Astrocytic function and its alteration in the epileptic brain


Address correspondence to Dr. Christian Steinhäuser, PhD, Institute of Cellular Neurosciences, Medical Faculty, University of Bonn, Sigmund-Freud-Str. 25, D-53105 Bonn, Germany. E-mail:


Currently available anticonvulsant drugs and complementary therapies are insufficient to control seizures in about a third of epileptic patients. Thus, there is an urgent need for new treatments that prevent the development of epilepsy and control it better in patients already afflicted with the disease. A prerequisite to reach this goal is a deeper understanding of the cellular basis of hyperexcitability and synchronization in the affected tissue. Epilepsy is often accompanied by massive reactive gliosis. Although the significance of this alteration is poorly understood, recent findings suggest that modified astroglial function may have a role in the generation and spread of seizure activity. Here we summarize properties of astrocytes as well as their changes that can be associated with epileptic tissue. The goal is to provide an understanding of the current knowledge of these cells with the long-term view of providing a foundation for the development of novel hypotheses about the role of glia in epilepsy.

Astrocyte Physiology

Heterogeneity and terminology

Before discussing details of membrane physiology of astrocytes it is important to note that there are different types of cells with astroglial properties within a given brain region, and that astrocyte properties may vary in different subregions. An approach to demonstrate this heterogeneity is the analysis of glial cells in appropriate transgenic animal models. For example, cells with astroglial properties are labeled in the living brain of mice with human glial fibrillary acidic protein (hGFAP) promoter-driven enhanced green fluorescent protein (EGFP) expression (Tg[hGFAP/EGFP] mice; (Nolte et al., 2001)). In the hippocampus of these mice, a coexistence of two distinct populations of hGFAP/EGFP-positive glial cells has been identified and investigated in detail (Matthias et al., 2003; Wallraff et al., 2004; Jabs et al., 2005). Figure 1 summarizes important properties of these cells. The two cell types were termed GluT and GluR cells for their segregated expression of glutamate transporters (GluT cells) and ionotropic glutamate receptors (GluR cells), respectively. Moreover, GluT cells are characterized by a significant resting K+ conductance, giving rise to very low-input resistances, mostly below 10 MΩ, and almost linear, “passive” whole-cell current patterns (Figs 1A2 and B3). GluT cells are intensively coupled via gap junctions, enwrap blood vessels with their endfeet and bear a sponge-like, heavily branched net of processes. The majority of GluT cells show immunoreactivity for GFAP and S100β but not for NG2. Hence, GluT cells resemble all properties of bona fide protoplasmic astrocytes. Investigation of the role of gap junction coupling for K+ buffering in the hippocampus of adult mice revealed two types of astrocytes differing in morphology and function (Wallraff et al., 2006).

Figure 1.

Postrecording analysis of hGFAP/EGFP-positive cells in the hippocampus. (A1) The morphology of a GluR cell was visualized by Texas Red dextran filling during whole cell recording. Subsequent confocal analysis and 2D projection of 32 optical sections (total depth 21 μm) allowed the detailed resolution of cellular process arborization. Note the typical nodules appearing as dots all along the fine processes. The current pattern of this GluR-type glial cell is given in the middle panel. Current responses were evoked by de- and hyperpolarizing the membrane between +20 and −160 mV (holding potential −80 mV), and capacitive artifacts were compensated offline (Vrest=−83 mV, Ri= 78 MΩ, CM= 37 pF). This cell showed sPSPs and ePSPs sensitive to NBQX and bicuculline. Postrecording immunostaining and triple fluorescence confocal analysis were applied to examine NG2 immunoreactivity. The middle panel shows the three separated color channels of one confocal plane. To improve visibility, Texas Red dextran labeling of the recorded cell is shown in green (g), NG2 immunoreactivity in red (r), and EGFP expression in blue (b). Note that the EGFP fluorescence remaining post-recording is only 16 % compared to surrounding cells (b). The superimposed RGB picture (right panel) shows the membrane-associated distribution of NG2 immunoreactivity of the recorded GluR cell (yellow details). (A2) In contrast to GluR cells, hGFAP/EGFP-positive GluT-type astrocytes predominantly expressed time- and voltage-independent currents (middle panel, stimulus protocol as in A1) and displayed a different morphology (left panel, see text for details; Vrest=−84 mV, Ri= 3 MΩ, Cm= 71 pF). The cell did not generate sPSCs. The EGFP fluorescence intensity determined postrecording reached 53 % of that measured in adjacent cells (b). The cell was NG2-negative (middle panel (r) and right panel). (B1–3) Analog to (A), GluR and GluT cells were tested postrecording for S100β immunoreactivity. The cells were recorded for exactly 1 min (see text). (B1, left) 2D projection of a GluR cell after TRITC dextran filling (16 optical sections, total depth 8.4 μm) reveals a typical morphology with thin, wide spanning, nodule-containing processes. (B1, middle) Artifact-compensated current pattern of the GluR cell (Vrest=−84 mV, Ri= 72 MΩ, Cm= 29 pF). In this cell, postrecording analysis did not detect S100ß immunoreactivity (S100ß, red (r); TRITC dextran, green (g)). (B2) Another GluR cell (Vrest=−83 mV, Ri= 270 MΩ, Cm= 24 pF) showed postrecording S100β labeling. (B3) Analysis of a GluT cell. Projection of EGFP fluorescence (left, 32 optical sections, total depth 19.5 μm) revealed its characteristic morphology. (B3, middle) Current pattern of the GluT cell (Vrest=−86 mV, Ri= 5.1 MΩ, Cm= 61 pF). The cell was filled with Texas Red dextran (g) during recording, and postrecording confocal analysis detected S100ß immunoreactivity (71 % fluorescence intensity as compared to surrounding S100ß-positive cells (r)). Scale bars in morphological pictures represent 10 μm; for current patterns, 1 nA and 10 ms, respectively. From (Jabs et al., 2005), with permission.

On the other hand, GluR cells lack dye coupling and are characterized by the expression of various voltage- and time-dependent ion channels and input resistances larger than 50 MΩ (Figs 1A1, B1, B2). Their processes often spread in a radial, spider web-like manner and apparently do not touch blood capillaries. GluR cells receive synaptic input from GABAergic interneurons and glutamatergic CA1 pyramidal neurons (Jabs et al., 2005). The majority of GluR cells is immunoreactive for NG2 but not for GFAP. In this article, we refer to both cell types as astrocytes, based on their GFAP promoter activity. However, it is clear that the functional impact of these two cell types with astroglial properties is different, although not yet completely understood. It will be a challenge for future work to define sets of parameters allowing for unequivocal identification and discrimination of glial cell types in the central nervous system (CNS).

Membrane physiology

Astrocytes express almost the same set of ion channels and receptors as neurons do (Verkhratsky & Steinhäuser, 2000; Seifert & Steinhäuser, 2004; Kettenmann & Steinhäuser, 2005) although the relative strength of expression varies between the two cell types. For example, in astrocytes, K+ channel density exceeds that of Na+ channels by far, preventing generation of glial action potentials. Nevertheless, a glia-specific, or at least preferential expression has been elucidated for some of these channels and carriers. Among them is Kir4.1, a subunit belonging to the family of inwardly rectifying K+ (Kir) channels. In the CNS, this channel is predominantly localized at distant astrocyte processes surrounding synapses or capillaries (Higashi et al., 2001). Recent work suggests a colocalization of Kir4.1 with the water channel aquaporin-4 (AQP4), which in the brain and spinal cord is also expressed by astrocytes but not neurons. Increasing evidence indicates that a coordinated action of both channels is required for the astrocytes to maintain K+ and water homeostasis in the CNS (Nielsen et al., 1997; Verkman, 2005). As illustrated in the following sections, dysfunction of these astroglial transmembrane channels appears to play a key role in epilepsy.

In contrast to the majority of mature neurons, astrocytes are usually coupled through gap junctions to form large intercellular networks. Astrocytic gap junctions are mainly formed by connexins 43 and 30 (Cx43 and Cx30) in a cell type-specific fashion. Through these networks astrocytes can dissipate molecules, such as K+ or glutamate, a process considered important to prevent their detrimental extracellular accumulation (Theis et al., 2005). Recent data suggest that the capacity of K+ clearance is only partially disturbed in the absence of astrocyte gap junctions, presumably because of the existence of “indirect” coupling of elongated astrocytic processes (Wallraff et al., 2006). Connexins also contribute to the propagation of intercellular Ca2+ waves, presumably by enhancing ATP release, rather than by providing an intercellular pathway for signal diffusion (Nedergaard et al., 2003). However, the pathological impact of disturbed astroglial gap junction expression is not well understood yet (Steinhäuser & Seifert, 2002; Seifert et al., 2006).

Another important function of astrocytes is the removal of neurotransmitters released by active neurons. Uptake of glutamate is accomplished by two glia-specific transporters, EAAT1 and EAAT2 (in rodents termed GLAST and GLT-1), the activity of which may shape the kinetics of receptor currents at synapses (Bergles et al., 1999; Danbolt, 2001). Compelling evidence suggests that disturbed glutamate uptake by astrocytes is directly involved in the pathogenesis of epilepsy, as discussed in the following sections.

Astrocytes can also release neuroactive agents, including neurotransmitters. Several studies revealed that such a release is critically dependent on an increase of astroglial [Ca2+]i. Astrocytes express a plethora of neurotransmitter receptors that are coupled through G-proteins (Gq) and phospholipase C to the release of Ca2+ from internal stores (Haydon, 2001). Stimulation of neuronal afferents induces Ca2+ elevations within astrocytes (Dani et al., 1992; Porter & McCarthy, 1996), which can spread to neighboring astrocytes, demonstrating the presence of an astrocyte-to-astrocyte network (Charles et al., 1991; Sul et al., 2004). Thus, although electrically inexcitable, astrocytes contain a chemically based form of excitability that is bidirectionally linked to neuronal activity (Haydon, 2001). Though initially discovered in 1994 (Parpura et al., 1994), the past decade has seen many studies demonstrating that astrocytes release chemical transmitters (“gliotransmitters”), including glutamate, ATP and D-serine (reviewed in Haydon, 2001; Volterra & Meldolesi, 2005). Although the mechanism(s) underlying astroglial transmitter release are open to debate, at least part of the release seems to occur through regulated, Ca2+-dependent exocytosis, a mechanism that in the CNS was previously thought to be exclusive to neurons.

What is the impact of transmitter release from astrocytes? Several reports suggested that gliotransmitters may activate receptors in neurons to modulate the strength of inhibitory and excitatory synaptic transmission (Bezzi et al., 1998; Kang et al., 1998; Parri et al., 2001; Yang et al., 2003; Zhang et al., 2003; Fiacco & McCarthy, 2004; Pascual et al., 2005; reviewed by Volterra and Meldolesi, 2005). Importantly, because the fine terminal processes of single astrocytes reach tens of thousands of synapses simultaneously (Bushong et al., 2002), the release of gliotransmitters may lead to the synchronization of neuronal firing (Angulo et al., 2004; Fellin et al., 2004). The different gliotransmitters that are released from astrocytes have quite distinct functions. Glutamate, the first identified gliotransmitter (Parpura et al., 1994), is able to modulate neuronal excitability (Angulo et al., 2004; Fellin et al., 2004; Tian et al., 2005) through actions on N-methyl-d-aspartate (NMDA) receptors, and can also modulate synaptic transmission (Fiacco & McCarthy, 2004). D-serine is a coagonist of the NMDA receptor. Released D-serine can bind to the glycine-binding site of NMDA receptors and, as a consequence, enhance neuronal NMDA receptor function. In the hypothalamus, the amount of astrocyte-derived D-serine supplied to synapses can regulate forms of synaptic plasticity. In conditions where too little of the coagonist is supplied, long-term synaptic depression can result, whereas when D-serine is released, long-term potentiation is induced (Panatier et al., 2006). Release of ATP from astrocytes can have a several effects. In cultures, release of ATP mediates the propagation of Ca2+ waves through paracrine actions on neighbors, inducing further Ca2+ signals and ATP release (Guthrie et al., 1999). In more intact preparations, because of the presence of a plethora of extracellular ectonucleotidases, newly ATP released is rapidly hydrolyzed to adenosine. As a consequence, ATP released from astrocytes leads to synaptic modulation mediated by adenosine (Pascual et al., 2005; Serrano et al., 2006). In the hippocampus, high-frequency activity of groups of synapses induces Ca2+ signals in neighboring astrocytes, which then release ATP and, after hydrolysis to adenosine, cause presynaptic inhibition of neighboring synapses through A1 receptors (Pascual et al., 2005; Serrano et al., 2006). In this manner, astrocytes may coordinate the strength of synaptic signaling. However, we still await an understanding of the functional consequences of gliotransmission on neural signalling, i.e., on processes such as learning and memory and ultimately behavior. Rapid forms of neuron-glia interactions also seem to be involved in the regulation of local blood flow as demonstrated in cortical brain slices where neuronal stimulation led to glutamate release, activation of metabotropic glutamate receptors (mGluRs) in astrocytes, and regulation of the tone of vessels contacted by processes of the stimulated astrocyte (Zonta et al., 2003; Mulligan & MacVicar, 2004; Takano et al., 2006). Moreover, activity-dependent astrocytic Ca2+ increase led to K+ release through astrocytic BK channels and to vasodilatation of arteriolar smooth muscle cells in the brain (Filosa et al., 2006). Emerging evidence suggests that disturbances of these mechanisms are involved in the pathogenesis of epilepsy, as discussed below.

Several important aspects of neuron-glia interactions are not yet understood. Thus, as outlined above, recent studies corroborated the finding that astrocytes are heterogeneous with respect to antigen profiles and functional properties but it is still unclear which type(s) of astroglial cells are activated and are capable of releasing transmitters, which transmitters can be released by astrocytes, which mechanisms these cells use for the release, and whether the efficiency of neuron-glia signaling changes during development. Intriguingly, a recent report presented evidence that a subtype of cells with astroglial properties even receives direct synaptic input from glutamatergic and GABAergic neurons (Jabs et al., 2005). The physiological impact of this type of interaction remains to be clarified.

Astrocyte Dysfunction in Temporal Lobe Epilepsy

Neurons have been the primary focus of attention in the study of the pathology of epilepsies, because ictal activity is generated by neurons. It is, however, becoming clear that glial cells, in particular astrocytes, may play a major role in the excitability generated at seizure foci (Binder & Steinhäuser, 2006; Seifert et al., 2006). Reexamination of the pathology of seizure foci may therefore indicate a significant astrocytic component in many seizure foci. The hippocampal seizure focus in temporal lobe epilepsy (TLE) has been studied the most, compared to other seizure foci in the brain. These studies suggested that astrocytes at sclerotic hippocampal seizure foci have unique structural and functional characteristics. In this section, these characteristics are reviewed to exemplify some of the mechanisms through which they may contribute to seizure generation, and compared to the features of astrocytes at other seizure foci.

Voltage gated Na+ and Ca2+ channels

Information about changes in Nav channel expression in experimental or human epilepsy is inconsistent. Comparative patch-clamp analyses were performed in hippocampal specimens with and without significant astroglial sclerosis, surgically removed from patients with intractable TLE. Two papers reported enhanced Na+ current densities in human astrocytes from sclerotic specimens (Bordey & Sontheimer, 1998; Bordey & Spencer, 2004), while no increase was found in another human study (Hinterkeuser et al., 2000) and in the hippocampus of kainate-treated rats, an animal model of TLE (Jabs et al., 1997). It is conceivable that this apparent discrepancy reflects subregional differences because the latter two studies investigated astrocytes in the CA1 region while the aforementioned focused on the hilus.

During seizure activity, the extracellular Ca2+ concentration decreases at the site of the seizure focus. Low [Ca2+]o, in turn, generates spontaneous epileptiform activity. In this regard, it is worth considering the observation that depletion of external Ca2+ can induce Ca2+ oscillations in astrocytes. An increase in [Ca2+]i is known to induce the release of the excitatory amino acid glutamate from astrocytes, a process that seems to contribute to seizure generation. Although decreases in [Ca2+]o probably arise from Ca2+ influx into neurons, Ca2+ channels in astrocytes might also contribute to the depletion. Notably, immunostaining revealed an up-regulation of astrocyte L-type Ca2+ channels both in the kainate model of epilepsy and sclerotic hippocampal specimens from TLE patients (Westenbroek et al., 1998; Djamshidian et al., 2002), suggesting enhanced glial uptake of Ca2+ in the lesioned CNS. Besides a rapid and direct contribution to seizure generation via depletion of [Ca2+]o, enhanced astroglial Ca2+ influx is likely to stimulate synthesis and release of transmitters, cytokines, and growth factors, which may modify the architecture and activity of neural circuitry in the long term. Indeed, molecular analysis of hippocampal tissue resected from patients suffering from Ammon's horn sclerosis (AHS) showed enhanced expression of genes encoding proteins involved in astrocytic glutamate release (Lee et al., 2007).

Under pathological conditions such as epilepsy, Ca2+ might also enter astrocytes through Kir4.1 channels. Recent data indicated that local depletion of [K+]o, which might occur during compensatory K+ undershoot through prolonged activity of Na+/K+ ATPase activity, renders astroglial Kir channels Ca2+-permeable (Dallwig et al., 2000; Härtel et al., 2007). This pathway might well add to astroglial Ca2+ oscillations and could contribute to glutamate release from astrocytes.

K+ channels and water channels

Seizure activity in vivo is characterized by elevations of [K+]o from 3 mM to a ceiling level of 10–12 mM while, on the other hand, high [K+]o levels are sufficient to trigger seizure-like events in acute brain slices. Because of its presumed importance in the regulation of excitability, properties of astroglial Kir channels have been investigated in experimental and human epilepsy. Evidence of an involvement of Kir channels in impaired K+ buffering in sclerotic human hippocampus came from measurements of [K+]o with ion-sensitive microelectrodes and patch-clamp studies. Heinemann's group compared the effect of Ba2+ on stimulus-induced changes in [K+]o in the CA1 region of hippocampal brain slices obtained from TLE patients with AHS or without sclerosis (non-AHS). In non-AHS tissue, Ba2+ significantly augmented rises in [K+]o while this effect was not observed in AHS specimens. Since Ba2+ is a blocker of Kir channels, these findings suggested impaired function of these channels in the sclerotic tissue (Kivi et al., 2000). Indeed, direct evidence for a down-regulation of Kir currents in the sclerotic human CA1 region of epilepsy patients came from a comparative patch-clamp study (Hinterkeuser et al., 2000; see also Bordey & Sontheimer, 1998). Together, these data indicate that in the sclerotic condition impaired K+ buffering through reduced expression of functional Kir channels contributes to or even initiates seizure generation. Preliminary data suggested an involvement of the Kir4.1 subunit in this process (Schröder et al., 2000).

Recent studies revealed a spatial overlap of Kir (subunit Kir4.1) and water channels (aquaporins) in astroglial endfeet contacting capillaries (Nielsen et al., 1997; Higashi et al., 2001) and suggested that buffering of K+ via Kir channels depends on concomitant transmembrane flux of water in the same cell. These parallel water fluxes are thought to be necessary to dissipate osmotic imbalances due to K+ redistribution. In agreement with this hypothesis, clearance of extracellular K+ is compromised upon reduction of the perivascular pool of astroglial AQP4 (Amiry-Moghaddam et al., 2003b), and impaired K+ buffering in concert with prolonged seizure duration, is observed in AQP4−/− mice (Binder et al., 2006). Hence, dysfunction of the blood-brain barrier seems to be involved in seizure generation. Indeed, a recent study suggests that transient opening of the blood-brain barrier is sufficient for focal epileptogenesis (Seiffert et al., 2004). The same group has now shown that albumin is taken up by astrocytes through a TGF-β receptor-mediated mechanism, resulting in down-regulation of Kir4.1 channels, impaired buffering of extracellular K+ and abnormal NMDA receptor-mediated hyperexcitability (Ivens et al., 2007). In the sclerotic hippocampus of TLE patients, AQP4 immunoreactivity of vasculature-associated astrocyte endfeet was lower compared with nonsclerotic human epileptic hippocampi (Eid et al., 2005). This loss of perivascular AQP4 is probably secondary, following disruption of the dystrophin complex that is necessary for anchoring of AQP4 (Amiry-Moghaddam et al., 2003a). In conclusion, in the sclerotic hippocampus of TLE patients' dislocation of water channels in concert with reduced expression of Kir channels in astrocytes probably underlie the impaired K+ buffering leading to increased seizure propensity.

Glutamate transporters and receptors

Several studies have suggested an involvement of glutamate transporters and receptors in seizure development and spread. Increased extracellular levels of glutamate have been found in epileptogenic foci (Glass & Dragunow, 1995). Glutamate transporters are expressed by several CNS cell types, but astrocytes are primarily responsible for glutamate uptake. Important studies, using mice with deletion (Tanaka et al., 1997) or antisense oligonucleotide-mediated inhibition of synthesis (Rothstein et al., 1996) of the astroglial transporter GLT-1, revealed that this subtype is responsible for the bulk of extracellular glutamate clearance in the CNS. Elimination of GLT-1 in knockout mice, but not antisense inhibition, resulted in the development of spontaneous seizures and hippocampal pathology resembling alterations in TLE patients with AHS. Pharmacological inhibition of GLT-1 reduced the threshold for evoking epileptiform activity (Campbell & Hablitz, 2004; see also Demarque et al., 2004). Reduced GLAST and GLT-1 expression was also observed in a tuberous sclerosis epilepsy model (Wong et al., 2003) but other animal studies were contradictory. Tessler and colleagues (Tessler et al., 1999) investigated transporter expression on the mRNA and protein levels in human TLE specimens and found changes neither for GLT-1 nor GLAST. However, two other groups reported decreased GLT-1 protein as well as reduced (Mathern et al., 1999) or increased (Proper et al., 2002) GLAST immunoreactivity in the sclerotic human hippocampus. The latter authors also noted an up-regulation of GLT-1 in the nonsclerotic epileptic hippocampus. These findings support the hypothesis that reduced or dysfunctional glial glutamate transporters in the hippocampus may trigger spontaneous seizures in AHS patients (During & Spencer, 1993), yet the underlying mechanisms are unclear. It has been proposed that the role of glutamate transporters in epilepsy may not be related directly to the control of excitation through synaptic glutamate concentration but rather to impaired glutamate metabolism (Maragakis & Rothstein, 2004). In this context, the finding of a loss of glutamine synthetase in the sclerotic versus nonsclerotic hippocampus of TLE patients (Eid et al., 2004) deserves further consideration. After uptake of glutamate into astrocytes, this enzyme rapidly converts the transmitter into glutamine that is then transported to neurons, where it is resynthesised to glutamate. Eid and coworkers did not observe epilepsy-related changes in the expression of GLT-1. They concluded that in the sclerotic tissue down-regulation of glutamine synthetase caused a slowing of glutamate-glutamine cycling, which resulted in the accumulation of the transmitter in astrocytes and in the extracellular space (Eid et al., 2004). This conclusion is compatible with findings in animal models of epilepsy and earlier data, demonstrating slowed glutamate-glutamine cycling in sclerotic human epileptic hippocampus with magnetic resonance spectroscopy (Petroff et al., 2002). Whether activation of glutamate transporters cells, e.g., through β-lactam antibiotics (Rothstein et al., 2005), might be beneficial in the treatment of epilepsies remains a matter of further investigation.

A few studies addressed the potential involvement of ionotropic glutamate receptors cells in seizure generation. Mouse mutants with deficient GluR2 Q/R editing develop early-onset epilepsy with spontaneous and recurrent seizure activity, suggesting that enhanced Ca2+ influx through the Q form of the GluR2 subunit of AMPA receptors reduces seizure threshold (Brusa et al., 1995). Astrocytes also carry the GluR2 subunit, but altered glial GluR2 editing does not seem to play a role in human TLE. Rather, combined functional and single-cell transcript analyses put forward the idea that enhanced expression of GluR1 flip variants account for the prolonged receptor responses observed in hippocampal astrocytes of TLE patients with AHS (Seifert et al., 2002; Seifert et al., 2004) (Fig. 2). This alteration in the splicing status of AMPA receptors predicts enhanced depolarization upon activation by endogenously released glutamate. Prolonged receptor opening will promote influx of Ca2+ and Na+ ions, and the latter plug astroglial Kir channels (Schröder et al., 2002). This, in turn, will further strengthen depolarization and reduce the K+ buffer capacity of astrocytes. It is yet unknown whether the changes in glial receptor function are causative of, or result from, the epileptic condition. Also, the question of how and to which extent alterations in glial GluR1 splicing contribute to seizure generation or spread requires further investigation. Astrocytes cultured from patients with Rasmussen's encephalitis, a rare form of childhood epilepsy, show spontaneous Ca2+ oscillations that are dependent on transmembrane influx of Ca2+ (Manning & Sontheimer, 1997). The authors speculated that these responses might add to neuronal hyperactivity, possibly due to autocrine ionotropic glutamate receptor stimulation by glutamate released from astrocytes. Another study suggested that the destruction of astrocytes by GluR3 antibodies plays a critical role in the progression of this autoimmune disorder (Whitney & McNamara, 2000).

Figure 2.

(A, B) Activation and desensitization properties of glutamate-evoked currents in acutely isolated human astrocytes. (A) Fast application of glutamate (1 mM) elicited inward currents in astrocytes from AHS and lesion specimens. The glutamate responses desensitized rapidly and almost completely, with a mono-exponential current decay. Time constants of current desensitization were as indicated. (B) Glutamate-induced currents of astrocytes located in lesion-associated human hippocampal tissue decay significantly faster than corresponding currents of cells from AHS patient tissue. The degree of desensitization is similar in both patient groups. Holding potential was always −70 mV. (C) AMPA receptor subunit expression and splicing by human astrocytes. After electrophysiological characterization, the cytoplasm of the respective cell was harvested and analyzed with single-cell RT-PCR. Cells from AHS (n= 16) and lesion specimens (n = 13) displayed similar subunit combinations. Restriction analysis revealed prevailing expression of flip versions. GluR1 flip was more abundant in AHS (asterisk). (D) Determination of AMPA-receptor flip/flop ratios in human astrocytes using semiquantitative real-time PCR. Amplification plots of two individual astrocytes from AHS (filled symbols) and lesion specimens (open symbols) are shown. Note the higher cycle number necessary to amplify flop transcripts (circles) in the AHS cell. The threshold for detection was set at ΔRn= 0.17 (indicated by the dashed line). Modified from (Seifert & Steinhäuser, 2004), with permission.

Ca2+ signaling in astrocytes

Under normal conditions, mGluR3 and mGluR5 are the predominant metabotropic glutamate receptor subtypes expressed by glial cells. Activation of these receptors affects cAMP accumulation and leads to an increase in intracellular Ca2+, respectively. The Ca2+ rise may oscillate and initiate Ca2+ wave propagation within the astrocyte network, activate Ca2+-dependent ion channels and induce glutamate release from astrocytes. In experimental epilepsy, reactive astrocytes of the hippocampus persistently up-regulate mGluR3, mGluR5, and mGluR8 protein (Steinhäuser & Seifert, 2002). Electron-microscopic inspection of hippocampal tissue from TLE patients revealed an expression of mGluR2/3, mGluR4, and mGluR8 in reactive astrocytes, suggesting an involvement of these receptors in gliosis (Tang & Lee, 2001). Up-regulation of astroglial mGluR2/3 and mGluR5 was also observed in epileptic specimens from patients with focal cortical dysplasia (Aronica et al., 2003b). Because their activation modulates the expression of GLAST and GLT-1 (Aronica et al., 2003a) and elevates [Ca2+]i, mGluRs in astrocytes might be involved in the generation of seizure foci.

Over the past decade, Ca2+ signalling mechanisms in astrocytes have received considerable attention. Of particular importance for this discussion is the novel observation that astrocytes exhibit Ca2+-induced release of glutamate, which provides direct excitation to neighboring neurons. Because of the observed changes in protein expression within astrocytes following injuries that lead to the development of epilepsy, it is tempting to speculate that alterations in this glia-derived excitatory pathway in coordination with reductions in glutamate uptake might provide an excitatory drive underlying seizure disorders. Recent work suggested that in chemically induced, acute epilepsy models, astrocytic Ca2+ oscillations and glutamate release contribute to the generation of synchronized epileptiform activity (Tian et al., 2005). However, another study, using the same model, indicated that astrocytic glutamate is not necessary for the generation of epileptiform activity (Fellin et al., 2006). Future work will have to elaborate whether astrocytes provide sufficient excitation to contribute to seizures and whether they indeed represent new targets for the development of anti-epileptic treatments.


Astrocytes undergo morphological alterations in epilepsy, and recent evidence suggests that these structural changes are accompanied by modified cellular function. Since astroglial cells have been identified as direct communication partners of neurons, their dysfunction might be involved in the pathogenesis of the disease. In fact, surgical removal of sclerotic tissue often results in a significant improvement of the epileptic condition, suggesting that gliosis contributes to seizure generation. It remains an important issue to figure out which factors initiate the dysregulation of astrocytic signaling molecules, and whether these changes are causative for the development of the disease. Moreover, the specific role of different astroglial subtypes in epilepsy still has to be elaborated. If these cell types are affected differently in epilepsy, this will likely produce distinct consequences for the excitability of neural circuitry. In fact, cells with astroglial properties differ significantly in their morphological and functional properties, but most studies describing glial alterations in epilepsy have so far not identified the cellular subtype affected. Thus, it is important to appreciate that some of the changes discussed here may represent impaired properties of one astroglial subtype, and other changes could result from modified ratios of different subclasses of astrocytes. A further understanding of the diversity of “normal” astrocytes, by establishing critical parameter sets allowing unequivocal subclassification, will be an essential step in unraveling the role of astrocytes in epilepsy and other neurological disorders. Progress made in the field over the past years suggests that this should eventually open novel rational therapeutic approaches to those seizure disorders, which can only be poorly controlled so far.


Work of the authors are supported by Deutsche Forschungsgemeinschaft (grants RJ 942/2 to RJ, SE 774/3 to GS, SFB/TR3 C1, C3 to CS). We thank B. Bäcker for comments on the manuscript, and apologize to all those whose work could not be discussed due to space constraints.

Conflict of interest:  The contributing authors to this article have declared no conflicts of interest.