety linked to an acceptor nucleophile (Meister and Anderson 1983). Cysteinylglycine (Cys Gly) is a precursor for de novo GSH synthesis in both astrocytes and neurones (Dringen et al. 1997, 1999). Indeed, an increase in neuronal GSH concentration is observed when neurones are cocultured with astrocytes (Bolanos et al. 1996; Stewart et al. 1998).

Perturbed GSH metabolism within the brain has been implicated in the pathogenesis of several neurological diseases (Jenner and Olanow 1998; Marcus et al. 1998; Schulz et al. 2000). In Parkinson's disease (PD), a 40% loss of GSH has been reported in the substantia nigra (Sian et al. 1994a). In postmortem brains the depletion of reduced GSH does not appear to be due to altered GCL activity (Sian et al. 1994b). However, an increase in γ-GT activity has been observed in the substantia nigra in PD brains (Sian et al. 1994b). A similar loss of GSH has been reported in the substantia nigra of individuals with Incidental Lewy body disease (thought to be pre-symptomatic PD; Dexter et al. 1994). The loss of GSH precedes other hallmarks of PD (e.g. loss of mitochondrial complex I activity, altered iron metabolism), indicating that the loss of GSH may be an early and important process in the pathogenesis of PD (Dexter et al. 1994).

There is also evidence for nitric oxide (NO) involvement in PD. NADPH diaphorase activity, a putative marker for NO synthase, is observed in glial cells in postmortem PD brains (Hunot et al. 1996). Furthermore, the MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine) model of PD in mice is associated with a significant up-regulation of inducible NO synthase in the substantia nigra (Liberatore et al. 1999). In addition, mice lacking inducible NO synthase were more resistant to MPTP associated dopaminergic neurodegeneration (Liberatore et al. 1999). Inhibition of complex I of the mitochondrial electron transport chain has been observed in PD (Schapira et al. 1990). Exposure to high concentrations of NO decreases GSH in neurones (Bolanos et al. 1996). Furthermore, the reduction in GSH during prolonged exposure to NO is accompanied by a gradual and persistent inhibition of complex I in J774 and Jurkat cells (Clementi et al. 1998; Beltran et al. 2000).

We have therefore investigated the effect of NO on GSH metabolism, and the consequences of this on the function of the mitochondrial respiratory chain and on cell viability, in astrocytes and neurones. Given the importance of astrocytes in providing GSH precursors for neuronal GSH synthesis (Dringen et al. 1999), the efflux of GSH and activity of γ-GT in astrocytes exposed to NO was also investigated.

Materials and methods

Materials

Minimal essential medium, fetal bovine serum (FBS), and tissue culture plastics were purchased from Life Technologies (Renfrewshire, UK). (Z)-1-[2-Aminoethyl)-N-(2-ammonioethyl)amino]diazen-1-ium-1,2-diolate (DETA-NO) was purchased from Alexis Biochemicals (Nottingham, UK). Centrifugal filter devices were purchased from Amicon (Watford, UK). γ-glutamylcysteine standards were supplied by Bachem Feinchemikalien AG (Bubendorf, Switzerland) and were a gift from Dr Ralf Dringen (University of Tubingen, Germany). The γ-glutamyltranspeptidase assay kit was purchased from Sigma Chemicals (Poole, UK). All other chemicals were bought from either Sigma Chemicals or Boehringer Mannheim (Lewes, UK).

Cell culture

Wistar rats were purchased from A. J. Tuck and Sons Ltd (Rayleigh, Essex) and humanely killed as described under Schedule 1 of the UK Animal (Scientific Procedures) Act 1986. Primary cortical astrocyte cultures were prepared from Wistar rat neonates (0–2 days) as previously described by Bolanos et al. (1995). Cells were cultured in d-valine-based minimal essential medium (Cholewinski et al. 1989) supplemented with 10% (v/v) FBS and 2 mm l-glutamine for 7 days in 80-cm² flasks in an incubator at 37°C (95% air/5% CO₂). The cells were then cultured in l-valine-based minimal essential medium (supplemented as above and referred to as astrocyte media in text) for a further 6 days. Astrocytes at day 13 in vitro (24 h prior to treatment) were removed from the flasks with trypsin, and plated at a density of 1 × 10⁶ cells/well onto poly-l-lysine-coated 6-well plates. Primary neurone cultures were prepared from Wistar rats (embryonic day 17) as described by Bolanos et al. (1995). Cells were cultured in d-valine-based minimal essential medium for 3 days supplemented with 10% (v/v) FBS, 25 mm KCl and 2 mm l-glutamine (referred to as neuronal media in text) in 6-well plates coated with poly-L-ornithine. Addition of 25 mm KCl enhances neuronal survival in the presence of serum (Lasher and Zagon 1972). On day 3, the medium was changed for fresh neuronal medium containing 10 µm cytosine arabinoside and maintained for a further 3 days prior to experimental procedures.

Treatment of astrocytes and neurones

Astrocytes and neurones were incubated with 1 mL of astrocyte or neuronal media, respectively, in the absence/presence of 0.5 mm DETA-NO on day 14 and day 6, respectively. Decomposed DETA-NO, which does not release any detectable NO, was also used as a control to demonstrate that the effects observed were due to the release of NO. Following exposure of astrocytes/neurones to DETA-NO for a set period, media was removed, and the cells washed twice with Hanks' Balanced Salt Solution (HBSS) to remove dead cells. The cells were then trypsinized, pelleted and resuspended in isolation medium (320 mm sucrose, 10 mm Tris, 1 mm EDTA, pH 7.4), and stored at − 70°C until required except where stated.

GSH quantification

GSH levels were determined electrochemically following extraction of GSH into 15 mm ortho-phosphoric acid and separation by reverse-phase high performance liquid chromatography (HPLC; Riederer et al. 1989).

GCL assay

GCL (EC 6.3.2.2) activity was determined by reverse-phase HPLC with electrochemical detection as previously described (Gegg et al. 2002). GCL activity was calculated by measuring the amount of γ-glutamylcysteine (γ-GC) synthesized over a defined period of time and related to the protein content of the sample assayed. Briefly, cell homogenates were centrifuged through 10 kDa molecular mass cut off filters to remove cellular GSH/endogenous substrates. The retained protein was then incubated with ATP, cysteine (both 10 mm), glutamate (40 mm) and 220 µm acivicin in assay buffer (100 mm Tris, 150 mm KCl, 20 mm MgCl₂ and 2 mm EDTA; pH 8.2) at 37°C. γ-GC was extracted into 15 mm ortho-phosphoric acid and resolved by reverse-phase HPLC using the electrochemical conditions described by Riederer et al. (1989).

GSH release from astrocytes

Astrocytes (1 × 10⁶ cells/well) were incubated in astrocyte medium for 24 h in the absence or presence of 0.5 mm DETA-NO. The medium was removed and the cells washed twice in 2 mL HBSS. GSH release into 1 mL of minimal medium (44 mm NaHCO₃, 110 mm NaCl, 1.8 mm CaCl₂, 5.4 mm MgSO₄, 0.92 mm NaH₂PO₄, 5 mm glucose, adjusted with CO₂ to pH 7.4 as described by Dringen et al. 1997) after 1, 2, or 4 h was measured. GSH was extracted into one volume of 15 mm ortho-phosphoric acid and quantified as described above.

γ-GT assay

γ-GT (EC 2.3.2.2) activity was measured using the Sigma assay kit. Following treatment, astrocytes were scraped into HBSS. Activity was calculated by measuring the production of 5-amino-2-nitrobenzoate at an absorbance of 405 nm following addition of the substrates l-γ-glutamyl-3-carboxy-4-nitroanilide and glycylglycine to the astrocyte preparation. Enzyme activity was measured at 37°C and was totally abolished when incubated with the γ-GT inhibitor acivicin (220 µm; Stole et al. 1990).

Determination of mitochondrial respiratory chain enzyme activities

Cell suspensions were freeze-thawed three times and activities determined using a Uvikon 940 spectrophotometer (Kontron Instruments, Watford, UK). NADH-CoQ1 reductase (complex I, EC 1.6.5.3), succinate-cytochrome c reductase (complex II + III, EC 1.3.5.1 and EC 1.10.22, respectively), cytochrome c oxidase (complex IV, EC 1.9.3.1), and citrate synthase (EC 4.1.3.7) were measured as described by Bolanos et al. (1995). Neuronal complex activity was also expressed against citrate synthase activity to account for the possible decrease in mitochondrial number.

Lactate dehydrogenase release

The percentage of lactate dehydrogenase (LDH, EC 1.1.1.27) released into minimal essential medium was calculated by: LDH activity in minimal essential medium/(Total LDH activity in media + cells). The oxidation of NADH was measured spectrophotometrically at 340 nm as described by Bolanos et al. (1995).

RNA isolation and northern blot analysis

Total RNA was isolated from astrocytes (∼ 5 × 10⁶ cells) as previously described (Chomczynski and Sacchi 1987). RNA was loaded (20 µg/lane) on a 1% (w/v) agarose-formaldehyde gel, separated by electrophoresis, transferred to a GeneScreen Plus membrane (NEN Life Science, Boston, USA) and cross-linked with ultraviolet irradiation. Membranes were incubated in hybridization solution [1% (w/v) sodium dodecyl sulphate, 1 m NaCl, 10% (w/v) dextran sulphate] containing ³²P-labelled GCL_h, GCL_l, or cyclophilin cDNA probes for 18 h at 60°C. The rat 1.1 kb GCL_h and 0.9 kb GCL_l cDNA fragments (EMBL accession numbers J05181 and S65555, respectively) were cloned from rat brain RNA (1 µg) by RT-PCR using the primers: 5′-CCGGAATTCGCCATGGGGCTGCTG-3′ (5′ position 24) and 5′-TGCCAGAAGGTGATCGATGCCTT-3′ (3′ position 1117) for GCL_h; and 5′-CGCGGATCCCCTCGGGCGGCAGCT-3′ (5′ position 24) and 5′-CGCGGATCCTAAATACAAGGCCCCTGAG-3′ (3′ position 905) for GCL_l. A 0.7-kb cDNA fragment of rat cyclophilin gene (generously donated by Dr Dionisio Martín-Zanca, Universidad de Salamanca, Spain) was used to control for total RNA loaded in each lane. After hybridization, the membrane was washed and exposed to Kodak XAR-5 film for 2–3 days at − 70°C. Autoradiograms were scanned, and the density of the GCL_h, GCL_l and cyclophilin mRNA bands were quantified using image-analyser software (NIH Image, National Institutes of Health, USA). The density of the GCL_{h or l} mRNA bands were expressed as a ratio against cyclophilin mRNA band density. The GCL_{h or l}/cyclophilin mRNA ratio at 0 h was arbitrarily given a value of 1, and the GCL_{h or l}/cyclophilin mRNA ratios at 9 and 24 h compared to this.

Measurement of NO

The steady state concentration of NO generated by 0.5 mm DETA-NO in astrocyte/neurone medium was measured using an ISO NO electrode following incubation at 37°C for 0.5, 4 and 24 h (WPI, Sarasota, FL, USA; Brown et al. 1995).

Protein determination

Protein concentration was determined by the method of Lowry et al. (1951).

Statistical analysis

Results are expressed as mean ± SEM values for the number of independent cell culture preparations indicated. The Student's t-test was used to analyse data from DETA-NO treated cells and their time-matched controls, and one-way anova followed by the least significance difference test was used for comparisons between multiple groups. Enzyme activity expressed against the mitochondrial marker enzyme citrate synthase was transformed using the following equation: arcsin (√enzyme activity/citrate synthase activity). Data were transformed in this way because it minimizes the negative skew of distribution produced when expressing proportions (personal communication from Professor Richard Lowry, Professor of Psychology, Vassar College, New York, USA). Following transformation, the statistical significance of the data was determined by the Student's t-test. In all cases, p < 0.05 was considered significant.

Results

Measurement of NO generated by DETA-NO

DETA-NO (0.5 mm) in either astrocyte or neuronal medium generated a steady state NO concentration that was constant for at least 24 h. The concentration of NO in astrocyte or neuronal medium following incubation at 37°C was estimated to be 0.95 ± 0.05 µm after 30 min, 0.90 ± 0.15 µm after 4 h, and 0.93 ± 0.07 µm after 24 h (n = 3–5).

GSH levels in astrocytes and neurones following exposure to NO

Astrocytes were treated with 0.5 mm DETA-NO for 6, 9, and 24 h and the cellular GSH concentration determined at each time point (Fig. 1a). The GSH concentration in astrocytes was significantly elevated following exposure to NO for 24 h compared to time-matched controls. Note that GSH levels in both control and DETA-NO treated astrocytes incubated for 9 h were also significantly greater compared to astrocytes prior to incubation with fresh media (control 0; Fig. 1a). Exposure of astrocytes to DETA-NO for 24 h had no effect on viability [control, 5.0 ± 1.2; NO-treated, 6.5 ± 1.1% LDH released into media (n = 5)].

Treatment of neurones with 0.5 mm DETA-NO for 24 h resulted in extensive neuronal death (∼ 95–100% determined by morphological examination). This was not apparent (determined as percentage of LDH release) following 9 h of exposure to NO [control, 5.0 ± 2.4; NO-treated, 7.5 ± 2.3% (n = 4)]; however, after 18 h, a significant increase in LDH release was observed [control, 4.9 ± 1.7; NO-treated, 12.2 ± 2.3% (n = 4), p < 0.05]. GSH concentrations in neurones were unaltered after 6 or 9 h of exposure to NO, but were decreased by 44% in surviving cells following 18 h of exposure (Fig. 1b). Neurones were also exposed to 0.05, 0.1 mm, and 0.25 mm DETA-NO for 18 h to determine whether neurones increased GSH levels when exposed to lower NO concentrations (e.g. 0.05 mm DETA-NO generated a steady state NO concentration of ∼ 100 nm). GSH levels were similar to control levels [control; 10.1 ± 0.8; 0.05 mm, 8.7 ± 0.4; 0.1 mm, 9.7 ± 0.6; 0.25 mm, 8.3 ± 1.2 nmol GSH/mg protein (n = 4)]. LDH release was unaffected in neurones treated with these lower concentrations of DETA-NO (results not shown). Neuronal media, unlike astrocyte media, contains 25 mm KCl. However, GSH levels were still significantly elevated in astrocytes exposed to 0.5 mm DETA-NO when grown in neuronal media (control, 20.0 ± 0.2; NO-treated, 31.5 ± 0.8 nmol GSH/mg protein, p < 0.05), suggesting that the presence of KCl in the media does not prevent elevation of GSH levels.

Cysteine is thought to be the rate-limiting substrate for GSH synthesis in cultured astrocytes and neurones (Kranich et al. 1996; Dringen et al. 1999). Astrocytes are thought to prefer cystine as the precursor for GSH synthesis (Kranich et al. 1996), whereas neurones rely on cysteine for de novo GSH synthesis (Sagara et al. 1993; Dringen et al. 1999). Therefore, perhaps the autooxidation of cysteine in neuronal media may cause the depletion of neuronal GSH levels upon exposure to NO. Neurones were therefore incubated in the absence or presence of 0.5 mm DETA-NO for 16 h. The neuronal media of both control and NO-exposed neurones was then supplemented with 350 µm cysteine, a concentration that has previously been shown to be non-toxic to neurones and to elevate GSH levels within 1 h in cysteine-starved neurones (Kranich et al. 1996; Dringen et al. 1999), and the neurones incubated for a further 2 h. GSH levels in neurones exposed to NO were still significantly depleted, compared to control neurones in the presence of supplemental cysteine [control + cysteine, 11.5 ± 0.6; NO-treated + cysteine, 8.2 ± 0.6 nmol GSH/mg protein, p < 0.01 (n = 4)]. Control neurones in the absence of supplemental cysteine had 11.0 ± 1.5 nmol GSH/mg protein (n = 4). These results suggest that the cysteine concentration in neuronal media does not limit GSH levels in neurones.

The effect of DETA-NO on mitochondrial respiratory chain enzyme activities in astrocytes and neurones

There was greater inhibition of the mitochondrial respiratory chain in neurones than in astrocytes exposed to DETA-NO (Table 1). There was also a 26% loss of citrate synthase activity in neurones treated with DETA-NO for 18 h, indicating a possible loss of mitochondrial number. Citrate synthase activity was unaffected in astrocytes exposed to NO. A 39% and 35% loss of neuronal complex II + III and IV activity, respectively, was observed following 9 h of exposure to NO compared to time-matched controls. Loss of neuronal complex I, II + III and IV activity (31%, 52% and 64%, respectively) was observed after 18 h of DETA-NO treatment. When neuronal complex activity is expressed against citrate synthase activity, significant inhibition of complex II + III [control, 0.29 ± 0.01; NO-treated, 0.23 ± 0.01 (n = 6) p < 0.05] and IV activity [control, 0.11 ± 0; NO-treated, 0.08 ± 0.01 (n = 6), p < 0.05] persists, whereas loss of complex I activity is not observed. No loss of complex activity was observed in astrocytes treated with DETA-NO for 9 h, whereas a 23% and 29% loss of complex II + III and IV, respectively, was observed in astrocytes treated for 24 h. These findings are comparable to the pattern of inhibition observed following induction of inducible NO synthase in astrocytes (Bolanos et al. 1994).

GCL activity in NO-exposed astrocytes and neurones

The activity of GCL, the rate-limiting enzyme in GSH synthesis, was measured in astrocytes treated with DETA-NO to determine whether an alteration in GCL activity contributes to the elevation in GSH levels. GCL activity was significantly increased in astrocytes treated with DETA-NO for 9 and 24 h (Fig. 2a). Note that GCL activity in control astrocytes following 6 and 9 h of incubation with fresh media was also significantly greater than prior to change of media (control 0). This may explain why GSH levels were also significantly greater in control astrocytes after 9 h (Fig. 1a). Neurones exposed to DETA-NO (0.5 mm) did not change GCL activity compared to control values after 6, 9 or 18 h of incubation (Fig. 2b). The toxicity of this concentration of DETA-NO to neurones may, for example, inhibit protein synthesis, and account for lack of increased GCL activity in neurones. However, GCL activity was unchanged in neurones exposed to 0.05 or 0.25 mm DETA-NO for 18 h, compared to control [0.05 mm; 0.97 ± 0.11; 0.25 mm; 1.09 ± 0.05 nmol γ-GC synthesized/min/mg protein (n = 3)]. These concentrations of DETA-NO had no effect on LDH release and caused much less damage to the ETC (results not shown). Note that as in astrocytes, GCL activity in control neurones was significantly higher 9 h after change of media (Fig. 2b). However, GSH levels in control neurones at 9 h although higher, were not significantly greater compared to cells prior to change of media (Fig. 1b).

GCL mRNA levels in astrocytes treated with DETA-NO

RNA was extracted from astrocytes treated with DETA-NO for 9 or 24 h to determine whether there was an increase in GCL_h and GCL_l mRNA levels. GCL_h mRNA was elevated 1.5-fold following 9 h of DETA-NO exposure (Fig. 3). No increase in GCL_h mRNA was observed at 24 h. GCL_l mRNA levels were elevated following DETA-NO exposure for both 9 and 24 h by 1.9-fold and 1.6-fold, respectively (Fig. 3).

GSH efflux in astrocytes treated with DETA-NO

Astrocytes were treated with DETA-NO for 24 h, washed in HBSS to remove NO, and the rate of GSH release into minimal medium was measured (see Materials and methods), to determine whether exposure to NO affected GSH efflux. In astrocytes exposed to NO, the rate of GSH efflux was significantly greater than in untreated cells for at least 4 h after NO was removed (Fig. 4a). Intracellular GSH levels were maintained in control cells during this period and for the first 2 h in NO-treated astrocytes (Fig. 4b). However, GSH levels were not maintained in NO-treated astrocytes 4 h into the experiment. No increase in LDH release was observed from NO-treated astrocytes throughout the course of the GSH release study (results not shown).

γ-GT activity in astrocytes exposed to NO

Astrocytic-released GSH is cleaved by the ectoenzyme γ-GT. We therefore investigated the activity of γ-GT in astrocytes exposed to NO. Activity was unchanged following 1, 3 or 9 h of NO exposure. However, γ-GT activity was elevated in astrocytes treated with DETA-NO for 18 and 24 h by 28% and 42%, respectively (Fig. 5).

Discussion

When compared to astrocytes, the data presented here suggests that neurones are more susceptible, as judged by mitochondrial respiratory chain enzyme activities and LDH release, to NO exposure. This data is in accordance with previous studies (Bolanos et al. 1995; Almeida et al. 2001). Although these studies showed that astrocytes are able to invoke a glycolytic response that is fundamental to early protection against NO induced inhibition of respiration, our study shows that astrocytes, but not neurones, are able to increase the concentration of GSH upon exposure to NO. Following prolonged exposure to NO, GSH in astrocytes was approximately doubled, whereas GSH was depleted by up to 44% in neurones. The increased GSH concentration in astrocytes could be attributed to the increase in GCL activity observed following exposure to NO. We also observed increased mRNA levels for both GCL_h (catalytic) and GCL_l (modifier) subunits in NO-treated astrocytes. This suggests that the increase in GCL activity observed in NO-treated astrocytes is due, at least in part, to increased expression of the enzyme.

It cannot be discounted that the presence of DETA-NO may also reduce the decline in GSH levels and GCL activity usually observed between 9 and 24 h in control astrocytes. The reason for the significant increase in GCL activity in control astrocytes 6 and 9 h after a change of media is unclear. Since the amount of GCL_h and GCL_l mRNA was the same at 9 h as it was prior to feeding (control 0) and after 24 h, this suggests that post-translational modification of GCL or an increase in protein half-life may transiently increase enzyme activity following a change of media. Dephosphorylation of GCL has been reported to increase enzyme activity (Sun et al. 1996).

GCL activity was not increased in neurones upon exposure to NO. The activity of GCL in untreated astrocytes was approximately nine-fold greater than in neurones, supporting previous findings (Makar et al. 1994). The lower GCL activity in untreated neurones, compared to astrocytes, may contribute to the lower basal concentration of GSH in these cells. Furthermore, the low GCL activity in neurones could mean that, upon acute exposure to NO (e.g. 18 h), the rate of GSH depletion may be greater than the rate of de novo GSH synthesis, resulting in a net loss of GSH. Since no inhibition of GCL activity was observed in neurones treated with NO, the low neuronal GCL activity, and the inability of neurones to increase GCL activity upon exposure to NO, could contribute towards the greater susceptibility of neurones to oxidative stress.

The observation that GSH levels were not significantly increased in either control or DETA-NO treated neurones 9 h after change of media, despite a significant increase in GCL activity in both, may suggest that it is substrate availability in neuronal media, rather than GCL activity as described above, that is limiting neuronal GSH synthesis in both control and NO-exposed cells. However, this does not appear to be the case because supplementation of neuronal media with cysteine did not (i) prevent the NO-mediated depletion of neuronal GSH levels, or (ii) increase GSH levels in control neurones when compared to untreated neurones not incubated with extra cysteine. Previous studies have shown that it is only cysteine in cell culture media that is the rate-limiting substrate for de novo GSH synthesis in cultured astrocytes and neurones, and not glycine or glutamate (Yudkoff et al. 1990; Dringen et al. 1999). A small lag possibly occurs between increased GCL activity and significantly elevated GSH levels and may explain the observed discrepancy. Indeed, astrocytes show such a lag between increased GCL activity and significantly elevated GSH levels.

It should be noted that it is not known whether the effects described in this study on the various facets of GSH metabolism were due to NO alone. The reactive nature of NO may mean that metabolites such as peroxynitrite, NO^– or NO⁺ may also be involved in the effects observed. No increase in GSH levels have been observed in astrocytes treated with peroxynitrite (Bolanos et al. 1995; Barker et al. 1996). However, in these two studies, peroxynitrite was administered as a bolus for 30 min and GSH levels determined 24 h later, rather than incubated for a prolonged period with peroxynitrite. Also GCL activity was not measured in these astrocytes to see if enzyme activity was elevated to maintain GSH levels. Induction of astrocytic NO synthase by cytokines has also been reported to have no significant effect on GSH levels (Bolanos et al. 1994; Chatterjee et al. 2000). Once again GCL activity in these cells was not measured. The wide range of effects invoked by cytokine stimulation, in addition to induction of NO synthase, may also lead to the modulation of different pathways upon exposure to NO that are able to protect astrocytes from oxidative stress.

A variety of chemical and physical treatments (e.g. metals, oxidants, GSH depletion) have been shown to induce transcription of either, or both, the GCL_h and GCL_l genes in endothelial, muscle and astrocyte cells (reviewed by Soltaninassab et al. 2000; Iwata-Ichikawa et al. 1999; Wild and Mulcahy 2000). Our results in astrocytes support those of Iwata-Ichikawa et al. (1999), who showed that astrocytes, but not neurones, are capable of increasing expression of GCL mRNA when exposed to hydrogen peroxide or 6-hydroxydopamine. The results in the present study also suggest that astrocytes are similar to both smooth muscle and endothelial cells, which induce expression of both the GCL_h and GCL_l genes following exposure to NO (Moellering et al. 1998, 1999).

Cloning of the 5′ flanking regions of both GCL genes has identified putative antioxidant response elements (also referred to as electrophile response element, EpRE; Mulcahy and Gipp 1995; Moinova and Mulcahy 1998). Induction of detoxifying enzymes such as glutathione S-transferases and NAD(P)H quinone-oxidoreductase has been shown to be mediated by EpREs (Wild and Mulcahy 2000). In the case of glutathione S-transferase, EpRE-mediated expression occurred following exposure to peroxynitrite (Kang et al. 2002). Recently, Murphy et al. (2001) showed that induction of EpRE-mediated gene expression is largely restricted to astrocytes. This could explain why astrocytes, but not neurones, exhibit increased GCL activity and GSH levels upon exposure to NO.

The differences in the in vitro age of astrocytes and neurones, or the differing ages at which these cells were isolated from the brain, cannot be discounted as a reason for the differential susceptibility of these two cell types. However, astrocytes but not neurones have been shown to induce transcription of GCL upon exposure to hydrogen peroxide, when isolated from rat brains at the same embryonic stage (Iwata-Ichikawa et al. 1999). Furthermore, the induction of EpRE-mediated transcription in astrocytes discussed above was compared to neurones also isolated from the same embryonic stage (Murphy et al. 2001). Differences in the culture conditions between astrocytes and neurones can probably be discounted, since GSH levels were still elevated in astrocytes cultured in neuronal media.

The observed increased rate of GSH efflux and activity of γ-GT by astrocytes exposed to NO for a short period could have important implications for neuroprotection in vivo. The maintenance of intracellular GSH levels in astrocytes exposed to NO during the first 2 h of the GSH release study, coupled with no increase in LDH release throughout the course of the experiment, suggests that the increased rate of GSH efflux observed following exposure to NO is not due to increased permeability of the plasma membrane, but a regulated response as previously reported (Sagara et al. 1996; Stewart et al. 2002). Since the minimal medium used in the GSH release experiments contains no cysteine and cystine, the only source of cysteine/cystine will be the recycling of released GSH by γ-GT. Therefore the reason why intracellular GSH levels begin to fall in NO-treated astrocytes after 4 h of GSH release maybe because the amount of GSH recycling is not sufficient to maintain the greater rate of GSH synthesis in these cells. Sagara et al. (1996) have previously shown that the release of GSH by astrocytes is dependent on cellular GSH concentration. Therefore the increased GSH levels observed in astrocytes exposed to NO may be responsible for stimulating the increased rate of GSH release in these cells.

As stated above, cysteine is the rate-limiting substrate for GSH synthesis, with both astrocyte and neuronal GSH concentration determined by the availability of cysteine or cystine in the culture medium (Kranich et al. 1996; Dringen et al. 1999). Astrocytes are thought to prefer cystine as the precursor for GSH synthesis (Kranich et al. 1996), whereas neurones rely on cysteine for de novo GSH synthesis (Sagara et al. 1993; Dringen et al. 1999). The preferred cysteine precursor for neuronal GSH synthesis appears to be cysteinylglycine (Dringen et al. 1999) and this is generated by the metabolism of extracellular GSH by γ-GT (Meister and Anderson 1983). Neurones cocultured with astrocytes approximately double their GSH concentration (Bolanos et al. 1996; Dringen et al. 1999). This increase in neuronal GSH concentration is abolished if astrocytes are incubated in the presence of acivicin, an inhibitor of γ-GT (Dringen et al. 1999). Therefore, the increased release of GSH from astrocytes, coupled with the increased rate of GSH metabolism to cysteinylglycine by γ-GT in astrocytes exposed to NO, could result in increased trafficking of GSH precursors to neurones (Fig. 6). This in turn may elevate neuronal GSH levels, thus giving greater protection against acute NO exposure. Indeed, neurones cocultured with NO-generating astrocytes have an increased GSH concentration, and appear more resistant to oxidative stress compared to neurones cultured alone (Bolanos et al. 1996; Stewart et al. 1998).

The reason for increased astrocytic γ-GT activity (when compared to appropriate time-matched controls) following 18 and 24 h of exposure to NO is unclear. As in the case of GCL, exposure to NO may increase the expression of γ-GT in astrocytes, either by stimulating increased transcription, or increasing the half-life of γ-GT mRNA. Alternatively, prolonged DETA-NO exposure may alter the rate of trafficking of γ-GT to/from the membrane or increase the protein half-life, thereby increasing the amount of enzyme capable of metabolizing extracellular GSH. Since increased γ-GT activity was not observed at 1, 3 or 6 h, this suggests that a post-translational modification (e.g. nitrosylation, phosphorylation) that directly affects enzyme activity was probably not responsible for the observed results.

It should be noted that increased γ-GT activity and evidence of increased NO production has been found in the substantia nigra of PD brains (Sian et al. 1994b; Hunot et al. 1996). The increased activity of γ-GT in PD could be a mechanism to increase the availability of GSH precursors in an attempt to protect neuronal cells from NO-mediated damage.

The observation that exposure of astrocytes to DETA-NO resulted in an increased extracellular GSH concentration despite γ-GT activity also being increased, appears initially paradoxical. However, it should be noted that the activities of γ-GT quoted here are maximal rates determined spectrophotometrically and under optimized conditions. Consequently, our data suggest that for astrocytes in culture the rate of GSH release, in the presence of DETA-NO, is greater than the rate of degradation by γ-GT, despite an increase in specific activity. Such a scenario would still theoretically increase the availability of CysGly for utilization by neurones. Furthermore, GSH that is not cleaved by γ-GT would be used to enhance the extracellular antioxidant environment.

The increased astrocytic GSH levels observed in this study are at odds with the GSH depletion observed in PD. The observed elevation of GSH may be a transient protective mechanism. A prolonged time-course (i.e. > 24 h) is needed to determine for how long astrocytes can increase their GSH levels upon exposure to NO. Indeed, GCL_h mRNA levels were no longer elevated following 24 h of exposure, whereas GCL_l mRNA levels were still greater compared to controls, but lower than at 9 h. Prolonged exposure to NO, which may occur during the progression of PD, may result in the protective mechanisms in astrocytes failing, and therefore leading to loss of GSH.

In conclusion, astrocytes increase cellular GCL activity upon acute exposure to NO, possibly due to increased expression of both GCL genes, resulting in an increased cellular GSH concentration. This may help to protect astrocytes from NO-mediated damage (e.g. damage to the mitochondrial respiratory chain). Neurones in culture on their own are unable to increase GCL activity upon exposure to stress, and are therefore more susceptible to the effects of exposure to NO. The increased intracellular GSH concentration, rate of GSH release, and γ-GT activity in astrocytes exposed to NO may help protect neurones in vivo by supplying more GSH precursors, and thus elevating neuronal GSH levels.

Acknowledgements

We are grateful to The Brain Research Trust (MG) and The Medical Research Council (SJRH, SM and JBC) for funding this work. In part, this work is also the result of a collaboration between SJRH and JPB funded by the Ministerio De Ciencia Y Tecnologia (Spain: SAF2001-1961). The authors would like to thank Annie Higgs for her help during the writing of this paper.