Cytosine arabinofuranoside-induced activation of astrocytes increases the susceptibility of neurons to glutamate due to the release of soluble factors

Activation of astrocytes occurs during many forms of CNS injury, but its importance for neuronal survival is poorly understood. When hippocampal cultures of neurons and astrocytes were treated from day 2–4 in vitro (DIV 2–4) with 1 μM cytosine arabinofuranoside (AraC), we observed a stellation of astrocytes, an increase in glial fibrillary acidic protein (GFAP) level as well as a higher susceptibility of the neurons to glutamate compared with cultures treated from DIV 2–4 with vehicle. To find out whether factors released into the culture medium were responsible for the observed differences in glutamate neurotoxicity, conditioned medium of AraC-treated cultures (MCMAraC) was added to vehicle-treated cultures and conditioned medium of vehicle-treated cultures (MCMvh) was added to AraC-treated cultures 2 h before and up to 18 h after the exposure to 1 mM glutamate for 1 h. MCMAraC increased glutamate neurotoxicity in vehicle-treated cultures and MCMvh reduced glutamate neurotoxicity in AraC-treated cultures. Heat-inactivation of MCMvh increased, whereas heat-inactivation of MCMAraC did not affect glutamate toxicity suggesting that heat-inactivation changed the proportion of factors in MCMvh inhibiting and exacerbating the excitotoxic injury. Similar findings were obtained using conditioned medium of pure astrocyte cultures of DIV 12 treated from DIV 2–4 with vehicle or 1 μM AraC suggesting that heat-sensitive factors in MCMvh were mainly derived from astrocytes. Treatment of hippocampal cultures with 1 mM dibutyryl-cAMP for 3 days induced an activation of the astrocytes similar to AraC and increased neuronal susceptibility to glutamate. Our findings provide evidence that activation of astrocytes impairs their ability to protect neurons after excitotoxic injury due to changes in the release of soluble and heat-sensitive factors.     

Long-term culture of mouse cortical neurons as a model for neuronal development, aging, and death

A long-term cell culture system was used to study maturation, aging, and death of cortical neurons. Mouse cortical neurons were maintained in culture in serum-free medium (Neurobasal supplemented with B27) for 60 days in vitro (DIV). The levels of several proteins were evaluated by immunoblotting to demonstrate that these neurons matured by developing dendrites and synapses and remained continuously healthy for 60 DIV. During their maturation, cortical neurons showed increased or stable protein expression of glycolytic enzyme, synaptophysin, synapsin IIa, alpha and beta synucleins, and glutamate receptors. Synaptogenesis was prominent during the first 15 days and then synaptic markers remained stable through DIV60. Very early during dendritic development at DIV3, beta-synuclein (but not alpha-synuclein) was localized at the base of dendritic growth cones identified by MAP2 and alpha-amino-3-hydroxy-5-methyl-4-isoxazole (AMPA) receptor GluR1. In mature neurons, alpha and beta synucleins colocalized in presynaptic axon terminals. Expression of N-methyl-D-aspartate (NMDA) and AMPA receptors preceded the formation of synapses. Glutamate receptors continued to be expressed strongly through DIV60. Cortical neurons aging in vitro displayed a complex profile of protein damage as identified by protein nitration. During cortical neuron aging, some proteins showed increased nitration, while other proteins showed decreased nitration. After exposure to DNA damaging agent, young (DIV5) and old (DIV60) cortical neurons activated apoptosis mechanisms, including caspase-3 cleavage and poly(ADP)-ribose polymerase inactivation. We show that cultured mouse cortical neurons can be maintained for long term. Cortical neurons display compartmental changes in the localization of synucleins during maturation in vitro. These neurons sustain protein nitration during aging and exhibit age-related variations in the biochemistry of neuronal apoptosis.     

Dysfunction of astrocytes in senescence-accelerated mice SAMP8 reduces their neuroprotective capacity

Early onset increases in oxidative stress and tau pathology are present in the brain of senescence-accelerated mice prone (SAMP8). Astrocytes play an essential role, both in determining the brain's susceptibility to oxidative damage and in protecting neurons. In this study, we examine changes in tau phosphorylation, oxidative stress and glutamate uptake in primary cultures of cortical astrocytes from neonatal SAMP8 mice and senescence-accelerated-resistant mice (SAMR1). We demonstrated an enhancement of abnormally phosphorylated tau in Ser(199) and Ser(396) in SAMP8 astrocytes compared with that of SAMR1 control mice. Gsk3beta and Cdk5 kinase activity, which regulate tau phosphorylation, was also increased in SAMP8 astrocytes. Inhibition of Gsk3beta by lithium or Cdk5 by roscovitine reduced tau phosphorylation at Ser(396). Moreover, we detected an increase in radical superoxide generation, which may be responsible for the corresponding increase in lipoperoxidation and protein oxidation. We also observed a reduced mitochondrial membrane potential in SAMP8 mouse astrocytes. Glutamate uptake in astrocytes is a critical neuroprotective mechanism. SAMP8 astrocytes showed a decreased glutamate uptake compared with those of SAMR1 controls. Interestingly, survival of SAMP8 or SAMR1 neurons cocultured with SAMP8 astrocytes was significantly reduced. Our results indicate that alterations in astrocyte cultures from SAMP8 mice are similar to those detected in whole brains of SAMP8 mice at 1-5 months. Moreover, our findings suggest that this in vitro preparation is suitable for studying the molecular and cellular processes underlying early aging in this murine model. In addition, our study supports the contention that astrocytes play a key role in neurodegeneration during the aging process.     

Increase in glutamate-induced neurotoxicity by activated astrocytes involves stimulation of protein kinase C

Activation of astrocytes is a common feature of neurological disorders, but the importance of this phenomenon for neuronal outcome is not fully understood. Treatment of mixed hippocampal cultures of neurones and astrocytes from day 2-4 in vitro (DIV 2-4) with 1 micro m cytosine arabinofuranoside (AraC) caused an activation of astrocytes as detected by a stellate morphology and a 10-fold increase in glial fibrillary acidic protein (GFAP) level compared with vehicle-treated cultures. After DIV 12, we determined 43% and 97% damaged neurones 18 h after the exposure to glutamate (1 mm, 1 h) in cultures treated with vehicle and AraC, respectively. Dose-response curves were different with a higher sensitivity to glutamate in cultures treated with AraC (EC50 = 0.01 mm) than with vehicle (EC50 = 0.12 mm). The susceptibility of neurones to 1 mm glutamate did not correlate with the percentage of astrocytes and was insensitive to an inhibition of glutamate uptake. In cultures treated with vehicle and AraC, glutamate-induced neurotoxicity was mediated through stimulation of the NR1-NR2B subtype of NMDA receptors, because it was blocked by the NMDA receptor antagonist MK-801 and the NR1-NR2B selective receptor antagonist ifenprodil. Protein levels of the NR2A and NR2B subunits of NMDA receptor were similar in cultures treated with vehicle or AraC. AraC-induced changes in glutamate-induced neurotoxicity were mimicked by activation of protein kinase C (PKC), whereas neuronal susceptibility to glutamate was reduced in cultures depleted of PKC and treated with AraC suggesting that the increase in glutamate toxicity by activated astrocytes involves activation of PKC.     

Developmental expression and activity of high affinity glutamate transporters in rat cortical primary cultures

The expression and activity of glutamate transporters (EAAC1, GLAST and GLT1) were examined during the development of cortical neuron-enriched cultures. Protein content and mitochondrial respiration both increased during the first 7 days, later stabilized and decreased from DIV14. Glutamate transport and extracellular concentration were relatively constant from DIV3 to 18. The kinetic parameters of glutamate transport were at DIV7:K_m=19±3 μM and V_max=1068±83 pmol/mg protein/min and at DIV14: K_m=40.8±9.3 μM and V_max=1060±235 pmol/mg protein/min. The shift in K_m towards higher values suggest a more important participation of GLAST after DIV14. At DIV7 and 14, glutamate transport was poorly sensitive to dihydrokaïnate (DHK) suggesting a weak participation of GLT1 in glutamate transport. Western blot experiments and immunocytochemistry showed that EAAC1 was expressed by neurons whatever the stage of the culture. GLAST was found in astrocytes as soon as DIV3 and labeling increased during the development of the culture. There was little neuronal GLT1 immunoreactivity at DIV7, only detected by immunocytochemistry. From DIV10 to 18, an increasing astrocytic expression of GLT1 was observed, also detected by Western blotting. These results show that: (1) glutamate uptake remains stable all along the development of the cultures although the pattern of expression of the different transporters is changing, suggesting that glutamate transport is highly regulated; (2) neuronal EAAC1 may play a critical role during the early stages of the culture when it is expressed alone; and (3) the developmental expression pattern of glutamate transporters in cortical neuron-enriched cultures is quite similar to that observed in vivo during early postnatal development.     

Astrocyte Mitochondrial Mechanisms of Ischemic Brain Injury and Neuroprotection

Research on ischemic brain injury has established a central role of mitochondria in neuron death. Astrocytes are also damaged by ischemia, although the participation of mitochondria in their injury is ill defined. As astrocytes are responsible for neuronal metabolic and trophic support, astrocyte dysfunction will compromise postischemic neuronal survival. Ischemic alterations to astrocyte energy metabolism and the uptake and metabolism of the excitatory amino acid transmitter glutamate may be particularly important. Despite the significance of ischemic astrocyte injury, little is known of the mechanisms responsible for astrocyte death and dysfunction. This review focuses on differences between astrocyte and neuronal metabolism and mitochondrial function, and on neuronal-glial interactions. The potential for astrocyte mitochondria to serve as targets of neuroprotective interventions is also discussed.     

Astroglial-Conditioned Media and Growth Factors Modulate Proliferation and Differentiation of Astrocytes in Primary Culture

Astroglial conditioned media (ACM) influence the development and maturation of cultured nerve cells and modulate neuron-glia interaction. To clarify mechanisms of astroglial cell proliferation/differentiation in culture, incorporation of [methyl-³H]-thymidine or [5,6-³H]-uridine in cultured astrocytes was assessed. Cultures were pre-treated with epidermal growth factor (EGF), insulin (INS), insulin-like growth factor-I (IGF-I), and basic fibroblast growth factor (bFGF) and subsequently with ACM. DNA labeling revealed a marked stimulatory effect of ACM from 15 days in vitro (DIV) cultures in 30 DIV astrocytes after12 h pre-treatment with growth factors. The main effects were found after INS or EGF pre-treatment in 30 DIV cultures. ACM collected from 15 or 60 or 90 DIV increased RNA labeling of 15 and 30 DIV astrocyte cultures, being the highest value that of 30 DIV cultures added with ACM from 90 DIV. The findings of increased DNA labeling after EGF or INS pre-treatment in 30 DIV cultures, followed by addition of ACM from 15 DIV cultures, suggest that these phenomena may depend by extra cellular signal-regulated kinase 1 (ERK1) activation.     

Dependence of neurones on astrocytes in a coculture system renders neurones sensitive to transforming growth factor beta1-induced glutamate toxicity

Transforming growth factor beta1 (TGF-beta1) has been implicated in formation of astrocyte scars, which prevents axonal regeneration. A coculture system of astrocytes and cerebellar cells was used to investigate possible neurotoxic effects of TGF-beta1. Although not directly neurotoxic, TGF-beta1 was toxic to cerebellar cells in the presence of astrocytes. This toxicity is based on an effect of the cytokine on astrocytes, as conditioned medium from astrocyte cultures treated with TGF-beta1 was more toxic by a similar mechanism. This neurotoxicity was mediated by glutamate present in the culture medium as demonstrated by inhibition by MK-801. Astrocytic ability to metabolise glutamate was compromised by TGF-beta1, as this cytokine increased glutamate concentration. The astrocytes in the coculture system responded to the presence of neurones by secreting neuroprotective interleukin-6, which was partly protective against the TGF-beta1-induced toxicity. In the coculture system, neurones responded to the presence of astrocytes by a reduction in resistance to glutamate toxicity. On addition of TGF-beta1, which compromised astrocytic clearance of glutamate, this reduction in resistance to glutamate toxicity led to a reduction in neuronal survival. These results suggest that when neurones are cocultured with astrocytes they become dependent on astrocytes for survival. This dependence makes neurones susceptible to damage when astrocytes are activated by substances such as TGF-beta1.     

Amyloid β Peptide (25–35) Inhibits Na+-Dependent Glutamate Uptake in Rat Hippocampal Astrocyte Cultures

Abstract: Large numbers of neuritic plaques surrounded by reactive astrocytes are characteristic of Alzheimer's disease (AD). There is a large body of research supporting a causal role for the amyloid β peptide (Aβ), a main constituent of these plaques, in the neuropathology of AD. Several hypotheses have been proposed to explain the toxicity of Aβ including free radical injury and excitotoxicity. It has been reported that treatment of neuronal/astrocytic cultures with Aβ increases the vulnerability of neurons to glutamate-induced cell death. One mechanism that may explain this finding is inhibition of the astrocyte glutamate transporter by Aβ. The aim of the current study was to determine if Aβs inhibit astrocyte glutamate uptake and if this inhibition involves free radical damage to the transporter/astrocytes. We have previously reported that Aβ can generate free radicals, and this radical production was correlated with the oxidation of neurons in culture and inhibition of astrocyte glutamate uptake. In the present study, Aβ (25–35) significantly inhibited l -glutamate uptake in rat hippocampal astrocyte cultures and this inhibition was prevented by the antioxidant Trolox. Decreases in astrocyte function, in particular l -glutamate uptake, may contribute to neuronal degeneration such as that seen in AD. These results lead to a revised excitotoxicity/free radical hypothesis of Aβ toxicity involving astrocytes.     

Glutamate/glutamine metabolism coupling between astrocytes and glioma cells: Neuroprotection and inhibition of glioma growth

Glioma glutamate release has been shown to promote the growth of glioma cells and induce neuronal injuries from epilepsy to neuronal death. However, potential counteractions from normal astrocytes against glioma glutamate release have not been fully evaluated. In this study, we investigated the glutamate/glutamine cycling between glioma cells and astrocytes and their impact on neuronal function. Co-cultures of glioma cells with astrocytes (CGA) in direct contact were established under different mix ratio of astrocyte/glioma. Culture medium conditioned in these CGAs were sampled for HPLC measurement, for neuronal ratiometric calcium imaging, and for neuronal survival assay. We found: (1) High levels of glutaminase expression in glioma cells, but not in astrocytes, glutaminase enables glioma cells to release large amount of glutamate in the presence of glutamine. (2) Glutamate levels in CGAs were directly determined by the astrocyte/glioma ratios, indicating a balance between glioma glutamate release and astrocyte glutamate uptake. (3) Culture media from CGAs of higher glioma/astrocyte ratios induced stronger neuronal Ca²⁺ response and more severe neuronal death. (4) Co-culturing with astrocytes significantly reduced the growth rate of glioma cells. These results indicate that normal astrocytes in the brain play pivotal roles in glioma growth inhibition and in reducing neuronal injuries from glioma glutamate release. However, as tumor growth, the protective role of astrocytes gradually succumb to glioma cells.     

Neuroprotectant FK506 inhibits glutamate-induced apoptosis of astrocytes in vitro and in vivo

Neuron-astrocyte interactions are critical for signalling, energy metabolism, extracellular ion and glutamate homeostasis, volume regulation and neuroprotection in the CNS. Glutamate uptake by astrocytes may prevent excitotoxic glutamate elevation and determine neuronal survival. However, an excess of glutamate can cause the death of astrocytes. FK506, an inhibitor of calcineurin, and an immunosuppressive drug, is neuroprotective in animal models of neurologic diseases, including focal and global ischaemia. In the present work, we demonstrate that a single injection of FK506 60 min after a transient middle cerebral artery occlusion (MCAo) significantly decreases the number of terminal deoxynucleotidyl transferase nick-end labelling (TUNEL)-positive cells in the ischaemic cortex and striatum. Using 3-D confocal microscopy we found that, 24 h after MCAo, many TUNEL-positive cells in the ischaemic striatum and cortex are astrocytes. Furthermore, we demonstrate that exposure of cultured cortical astrocytes to 50-100 mM Glu for 24 h induces apoptotic alterations in nuclear morphology, DNA fragmentation, dissipation of mitochondrial transmembrane potential (DeltaPsi) and caspase activation. FK506 (1 muM) efficiently inhibits Glu-induced apoptosis of cultured astrocytes, DNA fragmentation and changes in mitochondrial DeltaPsi. Our findings suggest that modulation of glutamate-induced astrocyte death early after reperfusion may be a novel mechanism of FK506-mediated neuroprotection in ischaemia.     

Neuroprotective immunity: T cell-derived glutamate endows astrocytes with a neuroprotective phenotype

A well-controlled T cell response to CNS injury may result in increased neuronal survival. However, the precise mechanism of T cell-induced neuroprotection is unknown. In this study, we report the unexpected finding that during culture of T cells, high levels of glutamate accumulate, which are efficiently cleared if T cells are cocultured with astrocytes. The T cell-derived glutamate elicits in turn, the release of neuroprotective thiols (cysteine, glutathione, and cysteinyl-glycine) and lactate from astrocytes. Media obtained from astrocytes conditioned in the presence of T cells reduce neuronal apoptosis induced by oxidative stress in primary neuronal cultures from 48 +/- 14 to 9 +/- 4% (p < 0.001). Inhibition of glutamate-dependent signaling during astrocyte-T cell cocultivation by a glutamate uptake inhibitor, l-aspartic acid beta-hydroxamate, abolishes this neuroprotective effect. The ability of astrocytes to clear extracellular glutamate is impaired under conditions of oxidative stress. We demonstrate that T cells, via secreted cytokines, restore glutamate clearance capacity of astrocytes under oxidative conditions. Furthermore, under normoxic conditions, glutamate-buffering capacity of astrocytes is increased upon cocultivation with T cells. It is known that, following CNS injury, astrocytes can respond with beneficial or destructive effects on neurons. However, the context and signaling mechanisms for this dual astrocytic response are unknown. Our results implicate T cells as potential determinants of the context that elicits a protective role for astrocytes in the damaged CNS.     

Transient increase in the high affinity [3H]-L-glutamate uptake activity during in vitro development of hippocampal neurons in culture

The glial GLAST and GLT-1 glutamate transporters are transiently expressed in hippocampal neurons as shown by immunocytochemistry (Plachez et al., 2000. J. Neurosci. Res., 59, 587-593). In order to test if this transient expression is associated to a transient glutamate uptake activity, [3H]-glutamate uptake was studied during the in vitro development of embryonic hippocampal neurons cultured in a defined (serum free) medium. In these cultures, the ratio of the number of glial cells to the number of neurons increased from 1.7 to 11.3% during the first 10 days of culture, while 77% of the neurons died. The number of neurons then remains stable up to 23 days of culture. The initial glutamate uptake velocity at 20 and 200 microM [3H]-glutamate usually increased about five times between 1 and 10 days in vitro (DIV). Interestingly, at 2 microM [3H]-glutamate, the uptake initial velocity showed a biphasic pattern, with a transient peak between 1 and 6 DIV, the maximum being reached at 2 DIV and a delayed regular increase from 8 to 23 DIV. The concentration-dependent curves were best fitted with two saturable sites high and low affinities, at both 2 and 10 DIV. To pharmacologically characterize the transient increased glutamate uptake activity, four uptake inhibitors, L-threo-3-hydroxy-aspartic acid (THA), L-trans-pyrrolidine-2,4-dicarboxylic acid (L-trans-2,4-PDC), dihydrokainate (DHK), and DL-threo-beta-benzyloxyaspartate (TBOA) were tested. THA, L-trans-2,4-PDC and DL-TBOA inhibited glutamate uptake both at 2 and 10 DIV, while the GLT-1 selective uptake inhibitor DHK neither strongly affected the uptake at 2, nor at 10 DIV. These data indicated that, besides the regular increase in the glial-dependent glutamate uptake activity, a transient high-affinity, DHK insensitive, glutamate transport activity in hippocampal neurons in culture is present. This latter activity could potentially be related to the transient expression of the glial GLAST transporter in neurons.     

Role of the Cellular Stress Response in the Biogenesis of Cysteamine-Induced Astrocytic Inclusions in Primary Culture

Abstract— Cysteamine (CSH; 2-mercaptoethylamine) stimulates the accumulation of peroxidase-positive inclusions in cultured astroglia akin to those observed in the aging periventricular brain. Because CSH induces the synthesis of a stress protein (heme oxygenase) in rat liver, we hypothesized that aspects of the cellular stress response may play a role in the biogenesis of CSH-induced astro-cyte granules. In the present study, we performed indirect immunofluorescent staining and immunoblotting for various stress proteins in rat neuroglial cultures. Exposure of astrocyte cultures to CSH enhanced immunostaining for heme oxygenase-1 (HO-1) and heat-shock proteins 27, 72, and 90, but not glucose-regulated protein 94, relative to untreated cultures. CSH-pretreated astrocytes exhibited enhanced tolerance to H₂O₂ toxicity relative to untreated cells, providing physiological evidence of an antecedent stress response in the former. In addition, exposure for 12 days to H₂O₂, a known inducer of the stress response, elicited astrocyte granulation similar to that observed with CSH. Chronic induction of HO-1 and other stress proteins may participate in the biogenesis of metal-loporphyrin-rich inclusions in CSH-treated astroglial cultures and in astrocytes of the aging periventricular brain.     

Long‐term culture of mouse cortical neurons as a model for neuronal development,aging, and death

A long‐term cell culture system was used to study maturation, aging, and death of cortical neurons. Mouse cortical neurons were maintained in culture in serum‐free medium (Neurobasal supplemented with B27) for 60 days in vitro (DIV). The levels of several proteins were evaluated by immunoblotting to demonstrate that these neurons matured by developing dendrites and synapses and remained continuously healthy for 60 DIV. During their maturation, cortical neurons showed increased or stable protein expression of glycolytic enzyme, synaptophysin, synapsin IIa, α and β synucleins, and glutamate receptors. Synaptogenesis was prominent during the first 15 days and then synaptic markers remained stable through DIV60. Very early during dendritic development at DIV3, β‐synuclein (but not α‐synuclein) was localized at the base of dendritic growth cones identified by MAP2 and α‐amino‐3‐hydroxy‐5‐methyl‐4‐isoxazole (AMPA) receptor GluR1. In mature neurons, α and β synucleins colocalized in presynaptic axon terminals. Expression of N‐methyl‐D ‐aspartate (NMDA) and AMPA receptors preceded the formation of synapses. Glutamate receptors continued to be expressed strongly through DIV60. Cortical neurons aging in vitro displayed a complex profile of protein damage as identified by protein nitration. During cortical neuron aging, some proteins showed increased nitration, while other proteins showed decreased nitration. After exposure to DNA damaging agent, young (DIV5) and old (DIV60) cortical neurons activated apoptosis mechanisms, including caspase‐3 cleavage and poly(ADP)‐ribose polymerase inactivation. We show that cultured mouse cortical neurons can be maintained for long term. Cortical neurons display compartmental changes in the localization of synucleins during maturation in vitro. These neurons sustain protein nitration during aging and exhibit age‐related variations in the biochemistry of neuronal apoptosis. © 2002 Wiley Periodicals, Inc. J Neurobiol 51: 9–23, 2002     

Neuroligand-Evoked Calcium-Dependent Release of Excitatory Amino Acids from Cultured Astrocytes

Abstract: The release of excitatory amino acids (EAAs) from neuron-free cultures of neocortical astrocytes was monitored using HPLC. The neuroligand bradykinin caused a dose-dependent receptor-mediated increase in release of the EAAs glutamate and aspartate from type 1 astrocyte cell cultures obtained from rat cerebral cortex. Removal of calcium from the extracellular fluid prevented the bradykinin-induced release of EAAs from astrocytes. The addition of the calcium ionophore ionomycin caused a calcium-dependent release of EAAs. Inhibitors of the glutamate transporters p -chloromercuriphenylsulfonic acid, l - trans -pyrrolidine-2,4-dicarboxylate, and dihydrokainate failed to impair the ability of bradykinin to stimulate glutamate release from astrocytes. α-Latrotoxin, an active compound of black widow spider venom, caused a significant increase of the release of glutamate in calcium-containing saline. In calcium-depleted saline, α-latrotoxin produced an initial increase in the concentration of glutamate followed by a decline in the concentration of glutamate indicating stimulation of exocytosis coupled with low calcium-induced inhibition of endocytosis. Taken together, these data suggest that astrocytes may release neurotransmitter through a mechanism that is similar to the neuronal secretory process. Given the important role of glutamate in the induction of long-term potentiation, learning, memory, and excitotoxicity, it will be important to determine external signals that control both the uptake and release of glutamate by astrocytes.     

Astrocyte influences on ischemic neuronal death

Glutamate excitotoxicity, oxidative stress, and acidosis are primary mediators of neuronal death during ischemia and reperfusion. Astrocytes influence these processes in several ways. Glutamate uptake by astrocytes normally prevents excitotoxic glutamate elevations in brain extracellular space, and this process appears to be a critical determinant of neuronal survival in the ischemic penumbra. Conversely, glutamate efflux from astrocytes by reversal of glutamate uptake, volume sensitive organic ion channels, and other routes may contribute to extracellular glutamate elevations. Glutamate activation of neuronal N-methyl-D-aspartate (NMDA) receptors is modulated by glycine and D-serine: both of these neuromodulators are transported by astrocytes, and D-serine production is localized exclusively to astrocytes. Astrocytes influence neuronal antioxidant status through release of ascorbate and uptake of its oxidized form, dehydroascorbate, and by indirectly supporting neuronal glutathione metabolism. In addition, glutathione in astrocytes can serve as a sink for nitric oxide and thereby reduce neuronal oxidant stress during ischemia. Astrocytes probably also influence neuronal survival in the post-ischemic period. Reactive astrocytes secrete nitric oxide, TNFalpha, matrix metalloproteinases, and other factors that can contribute to delayed neuronal death, and facilitate brain edema via aquaporin-4 channels localized to the astrocyte endfoot-endothelial interface. On the other hand erythropoietin, a paracrine messenger in brain, is produced by astrocytes and upregulated after ischemia. Erythropoietin stimulates the Janus kinase-2 (JAK-2) and nuclear factor-kappaB (NF-kB) signaling pathways in neurons to prevent programmed cell death after ischemic or excitotoxic stress. Astrocytes also secrete several angiogenic and neurotrophic factors that are important for vascular and neuronal regeneration after stroke.