Neurotherapeutics: The Journal of the American Society for Experimental NeuroTherapeutics

Astrocytes and Epilepsy

Nihal C. de Lanerolle,* Tih-Shih Lee,† and Dennis D. Spencer*

*Department of Neurosurgery, Yale School of Medicine, New Haven, Connecticut 06520; †Department of Psychiatry,
Duke University, Durham, North Carolina 27708

Summary: Astrocytes form a significant constituent of seizure capacity to generate action potentials. They also have reduced
foci in the human brain. For a long time it was believed that glutamine synthetase and increased glutamate dehydrogenase
astrocytes play a significant role in the causation of seizures. activity. The molecular interface between the astrocyte and
With the increase in our understanding of the unique biology of microvasculature is also changed. The astrocytes are also as-
these cells, their precise role in seizure foci is receiving re- sociated with increased expression of many molecules nor-
newed attention. This article reviews the information now mally concerned with immune and inflammatory functions.
available on the role of astrocytes in the hippocampal seizure A speculative mechanism postulates that neuron glia-2-like
focus in patients with temporal lobe epilepsy with hippocampal cells may be involved in creating a high glutamate environ-
sclerosis. Our intent is to try to integrate the available data. ment, whereas the function of more typical reactive astro-
Astrocytes at seizure foci seem to not be a homogeneous pop- cytes contribute to maintain high extracellular K⫹ levels;
ulation of cells, and in addition to typical glial fibrillary acidic both factors contributing to the hyperexcitability of subicu-
protein, positive reactive astrocytes also include a population of lar neurons to generate epileptiform activity. The functions
neuron glia-2-like cells The astrocytes in sclerotic hippocampi of the astrocyte vascular interface may be more critical to
differ from those in nonsclerotic hippocampi in their membrane the processes involved in epileptogenesis. Key Words: As-
physiology, having elevated Na⫹ channels and reduced in- trocytes, temporal lobe epilepsy, hippocampal sclerosis, NG2
wardly rectifying potassium ion channels, and some having the cells, seizures.

INTRODUCTION on the role of astrocytes in epilepsy carried out mostly on

gliotic scars induced by topical application of chemicals
It is generally recognized that the immediate cause of
such as alumina or cobalt to the cortex of animal brains
epilepsy is an over-excitation of populations of neurons
in the brain and the spread of such excitation throughout are reviewed by Tiffany-Castiglioni and Castiglioni.7 In
the brain resulting in behavioral seizures. However, for a conclusion to their review, the authors write that “astro-
considerable period, it was believed that astrocytes play cytes can hypothetically contribute to epileptogenesis by
a significant role in the causation of seizure activity. any of three routes: (1) the initiation of neuronal hyper-
Several brain foci associated with seizure generation are activity in previously normal neurons, (2) the promotion
populated by increased numbers of astrocytes. Such foci of epileptic bursting in abnormal neurons, and (3) the
include hippocampal seizure foci in temporal lobe epi- failure to neutralize and arrest neuronal hyperactivity.”
lepsy (TLE), several types of mass lesions in the brain However, they go on to add that “evidence that astro-
(low-grade astrocytomas, oligodendrogliomas, arterio- cytes cause seizures is almost completely lacking, except
venous malformations) and tubers in tuberous sclerosis.1 for the finding that glial scars mechanically distort neu-
Gliotic scar formation is a prominent feature of human ronal morphology.” In the years since this review by
epilepsy.2,3 The almost invariable presence of gliotic Tiffany-Castiglioni and Castiglioni,7 our understanding
scars in chronic focal epilepsy has led many to suggest a of the biology and physiology of astrocytes has ex-
physiological role for glia in the disease.4-6 Early studies ploded. As a consequence, so has our understanding of
the role of astrocytes in epilepsy.
In this article, we review the information available on
Address correspondence and reprint requests to: Nihal C. de Laner- the probable contribution of astrocytes to human epi-
olle, D.Phil., D.Sc., Department of Neurosurgery FMB 414, Yale
School of Medicine, 333 Cedar St, New Haven, CT 06520-8082. E-mail: lepsy, based on the information gathered on hippocampal
Nihal.delanerolle@yale.edu. seizure foci in patients with TLE, which is a naturally

424 Vol. 7, 424 – 438, October 2010 © The American Society for Experimental NeuroTherapeutics, Inc.
 ASTROCYTES AND EPILEPSY 425

occurring seizure disorder. TLE seizures originate from

the temporal lobe, particularly the hippocampus.8 The
seizure foci are surgically removed for the control of
medically intractable seizures and thus provide for a
careful study of their biology.
The examination of hippocampal specimens from pa-
tients who had undergone surgery for the control of
medically intractable epilepsy has revealed approxi-
mately 40% to 65% of cases with hippocampal sclero-
sis.9 Hippocampal excision surgery in this type of patient
has a better surgical outcome than in those without scle-
rosis. Analysis of the pathology and electrophysiology of
151 hippocampi removed in the Yale surgical series re-
vealed that after surgery, 84% of patients with sclerotic
hippocampi had an excellent (Engel class I) outcome.9 In
these sclerotic hippocampi, in addition to a greater than
50% loss of neurons, the glial density is increased by
approximately 80% compared to nonsclerotic hip-
pocampi and neurologically normal autopsy controls10,11
(FIG. 1). These astrocytes exhibit several unusual prop-
erties.

ASTROCYTE RECEPTORS
Astrocytes express similar sets of receptors as do neu-
rons, but with varying relative strength of expression.
These receptors can be activated by synaptically released
neurotransmitters, by “glio” transmitters, or by mole-
cules diffusing in the brain extracellular space.12 Among FIG. 1. A: Photomicrograph of a section of sclerotic human
the receptors expressed on astrocytes are glutamate re- hippocampus immunostained for glial fibrillary acidic protein
(GFAP). The clear regions (dentate granule cell layer, CA2, and
ceptors, and both ionotropic and metabotropic receptors. subiculum) are those with little or no neuronal loss and thus weak
Ionotropic glutamate receptors of the ␣-amino-3-hy- GFAP immunoreactivity. The hilus, CA3, and CA1 regions have
droxy-5-methyl-4-isoxazolepropionic acid (AMPA) sub- considerable neuronal loss strong GFAP positivity. CA ⫽ Cornu
Ammonis or Ammon’s horn. (Courtesy of Dr. Jung H. Kim, De-
type made up of subunits GluR1 to GluR4 are found on partment of Pathology, Yale University). B: Glial density as ex-
astrocytes. An elevated flip-to-flop mRNA ratio of the pressed by mean glial number per mm3. Error bar ⫽ standard
GluR1 splice variant is reported in reactive astrocytes error of the mean. Yellow bar ⫽ autopsy control (n ⫽ 12); red
bar ⫽ sclerotic hippocampi (n ⫽ 83); green bar ⫽ temporal lobe
from sclerotic hippocampi, suggesting an increase in re- epilepsy with extrahippocampal pathology (nonsclerotic). (Cour-
sponsiveness of these astrocytes to glutamate.13,14 Im- tesy of Dr. Jung H. Kim, Dept. of Pathology, Yale University.)
munoelectron microscopic examination of sclerotic hip-
pocampi has revealed the expression of mGluR2/3,
mGluR4, and mGluR8 in reactive astrocytes. In the same normal activity of the brain, astrocytes play a major role
manner, mGluR3, mGluR5, and mGluR8 are reported to in the clearance of glutamate released by synapses into
be up-regulated in the hippocampus in experimental an- the extracellular space. Astrocytes achieve this through
imal models of TLE.15 Activation of these receptors the activity of two glutamate transporter molecules ex-
leads to an increase in intracellular Ca2⫹ and Ca2⫹ wave citatory amino acid transporter (EAAT)1 and EAAT2.
propagation, leading to the release of glutamate from Increased levels of extracellular glutamate are detected
astrocytes according to some investigators.16 However, by in vivo dialysis in sclerotic seizure foci.18,19 Some
whether Ca2⫹ waves or oscillations lead to glutamate authors have reported a down regulation of EAAT1 and
release remains controversial.17 EAAT2 glutamate transporters20,21 in sclerotic hip-
pocampi and propose that this accounts for the observed
increase in extracellular glutamate. However, others
ASTROCYTE TRANSPORTER MOLECULES
have been unable to confirm such an observation.22 A
On their cell membrane, astrocytes have a variety of careful review of these discrepant data lead to the con-
transporter molecules through which they exchange a clusion that astrocyte glutamate transporter differences
variety of molecules with the extracellular space. In the in sclerotic versus nonsclerotic hippocampi are an inad-

Neurotherapeutics, Vol. 7, No. 4, 2010

426 DE LANEROLLE ET AL.

equate explanation for high extracellular glutamate lev- decrease in the expulsion of water from the neuropil out
els observed by in vivo dialysis in sclerotic seizure foci.23 to the blood vessel lumen.25
The ␥-amino butyric acid (GABA) transporter ␥-ami-
nobutyric acid transporter (GAT)3 is usually only
MEMBRANE ION CHANNELS
weakly expressed, if at all, on astrocytes. The expression
of GAT3 on astrocytes in the sclerotic hippocampus is Several studies involving patch and voltage-clamp
increased.24 GAT3 expression is confined to cells resem- techniques demonstrate the presence of voltage-depen-
bling protoplasmic astrocytes, which are located in re- dent Na⫹, K⫹, and Ca2⫹ ion and anion channels on
gions of relative neuronal sparing, such as the dentate astrocytes.26-31 Comparative patch-clamp studies were
gyrus and hilus of the sclerotic hippocampus.24 In vivo carried out on astrocytes in hippocampal specimens with
microdialysis studies in human hippocampal seizure foci and without significant sclerosis removed in epilepsy
have demonstrated reduced levels of extracellular GABA surgery. In a study from our laboratory, primary cultures
in the epileptogenic seizure focus during the ictal state.18 of astrocytes established from the hippocampus (dentate
The increased expression of GAT3 in these hippocampi gyrus and Ammon’s horn) and parahippocampus (ento-
may contribute to excess removal of GABA and thus rhinal region) of sclerotic hippocampi displayed much
reduced extracellular levels in the ictal state. larger tetrodotoxin (TTX)-sensitive Na⫹ currents with
Aquaporin 4 (AQP4) is a water transporter molecule ⬃66-fold higher Na⫹ channel density than in astrocytes
that is found on astrocytes in the hippocampus. The from nonsclerotic hippocampi and comparison temporal
distribution of these transporter molecules shows a dis- neocortex of the same patient32 (FIG. 3). Two other
tinct polarity on the astrocyte membrane being more studies have confirmed enhanced Na⫹ current densities
densely expressed on the perivascular astrocytic end feet in human astrocytes in situ in sclerotic hippocampi.33,34
than on the membrane facing the neuropil (FIG. 2). In A third study found no increase.35
sclerotic hippocampi the expression of AQP4 on the The expression of voltage-dependent calcium channel
perivascular membrane of the astrocyte is reduced, a1 subunits was examined by immunohistochemistry in
whereas its expression remains unchanged on the mem- sclerotic hippocampi from patients with TLE and com-
brane facing the neuropil, thus suggesting a probable pared to their expression in autopsy controls.36 A signif-

FIG. 2. Quantitative ImmunoGold electron microscopy reveals significant loss of AQP4 from the perivascular astrocyte membrane in
mesial temporal lobe epilepsy (MTLE) versus non-MTLE hippocampi. ImmunoGold electron microscopy of the endothelial-astrocyte
interface in CA1 of a representative non-MTLE hippocampus demonstrates that AQP4 (arrows) is enriched on the astrocyte membrane
facing the endothelial cell. Considerably less labeling is present on the astrocyte membrane facing the neuropil (arrowheads). Inset:
Quantitation of gold particle densities in random fields from area CA1 of six non-MTLE (n-MTLE) and six MTLE hippocampi. Gold particle
counts for MTLE are given as percent of non-MTLE ⫾ standard error of the mean: astrocyte membranes facing the endothelial cell
(⫹cap; particles per ␮m), 56 ⫾ 16% (*p ⫽ 0.01); astrocyte membranes facing the neuropil (-cap; particles per ␮m), no change. The
number of gold particles per unit area of neuropil (particles per ␮m2) was 273 ⫾ 23% (**p ⫽ 0.013). A two-tailed Mann-Whitney U test
was used for statistical analysis. bm ⫽ basal lamina; ns ⫽ not significant. (Scale bar, 100 nm.) (From T. Eid et al. [2005] PNAS 102:
p. 1196).

Neurotherapeutics, Vol. 7, No. 4, 2010

 ASTROCYTES AND EPILEPSY 427

FIG. 3. Na⫹ channel densities and action potential (AP)-like responses in human astrocyte cultures. A: Peak currents (at ⫺40 to ⫺20
mV) were divided by membrane capacitance, and mean values plotted. n ⫽ number of cells recorded. *p ⬍ 0.01 as evaluated by
student’s t test. B: AP-like response to current injection in a hippocampal astrocyte from a patient with mesial temporal lobe epilepsy
(MTLE). Current injections of increasing value (0.6–2.0 nA) were applied to a cell with a resting potential of ⫺52 mV (holding potential
was ⫺75 mV), and the change in membrane potential was recorded. Bar indicates current injection (2 ms). (From E. R. O’Connor et al.
Epilepsia 1998;39:351). MaTLE ⫽ mass associated temporal lobe epilepsy; PTLE ⫽ paradoxical temporal lobe epilepsy.

icant finding of this study is the strong up-regulation of Astrocytes play a major role in K⫹ homeostasis in the
the a1C subunit on astrocytes throughout the sclerotic brain. They help move K⫹ away from regions of high
hippocampus compared to controls in which only occa- concentration to restore normal extracellular levels. Dur-
sional white matter astrocytes were stained. The a1C sub- ing seizure activity, the [K⫹]o significantly increases.
unit contributes to the L-type calcium currents, suggest- The inwardly rectifying potassium ion (Kir) channels on
ing the astrocytes in sclerotic hippocampi have a astrocytes play a major role in the removal of K⫹ from
significant change in their current characteristics, al- the extracellular space. Impaired K⫹ buffering in astro-
though the functional significance of this change is un- cytes in sclerotic hippocampi from patients with TLE
clear. It may suggest an enhanced astrocytic uptake of was first detected in studies of the effects of barium (a
Ca2⫹. blocker of Kir channels) on stimulus induced changes in

Neurotherapeutics, Vol. 7, No. 4, 2010

428 DE LANEROLLE ET AL.

[K⫹]o in the dentate gyrus and area CA1 using potassium pense of ␣-ketoglutarate from the tricarboxylic acid cy-
selective microelectrodes. Ba2⫹ augmented rises in cle, which could be replaced by anaplerosis in astro-
[K⫹]o in the dentate gyrus, but did not do so in the cytes.46 Astrocytes can use glutamate as a substrate for
sclerotic CA1 region37 suggesting defective Kir channels energy metabolism when extracellular glutamate levels
in the sclerotic region. Functional properties of astro- increase. Aspartate amino transferase (also called aspar-
cytes in acute hippocampal brain slices from patients tate transaminase) facilitates the conversion of aspartate
with and without hippocampal sclerosis who were stud- and ␣-ketoglutarate to oxaloacetate and glutamate and
ied with the patch-clamp technique provided more direct vice versa. The functional significance of its reduced
evidence of defective Kir channels.34,35,38 Hyperpolariz- levels in the sclerotic hippocampus is unclear.
ing voltages elicited inward rectifier currents that inacti- Lactate dehydrogenase (LDH) catalyzes the intercon-
vated at membrane potentials ⫺130 mV in astrocytes version of pyruvate and lactate with concomitant inter-
from nonsclerotic hippocampi, which were larger than conversion of nicotinamide adenine dinucleotide dehy-
those in the sclerotic tissue. Also the ratio of inward to drogenase (NADH) and NAD⫹. Lactate dehydrogenase
outward K⫹ conductances showed that they were signif- activity levels, normalized to citrate synthase activity
icantly smaller in astrocytes from the sclerotic hip- levels, are decreased in the sclerotic hippocampus.45
pocampi compared to those from the nonsclerotic ones.35 However, interictal extracellular levels of lactate are el-
The Kir4.1 subunit is involved in this defect.38 evated in the epileptogenic (sclerotic) hippocampus. Ex-
tracellular lactate levels measured by in vivo dialysis rise
rapidly and clear slowly in the epileptogenic hippocam-
ASTROCYTES SPECIFIC ENZYMES
pus during spontaneous seizures.47,48 Due to the predom-
Astrocytes play an important role in the metabolism of inance of astrocytes rather than neurons in the sclerotic
the brain through their unique degradation of both glu- hippocampus, they may be the source of such increased
cose and glutamate (reviewed39). The importance of the lactate, even though LDH levels are reduced. Two dis-
astrocyte in these processes lies in their possession of tinct subunits, LDH1 and LDH5, combine to form the
some key enzymes not normally found in neurons. isoenzyme types of LDH. Of these LDH5 is found ex-
In the sclerotic hippocampus, there is a marked loss of clusively in astrocytes.49 Whether these subunits are al-
glutamine synthetase immunoreactivity. This loss is par- tered in the sclerotic hippocampus is not yet known. The
ticularly in prominent regions of major neuronal loss and changes in pyruvate carboxylase and cytosolic malic en-
increased glial density, such as CA1 and CA3. Western zyme (two other enzymes exclusive to astrocytes in the
blot analysis confirmed these immunohistochemical sclerotic hippocampus) have not been investigated yet.
findings, whereas in vitro enzymology confirmed that
glutamine synthetase activity was reduced as well.40,41
ASTROCYTES GENE EXPRESSION
Glutamine synthetase catalyzes the conversion of gluta-
mate to glutamine in astrocytes in a process that uses Gene expression analysis studies have revealed the
ammonia. The addition of ammonia, as ammonium chlo- increased expression of several genes that may be asso-
ride to in vitro slices of the sclerotic hippocampus only ciated with astrocytes in the sclerotic hippocampus.50
minimally increased glutamine labeling, but significantly These included the characteristic astrocytic marker glial
increased labeling in slices of (normal) temporal neocor- fibrillary acidic protein (GFAP) as observed in other
tex from the same patients.42 Thus, astrocytes in the expression studies as well,51,52 along with vimentin des-
sclerotic hippocampus seem to have a reduced capacity moyokin (AHNAK), CD44 antigen (CD44), crystallin
for glutamine synthesis and ammonia detoxification. In- alpha B, calpain 2, (m/II) large subunit (CAPN2), cho-
deed, extracellular glutamine levels43 and cellular glu- ndroitin sulfate proteoglycan 2 (versican), paladin
tamine levels10,44 are reported reduced in the epilepto- (KIAA0992), moesin, presenilin 1, plectin 1, radixin
genic (sclerotic) hippocampus. (RDX), calcyclin (S100A6), neural (S100B), tenascin C
Malthankar-Phatak et al.45 have also demonstrated (TNC), titin, villin, and AQP4. The up-regulation of
that glutamate dehydrogenase (GDH) activity is in- astrocyte-related genes are consistent with the increased
creased while aspartate amino transferase activity is re- gliosis of the sclerotic hippocampus.50 Some of the pre-
duced in the sclerotic hippocampus when activity levels viously mentioned genes are involved in changing the
were normalized to the cortical levels.45 Although neu- morphology of the astrocytes.
ron specific isoforms are known, GDH is expressed pri- Another group of genes whose up-regulation is inter-
marily in astrocytes. GDH catalyzes the conversion of esting are those involved with immune and inflammatory
glutamate to a-ketoglutarate and in the process produces responses.50 These include those regulating chemokines
ammonia. Nicotinamide adenine dinucleotide (NAD⫹) and their receptors (chemokine C-C motif ligand 2
is a cofactor for this reaction. GDH can also synthesize [CCL2], CCL3, CCL4, chemokine C-X-C motif receptor
glutamate, thereby removing toxic ammonia at the ex- 4 [CXCR4]), cytokines, and their receptors (fibroblast

Neurotherapeutics, Vol. 7, No. 4, 2010

 ASTROCYTES AND EPILEPSY 429

growth factor-1) (FGF1), FGF2, FGF3, tumor necrosis and maintenance of tight junctions between endothelial
factor ligand superfamily member 7 (TNFSF7), signal cells, as well as the expression of endothelial transport
transduction protein (calmodulin-1 [CALM1], CALM3, molecules. They also play a role in the movement of
protein phosphatase 3, catalytic subunit alpha isoform water and other molecules between the blood and brain
(calcineurin A alpha), PPP3CA, PPP3R1, protein ty- parenchyma.58 It was reported more than 100 years ago
rosine phosphatase receptor type D (PTPRD), PTPRG, that there is a proliferation of the microvasculature in the
PTPRN, PTPRO, transcription factors (FK506 binding sclerotic hippocampus,59 an observation that has been
protein 1B, 12.6 kDa-FKBP1B, FKPB1A), transcription confirmed since then.60,61 More recently, it has been
factor-related genes (AGT, COL1A1, COL21A1, reported that the blood-brain barrier may become leaky
NCAM1, VCAM1, CD44, IL11RA, IL13RA, IL15), during seizures, resulting in the passage of substances
complement (C1QB, C3, C4), and class II major histo- from the blood to the brain. Immunohistochemical local-
compatibility complex antigen genes (HLA-DPA1, ization of albumin in resected hippocampi from TLE
HLA-DQA1, HLA-DRB1, and HLA-DRB3). Astrocytes patients revealed strong albumin immunoreactivity in the
are shown to contribute to the inflammatory response of parenchyma throughout the hippocampus next to blood
the central nervous system. In vivo and in vitro studies vessels. Neurons and astrocytes located around the ves-
have shown that astrocytes can produce a range of im- sels were also albumin positive. Such extravasations of
munologically relevant molecules, including class II ma- albumin are not observed in hippocampi of autopsy con-
jor histocompatibility complex antigens, many cyto- trols.62 Studies on experimentally induced status epilep-
kines, and chemokines, such as those seen in the sclerotic ticus in animals confirm that the vasculature became
hippocampus.53,54 Immunohistochemical analysis of permeable shortly after SE and continued into the latent
sclerotic hippocampi from TLE patients has revealed and chronic epileptic phase.62 Albumin released into the
increased expression of the NFkB-p65 subunit in astro- brain through vascular permeability is reported in animal
cytes.55 Several genes regulated through the NFkB path- studies to be taken up by astrocytes through transforming
way are among those up-regulated in the sclerotic hip- growth factor-b receptors (TGF-bR), and the transcrip-
pocampus (S100B, ezrin/radixin/moesin, the chemokines tional activation of downstream pathways63 resulting in
[CCL2, CCL3, CCL4, CXCL1, and the chemokine re- the down regulation of inward rectifying potassium (Kir
ceptor CXCR4]). 4.1) channels in astrocytes, astrocytic activation, in-
Immunohistochemical localization of interleukin creased inflammation, and reduced inhibitory transmis-
(IL)-1␤ and IL-1R in hippocampi of TLE patients with sion.64
hippocampal sclerosis reveals increases IL-1␤ and IL-1R Several changes in molecular expression have been
immunoreactivity in astrocytes in areas of prominent reported in the blood-brain barrier, the astrocyte en-
gliosis and neuronal loss (CA1–CA3 and hilus) with dothelial cell interface. The erythropoietin receptor
expression in perivascular end feet.56 No such immuno- (EPO-r) is strongly expressed on the capillaries of the
reactivity is observed in nonsclerotic hippocampi. Immu- sclerotic hippocampus,60 particularly in regions of ex-
nohistochemical localization of complement C1q, C3c, tensive neuronal loss and gliosis (CA3, CA1, and den-
C3d, and C5b–C9 in sclerotic and nonsclerotic hip- tate hilus). High resolution immunogold electron mi-
pocampi from TLE patients show no expression of com- croscopy revealed that the capillary EPO-r was
plement component proteins in nonsclerotic hippocampi. localized to the luminal and abluminal plasma mem-
However, in the sclerotic hippocampi immunoreactivity branes of endothelial cells to endosome-like structures
for complement component proteins was increased in of these cells and to pericapillary astrocytic end feet.60
regions of neuronal loss (CA1–CA3 and hilus). Strong The enrichment of EPO-r in these locations suggests a
C1q, C3c, and C3d immunoreactivity is observed in vi- highly efficient uptake of plasma EPO into the hip-
mentin positive astrocyte-like cells. Astrocytes are not pocampus. Such expression of EPO-r is not found in
stained for C5–C9 protein components.57 These immu- the nonsclerotic hippocampus.
nohistochemical observations have been confirmed by The multidrug resistance gene-1 (MDR1) encodes
quantitative polymerase chain reaction.57 for P-glycoprotein, an energy-dependent efflux pump
that exports planar hydrophobic molecules from the
cell. A ⬎ 10-fold ratio of multiple drug resistance
gene-1 mRNA is reported in 9 of 14 medically intrac-
ASTROCYTES AND VASCULAR CHANGES
table TLE patients. Immunohistochemistry for P-gly-
Astrocytes in the normal brain have a close association coprotein shows increased staining in the capillary
with the microvasculature. Astrocyte end feet wrap endothelial cells in hippocampi in all TLE patients
around the blood-brain barrier formed by the tight cou- with high mRNA ratios and in comparison to normal
pling of endothelial cells. Astrocyte end feet ensheathing autopsy controls and in astrocytes of 8 of 9 patients
blood vessels release signals that support the formation with high mRNA ratios.65

Neurotherapeutics, Vol. 7, No. 4, 2010

430 DE LANEROLLE ET AL.

FIG. 4. Photomicrographs of the localization in area CA1 pyramidal layer of non-sclerotic (non-mesial temporal lobe epilepsy
[non-MTLE]) (A, C, E, G, and I) and sclerotic (MTLE) (B, D, F, H, and J) hippocampi of some proteins whose genes showed changes
in expression in this study. A, B: Immunolocalization of Aquaporin 4 (AQP4). Note the reduction in immunoreactivity in astrocytic
foot processes around blood vessels in MTLE and increase in astrocytic processes throughout the pyramidal layer. C, D: Reduction
in Dystrophin (DMD) around blood vessels in MTLE. E, F: Localization of CD44 antigen. In the nonsclerotic hippocampus (E),
immunoreactivity is seen on fiber-like processes (arrow) extending into the pyramidal layer from the stratum oriens. Scattered
among these fibers are CD44 positive astrocyte cell bodies (not shown). In MTLE, many more astrocytes are immunoreactive and
form a dense network throughout the region with more intense staining in perivascular foot processes (arrow in F). G, H: Plectin
1 (pectin 1, intermediate filament binding protein 500 kDa) immunoreactivity, which is increased in expression in astrocyte cell
bodies and perivascular foot processes (arrow). I, J: Increase in CXCR4 (Chemokine [C-X-C motif] receptor 4) expression on
microglia in the sclerotic hippocampus (arrow). CXCR4 is increased on a small proportion of astrocytes as well. (From T.S. Lee et
al. Mol Med 2007;13:10.)

Several other molecules located on the perivascular sclerotic hippocampus. Brain microvessels from human
astrocyte end feet also show changes in the sclerotic hippocampi removed in epilepsy surgery express the che-
hippocampus; AQP4 is reduced along with dystrophin, mokine receptors CCR1 and CCR2 on the parenchymal
whereas CD44 and plectin 1 expression are increased50 surface of the microvessels.66 These receptor sites are also
(FIG. 4). found on human astrocytes.67 The ligands for these recep-
In addition, microarray analysis demonstrates that mol- tors CCL3 and CCL2 are up-regulated in the sclerotic hip-
ecules such as CCL2 and CCL3 are up-regulated in the pocampus.50 The binding of these chemokines is modulated

Neurotherapeutics, Vol. 7, No. 4, 2010

 ASTROCYTES AND EPILEPSY 431

in reponse to astrocyte stimulation by pro-inflammatory to hippocampal astrocytes, with both being stronger than
cytokines IL-1b and tumor necrosis factor-␣, indicating that cortical astrocytes. There were larger numbers of in-
these binding sites are subject to regulation.67 tercellular Ca2⫹ waves in the parahippocampal astro-
cytes cultures compared to hippocampal and cortical as-
trocytes.68 The same investigators carried out time lapse
ASTROCYTE PHYSIOLOGY
imaging of fluorescence recovery after laser bleach on
Membrane physiology astrocytes loaded with 6-carboxyfluorescein diacetate
Physiological studies of astrocytes from hippocampi succinimidyl ester, to assess functional gap junctional
surgically removed from patients with medically intrac- coupling in the cultures. Gap junction coupling was more
table epilepsy reveal several altered functional proper- pronounced in the parahippocampal (entorhinal cortex)
ties. Primary astrocyte cultures were established from the cultures from sclerotic patients, having a faster and more
hippocampus (dentate gyrus and CA1–CA3), parahip- complete fluorescence recovery after bleaching than as-
pocampus (entorhinal cortex) and temporal neocortex of trocytes from the less excitable temporal neocortex from
each patient operated for intractable TLE. One group of the same patient.68
patients had hippocampal sclerosis (MTLE) whereas two
others (mass associated temporal lobe epilepsy and par- Glutamate glutamine cycling
adoxical temporal lobe epilepsy) had no sclerosis.32 A In the normally functioning brain, glutamate released
large proportion (⬃60%) of astrocytes in primary cul- by neurons into the extracellular space is cleared by
tures derived from sclerotic epileptogenic hippocampal uptake into astrocytes and converted to glutamine. This
seizure foci are capable of generating action potential- glutamine is released by the astrocyte into the extracel-
like responses after depolarizing currents in comparison
lular space from where it is taken up by neurons and used
to astrocytes from nonsclerotic foci and neocortex from
to synthesize glutamate. This process is known as the
which such action potential-like responses could not be
glutamate-glutamine cycle.69 The glutamate-glutamine
elicited32 (FIG. 3). Such responses in these astrocytes
cycle during the interictal period is significantly reduced
from sclerotic foci may be facilitated by their signifi-
cantly depolarized resting membrane potentials (approx- in the sclerotic hippocampus, whereas it is normal in the
imately ⫺55 mV) compared to astrocytes from nonscle- nonsclerotic hippocampus.70 In vivo dialysis studies
rotic foci (approximately ⫺75 to ⫺80 mV) and higher show that glutamate release during a seizure is increased
tetrodotoxin sensitive Na⫹ channel density. Whole cell and its clearance prolonged, in the epileptogenic (scle-
patch clamp recordings from acute slices taken from rotic) hippocampus compared to the contralateral non-
similar patient tissue confirm the findings on cultured sclerotic one.18 It was subsequently shown that interictal
cells.34 extracellular glutamate levels are also higher in the epi-
leptogenic hippocampus compared to nonepileptogenic
Calcium release ones.43 The increased interictal basal extracellular gluta-
LEE et al68 carried out Ca2⫹ imaging using time- mate levels negatively correlate with hippocampal
lapsed confocal scanning laser microscopy in primary volume being significantly increased in the atrophic43
astrocyte cultures from hippocampus, parahippocampus, hippocampus compared to nonatrophic ones.71 The hip-
and temporal neocortex of patients with both sclerotic pocampal volume negatively correlates with seizure fre-
and nonsclerotic hippocampi, before and after the appli- quency.71 Magnetic resonance spectroscopic measure-
cation of glutamate. These studies showed that in astro- ments show that cellular glutamate levels are also lower
cytes from the hippocampal region of nonsclerotic hip-
in the atrophic (sclerotic) hippocampus compared to the
pocampi and temporal neocortex of both groups,
nonatrophic one.72 However, tissue glutamate levels
exposure to glutamate resulted in an initial peak of in-
measured by proton magnetic resonance spectroscopy of
tracellular Ca2⫹, with subsequent small oscillations and
perchloric acid tissue extracts show that the cellular glu-
elevations of Ca2⫹ representing intracellular and inter-
cellular Ca2⫹ waves.68 Astrocytes from hippocampus tamate content is increased above normal levels in the
and parahippocampus from patients with hippocampal epileptogenic human hippocampus.44 In this study, ele-
sclerosis responded to glutamate with a strongly elevated vated glutamate content is found in hippocampi from
intracellular Ca2⫹ response, which included an initial TLE patients with normal clinical MRI appearance, and
peak, followed by an increase in the number of intracel- even higher levels in almost half the sclerotic cases with
lular Ca2⫹ oscillations. These oscillations continue even greatest neuron loss.44 This suggests that in both groups
after glutamate is removed with superfusion of saline. in addition to neurons astrocytes may contain glutamate,
Often spontaneous calcium spikes were seen in parahip- and the release of this glutamate during a seizure con-
pocampal astrocytes prior to glutamate. In general, these tributes to the elevated extracellular glutamate levels
responses were stronger in parahippocampal compared measured.18,43

Neurotherapeutics, Vol. 7, No. 4, 2010

432 DE LANEROLLE ET AL.

ASTROCYTE TYPES have identified two classes of NG2 cells in the CNS
white matter of the rat. One class of cells expressed
Astrocytes are star-shaped cells that are normally
voltage gated Na⫹ currents, which are reversibly
characterized by the presence of GFAP. Three types of
blocked by TTX, along with voltage gated K⫹ currents,
astrocytes have been recognized based on their morpho-
whereas the other class of cells did not express INa cur-
logical features (i.e., radial astrocytes, fibrous astrocytes, rents. However, morphologically the two cell types are
and protoplasmic astrocytes).73 More recently, the coex- similar.79 In those NG2 cells expressing voltage gated
istence of distinct types of cells with the astrocyte spe- Na⫹ channels, current clamp recordings injecting depo-
cific marker GFAP, but with diverse morphological, mo- larizing currents evoked action potentials, which were
lecular, and functional profiles has been identified in the reversibly blocked by TTX. In a small number of cells
hippocampus. These were first identified in transgenic spontaneous action potentials are observed.79 Many cells
mice with GFAP-promoter controlled, enhanced green expressing Na⫹ channels receive synaptic inputs.79
fluorescent protein expression combined with patch In addition to Na⫹ and K⫹ channels, NG2 cells ex-
clamp recordings and single cell reverse transcription press voltage gated Ca2⫹ channels.80 Neuronal activity
studies.74 One group of cells was weekly fluorescent can depolarize NG2 cells allowing influx of Ca2⫹ into
with short thin processes. Outward K⫹ currents domi- the cell, with changes in intracellular Ca2⫹, which can
nate these cells, whereas the inward K⫹ currents (IKir) lead to changes in gene expression and glutamate release.
were much less pronounced. Their resting potentials (Vr) In the sclerotic hippocampus of TLE patients, several
were ⫺31 ⫾ 7 mV. Most of these cells have TTX- lines of evidence suggest the presence and accumulation
sensitive Na⫹ currents after depolarization beyond ⫺50 of NG2 cells. cDNA microarray studies show elevated
mV, but no action potentials were produced in the cur- CSPG expression in area CA1 of the sclerotic hippocam-
rent clamp mode. The second group of cells were more pus.50 In primary cultures derived from sclerotic hip-
intensely fluorescent and resembled protoplasmic astro- pocampi, we observed two morphological types of cells.
cytes having irregularly shaped somata bearing expanded One cell type was strongly immunopositive for GFAP
branched nets of processes. They have a more negative and possessed long fibrous processes, whereas the other
resting potential (Vr ⫽ ⫺70 mV). The inward potassium cell type was only weakly positive for GFAP and was
currents (IKir) were present in these cells.74 The weakly flatter and did not have fibrous processes (FIG. 5).
fluorescent cells express AMPA-type glutamate recep- In Ca2⫹ imaging studies with FURA 2, it was the flat
tors (GluRs), but not glutamate transporter currents, cells that showed increases and oscillations in intracel-
whereas the brightly fluorescent cells express glutamate lular Ca2⫹ levels, which dramatically increase on appli-
transporters (GluT) but lack GluR currents.74 These two cation of glutamate, especially in hippocampal and para-
types of enhanced green fluorescent protein-positive cells hippocampal cultures50 (and unpublished observations).
can be recognized also in acute slices from the hippocam- Our electrophysiological studies revealed that these as-
pus. These two types of astrocytes are nonoverlapping pop- trocytes such as NG2 cells had significantly more depo-
ulations. The GluR cells completely lack gap junction cou-
pling, whereas the GluT cells are extensively coupled.75
Electron microscopic studies have identified synapse-like
structures onto GluR cells and receive spontaneous synaptic
inputs from GABAergic and glutamatergic neurons in acute
hippocampal slice preparations.76
The GluR cell described previously also closely re-
sembles a cell type identified as a third class of macroglia
and variously named as neuron glia-2 (NG2) cells, oli-
godendrocyte precursor cells, polydendrocytes, synanto-
cyte, or complex cells.77 This star-shaped glial cell is
characterized by the expression of the protein chon-
droitin sulphate proteoglycan, and does not express
GFAP and does not have any glutamate transporters or
gap junctions.77 As with the GluR astrocytes, NG2 cells
express AMPA glutamate receptors, and additionally ex- FIG. 5. A photomicrograph of a primary culture of astrocytes,
made from a sclerotic hippocampus, immunostained for glial
press GABA receptors and receive synaptic terminals. fibrillary acidic protein (GFAP). The strongly GFAP positive fi-
The membrane properties of GluR cells and NG2 cells brous astrocytes (arrow) probably correspond to GluT type cells.
are indistinguishable.77 Immunostaining of GluR cells The weekly stained cell (arrowhead) probably corresponds to the
GluR type cells. These are the cells that show strong oscillations
show the presence of chondroitin sulfate proteoglycan in intracellular Ca2⫹, spontaneously and increasing with the
(CSPG)-NG2 protein.78 Most recently, Káradóttir et al.79 application of glutamate.

Neurotherapeutics, Vol. 7, No. 4, 2010

 ASTROCYTES AND EPILEPSY 433

larized resting membrane potential (⫺57 ⫾ 8.5 mV; studies report a series of reorganizational changes in the
range, ⫺37 to ⫺70 mV), increased Na⫹ channel densi- dentate gyrus that result in hyperexcitable dentate gran-
ties, and expressed large inward Na⫹ currents, which ular neurons.8 The subicular region of the sclerotic hip-
were blocked by the application of TTX.32 Action po- pocampus is remarkable in that it does not show signif-
tential like responses was obtained in approximately icant neuronal loss,84,85 although more subtle changes in
60% of hippocampal and parahippocampal astrocytes receptor expression and synaptic inputs may occur. This
from sclerotic hippocampi when depolarized by a series is quite different from other neurological disorders in-
of current steps of increasing amplitude.32 These obser- volving the hippocampus, such as Alzheimer’s disease,
vations strongly suggest the presence of NG2/GluR cells schizophrenia, and so forth, where the subiculum shows
in the sclerotic hippocampus. Seifert et al.14 provide significant neuronal injury. The entorhinal cortex is also
more direct evidence. They found that cells similar to reported to have neuronal loss with some gliosis, espe-
GluR and GluT are also present in the human hippocam- cially in layers three and to a lesser extent in layer two.86
pus. However, although both types are found in the non- Studies with depth electrodes in the hippocampus of TLE
sclerotic hippocampi from TLE patients, they report an patients show that seizure activity originates from such
almost complete loss of GluT cells in the CA1 region of hippocampi, and especially from the most sclerotic re-
the sclerotic hippocampus.35 Furthermore, it was the gions within them.87 How does this happen? How does
GluR cells in the sclerotic tissue but not in the nonscle- the hyperexcitability of the dentate granule cells88 spread
rotic hippocampus that had increased levels of the Flip to the subiculum, through the neuron-depleted area CA1?
isoform of GluR1.14 What triggers the injury of the hippocampus? The altered
function of astrocytes may provide answers to such ques-
WHAT ROLE DO ASTROCYTES PLAY IN tions.
THE HIPPOCAMPAL SEIZURE FOCUS?
Astrocytes may contribute to the high glutamate
Based on the information reviewed previously, astro- levels at the seizure focus
cytes in the sclerotic hippocampus could influence sei- Several lines of information point to this conclusion.
zure generation through several means. To place in per- The down regulation of the enzyme glutamine synthetase
spective on how some of these changes in astrocytes in a activity in the astrocytes, may account for the increase
sclerotic hippocampus may contribute to establishing an the extracellular glutamate levels.41 In addition, increase
epileptic focus, it is helpful to understand the organiza- in GDH may lead to increase in cellular glutamate ob-
tion of the hippocampus and its unique pathology in served.46 Evidence of intracellular Ca2⫹ release and
TLE. The principal source of inputs to the hippocampus Ca2⫹ oscillations in NG2-like cells in the sclerotic focus,
is from the entorhinal cortex, which is a part of the as argued for previously, along with evidence of the
parahippocampal gyrus. Efferents from the entorhinal up-regulation of synaptosome associated protein 23 ex-
cortex project into the hippocampus as the perforant path pression gathered from microarray analysis studies50
and synapse on the apical dendrites of the dentate gran- suggest the possibility that in the sclerotic hippocampus
ule cells, with some collaterals to area CA1 via a tem- are cells that may be capable of Ca2⫹-dependent exocy-
poro-amonic path. The axons of the granule cells (mossy totic release of glutamate.89 The intracellular Ca2⫹ re-
fiber axons) extend to synapse on CA3 neurons with lease pathway may be activated by the transcription fac-
collaterals to hilar neurons, most prominently the mossy tor NFkB, which is increased in the sclerotic
cells. Neurons in area CA3 project collaterals of their hippocampus.50,55 Experimental demonstration of the ac-
axons, known as the Schaffer collaterals, to area CA1 tivation of the NFkB activated cycloxygenase-2/pros-
and synapse on neurons there. The axons from CA1 toglandin 2 mediated Ca2⫹ release pathway in the human
neurons in turn project to the subiculum, with the sub- sclerotic hippocampus is still lacking. Another mecha-
iculum being the major source of efferents from the nism by which intracellular glutamate may be released is
hippocampus to other regions of the brain. The “tri syn- due to astrocyte swelling.90 Alterations in the distribu-
aptic pathway” is a well understood system of projec- tion of AQP4 on astrocytes in sclerotic hippocampi may
tions through the hippocampus, but there are other less result in accumulation of metabolic water within the
well understood anatomical interconnections between astrocyte as AQP4 transporters on the cell body are
the hippocampal regions.81 In the sclerotic hippocampus increased, but the release of water into the blood through
significant neuronal loss is reported in both the dentate the astrocyte end feet may be reduced due to reduced
gyrus and hippocampus proper (Ammon’s horn).9,82 expression of AQP4 on the perivascular astrocyte mem-
Area CA1 neurons are the most susceptible to injury and brane, thus resulting in astrocyte swelling.
was so noted more than a decade ago by Sommer.83 The Diffusion of glutamate, released into the extracellular
region is filled with reactive astrocytes, resulting in the space, from the Ammon’s horn region to the relatively
hardening or sclerotic nature of the hippocampus. Recent intact subicular region could trigger excitation of the

Neurotherapeutics, Vol. 7, No. 4, 2010

434 DE LANEROLLE ET AL.

neurons in the latter, and thus the spread of seizure Excitable astrocytes may contribute to the
activity out of the hippocampus resulting in behavioral spread of excitation through the hippocampus
seizures. As discussed previously, the sclerotic hippocampus
It is suggested that pathologic changes in the ento- may also be populated with a large number of NG2-like
rhinal cortex (EC) may underlie epileptogenesis in the cells. These cells in addition to contributing to Ca2⫹-
sclerotic hippocampus. Abnormal epileptiform activ- dependent glutamate release may also directly contribute
ity has been recorded from the EC region91,92 and is to the excitability of the seizure focus. NG2-like cells
reported to be the lead structure in the emergence of have been demonstrated in animal studies to be capable
tonic discharges in mesial structures in a majority of of being depolarized to yield action potentials, some-
patients with EC atrophy.93,94 Some studies have re- times even producing spontaneous action potentials.79
ported neuronal loss and gliosis in superficial layers of Cells capable of being depolarized to generate actions
the EC,86 whereas others find no significant difference potentials32 and GluR1 receptors with elevated flip to
in neuronal densities in the EC of patients with and flop ratios, which would facilitate and prolong depolar-
ization14 are demonstrated in the human hippocampus.
without hippocampal sclerosis.95 However, gliosis
Such cells may facilitate the generation or spread of
was a common finding in the EC of sclerotic and
waves of depolarization from granule cells to the subic-
nonsclerotic patients. The evidence for increased Ca2⫹
ulum without synaptic pathways between these two re-
release and Ca2⫹ oscillations in NG2-like cells from
gions. However, even in the most sclerotic of hip-
the EC (parahippocampus) in TLE patients,68 if asso- pocampi there is a intricate network of neurons in the
ciated with glutamate release, suggests a mechanism stratum oriens that strongly express the glutamate syn-
by which neurons in the entorhinal cortex may be thesizing enzyme phosphate activated glutaminase,97
excited and thus provide strong excitatory inputs to the along with other transmitter molecules such as neuropep-
hippocampus. Although such a mechanism has been tide Y, somatostatin, and GABA among other molecules.
postulated as the trigger for hippocampal seizure ac- It is interesting to speculate if this network of neurons,
tivity, how this occurred has remained a puzzle. The which runs along the CA1 region provides synaptic in-
activity of astrocytes may provide the answer, and put to NG2-like neurons to excite them.98 If so, a weak
merits further investigation in this context. wave of depolarization from surviving granule cells in
the sclerotic hippocampus could be transmitted along the
Astrocytes may contribute to increased stratum oriens neural network and “amplified” via the
extracellular Kⴙ in the seizure focus NG2 cell substrate into a stronger wave of depolarization
K⫹ buffering capacity is diminished in the sclerotic then excites neurons in the subiculum to generate sei-
hippocampus compared to a nonsclerotic one, most zures.
prominently in areas such as CA1.34 The impaired in- Alternatively, the NG2 cell action potentials may just
wardly rectifying K⫹ channels in sclerotic hippocampi serve as Na⫹ ion influx to increase astrocytic [Na⫹]I that
may be one contributory factor.38 The loss AQP4 from stimulates the activity of Na/K ATPase. An increase in
the astrocyte perivascular membrane may contribute to Na/K ATPase in astrocytes may play a role in buffering
this. The buffering of K⫹ by the inwardly rectifying Kir extracellular K⫹, compensating to some degree for the
channel depends on a parallel flux of water through the decreased Kir function in these astrocytes.
plasma membranes of these cells. K⫹ and water are taken
up by the astrocyte membrane facing the neuropil and Astrocytes modulate and are modulated by the
microvasculature
pushed into the blood and CSF through the end foot
Immunohistochemical and gene expression studies re-
membranes under conditions of high neuronal activity.96
viewed above demonstrate that a variety of molecules are
A loss of AQP4 from the perivascular membrane could
reorganized in the perivascular end feet on blood vessels.
impair water movements25 and lead to increased extra-
Some of these such as AQP4 and dystrophin are down
cellular concentrations of K⫹, which can depolarize neu- regulated and associated with water and K⫹ buffering.
rons in adjacent regions such as the subiculum. This Others, such as plectin 1, CD44, CXCR4, and erythro-
buffering of K⫹ is probably performed by the GFAP poietin receptor, are up regulated. Their functions are not
positive subset of GluT-like astrocytes. Although very clear. The monocyte chemoattractant protein-1 (MCP-1
preliminary data presented by Seifert et al.14 suggest that or CCL2) and macrophage chemoattractant protein-1a
such cells are absent in the sclerotic regions of the hip- (MIP-1␣ or CCL3) are up regulated in the sclerotic hip-
pocampus, they may be a reduced population rather than pocampus. Both these molecules are synthesized by as-
absent. The relative contribution of GluT-like cells to a trocytes and are expressed in astrocyte perivascular
human sclerotic hippocampal seizure focus needs further membranes that contact the abluminal surface of brain
analysis. microvessels.67 The receptors for these compounds

Neurotherapeutics, Vol. 7, No. 4, 2010

 ASTROCYTES AND EPILEPSY 435

FIG. 6. A diagram to represent the postulated mechanisms involving astrocyte-like cells in the human sclerotic hippocampus, showing
their altered functional states. (Downward small arrow ⫽ down regulated; long arrows indicate direction of flow.)

CCR2 and CCR1, respectively, have been identified on scribed by the transcription factor NFkB, and the result-
microvessels from epileptogenic brain tissue.66 It is be- ant molecules may be involved in several functions, such
lieved that these glial-derived chemokines guide circu- as changing astrocyte morphology (GFAP, vimentin,
lating leukocytes through endothelial junctions into un- Ezrin, Radixin, Moesin), triggering intracellular Ca2⫹
derlying brain tissue. They may also facilitate the release (S100B, CXCR4, COX2/PGE2 pathway), and
extravasation of molecules, such as albumin into the influencing the functioning of the microvasculature
brain through a leaky blood-brain barrier. Albumin re- (CCL2, CCL3, EPOR) and down regulating critical
leased into the brain parenchyma binds to TGF␤R2 re- membrane channels and transporters (Kir 4.1, AQP4).50
ceptors on astrocytes and modulates genes associated Components of the complement cascade increase vascu-
with the TGF␤ pathway,99 including the COX2/PGE2 lar permeability.100 The role in epilepsy of these factors
pathway for Ca2⫹ release and the down regulation of the associated with immune functions is still poorly under-
Kir channels.
stood, but may have a significant place in the process of
Astrocytes and immune and inflammatory factors epileptogenesis.
The association of inflammatory and immune markers FIG. 6 is a summary of the mechanisms involving
(chemokines, cytokines, class II major histocompatibility astrocytes that we speculate are operative in the human
complex markers and complement) in association with sclerotic hippocampal seizure focus. Astrocytes may
astrocytes, pose the question as to their role in epilepsy. play several roles in epilepsy, which include participat-
Factors such as IL-1␤ is shown to be able through bind- ing in the genesis of a seizure focus by structural alter-
ing to IL receptors on astrocytes activate them to produce ations to brain regions that become epileptogenic to
several of the factors observed in micro-array studies to mechanisms that maintain seizure activity through re-
be regulated in astrocytes.54 Many of these are tran- lease of glutamate and poor clearance of K⫹. The careful

Neurotherapeutics, Vol. 7, No. 4, 2010

436 DE LANEROLLE ET AL.

study of each of these roles of the astrocytes during the 18. During MJ, Spencer DD. Extracellular hippocampal glutamate
and spontaneous seizure in the conscious human brain. Lancet
process of epileptogenesis may offer insight for more 1993;341:1607–1610.
targeted pharmacotherapy for epilepsy. 19. Cavus I, Abi-Saab WM, Cassadey M, et al. Basal glutamate,
gamma-aminobutyric acid, glucose , and lactate levels in the
epileptogenic and non-epileptogenic brain site in neurosuergery
Acknowledgements: We thank our collaborators, especially patients. Epilepsia 2002;43:247.
Tore Eid, Jung Kim, Michael Brines Idil Cavus, Anne Wiliam- 20. Proper EA, Hioogland G, Kappen SM, et al. Distribution of
son, and Hitten Zaveri, who participated in the many studies on glutamate transporters in the hippocampus of patients with phar-
the pathophysiology of human temporal lobe epilepsy, which maco-resistant temporal lobe epilepsy. Brain 2002;125:32– 43.
have contributed to this review. We thank Dr. Ognen Petroff 21. Mathern GW, Mendoza D, Lozada A, et al. Hippocampal GABA
for many stimulating discussions, which have helped us greatly and glutamate transporter immunoreactivity in patients with tem-
in clarifying our own understanding of the role of astrocytes in poral lobe epilepsy. Neurology 1999;52:453– 472.
epilepsy. Ilona Kovacs has provided outstanding technical sup- 22. Tessler S, Danbolt NC, Faull RLM, Storm-Mathisen J, Emson P.
port for many years. The National Institutes of Health and the Expression of the glutamate transporters in human temporal lobe
epilepsy. Neuroscience 1998;88:1083–1091.
Lee Foundation supported our research.
23. Bjørnsen LP, Eid T, Holmseth S, Danbolt NC, Spencer DD, de
Lanerolle NC. Changes in glial glutamate transporters in human
REFERENCES epileptogenic hippocampus: inadequate explanation for high ex-
tracellular glutamate during seizures. Neurobiol Dis 2006;25:
1. Steinhäuser C, Haydon PG, de Lanerolle NC. Astroglial mecha- 319 –330.
nisms in epilepsy. In: Engel JJ, Pedley TA, eds. Epilepsy: a 24. Lee T-S, Bjornsen LP, Paz C, et al. GAT1 and GAT3 expression
comprehensive textbook. Philadelphia: Lipincott, Williams & are differently localized in the human epileptogenic hippocam-
Wilkins, 2008:277–288. pus. Acta Neuropathol 2006;111:351–363.
2. Penfield W, Humphreys S. Epileptogenic lesions of the brain. A 25. Eid T, Lee T-SW, Thomas MJ, et al. Loss of perivascular aqua-
histologic study. Arch Neurol Psychiatry 1940;43:240 –259. porin 4 inderlie deficient water and K⫹ homeostasis in the human
3. Foerster O, Penfield W. The structural basis of traumatic epilepsy epileptogenic hippocampus. Proc Natl Acad Sci U S A 2005;102:
and results of radical operations. Brain 1930;53:99 –119. 1193–1198.
4. Ward AA. Glia and epilepsy. In: Schoffeniels E, Frank G, Towers 26. Tse FW, Fraser DD, Duffy S, MacVicar BA. Voltage-activated
DB, Hertz L, eds. Dynamic PROPERTIES OF GLIA CELLS. K⫹ currents in acutely isolated hippocampal astrocytes. J Neu-
New York: Pergamon, 1977:413– 427. rosci 1992;12:1781–1788.
5. Pollen DA, Trachtenberg MC. Neuroglia: gliosis and focal epi- 27. Sontheimer H, Waxman SG. Expression of voltage-activated ion
lepsy. Science 1970;167:1252–1253. channels by astrocytes and oligodendrocytes in the hippocampal
6. Harris AB. Cortical neuroglia in experimental epilepsy. Exper slice. J Neurophysiol 1993;70:1863–1873.
Neurol 1975;49:691–715. 28. Sontheimer H, Ransom B, Cornell-Bell A, Black J, Waxman S.
7. Tiffany-Castiglioni E, Castiglioni AJJ. Astrocytes in epilepsy. In: Sodium current expression in rat hippocampal astrocytes in vitro:
Fedoroff S, Vernadarkis A, eds. Astrocytes: cell biology and Alterations during development. J Neurophysiol 1991;65:3–19.
pathology of astrocytes. New York: Academic Press, 1986:401– 29. Bevan B, Chiu S, Gray P, Ritchie J. The presence of voltage gated
424. sodium, potassium, and chloride channels in rat cultured astro-
8. de Lanerolle NC, Lee TS. New facets of the neuropathology and cytes. Proc R Soc Lond 1985;225:299 –313.
molecular profile of human temporal lobe epilepsy. Epilepsy Be-
30. Barres BA, Chun LLy, Corey DP. Ion channels in vertebrate glia.
hav 2005;7:190 –203.
Ann Rev Neurosci 1990;13:441– 474.
9. de Lanerolle NC, Kim JH, Williamson A, et al. A retrospective
31. Barres BA, Chun LLY, Corey DP. Ion channel expression by
analysis of hippocampal pathology in human temporal lobe epi-
white matter glia: I. type 2 astrocytes and oligodendrocytes. Glia
lepsy: evidence for distinctive patient subcategories. Epilepsia
1988;1:10 –30.
2003;44:677– 687.
10. Petroff OA, Errante LD, Kim JH, Spencer DD. N-acetyl-aspar- 32. O’Connor ER, Sontheimer H, Spencer DD, de Lanerolle NC.
tate, total creatine, and myo-inositol in the epileptogenic human Astrocytes from human hippocampal epileptogenic foci exhibit
hippocampus. Neurology 2003;60:1646 –1651. action potential-like responses. Epilepsia 1998;39:347–354.
11. Cohen-Gadol AA, Pan JW, Kim JH, Spencer DD, Hetherington 33. Bordey A, Spencer DD. Distinct electrophysiological alterations
HH. Mesial temporal lobe epilepsy: a proton magnetic resonance in dentate gyrus versus CA1glial cells from epileptic humans with
spectroscopy study and a histopathological analysis. J Neurosurg temporal lobe sclerosis. Epilepsy Res. 2004;59:107–122.
2004;101:613– 620. 34. Bordey A, Sontheimer H. Properties of human glial cells associ-
12. Verkhratsky A. Neurotransmitter receptors in astrocytes. In: Par- ated with epileptic tissue. Epilepsy Res 1998;32:286 –303.
pura V, Haydon PG, eds. Astrocytes in (patho)physiology of the 35. Hinterkeuser S, Schröder W, Hager G, et al. Astrocytes in the
nervous system: Springer Science, 2009. hippocampus of patients with temporal lobe epilepsy display
13. Seifert G, Schroder W, Hinterkeuser S, Schumacher T, Schramm changes in potassium conductances. Eur J Neurosci 2000;12:
J, Steinhauser C. Changes in flip/flop splicing of astroglial AMPA 2087–2096.
receptors in human temporal lobe epilepsy. Epilepsia 2002; 36. Djamshidian A, Grassl R, Seltenhammer M, et al. Altered ex-
43(Suppl. 5):162–167. pression of voltage-dependent calcium channel a1 subunits in
14. Seifert G, Hüttmann K, Schramm J, Steinhäuser C. Enhanced temporal lobe epilepsy with Ammon’s horn sclerosis. Neuro-
relative expression of glutamate receptor 1 flip AMPA receptor science 2002;111:57– 69.
subunits in hippocampal astrocytes of epilepsy patients with Am- 37. Gabriel S, Eilers A, Kivi A, et al. Effects of barium on stimulus
mon’s horn sclerosis. J Neurosci 2004;24:1996 –2003. induced changes in extracellular potassium concentration in area
15. Steinhäuser C, Seifert G. Glial membrane channels and receptors CA1 of hippocampal slices from normal and pilocarpine treated
in epilepsy: impact for generation and spread of seizure activity. rats. Neurosci Lett 1998;242:9 –12.
Eur J Pharmacol 2002;447:227–237. 38. Schroder W, Hinterkeuser S, Seifert G, et al. Functional and
16. Volterra A, Meldolesi J. Astrocytes, from brain glue to commu- molecular properties of human astrocytes in acute hippocampal
nication elements: the revolution continues. Nat Rev Neurosci slices obtained from patients with temporal lobe epilepsy. Epi-
2005;6:626 – 640. lepsia 2000;41:S181–184.
17. Fiacco TA, Agulhon C, Taves SR, et al. Selective stimulation of 39. Hertz L, Dringen R, Schousboe A, Robinson SR. Astrocytes:
astrocyte calcium in situ does not affect neuronal excitatory syn- glutamate producers for neurons. J Neurosci Res 1999;57:417–
aptic activity. Neuron 2007;54:611– 626. 428.

Neurotherapeutics, Vol. 7, No. 4, 2010

 ASTROCYTES AND EPILEPSY 437

40. van der Hel WS, Notenboom RG, Bos IW, van Rijen PC, van 63. Cacheaux LP, Ivens S, David Y, et al. Transcriptome profiling
Veelen CW, de Grann PN. Reduced glutamine synthetase in reveals TGF-b signalling involvement in epileptogenesis. J Neu-
hippocampal areas with neuron loss in temporal lobe epilepsy. rosci 2009;29:8927– 8935.
Neurology 2005;64:326 –333. 64. Ivens S, Kaufer D, Flores LP, et al. TGF-b receptor-mediated
41. Eid T, Thomas MJ, Spencer DD, et al. Loss of glutamine syn- albumin uptake into astrocytes is involved in neocortical epilep-
thetase in the human epileptogenic hippocampus: a possible togenesis. Brain 2007;130:535–547.
mechanism for elevated exttracellular glutamate in mesial tem- 65. Tishler DM, Weinberg KI, Hinton DR, Barbaro N, Annett GM,
poral lobe epilepsy. Lancet 2004;363:28 –37. Raffel C. MDR1 gene expression in brain of patients with med-
42. Eid T, Williamson A, Lee T-S, Petroff OA, de Lanerolle NC. ically intractable epilepsy. Epilepsia 1995;36:1– 6.
Glutamate and astrocytes: key players in human mesial temporal 66. Andjelkovic AV, Pachter JS. Characterization of binding sites for
lobe epilepsy. Epilepsia 2008;49 42–52. chemokines MCP-1 and MIP-1a on human brain microvessels.
43. Cavus I, Kasoff WS, Cassaday MP, et al. Extracellular metabo- J Neurochem 2000;75:1898 –1906.
lites in the cortex and hippocampus of epileptic patients. Ann 67. Andjelkovic AV, Kerkovich D, Shanley J, Pulliam L, Pachter JS.
Neurol 2005;57:226 –235. Expression of binding sites for b-chemokines on human astro-
44. Petroff OA, Errante LD, Rothman DL, Kim JH, Spencer DD. cytes. Glia 1999;28:225–235.
Neuronal and glial metabolite content of the epileptogenic human 68. Lee SH, Magge S, Spencer DD, Sontheimer H, Cornell-Bell A.
hippocampus. Ann Neurol 2002;52:635– 642. Human epileptic astrocytes exhibit increased gap junction cou-
45. Malthankar-Phatak GH, de Lanerolle NC, Eid T, et al. Differen- pling. Glia 1995;15:195–202.
tial glutamate dehydrogenase (GDH) activity profile in patients 69. Petroff OAC, Spencer DD. MRS studies of the role of altered
with temporal lobe epilepsy. Epilepsia 2006;47:1292–1299. glutamate and GABA neurotransmitter metabolism in the patho-
46. Petroff OAC. Metabolic biopsy of the brain. In: Waxman SG, ed. physiologyof epilepsy. In: Shulman RG, Rothman DL, eds. Brain
Molecular neurology. New York: Elsevier, 2007:77–100. energetics and neuronal activity:Applications to fMRI and med-
47. During MJ, Itzhak F, Leone P, Katz A, Spencer DD. Direct icine. New York: Wiley, 2004.
measurement of extracellular lactate in the human hippocampus 70. Petroff OA, Errante LD, Rothman DL, Kim JH, Spencer DD.
during spontaneous seizures. J. Neurochem. 1994;62:2356 –2361. Glutamate-glutamine cycling in the epileptic human hippocam-
48. Cendes F, Stanley JA, Dubeau F, Andermann F, Arnold DL. pus. Epilepsia 2002;43:703–710.
Proton magnetic resonancenspectroscopic imaging for discrimi- 71. Cavus I, Pan JW, Hetherington HP, et al. Decreased hippocampal
nation of absence and complex partial seizures. Ann Neurol 1997; volume on MRI is associated with increased extracellular gluta-
41:74 – 81. mate in epilepsy patients. Epilepsia 2008;49:2358-1366.
49. Bittar PG, Charnay Y, Pellerin L, Bouras C, Magistretti PJ. 72. Pan JW, Venkatraman T, Vives KP, Spencer DD. Quantitative
Selective distribution of lactate dehydrogenase isoenzymes in glutamate spectroscopic imaging of the human hippocampus.
neurons and astrocytes of human brain. J Cereb Blood Flow NMR Biomed 2006;19:209 –216.
Metab 1996;16:1079 –1089. 73. Privat A, Gimenez-Ribotta M, Ridet J-L. Morphology of astro-
50. Lee T-S, Mane S, Eid T, et al. Gene expression in temporal lobe cytes. In: Kettenmann H, Ransom BR, eds. Neuroglia. New York:
epilepsy is consistent with increased release of glutamate by Oxford University Press, 1995:3–22.
astrocytes. Mol Med 2007;13:1–13. 74. Matthias K, Kirchhoff F, Seifert G, et al. Segregated expression
51. Özbas-Gerceker F, Redeker S, Boer K, et al. Serial analysis of of AMPA-type glutamate receptors and glutamate transporters
gene expression in the hippocampus of patients with mesial tem- defines distinct astrocyte populations in the mouse hippocampus.
poral lobe epilepsy. Neuroscience 2006;138:457– 474. J Neurosci 2003;23:1750 –1758.
52. Becker AJ, Chen J, Paus S, et al. Transcriptional profiling in 75. Wallraff A, Odermatt B, Willecke K, Steinhäuser C. Distinct
human epilepsy: expression array and single cell real-time qRT- types of astroglial cells in the hippocampus differ in gap junction
PCR analysis reveal distinct cellular gene regulation. NeuroReport coupling. Glia 2004;48:36 – 43.
2002;13:1327–1333. 76. Jabs R, Seifert G, Steinhäuser C. Astrocytic function and its
53. Dong Y, Benveniste EN. Immune function of astrocytes. Glia alteration in the epileptic brain. Epilepsia 2008;49:3–12.
2001;36:180 –190. 77. Paukert M, Bergles DE. Synaptic communication between neu-
54. John GR, Lee SC, Song X, Rivieccio M, Brosnan CF. IL-1- rons and NG2 cells Curr Opinion in Neurobiol 2006;16:515-521.
Regulated responses in astrocytes: relevance to injury and recov- 78. Schools GP, Zhou M, Kimmelberg HK. Electrophysiolohically
ery. Glia 2005;49:161–176. “complex” glial cells fresshly isolated from the hippocampus are
55. Crespel A, Coubes P, Rousset M-C, et al. Inflammatory reactions immunopositive for chondroitin sulphate proteoglycan NG2.
in human medial temporal lobe epilepsy with hippocampal scle- J Neurosci Res 2003;73:765–777.
rosis. Brain Res 2002;952:159 –169. 79. Káradóttir R, Hamilton NB, Bakiri Y, Attwell D. Spiking and
56. Ravizza T, Gagliardi B, Noé F, Boer K, Aronica E, Vezzani A. non-spiking classes of oligodendrocyte precursor glia in CNs
Innate and adaptive immunity during epileptogenesis and spon- white matter. Nature Neurosci 2008;11:450 – 456.
taneous seizures: Evidence from experimental models and human 80. Akiopian G, Kressin K, Derouiche A, Steinhäuser C. Identified
temporal lobe epilepsy. Neurobiol Dis 2008;29:142–160. glial cells in the early postnatal mouse hippocampus display
57. Aronica E, Boer K, van Vilet EA, et al. Complement activation in different types of Ca2⫹ currents. Glia 1996;17:181–194.
experimental and human temporal lobe epilepsy. Neurobiol Dis 81. Witter MP, Groenewegen HJ, Lopes da Silva FH, Lohman AHM.
2007;26:497–511. Functional organization of the extrinsic and intrinsic circuitry of
58. Abbott NJ. Astrocyte-endothelial interactions and blood-brain the parahippocampal region. Prog Neurobiol 1989;33:161–253.
barrier permiability. J Anat 2002;200:629 – 638. 82. Babb TL, Brown WJ, Pretorius J, Davenport C, Lieb JP, Crandall
59. Bratz E. Ammonshornbefunde bei Epileptikern. Arch. Psychiatr PH. Temporal lobe volumetric cell densities in temporal lobe
Nervenkr 1899;32:820 – 835. epilepsy. Epilepsia 1984;25:729 –740.
60. Eid T, Brines M, Cerami A, et al. Increased expression of eryth- 83. Sommer W. Erkrankung des Ammonshorns als aetiologisches
ropoietin receptor on blood vessels in the human epileptogenic Moment der Epilepsie. Arch Psychiatr Nervenkr 1880;10:631–
hippocampus with sclerosis. J Neuropathol Exptl Neurol 2004; 675.
63:73– 83. 84. Fisher PD, Sperber EF, Moshe SL. Hippocampal sclerosis revis-
61. Rigau V, Morin M, Rousset M-C, et al. Angiogenesis is associ- ited. Brain Dev 1998;20:563.
ated with blood-brain barrier permeability in temporal lobe epi- 85. Kim JH, Guimaraes PO, Shen M-Y, Masukawa LM, Spencer DD.
lepsy. Brain 2007;130:1942–1956. Hippocampal neuronal density in temporal lobe epilepsy with and
62. Van Vliet EA, da Costa Araújo S, Redeker S, van Schaik R, without gliomas. Acta Neuropathol 1990;80:41– 45.
Aronica E. Blood-brain barrier leakage may lead to progression 86. Du F, Eid T, Lothman EW, Kohler C, Schwarcz R. Preferential
of temporal lobe epilepsy. Brain 2007;130:521–534. neuronal loss in layer III of the medial entorhinal cortex in rat

Neurotherapeutics, Vol. 7, No. 4, 2010

438 DE LANEROLLE ET AL.

models of temporal lobe epilepsy. J Neurosci 1995;15: 94. Bartolomei F, Khalil M, Wendling F, et al. Entorhinal cortex
6301– 6313. involvement in human mesial temporal lobe epilepsy: an elec-
87. Babb TL, Lieb JP, Brown WJ, Pretorius J, Crandall PH. Distri- trophysiologic and volumetric study. Epilepsia 2005;46:677–
bution of pyramidal cell density and hyperexcitability in the 687.
epileptic human hippocampal formation. Epilepsia 1984;25: 95. Dawodu S, Thom M. Quantitative neuropathology of the ento-
721–728. rhinal cortex region in patients with hippocampal sclerosis and
88. Williamson A, Spencer SS, Spencer DD. Depth electrode studies temporal lobe epilepsy. Epilepsia 2005;46:23–30.
and intracellular dentate granule cell recordings in temporal lobe 96. Paulson OB, Newman EA. Does the release of potassium from
epilepsy. Ann Neurol 1995;38:778 –787. astrocyte endfeet regulate cerebral blood flow. Science 1987;237:
89. Malarkey EB, Parpura V. Mechanisms of glutamate release from 896 – 898.
astrocytes. Neurochem International 2008;52:142–154.
97. Eid T, Hammer J, Runden-Pran E, et al. Increased expression of
90. Kimelberg HK, Mongin AA. Swelling-activated release of exci-
phosphate activated glutaminase in hippocampal neurons in hu-
tatory amino acids in the brain: relevance for pathophysiology.
Contrib Neprhol 1998;123:240 –257. man mesialtemporal lobe epilepsy. Acta Neuropathol 2007;113:
91. Bragin A, Wilson CL, Almajano J, Mody I, Engel J. High-fre- 137–152.
quency oscillations after status epilepticus: epileptogenesis and 98. Jabs R, Pivneva T, Hüttmann K, et al. Synaptic transmission onto
seizure genesis. Epilepsia 2004;45:1017–1023. hippocampal glial cells with hGFAP promoter activity. J Cell Sci
92. De Guzman P, D’Antuono M, Avoli M. Initiation of electrtoen- 2005;118:3791–3803.
cephalographic seizures by neural networks in entorhinal and 99. Friedman A, Kaufer D, Heinemann U. Blood-brain barrier break-
perirhinal cortices in vitro. Neurosci 2004;123:875– 886. down-inducing astrocytic transformation: Novel targets for the
93. Bar-Peled O, Ben-Hur H, Biegon A, et al. Distribution of gluta- prevention of epilepsy. Epilepsy Res 2009;85:142–149.
mate transporter subtypes during human brain development. 100. Lucas SM, Rothwell NJ, Gibson RM. The role of inflammation in
J Neurochem 1997;69:2571–2580. CNS injury and disease. Br J Pharmacol 2006;147:232–240.

Neurotherapeutics, Vol. 7, No. 4, 2010