SEARCH

SEARCH BY CITATION

Keywords:

  • buthionine sulfoximine;
  • excitotoxicity;
  • glia;
  • glutamate;
  • Transwell

Abstract

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Nitric oxide (NO) contributes to neuronal death in cerebral ischemia and other conditions. Astrocytes are anatomically well positioned to shield neurons from NO because astrocyte processes surround most neurons. In this study, the capacity of astrocytes to limit NO neurotoxicity was examined using a cortical co-culture system. Astrocyte-coated dialysis membranes were placed directly on top of neuronal cultures to provide a removable astrocyte layer between the neurons and the culture medium. The utility of this system was tested by comparing neuronal death produced by glutamate, which is rapidly cleared by astrocytes, and N-methyl-d-aspartate (NMDA), which is not. The presence of an astrocyte layer increased the LD50 for glutamate by approximately four-fold, but had no effect on NMDA toxicity. Astrocyte effects on neuronal death produced by the NO donors S-nitroso-N-acetyl penicillamine and spermine NONOate were examined by placing these compounds into the medium of co-cultures containing either a control astrocyte layer or an astrocyte layer depleted of glutathione by prior exposure to buthionine sulfoximine. Neurons in culture with the glutathione-depleted astrocytes exhibited a two-fold increase in cell death over a range of NO donor concentrations. These findings suggest that astrocytes protect neurons from NO toxicity by a glutathione-dependent mechanism.

Abbreviations used
BSO

buthionine sulfoximine

BSS

physiological balanced salt solution

CFDA

5(6)-carboxyfluorescein diacetate

GCM

glial conditioned medium

DTNB

5,5′-dithio-bis(2-nitrobenzoic) acid

FBS

fetal bovine serum

GFAP

glial fibrillary acidic protein

GSH

glutathione

GSNO

nitrosoglutathione

LDH

lactate dehydrogenase

MAP2

microtubule-associated protein 2

MEM

minimal essential medium

NMDA

N-methyl-d-aspartate

NO

nitric oxide

SNAP

S-nitroso-N-acetyl-dl-penicillamine

Spermine NONOate

N-[4-(-1-[3-aminopropyl]-2-hydroxy-2-nitrosohydrazino)butyl]-1,3-propanediamine.

Nitric oxide (NO) is a signaling molecule in the CNS, acting as a neurotransmitter and a vasodilator (Garthwaite 1991; Moncada et al. 1991). However, NO can also contribute to neuronal death in cerebral ischemia (Huang et al. 1994; Samdani et al. 1997) and probably other disorders (Dawson and Dawson 1998). NO reacts with O2· to form peroxynitrite and its reactive intermediates, which can react indiscriminately with proteins, DNA, and other cell constituents (Beckman and Koppenol 1996; Squadrito and Pryor 1998; Szabo and Dawson 1998). NO can also damage cells by reacting with protein thiol groups and protein–iron complexes (Stamler et al. 1992; Kim et al. 1995).

Glutathione (GSH) is important in limiting and repairing the deleterious actions of NO. GSH reacts directly with NO to produce nitrosoglutathione (GSNO), which acts as a sink for NO in cells (Clancy et al. 1994; Hogg et al. 1996; Singh et al. 1996). In addition, protein nitrosothiol groups formed by NO and sulfhydryl moieties undergo transnitrosation reactions with GSH to form nitrosoglutathione (GSNO) and regenerate the native protein thiols (Singh et al. 1996; Padgett and Whorton 1998). The formation of GSNO over other nitrosothiols is favored by the relative stability of GSNO and the normally high abundance of intracellular GSH (Singh et al. 1996). However, during exposure to high concentrations of NO, GSH is quickly consumed and cells lose this defense against NO toxicity (Clancy et al. 1994; Padgett and Whorton 1998). In addition, GSNO itself can contribute to nitrosation of protein sulfhydryl groups when the ratio of GSNO : GSH becomes elevated (Singh et al. 1996).

Interaction between astrocytes and neurons has been proposed as an important factor limiting neuronal death from excitotoxins (Rosenberg and Aizenman 1989), oxidants (Wilson 1997; Tanaka et al. 1999; Dringen et al. 2000), and other stressors, but unambiguous support for specific mechanisms for these astrocyte–neuron interactions has been difficult to provide. Here we introduce a new method for studying the effects of different cell populations in a co-culture system. Astrocytes are cultured on dialysis membranes for placement directly on top of neurons cultured in tissue culture wells, such that the astrocytes are interposed between the neurons and the culture medium. The method differs from standard co-culture systems in that compounds added to the culture medium can gain access to neurons only by passing through the astrocyte layer. The astrocyte-coated dialysis membrane can be removed or replaced with pharmacologically treated astrocytes in order to investigate mechanisms of astrocyte–neuronal interactions. Using this approach, we confirm a role for astrocyte glutamate uptake in limiting neuronal sensitivity to glutamate excitotoxicity and show that astrocytes protect neurons from NO toxicity by a glutathione-dependent mechanism.

Materials and methods

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Reagents were obtained from Sigma Chemicals (St Louis, MO, USA) except where noted.

Astrocyte-neuron co-cultures

Primary cortical astrocyte cultures were prepared from 1-day-old mice. Brains were removed after decapitation under deep isoflurane anesthesia, and the meninges were stripped away. The forebrain cortices were collected and dissociated by incubation in papain/DNAase followed by trituration. The dissociated cells were washed, suspended in a culture medium consisting of Eagle's minimal essential medium (MEM) with 10% fetal bovine serum (FBS) (Hyclone, Ogden, UT, USA) and glutamine (2 mm), and plated in 75 cm2 cell culture flasks. The medium was exchanged with fresh culture medium at day 7. At confluence (day 12–15), the astrocytes were trypsinized and re-plated onto 1.3-cm diameter discs of Spectra/Por 14-kDa cut-off regenerated cellulose dialysis membrane (Spectrum Laboratories, Rancho Dominguez, CA, USA) that had been sterilized by soaking in 70% ethanol and precoated with human placenta collagen (Sigma), 0.1 mg/mL in sterile distilled water. The culture medium was exchanged with fresh medium after two days. The astrocytes were used for co-cultures 7 days after replating, at which time the astrocytes formed a confluent layer across the surface of the dialysis membranes.

Cortical neuron suspensions were prepared using the same procedure as described for the astrocyte cultures except that the cortices were harvested from embryonic day 15 mice, as described previously (Ying et al. 1999). The neurons were suspended in glial conditioned medium (GCM) freshly collected from confluent astrocyte flasks that had been fed at least 3 days previously with MEM containing 10% FBS. The neurons were plated into 24-well culture plates coated with poly d-lysine. On the following day, astrocytes on dialysis membranes were placed on top of the neuronal layer such that the astrocyte-coated surface was resting on and facing the neurons (Fig. 1). Ceramic ring washers (1.25 cm in diameter; Small Parts Inc., Miami Lakes, FL, USA) were placed on top of the dialysis membranes to hold them down against the neuronal layers. Cytosine arabinoside (20 µm) was added to the cultures to stop cell proliferation on day 2, and on day 4 this medium was exchanged with fresh GCM. The co-cultures were used for experiments and immunocytochemical characterization between days 7 and 9 in vitro.

image

Figure 1. Schematic presentation of the co-culture model. (a) Cortical astrocytes cultures are plated onto discs cut from dialysis membranes. (b) Cortical neurons are cultured in 24-well plates. (c) The astrocyte-coated membrane discs are inverted and placed into the wells containing neurons, with the astrocyte layer facing the neuronal layer. (d) A ceramic ring is placed over the astrocyte-coated membrane. Compounds added to the culture medium can gain access to neurons only by passing through the astrocyte layer.

Download figure to PowerPoint

For some studies, neurons and astrocytes were cultured in Transwell (Costar) co-culture wells, as described by Dringen et al. (1999) with minor modifications. For these preparations, the cultures were prepared identically as described for the dialysis membrane co-cultures except that the astrocytes were grown to confluency on collagen-coated Transwell inserts, which are suspended approximately 2 mm above the culture well floor, instead of on dialysis membranes. The inserts with astrocytes were placed into the wells with neurons one day after the neurons were plated.

Immunocytochemistry

The astrocytes were removed from the co-cultures and placed in empty culture wells for fixation and staining. Astrocyte and neuronal preparations were fixed in 4% paraformaldehyde for 20 min. The fixed cells were incubated with mouse antibodies to microtubule-associated protein 2 (MAP2) (Boeringer-Mannheim, Indianapolis, IN, USA) diluted 1 : 800 to identify neurons, or with glial fibrillary acidic protein (GFAP; ICN, Costa Mesa, CA, USA) diluted 1 : 5000, or vimentin (DAKO, Carpenteria, CA, USA) diluted 1 : 2000 to identify astrocytes. After 18 h at 4°C, excess antibody was removed and the cultures were incubated with biotinylated horse anti-mouse IgG (Vector Laboratories, Burlingame, CA, USA) diluted 1 : 200 for 1 h at room temperature. After removal of excess antibody, cultures were treated with the ABC streptavidin detection system (Vector Laboratories, Burlingame, CA, USA) for 1 h, and the resulting horseradish peroxidase signal was detected using 3,3′-diaminobenzidene.

Exposure to NO donors and excitotoxins

The test reagents were prepared as concentrated stocks in physiological balanced salt solution (BSS) buffered at pH 7.2 with CO2/HCO3 during equilibrium with a 5% CO2 atmosphere. The BSS consisted of 3.1 mm KCl, 134 mm NaCl, 1.2 mm CaCl2, 1.2 mm MgSO4, 0.25 mm KH2PO4, 15.7 mm NaHCO3, and 2 mm glucose. Osmolarity was adjusted when necessary with small amounts of H2O or NaCl to attain a final value of 290–310 mOsm. Drugs were added from 20 × stocks, prepared and pH adjusted to pH 7.2 in BSS. Stocks of the NO donors S-nitroso-N-acetyl-dl-penicillamine (SNAP) and N-[4-(-1-[3-aminopropyl]-2-hydroxy-2-nitrosohydrazino)butyl]-1,3-propanediamine (spermine NONOate) (both from Alexis, San Diego, CA, USA) were prepared immediately before use and kept on ice until added to the cultures.

Experiments using the dialysis membrane co-cultures were preceded by removing the astrocyte-coated dialysis membranes from the co-culture wells and replacing them with either (i) astrocyte-coated dialysis membranes that had been treated for 24 h with 0.25 mm buthionine sulfoximine (BSO) in culture medium (to deplete astrocyte glutathione); (ii) control astrocyte-coated dialysis membranes (not treated with BSO); or (iii) with collagen-coated dialysis membranes that had not been plated with astrocytes. (All cultures underwent the replacement of the dialysis membranes to control for potential neuronal death caused by trauma from this exchange). A 1.25-cm outer diameter ceramic ring (Small Parts Inc., Miami Lakes, FL, USA) was placed over each membrane disc to ensure that substances placed in the culture media could gain access to the neurons only by first passing through the astrocyte layer. Experiments were begun by replacing the culture medium in each well with 0.5 mL CO2-equilibrated BSS. NMDA, glutamate, or NO donors were added from stock solutions and incubations were performed at 37°C in a 5% CO2 atmosphere. For studies of NO toxicity, the NO generating compounds remained in the BSS throughout the 24 h interval until assessment of neuron survival. In some studies the NO donors were hydrolyzed prior to use by incubating the stocks at 37°C for 5 h (spermine NONOate) or 24 h (SNAP). These hydrolyzed stocks were used to control for potential toxicity of products other than NO, namely NO2, NO3, and the parent donor adducts.

Experiments using Transwell co-cultures were begun by replacing the culture medium with 700 µL BSS. This total volume included 100 µL above the astrocytes on the Transwell insert. Glutamate, NMDA, or NO donors were added to the volume above the inserts at concentrations calculated to achieve the designated final concentrations in the wells assuming complete diffusion and equilibration through the Transwell inserts (i.e. at seven times the designated final concentrations).

Assessment of neuronal survival

Neuronal viability after excitotoxic and NO insults were assessed 24 h after addition of the test compounds by counting 5(6)-carboxyfluorescein diacetate (CFDA)-labeled cells. CFDA, like fluorescein diacetate, is a non-fluorescent hydrophobic dye that penetrates cell membranes (Bruning et al. 1980). Once inside the cell, the dye is cleaved by cytosolic esterases into the fluorescent product, 5(6)-carboxyfluorescein. The carboxyl group improves retention of the dye in viable neurons (Bruning et al. 1980; Petroski and Geller 1994). CDFA was incubated with the neurons for 30 min at a final concentration of 50 µm. Overlying dialysis membranes were not moved in order to prevent damage or displacement of the neurons. After the dye loading, cells were washed twice with BSS and fluorescing cells were counted using Nikon inverted fluorescence microscope equipped with appropriate filter set. Counts of 3–5 random 400 µm diameter fields in each culture well were averaged and normalized to counts obtained from four control wells.

Cell survival was assessed using CFDA staining of live cells rather than the more standard propidium iodide staining of dead cell nuclei (Ying et al. 1999) because typically 20–30% of the neurons initially plated subsequently died and fragmented during the 7–9 day culture period preceding the experiments. Since phagocytic cells were not present in the neuronal layers, fragmented nuclei from these cells were present at the onset of the experiments and these were often difficult to distinguish from nuclei of neurons killed by the experimental conditions. In some experiments, cell death determinations by the CFDA method were corroborated by measuring lactate dehydrogenase (LDH) activity in the cell culture medium (Swanson and Choi 1993). LDH activity was determined by the method of Koh and Choi (1987) using a kinetic plate reader (Molecular Devices, Menlo Park, CA, USA). In brief, 200 µL medium was mixed with 100 µL 500 mm potassium phosphate buffer (pH 7.5) containing 1.5 mm NADH and 7.5 mm sodium pyruvate. The A340 nm change was linear over 90 s. The CDFA and LDH methods were compared on the same set of culture wells after exposure to NMDA, using sham wash and 10 mm NMDA for the control and 100% neuronal death conditions, respectively. As shown in Fig. 2, the two methods provide similar dose–response curves for NMDA neurotoxicity. The LDH method was not used throughout the studies because of the possibility that the higher concentrations of NO donors could produce LDH release from astrocytes as well as from neurons.

image

Figure 2. The CDFA and LDH methods of determining cell death yield comparable results. Neuronal cultures were incubated 20 min with NMDA at the designated concentrations. Percent cell death was determined both by the LDH method and the CDFA method in the same culture wells. CDFA, 5(6)-carboxyfluorescein diacetate; LDH, lactate dehydrogenase; NMDA, N-methyl-d-aspartate.

Download figure to PowerPoint

Measurements of astrocyte glutathione content

Astrocytes cultured on dialysis membranes were placed in culture wells without neurons, then exposed to buthionine sulfoximine (BSO) or NO donors as described for the co-cultures. Total glutathione content (reduced glutathione content plus twice glutathione disulfide content) was measured by the cycling method of Baker et al. (1990). Astrocytes were lysed on ice in 1% sulfosalicylic acid, centrifuged to remove denatured protein, and brought to pH 7.0 with NaOH. A reaction mixture was added containing NADPH, glutathione reductase, and 5,5′-dithio-bis(2-nitrobenzoic) acid (DTNB). The rate of DTNB reduction was determined by following the rate of absorbance increase at 405 nm over 90 s on a kinetic plate reader. Results were calibrated against known glutathione standards that were treated in parallel with the cell extracts and normalized to protein content measured in sister culture wells by the bicinchonic acid (Pierce, Rockford, IL, USA) method (Smith et al. 1985), using bovine serum albumin standards. To confirm that low levels of glutathione were not simply due to cell death and lysis, medium from the culture wells was sampled and measured for LDH activity at the time the cells were harvested for glutathione measurements.

Statistics

Comparisons between two treatment groups were performed with Student's t-test. Comparisons between multiple groups were performed with anova followed by Tukey's test for multiple comparisons between groups or Dunnett's test for multiple comparisons against the control group.

Results

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

GFAP immunostaining of the astrocyte-coated dialysis membranes showed a confluent layer of astrocytes (Fig. 3a). Immunostaining of the neuronal cultures with antibody to MAP2 at 9 days in vitro showed dispersed neurons with rare clumping (Fig. 3b). The neuronal cultures had rare (< 1%) contaminating astrocytes as evidenced by staining for GFAP and vimentin: while most microscopic fields showed no astrocytes, some fields showed isolated astrocytes (Fig. 3c). The neurons could be maintained with at least 50% survival for up to 18 days as long as they remained under the astrocyte layer. After culture day 5, however, removal of the astrocyte layer for more than 3–4 h led to death of most of the neurons in the wells. The mechanism of this neuronal death was not further evaluated.

image

Figure 3. Immunocytochemical characterization of the cell cultures. (a) Astrocytes on dialysis membranes express glial fibrillary acidic protein (GFAP) and form a confluent monolayer. (b) Neurons on the culture plate surface beneath the astrocytes are identified by expression of microtubule-associated protein 2. (c) GFAP immunostaining of the neuronal culture layer shows rare astrocytes (most fields contained no astrocytes). Photograph (c) is at one-half magnification of (a) and (b).

Download figure to PowerPoint

Glutamate, but not NMDA, is rapidly taken up by astrocytes (Skerritt and Johnston 1981; Rosenberg et al. 1992; Tanaka 1993; Robinson and Dowd 1997). As shown in Fig. 4(a), neurons overlaid with astrocyte-coated dialysis membranes were more resistant to glutamate toxicity than were neurons overlaid with the dialysis membranes alone (without astrocytes) during the 20 minute glutamate exposures. As expected, the presence or absence of astrocytes in the co-cultures had no discernible effect on NMDA toxicity (Fig. 4b), consistent with rapid uptake as the mechanism by which astrocytes provide neuroprotection from glutamate. These results confirm that the astrocyte layer can have a significant effect on the composition of the medium reaching the neurons.

image

Figure 4. Effects of astrocytes on excitotoxic neuronal death. (a) Neuronal survival of glutamate exposure is increased in the presence of astrocytes. (b) Neuronal survival after incubation with NMDA, which is not taken up by astrocytes, is unaffected by the presence of astrocytes. n = 8, means ± SEM. **p < 0.01. vs. the ‘with astrocyte’ condition; NMDA, N-methyl-d-aspartate; ▪, with astrocytes; ○, without astrocytes.

Download figure to PowerPoint

Glutamate and NMDA neurotoxicity were also compared using the commercially available Transwell system. Transwell co-cultures, like the dialysis membrane co-cultures, allow cells to be plated on separate surfaces and share a common medium. The Transwell cultures differ from the dialysis membrane co-cultures in that the upper layer of cells (astrocytes in the present study) is plated on a rigid support that is suspended above the neuronal layer rather than directly on top of it. Diffusion through this support is slower than through a dialysis membrane, and unlike the dialysis membrane co-cultures, the cells on the upper layer are not interposed between the lower layer and the bulk of the culture medium. Glutamate and NMDA placed in the small fluid compartment above the Transwell insert at concentrations to achieve 100 µm with uniform diffusion through the culture medium produced no detectable neuronal death with incubations of up to 30 min (data not shown). The lack of neurotoxicity was, presumably, due to slow diffusion through the inserts, as 20 µm of either NMDA or glutamate added to the medium below the inserts caused death of more than 50% of the neurons. Since the NO donors employed have half-lives in the range of minutes to hours, the subsequent studies used only the dialysis membrane co-cultures.

Astrocyte glutathione levels were manipulated without affecting neuronal glutathione by placing astrocyte-coated dialysis membranes in fresh MEM containing 5% FBS, with or without 0.25 mm buthionine sulfoximine (BSO) to inhibit glutathione synthesis (Griffith and Meister 1979). As shown in Fig. 5, the BSO-treated astrocytes had glutathione levels of less than 5% of the non-treated, control astrocytes. The BSO-treated astrocytes were then compared to control astrocytes with respect to their effects on NO-mediated neurotoxicity. Astrocyte glutathione content was found to have a large effect on neuronal survival when the NO generating compounds were added to the co-cultures. Figure 6 shows representative photomicrographs of CDFA stained (live) neurons under various treatment conditions. In the absence of NO donors, neuronal survival was not significantly different in cultures with BSO-treated or control astrocytes. Treatment with 200 µm SNAP, however, produced much greater neuronal death in neurons overlaid with the BSO-treated (glutathione-depleted) astrocytes.

image

Figure 5. Effects of BSO and NO donors on astrocyte glutathione content and astrocyte viability. (a) Incubation with the NO donors produced dose- and time-dependent reductions in glutathione content. Near-total depletion was produced by extended incubations at the higher NO donor concentrations. BSO pre-incubation also produced near total glutathione depletion. (b) Measurements of extracellular lactate dehydrogenase (LDH) activity performed at the same time as the glutathione determinations showed that neither the BSO pretreatment nor the NO donors produced significant astrocyte death at the concentrations employed. BSO, buthionine sulfoximine; SNAP, S-nitroso-N-acetyl-dl-penicillamine; spermine NO, N-[4-(-1-[3-aminopropyl]-2-hydroxy-2-nitrosohydrazino)butyl]-1,3-propanediamine. n = 6, means ± SEM. **p < 0.01 vs. the control group at the same incubation interval.

Download figure to PowerPoint

image

Figure 6. Photomicrographs showing the effect of astrocyte BSO treatment on neuronal survival of NO exposure. Neuronal viability is established by 5(6)-carboxyfluorescein diacetate (CDFA) fluorescence, assessed 24 h after treatment with SNAP. Neurons were placed in co-culture with astrocytes that were sham treated (a and c) or pre-incubated with BSO to deplete glutathione (b and d). The co-cultures were then treated with 0 µm SNAP (a and b) or 200 µm SNAP (c and d), and photographed 24 h later. BSO, buthionine sulfoximine; SNAP, S-nitroso-N-acetyl-dl-penicillamine.

Download figure to PowerPoint

Figure 7 shows dose-response curves for the effects of SNAP and spermine NONOate on neuronal survival. Spermine NONOate and SNAP are both hydrolyzed in solution to form NO, but the rates of these hydrolyses are different: the half-life at pH 7.4, 37°C is approximately 0.6 h for spermine NONOate and greater than 4 h for SNAP (Bauer and Fung 1991; Maragos et al. 1991; Kroncke et al. 1993; Morley et al. 1993). Both compounds produced a dose-dependent decrease in neuronal viability, with spermine NONOate showing more potent effects than SNAP. Cultures incubated with hydrolyzed SNAP (1 mm) or hydrolyzed spermine NONOate (250 µm) did not increase neuronal death over control values (not shown). The primary finding shown here is that the neuronal death produced by the NO donors was markedly influenced by the astrocytes present in co-culture. Neurons in co-culture with astrocytes depleted of glutathione by BSO pretreatment exhibited significantly greater vulnerability to the NO donors than did neurons in co-culture with control astrocytes not pretreated with BSO.

image

Figure 7. Effects of astrocyte BSO treatment on neuronal survival after 24 hour exposures to NO generating compounds. (a) Co-cultures treated with SNAP exhibited decreased neuronal survival when the astrocytes were pretreated with BSO to deplete glutathione. (b) Similar results were observed in co-cultures treated with spermine-NONOate, which like SNAP produces NO in aqueous solution. n = 6–10, means ± SEM. *p < 0.05, **p < 0.01 vs. the control astrocyte condition. BSO, buthionine sulfoximine; SNAP, S-nitroso-N-acetyl-dl-penicillamine; spermine NONOate, N-[4-(-1-[3-aminopropyl]-2-hydroxy-2-nitrosohydrazino)butyl]-1,3-propanediamine; ▪, control astrocytes; ●, BSO-treated astrocytes.

Download figure to PowerPoint

If the astrocyte glutathione content is important in limiting NO-mediated neuronal death, astrocyte glutathione levels would be expected to decrease during NO exposure. Figure 5(a) shows that glutathione levels are depleted by the higher concentrations of NO donors employed, and also shows that BSO pretreatment causes nearly complete GSH depletion. Astrocyte survival was not significantly affected by the NO generators at these concentrations, as shown by the preserved intracellular LDH activity (Fig. 5b).

Discussion

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

These findings suggest that astrocytes can protect neurons from NO toxicity by a mechanism dependent on astrocyte glutathione content. In most regions of the CNS, astrocyte processes form a nearly continuous membrane around neuronal cell bodies and processes (Peters et al. 1976). Standard astrocyte–neuron co-cultures, in which the two cell types are intermingled on a common surface, do not fully replicate this anatomical relationship, since electron microscopy has shown that neuronal cell somas are exposed to the medium in these cultures (Harris and Rosenberg 1993). An important feature of the co-culture system presented here is that the astrocyte-coated dialysis membranes are interposed between the bulk of the culture medium and the neuronal layer. The aim of this arrangement was to allow the astrocytes to modify the composition of the medium reaching the neurons. This effect was confirmed by the studies comparing the effects of the layer on glutamate and NMDA toxicity. The astrocyte layer reduced the toxicity of glutamate, which is taken up by astrocytes, but not the toxicity of NMDA, which is cleared very slowly by astrocytes (Skerritt and Johnston 1981; Rosenberg et al. 1992; Tanaka 1993; Robinson and Dowd 1997). Of note, the astrocyte layer does not provide a impermeable barrier to NMDA. This resembles conditions in situ, where NMDA is a potent neurotoxin despite the presence of astrocyte processes around most neurons.

The glutamate sensitivity of cortical neurons during removal of the astrocyte layer in the present studies is comparable to the glutamate sensitivity previously observed in cortical neurons cultured under astrocyte-poor conditions, and much greater than the glutamate sensitivity of cortical neurons in astrocyte-rich cultures (Rosenberg and Aizenman 1989). Comparisons between astrocyte-poor and astrocyte-rich neuronal cultures have been used by several groups to investigate astrocyte–neuron interactions. These studies have provided support for the idea that astrocytes protect neurons from many excitotoxins and oxidants (Vibulsreth et al. 1987; Rosenberg and Aizenman 1989; Swanson and Choi 1993; Desagher et al. 1996; Lucius and Sievers 1996; Drukarch et al. 1997; Drukarch et al. 1998; Ye and Sontheimer 1998; Brown 1999; Tanaka et al. 1999; Xu et al. 1999). In studies comparing the two culture conditions, however, it is difficult to exclude the possibility that differences in neuronal survival might be due to intrinsic differences in the neurons cultured with and without abundant astrocytes, rather than to specific astrocyte functions during the toxin exposures. With studies of reactive oxygen species it is additionally difficult to exclude the possibility that the astrocytes in the astrocyte-rich cultures provide a non-specific protective effect by simply providing other targets for the oxidants. These problems were eliminated in the present study by (i) culturing all the neurons identically up to the start of experiments, and (ii) maintaining the presence of an astrocyte layer in all wells during the experiments.

NO diffuses freely across cell membranes (Garthwaite 1991; Moncada et al. 1991). Since astrocyte processes surround neurons in most brain regions (Peters et al. 1976), glutathione localized within astrocytes can trap NO and thereby reduce neuronal NO exposure. Glutathione has previously been shown to limit cell sensitivity to NO-mediated mitochondrial injury (Bolaños et al. 1996), and it is likely that glutathione can likewise limit other NO-mediated cytotoxic processes. The major finding presented here is that the ability to astrocytes to protect neurons from NO toxicity is dependent at least in part on astrocyte glutathione content. Astrocyte glutathione content was manipulated by incubation with BSO, an irreversible inhibitor of γ-glutamylcysteine synthetase (Griffith and Meister 1979). The plating of astrocytes on removable membranes allowed the astrocytes to be incubated with BSO without exposing the neurons to BSO, as performed previously by McNaught and Jenner (1999). Astrocytes depleted of GSH by prior incubation with BSO were less effective than control astrocytes in preventing NO-induced neuronal death. Moreover, death of neurons in co-culture with control astrocytes (not treated with BSO) occurred only at NO donor concentrations sufficient to significantly deplete astrocyte glutathione content. The most straightforward explanation for these observations is that astrocytes reduce the amount of NO reaching the neurons by trapping NO as GSNO within their cytoplasm. GSNO may also have intrinsic antioxidant properties (Konorev et al. 1995; Rauhala et al. 1998). Alternatively, it is possible that astrocyte glutathione fuels an exchange of reducing equivalents between astrocytes and neurons. For example, astrocytes can take up oxidized vitamin C (dehydroascorbic acid), reduce it to ascorbate at the expense of glutathione, and then release ascorbate to the extracellular fluid (Wilson 1997). Astrocytes can also release glutathione itself, but there is little evidence to suggest that neurons can directly import glutathione from the extracellular space (Yudkoff et al. 1990; Sagara et al. 1996; Dringen et al. 1999). Last, depletion of glutathione may facilitate production some reactive oxygen species by astrocytes (McNaught and Jenner 1999), and this could contribute to neuronal death during NO exposure.

The present findings do not exclude the presence of other, glutathione-independent mechanisms by which astrocytes limit NO neurotoxicity. In particular, Tanaka et al. (1999) have shown that neurons grown on an astrocyte-derived extracellular matrix have increased resistance to NO, and Dringen et al. (1999) have shown that astrocyte-derived precursors are taken up by neurons and used for neuronal glutathione synthesis. It is also possible that protein sulfhydryls and other reactive moieties within astrocytes can function, like glutathione, to bind or trap NO.

NMDA receptor activation stimulates NO production by neurons, and in many systems NMDA neurotoxicity is substantially reduced by blockade of NO production or scavenging of the NO produced (Dawson and Dawson 1998). However, astrocytes had no effect on neuronal vulnerability to NMDA in the present study despite a large effect of astrocyte glutathione content on neuronal vulnerability to exogenous NO donors. It is possible that NO is only minor aspect of NMDA toxicity in this culture system. As reviewed by Aizenman et al. (1998), the degree to which NMDA neurotoxicity is mediated by NO depends on several factors and varies considerably among different systems and conditions. It should also be noted that astrocytes in this system are well positioned to shield neurons from NO donors in the medium, but not to shield neurons from one another, and consequently the astrocyte layer may have little influence neuronal exposure to NO generated by other neurons. Astrocyte processes do separate neurons in situ, however, and this together with the present findings suggests that astrocyte glutathione content could influence the vulnerability of neurons to NMDA receptor-mediated death in situ.

Several reports suggest that concentrations of glutathione and glutathione synthesizing enzymes are generally higher in astrocytes than in neurons (Slivka et al. 1987); reviewed in (Makar et al. 1994). This arrangement is consistent with a significant role for astrocyte glutathione in brain oxidant scavenging, and reductions in astrocyte glutathione may therefore endanger neurons under oxidative stress. Glutathione depletion has previously been shown to exacerbate cerebral ischemic injury in rats (Mizui et al. 1992). In light of the current results, this exacerbation could be attributable in part to reduced capacity of astrocytes to limit NO neurotoxicity during ischemia. These results demonstrate a glutathione-dependent interaction between neurons and astrocytes and further support the idea that free radical scavenging is an important astrocyte function in the CNS.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

We thank Sophie Parmentier-Batteur for comments on the manuscript and Jill Guenza for expert technical support. This study was supported by grants from Saastamoinen Foundation, the Finnish Cultural Foundation of Northern Savo (NV), and the US National Institutes of Health (grant P50 NS 14543).

References

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
Footnotes
  1. 1Yongmei Chen and Nina E. Vartiainen contributed equally to this work.