David A. Sun, Sompong Sombati and Robert J. DeLorenzoof Stroke-Induced ''Epilepsy''

Induced Epileptogenesis in Hippocampal Neurons: An In Vitro Model−Glutamate Injury

Print ISSN: 0039-2499. Online ISSN: 1524-4628 Copyright © 2001 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Stroke doi: 10.1161/hs1001.097242

2001;32:2344-2350Stroke. 

http://stroke.ahajournals.org/content/32/10/2344World Wide Web at:

The online version of this article, along with updated information and services, is located on the

  http://stroke.ahajournals.org//subscriptions/

is online at: Stroke Information about subscribing to Subscriptions: 

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

  document. Permissions and Rights Question and Answer process is available in the

Request Permissions in the middle column of the Web page under Services. Further information about thisOnce the online version of the published article for which permission is being requested is located, click

can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.Strokein Requests for permissions to reproduce figures, tables, or portions of articles originally publishedPermissions:

by guest on February 20, 2014http://stroke.ahajournals.org/Downloaded from by guest on February 20, 2014http://stroke.ahajournals.org/Downloaded from

Glutamate Injury–Induced Epileptogenesisin Hippocampal Neurons

An In Vitro Model of Stroke-Induced “Epilepsy”

David A. Sun, BSE; Sompong Sombati, PhD; Robert J. DeLorenzo, MD, PhD, MPH

Background and Purpose—Stroke is the major cause of acquired epilepsy. The mechanisms of ischemia-inducedepileptogenesis are not understood, but glutamate is associated with both ischemia-induced injury and epileptogenesisin several models. The objective of this study was to develop an in vitro model of epileptogenesis induced by glutamateinjury in hippocampal neurons as observed during stroke.

Methods—Primary hippocampal cultures were exposed to 5mmol/L glutamate for various durations. Whole-cell currentclamp electrophysiology was used to monitor the acute effects of glutamate on neurons and chronic alterations inneuronal excitability up to 8 days after glutamate exposure.

Results—A single, 30-minute, 5-mmol/L glutamate exposure produced a subset of neurons that died and a larger populationof injured neurons that survived. Neuronal injury was characterized by prolonged reversible membrane depolarization,loss of synaptic activity, and neuronal swelling. Surviving neurons manifested spontaneous, recurrent, epileptiformdischarges in neural networks characterized by paroxysmal depolarizing shifts and high-frequency spike firing thatpersisted for the life of the culture.

Conclusions—This study demonstrates that glutamate injury produced a permanent epileptiform phenotype expressed asspontaneous, recurrent epileptiform discharges for the life of the hippocampal neuronal culture. These results suggesta novel in vitro model of glutamate injury–induced epileptogenesis that may help elucidate some of the mechanisms thatunderlie stroke-induced epilepsy.(Stroke. 2001;32:2344-2350.)

Key Words: epilepsyn excitotoxicity n glutamatesn stroke

Epilepsy is one of the most common neurological disor-ders, affecting an estimated 40 to 50 million people

worldwide.1 Approximately 30% to 50% of all epilepsy caseshave a known cause and are termed acquired epilepsy.2

Cerebral ischemia, or stroke, is the most common cause ofacquired epilepsy, accounting for'40% of these cases.3

Despite the important role of stroke in the development ofepilepsy, little is known concerning the mechanisms by whichan ischemic insult produces epilepsy.

Advances in the study of stroke, however, have demon-strated pathological events that may help elucidate the mech-anisms underlying ischemia induced epilepsy. It has beenshown that ischemia and anoxia during a stroke lead tomassive release of the excitatory amino acid neurotransmitterglutamate,4,5 causing excessive activation of postsynapticglutamate receptors believed to be the major cause of neuro-nal injury in stroke.6,7 In the core of a stroke, the majority ofneurons undergo an irreversible membrane depolarization,the anoxic depolarization,8 and die forming an infarct.9 In theperi-infarct penumbra, the injury is less severe and produces

a mixed population of neurons that undergo either irreversibleanoxic depolarization and die or transient, reversible depo-larization, the peri-infarct ischemic depolarization,10 andsurvive.9 Neurons that survive in the penumbra are theunderlying substrates for ischemia-induced epileptogenesis.

As with stroke, glutamate receptor activation has beenstrongly associated with epileptogenesis.11–17 Therefore, wehypothesized that a less severe glutamate insult that producesprolonged, reversible neuronal depolarization like the gluta-mate injury in the penumbra could induce epileptogenesis insurviving neurons. To develop an in vitro model of stroke-induced epilepsy, this study was initiated to determine if anexcitotoxic glutamate injury that produces prolonged, revers-ible depolarization could cause spontaneous, recurrent, epi-leptiform discharges (SREDs) that last for the life of hip-pocampal neurons in culture.

Materials and MethodsUnless otherwise noted, reagents were purchased from Sigma Chem-ical Co. Sodium pyruvate, minimum essential media containing

Received November 28, 2000; final revision received July 10, 2001; accepted July 10, 2001.From the Departments of Pharmacology and Toxicology (D.A.S., R.J.D.), Neurology (S.S., R.J.D.), and Biochemistry and Molecular Biophysics (R.J.D.) and

the Graduate Program in Neuroscience (D.A.S., R.J.D.), Medical College of Virginia, Virginia Commonwealth University, Richmond, Virginia.Correspondence to Robert J. DeLorenzo, MD, PhD, MPH, Medical College of Virginia, Virginia Commonwealth University, Box 980599, Richmond,

VA 23298. E-mail rdeloren@hsc.vcu.edu© 2001 American Heart Association, Inc.

Stroke is available at http://www.strokeaha.org

2344 by guest on February 20, 2014http://stroke.ahajournals.org/Downloaded from

Earle’s salts, fetal bovine serum, and horse serum were obtainedfrom Gibco-BRL.

Hippocampal Cell CulturePrimary hippocampal cultures were prepared as described andcharacterized previously by our laboratory18 and conformed toinstitutional guidelines. Briefly, hippocampal cells were preparedfrom 2-day postnatal Sprague-Dawley rats (Harlan) and plated onglial support layers at a density of 13105 cells/mL. Cultures weremaintained at 37°C in a 5% CO2/95% air atmosphere and fed twiceweekly with neuronal feed. Cultures were used for experiments from13 days in vitro through the life of the cultures (21 days). Glutamateconcentrations in neuronal feed before culture feeding and at time ofwashout were analyzed by enzymatic fluorometric assay with aCMA 600 Microdialysis Analyser (CMA/Microdialysis) and were,1 mmol/L.

Glutamate InjuryAt 13 days in vitro, neuronal media was removed, retained at 37°C,and replaced by a physiological treatment solution (145 mmol/LNaCl, 2.5 mmol/L KCl, 10 mmol/L N-[2-Hydroethyl]piperazine-N9[2-ethanesulfonic acid] (HEPES), 10 mmol/L glucose, 2 mmol/LCaCl2, 1 mmol/L MgCl2, 2 mmol/L glycine, pH 7.3, osmolarityadjusted to 325 mOsm with sucrose) in sham control cultures.Glutamate treatment was performed by supplementing this recordingsolution with 5mmol/L glutamate. All exposures were performed at37°C in the culture incubator. Glutamate exposure was terminated by3 washes with recording solution and return of the original culturemedia.

Neuronal Death AssayNeuronal death was assessed 24 hours after glutamate exposureusing fluorescein diacetate (FDA)–propidium iodide (PI) microflu-orometry.19 Neurons labeled with FDA or PI were quantified bymeans of the Ultraview image analysis software package (PerkinElmer Life Sciences), and percent neuronal death was calculated asthe number of neurons labeled by PI divided by the sum of thenumber of neurons labeled by PI and those labeled by FDA.Fluorescent images were compared with phase bright images toconfirm that only pyramid-shaped neurons were counted. Threerandomly selected fields were counted and averaged per culture(approximately 18 to 25 neurons per field).

Measurements of Neuronal InjuryNeuronal swelling was assessed before, during, and after exposure to5 mmol/L glutamate with FDA. Images of FDA-stained neuronswere captured as described above. The area of the neuronal somawas calculated with the use of the Ultraview image analysis softwarepackage. Changes in membrane potential and input resistance weremeasured as described below.

ElectrophysiologyWhole-cell current-clamp recordings were performed on pyramidal-shaped neurons by methods previously described in our laboratory.18

Cell culture media was replaced with a physiological recordingsolution (145 mmol/L NaCl, 2.5 mmol/L KCl, 10 mmol/L HEPES,10 mmol/L glucose, 2 mmol/L CaCl2, 1 mmol/L MgCl2, 0.5mmol/Lglycine, pH 7.3, osmolarity adjusted to 325 mOsm with sucrose).Patch microelectrodes of 3 to 7 MVresistance were pulled fromborosilicate glass capillaries (World Precision Instruments, Inc) witha Brown-Flaming P-80C electrode puller (Sutter Instruments) andfilled with an internal solution of 140 mmol/L K1 gluconate,1 mmol/L MgCl2, 10 mmol/L HEPES, 1.1 mmol/L Ethylene glycol-bis (b-aminoethyl ether)-N, N,N9,N9-tetraacetic acid (EGTA),4 mmol/L Na2 ATP, 15 mmol/L Tris Phosphocreatine, pH 7.2,osmolarity adjusted to 310 mOsm with sucrose. Recordings wereobtained by means of an Axopatch 200A amplifier or anAxoclamp-2A amplifier (Axon Instruments) in current clamp mode.Data were digitized and stored on videotape with a Neuro-corderDR-890 (Neurodata instruments Corp) and Sony VCR. In some

experiments, hyperpolarizing square pulses (20.5 nA, 200 ms) wereinjected every 5 seconds in the presence of 1mmol/L tetrodotoxin tocalculate input resistance. Data were later played back for analysis toa Dash IV chart recorder (Astro-Med Inc) or to a computer via aDigidata 1200 (Axon Instruments) and Strathclyde Electrophysiol-ogy Software (John Dempster, University of Strathclyde). Strath-clyde Electrophysiology Software in event detector mode was usedto calculate histograms of instantaneous spike frequency as definedas the inverse of the interval between 2 successive events.

SREDs or electrographic seizures were defined as bursts of spikefiring at a frequency of$3 Hz for durations of$20 seconds,analogous to electrographic seizures observed with EEG record-ings.20 Neurons were categorized as “epileptic” on manifestation of2 or more SREDs.

Statistical AnalysisData are reported as mean6SEM. Student’st test, 1-way ANOVA,or repeated-measures 1-way ANOVA and post hoc Tukey test wereapplied when appropriate, with SigmaStat 2.0 (Jandel Corp). Forcategoric data (percent of “epileptic” neurons),x2 tests were per-formed. A value ofP,0.05 was considered statistically significantfor all data analysis.

ResultsGlutamate Exposure Produces HippocampalNeuronal InjuryTo determine if neurons exposed to 5mmol/L glutamateunderwent irreversible or reversible membrane depolariza-tion, we performed whole-cell current-clamp experiments onneurons before, during, and after exposure to 5mmol/Lglutamate for 30 minutes. During glutamate exposure, neu-rons began to depolarize and spike briefly before succumbingto a larger prolonged depolarization (Figure 1A). This gluta-mate treatment did not produce continuous spiking butinduced a membrane depolarization analogous to the depo-larizations produced during ischemic or anoxic brain inju-ry.8,10 This depolarization was also associated with a loss ofsynaptic potentials (Figure 1A). Figure 1B provides a quan-titative analysis of the reversible depolarization. A smallergroup of neurons (16%, 3 of 19 neurons) did not return tobaseline potentials and manifested extended neuronal depo-larization.21 In the neurons undergoing reversible depolariza-tion (Figure 1, A and B), glutamate induced a significant andreversible decrease in membrane input resistance quantifiedin Figure 1C.

In addition, we evaluated neuronal swelling (Figure 1, Dthrough F) as a measure of reversible injury by directexamination of fluorescein diacetate–stained pyramidal-shaped neurons (see Materials and Methods). Neuronsswelled significantly, as assessed by increased somatic area(3768% compared with baseline area, n56 neurons,P,0.05). On washout of the glutamate, neuronal swellingdecreased to 1468% of baseline (n56 neurons) within hours,not significantly different from preexposure values.

Glutamate exposure produced a mixed population of in-jured and dead neurons. Using the FDA/PI technique,19 weassessed the excitotoxicity of a 5mmol/L glutamate exposureof increasing duration in hippocampal neuronal cultures 24hours after treatment (Figure 2A). Glutamate exposure of 5minutes resulted in 2764% (n58 cultures) neuronal death,not significantly different from sham controls (1862%, n513cultures). Percent neuronal death increased significantly for

Sun et al Glutamate Injury–Induced Epileptogenesis 2345

by guest on February 20, 2014http://stroke.ahajournals.org/Downloaded from

durations of 30 minutes (4764%, n510 cultures,P,0.05)and 90 minutes (6462%, n58 cultures,P,0.05). Over thecourse of 8 days after glutamate exposure, further neuronaldeath in experimental groups was not different from controls.Therefore, we were able to manipulate our excitotoxic con-ditions such that a small population of neurons died and alarger population of neurons recovered in response to gluta-mate exposure similar to the environment of the ischemicpenumbra.

Glutamate Induced Alterations inNeuronal ExcitabilityBecause of the association between stroke and epilepsy,3 wehypothesized that neurons surviving the 5-mmol/L glutamateinjury would display long-term changes in excitability. Totest this hypothesis, we obtained current-clamp recordingsfrom sham control neurons and neurons 1 to 8 days afterexposure to glutamate. In the 30-minute treatment group,SREDs were observed in 8664% of neurons per culture (n517 cultures). “Epileptic” neurons were not observed in thesham control group and were observed significantly more

often in the 30-minute treatment group than in the 5-minuteor 90-minute groups (Figure 2B). The percentage of “epilep-tic” neurons ranged from 77% to 100% over the course of 1to 8 days after exposure and was not significantly differentfrom day to day (Figure 2C). SREDs were not observedimmediately after glutamate exposure (Figure 1A) and werefully developed after a latency period of 24 hours after theinjury.

A typical recording from a sham control neuron is shownin Figure 3A. Control neurons demonstrated spontaneousaction potentials (spikes), excitatory postsynaptic potentials(EPSPs), and inhibitory postsynaptic potentials (IPSPs) typ-ical of normal synaptic activity. Quantitative analysis of spikedischarges in control neurons (n512 neurons, over 4 hours of

Figure 1. Application of 5 mmol/L glutamate caused hippocam-pal neuronal injury characterized by depolarization of membranepotential, reduction in membrane input resistance, and somaticswelling. A, Current-clamp recording of a neuron before, during,and after glutamate application of 30 minutes (representative of12 experiments). In presence of glutamate (black bar), this neu-ron depolarized from 262 mV to 217 mV and synaptic activityceased. On washout, the neuron repolarized to 247 mV andEPSPs returned. B, Neuronal membrane potential before(261.761.4 mV, n512 neurons), during (GLU, 26.962.9 mV,n512), and after (258.463.4 mV, n519 neurons) glutamateapplication (*P,0.05, repeated-measures ANOVA and post hocTukey test). C, Neuronal input resistance before (137.6617.4MV, n510 neurons), during (GLU, 37.265.0 MV, n510 neu-rons), and after (117.8611.2 MV, n510 neurons) glutamateapplication (*P,0.05, repeated-measures ANOVA and Tukeytest). D, Representative FDA-stained neuron before exposure to5 mmol/L glutamate for 30 minutes. E, Same FDA-stained neu-ron in presence of glutamate swelled, increasing in somatic areaby 31%. F, Same FDA-stained neuron within an hour of wash-out restored preexposure morphology, only 4% greater in areacompared with D.

Figure 2. Glutamate induced neuronal death and chronic epilep-tiform activity in cultured hippocampal neurons. A, Increasingglutamate exposure durations caused increasing neuronal death24 hours after injury induced by 5 mmol/L glutamate. Exposuredurations of 30 minutes and 90 minutes induced significantlymore neuronal death per culture than sham controls and5-minute exposure (mean6SEM, *P,0.05, ANOVA and post hocTukey test). B, Thirty-minute glutamate exposure induced epi-leptiform activity 24 hours after glutamate injury; 30-minutetreatment group (8664%, n517 cultures) manifested signifi-cantly more “epileptic” neurons per culture than sham control(060%, n58 cultures), 5-minute treatment (17611%, n55 cul-tures), and 90-minute treatment groups (22611%, n53 cultures,*P,0.05, ANOVA and Tukey’s test). C, Epileptiform activityinduced by 30-minute, 5-mmol/L glutamate exposure persistedfor the life of the culture. One through 8 days after glutamatetreatment 77% to 100% of neurons demonstrated epileptiformactivity. Percentage of neurons manifesting SREDs was signifi-cantly different from 0-day control (before exposure, *P,0.01, x2

test, n515 neurons). No statistical differences were observed inpercentage of “epileptic” neurons 1 through 8 days (n532, 48,12, 7, 2, 17, 18, 9 neurons, respectively) after glutamate treat-ment, demonstrating persistence of SREDs throughout the lifeof the culture (P50.841, x2 test).

2346 Stroke October 2001

by guest on February 20, 2014http://stroke.ahajournals.org/Downloaded from

recording) revealed that the great majority of spike firings(88%) occurred in a frequency range,3 Hz (Figure 4A).Occasionally, control neurons manifested spike discharges of.3 Hz. These bursts, however, never exceeded 10 seconds induration (Figure 4C). Therefore, control neurons never man-ifested SREDs (see Materials and Methods).

Neurons treated with glutamate manifested SREDs (Figure3B). The developing depolarizations (Figure 3C) startedabruptly and were typical of the paroxysmal depolarizingshifts (PDSs) characteristic of epileptiform discharges.22 Asepisodes began to terminate (Figure 3C), shorter, discretePDSs became apparent. These depolarizing shifts triggeredhigh-frequency spike discharges throughout the SRED (Fig-ure 3D). Quantitative analysis of 10 representative “epileptic”neurons (totaling over 7 hours of recording) revealed that.61% of the spikes in “epileptic” neurons had an instanta-neous frequency.3 Hz (Figure 4B). More than 60% of thesehigh-frequency discharges lasted longer than 20 seconds andwere therefore considered SREDs. Furthermore, there wereno significant differences in membrane potential, input resis-tance, or spike characteristics (amplitude, rise time, 50% rise,90% rise) between control and “epileptic” neurons. Theseresults demonstrated that SREDs, not seen in control neurons,were present in glutamate-treated neurons, although intrinsicmembrane properties were not significantly changed.

The SREDs or electrographic seizures in glutamate-injuredneurons occurred spontaneously, randomly, and recurrently.In the 2.5-hour recording shown in Figure 5, 9 independentSREDs occurred, ranging in duration from 1.08 to 4.83minutes. The average duration of SREDs and the averagetime interval between SREDs was 2.160.3 minutes and7.261.0 minutes, respectively (n510, over 7 hours ofrecording).

Neuronal Networks Displayed SynchronizedSREDs After Glutamate ExposureTo determine if neurons with glutamate injury–inducedSREDs were bursting in synchronized neural networks, weperformed whole-cell current-clamp recordings on pairs ofneurons 1 to 8 days after glutamate injury. Pairs of neuronsranged in distance from immediately adjacent to as far as 800mm apart. Epileptiform discharges occurred simultaneouslyin 90% of neuron pairs (n529 pairs). Both the onset and

Figure 3. After injury induced by glutamate (5 mmol/L, 30 min-utes), surviving hippocampal neurons manifested SREDs, notseen in control neurons. A, Representative current-clamp re-cording from control neuron with resting potential of 253 mV.Spontaneous action potentials, EPSPs, and IPSPs wereobserved, indicative of normal neurophysiology. B, One of 7SREDs from a neuron (resting potential of 263 mV) 24 hoursafter glutamate exposure. This SRED started and terminatedspontaneously, lasting '6 minutes. C, Regions at initiation andnear termination of SRED (bars) are displayed at faster timescale. SRED started abruptly as a sustained membrane depolar-ization of '20 mV. As the episode began to terminate, discretePDSs became apparent. D, Regions of high-frequency spikefiring throughout SRED (bars) are displayed at faster time scale.Initial depolarization of SRED triggered high-frequency spikefiring of '7 Hz. As the prolonged depolarization maximized,spike firing reached a frequency of 12 Hz, sufficient to reducespike amplitude, presumably as the result of sodium channelinactivation. As the SRED began to terminate, PDSs still main-tained high-frequency spike firing of 7 Hz.

Figure 4. Quantitative analysis of instantaneous spike frequencyand high-frequency burst discharge duration revealed the pres-ence of SREDs in glutamate treated neurons not seen in con-trols. A, Average histogram of instantaneous spike frequencywith bins of 0.25 Hz determined from individual histograms of12 control neurons making up .4 hours of recording. Majorityof spike firing (61%) occurred at spike frequencies ,0.25 Hz.Furthermore, 88% of all spikes had instantaneous frequency ,3Hz. B, Average histogram of instantaneous spike frequency withbins of 0.25 Hz determined from individual histograms of 10neurons manifesting epileptiform activity (.7 hours of record-ing); 61% of all spike firing had instantaneous frequency .3 Hz.C, Histogram of duration of high-frequency (.3 Hz) dischargesin 12 control neurons. Bursts of $3 Hz never occurred for .10seconds in control neurons. Therefore, no SREDs occurred incontrol neurons. D, Histogram of duration of high-frequency (.3Hz) discharges in 10 glutamate-treated neurons; 60% of thesedischarges occurred with durations .20 seconds. These SREDsmost often occurred with durations between 1 and 3 minutes.Glutamate-treated neurons also manifested high-frequency dis-charges of ,10 seconds (31% of high-frequency discharges)during the interval between SREDs.

Sun et al Glutamate Injury–Induced Epileptogenesis 2347

by guest on February 20, 2014http://stroke.ahajournals.org/Downloaded from

termination of the epileptiform discharges were highly syn-chronized (Figure 6B). Individual paroxysmal depolarizingshifts were also synchronized between neurons (Figure 6C).In addition, high-frequency spikes associated with both theprolonged initial depolarization and discrete PDSs near theend of SREDs occurred simultaneously (Figure 6D).

Phenobarbital But Not Ethosuximide InhibitedGlutamate-Induced SREDsPhenobarbital terminated an ongoing SRED and reversiblyinhibited the generation of SREDs (Figure 7A). On the otherhand, ethosuximide, a t-type, voltage-gated Ca21 channelblocker effective in treating generalized absence seizures,23

had no effect on SREDs in our model (Figure 7B). Thus, theSREDs induced by glutamate injury in our model respondedto therapeutically relevant concentrations of anticonvulsantsanalogous to the setting of generalized tonic clonic and partialcomplex seizures.24

DiscussionThe experiments in this study document a new model ofepileptogenesis in hippocampal neurons mediated by gluta-mate injury. Like the excitotoxic glutamate injuries associ-ated with both ischemic and anoxic events,6 the excitotoxicinjury in this model produced a mixed population of neuronscharacterized by both cell survival and cell death (Figure 2A).Glutamate exposure produced neuronal injury characterizedby prolonged reversible membrane depolarization, decreasedmembrane input resistance, loss of synaptic potentials, andneuronal swelling (Figure 1). As suggested by our initialhypothesis, neurons that survived the glutamate exposuredeveloped SREDS throughout the life of the cultures (Figure2, B and C). SREDS occurred in 8664% of the surviving

neurons per culture (n517 cultures) and thus represented amodel of glutamate injury–induced epileptogenesis.

SREDs expressed many characteristics of overt epilepticseizures. SREDs started and terminated spontaneously (Fig-ures 3 and 5) and were synchronized in nature (Figure 6). TheSREDs produced by excitotoxic glutamate injury manifestedPDSs that were typical of epileptiform activity22 with burstsof high-frequency spike firing (Figure 3).3 Hz for 20seconds or longer (Figures 4, B and D). Sham control neuronsnever displayed SREDs (Figure 3 and Figure 4, B and D, andFigure 6). Finally, the SREDs produced by glutamate injuryresponded to the anticonvulsant drug phenobarbital but not toethosuximide (Figure 7). These results demonstrated thathippocampal cultures subjected to injury by glutamate expo-sure could be transformed into neuronal networks manifest-ing SREDs for the life of the culture, producing an in vitromodel of epilepsy.

The involvement of glutamate in epileptogenesis has beenimplicated in whole animal,12–14 slice,15,16 and tissue culturemodels17 of epilepsy. To induce epileptogenesis, these mod-els all used continuous neuronal spiking produced by sei-

Figure 5. Epileptiform activity persisted for the duration of therecording session. Current-clamp recording from glutamate-treated neuron showed that SREDs occurred throughout therecording period. Five tracings shown were part of a continu-ous, 2.5-hour, current-clamp experiment. This neuron (266 mVresting potential) exhibited 9 independent SREDs during the re-cording period.

Figure 6. Epileptiform activity was synchronized in networks ofhippocampal neurons surviving glutamate injury. A, Simulta-neous current-clamp recording from 2 control neurons. Neuron1 (N-1) and neuron 2 (N-2) each had resting potentials of 260mV. Control neurons displayed normal neurophysiology withspontaneous action potentials, EPSPs, and IPSPs. Arrows indi-cate action potentials triggered by intracellular current injec-tions. B, Two neurons with synchronized epileptiform activityduring epileptiform discharge 24 hours after glutamate expo-sure. Neurons 1 and 2 (N-1 and N-2), with resting potentials of265 mV and 260 mV, manifested epileptiform discharges thatwere highly synchronized. C, Regions at initiation and near ter-mination of the SRED (bars) are displayed at faster time scale.Near beginning of the SRED, depolarization triggered high-frequency spike firing. As the SRED terminated, several discretedepolarizing bursts were closely synchronized between the twoneurons. D, Regions of recording from C (bars) at faster timescale. High-frequency spike firing associated with initial pro-longed depolarization (14 Hz) and PDSs at end (9 Hz) demon-strated synchronization.

2348 Stroke October 2001

by guest on February 20, 2014http://stroke.ahajournals.org/Downloaded from

zures,12 repeated high-frequency excitation,13,14,16 or lowextracellular magnesium environments15,17 rather than aglutamate-induced prolonged, reversible depolarization asused in this study. Many of these models have implicatedactivation of theN-methyl-D-aspartate receptor (NMDAR)for epileptogenesis.12–14,16,18Interestingly, epileptiform dis-charges have also been produced by growing neurons inculture in the presence of agents that block glutamate recep-tors and synaptic transmission (10 mmol/L Mg21 andkyurenic acid). Removal of these agents resulted in theexpression of seizure-like activity.25,26 This distinctly differ-ent culture model uses the inhibition of glutamate receptors toinduce hyperexcitability. Although the mechanism producinghyperexcitability in this model has not been fully delineated,it has been shown that inhibition of glutamate receptors inneurons in culture produces alterations in NMDAR subunitexpression that are regulated by synaptic activity duringdevelopment.27 It is possible that alterations in NMDARsubunit expression may underlie the development of hyper-excitability in this model. Though glutamate exposure may

induce changes in receptor subunit expression in the gluta-mate injury–induced epileptogenesis model, these potentialchanges probably occur through a separate mechanism.

Glutamate injury–induced epileptogenesis is clearly dis-tinct from the low magnesium model of epileptogenesis andrepresents a separate cause of acquired epilepsy. Thesemodels use distinctly different physiological inputs. Whereasthe low magnesium model uses continuous spike firinganalogous to the seizure activity in status epilepticus17 toproduce SREDs, the glutamate injury model used a pro-longed, reversible depolarization to induce epileptogenesis,analogous to the peri-infarct ischemic depolarizations inischemic stroke. Thus, two physiologically distinct mecha-nisms are used to induce epileptogenesis in these two modelsof epilepsy. In addition, the duration of the epileptogenicstimulus is very different between the glutamate injury modeland the low magnesium model. The low magnesium modelrequires 3 hours of spike activity to induce SREDs, whereasonly 30 minutes of glutamate exposure produces “epileptic”neurons in the glutamate injury mode. The element ofexcitotoxic neuronal death in the glutamate injury model isanother distinction from the low magnesium model. The lowmagnesium model is associated with little neuronal death.17

Thus, the induction mechanisms in the glutamate injurymodel and low magnesium model are as different as strokeand status epilepticus, known causes of acquired epilepsy.However, the inducers in these models may share commonunderlying mechanisms.

The low magnesium model requires calcium entry throughthe NMDAR as a second messenger for epileptogenesis.18

Results from our laboratory suggest that calcium may also actas a second messenger system in the glutamate injury model.Thus, although the models use the distinctly different physi-ological inputs of glutamate-induced depolarization and spikeactivity, they may converge on related molecular mechanismsfor inducing and maintaining the epileptic condition. Thepotential role of selective neuronal death in glutamate injury–induced epileptogenesis requires further investigation, espe-cially in light of the fact that inhibitory neurons are typicallyless vulnerable to excitotoxicity than excitatory neurons.28

Although differential cell death may affect the balancebetween the number of inhibitory and excitatory neurons,resulting in a larger number of surviving inhibitory neurons,28

the glutamate-induced injury produced “epilepsy” in theneurons despite the potential alterations in neuronal subpopu-lations. Further studies are needed to determine the role ofselective cell death in this model. In addition, the possibleroles of gap junctions,29 ischemia-induced alterations insecond-messenger systems, and gene changes30 in mediatingepileptogenesis represent important future directions for re-search that can be conveniently studied in this system.

The association between stroke and epilepsy has beendemonstrated clinically, and stroke is the most common causeof acquired epilepsy.3 However, the mechanisms by whichcerebral ischemia initiates epileptogenesis are not understood.The glutamate injury produced in this model of epileptogen-esis resembles some of the phenomena associated with stroke.Increases in extracellular glutamate,4,5 reversible depolariza-tion with loss of synaptic activity,10 acute neuronal swelling,

Figure 7. Phenobarbital but not ethosuximide inhibited epilepti-form activity in hippocampal neuronal cultures. A, Application of100 mmol/L phenobarbital terminated ongoing SRED andreversibly inhibited generation of SREDs. B, Application of1 mmol/L ethosuximide had no effect. C, Average number ofSREDs per hour for control (not exposed to anticonvulsant,n515 neurons), phenobarbital (n56 neurons), and ethosuximide(n56 neurons). *P,0.05, ANOVA and post hoc Tukey test, 0.1%ethanol vehicle not significantly different from control.

Sun et al Glutamate Injury–Induced Epileptogenesis 2349

by guest on February 20, 2014http://stroke.ahajournals.org/Downloaded from

and excitotoxic delayed neuronal death6 associated with theischemic penumbra7 are all present in this model. To ourknowledge, this study demonstrates, for the first time, spon-taneous, recurrent, epileptiform activity in hippocampal neu-rons induced by glutamate injury. This model of glutamateinjury–induced epileptogenesis may offer new insights intothe development and maintenance of the epileptic conditionafter a neurological trauma such as stroke and therefore mayprovide therapeutic strategies to develop both novel anti-epileptogenic and anticonvulsant agents to prevent stroke-induced epilepsy.

AcknowledgmentsThis study was supported by grants RO1-NS23350 (Dr DeLorenzo),P50-NS25630 (Dr DeLorenzo), and NSO7288 (D.A. Sun) and aGLENN/American Federation for Aging Research Scholarship (D.A.Sun). We are grateful to Drs M. Ross Bullock and Oscar L. Alves fortheir assistance with glutamate concentration measurements.

References1. Delgado-Escueta AV. Symptomatic lesional epilepsies. In:Jasper’s Basic

Mechanisms of the Epilepsies.Philadelphia, Pa: Lippincott Williams &Wilkins; 1999:433–436.

2. Hauser WA, Hesdorffer DC.Epilepsy: Frequency, Causes and Conse-quences.New York, NY: Demos; 1990.

3. Hauser WA, Annegers JF, Kurland LT. Prevalence of epilepsy inRochester, Minnesota: 1940–1980.Epilepsia. 1991;32:429–445.

4. Bullock R, Zauner A, Woodward J, Young HF, Massive persistent releaseof excitatory amino acids following human occlusive stroke.Stroke.1995;26:2187–2189.

5. Davalos A, Castillo J, Serena J, Noya M. Duration of glutamate releaseafter acute ischemic stroke.Stroke. 1997;28:708–710.

6. Choi D. Stroke.Neurobiol Dis.2000;7:552–558.7. Dirnagl U, Iadecola C, Moskowitz MA. Pathobiology of ischaemic

stroke: an integrated view.Trends Neurosci. 1999;22:391–397.8. Hansen AJ, Nedergaard M. Brain ion homeostasis in cerebral ischemia.

Neurochem Pathol. 1988;9:195–209.9. Lipton P. Ischemic cell death in brain neurons.Physiol Rev. 1999;79:

1431–1568.10. Hossmann KA. Viability thresholds and the penumbra of focal ischemia.

Ann Neurol. 1994;36:557–565.11. Dingledine R, McBain CJ, McNamara JO. Excitatory amino acid

receptors in epilepsy.Trends Pharmacol Sci. 1990;11:334–338.12. Rice AC, DeLorenzo RJ. NMDA receptor activation during status epi-

lepticus is required for the development of epilepsy.Brain Res. 1998;782:240–247.

13. Croucher MJ, Bradford HF, Sunter DC, Watkins JC. Inhibition of thedevelopment of electrical kindling of the prepyriform cortex by dailyfocal injections of excitatory amino acid antagonists.Eur J Pharmacol.1988;152:29–38.

14. Croucher MJ, Bradford HF. NMDA receptor blockade inhibits glutamate-induced kindling of the rat amygdala.Brain Res. 1990;506:349–352.

15. Anderson WW, Lewis DV, Swartzwelder HS, Wilson WA.Magnesium-free medium activates seizure-like events in the rat hip-pocampal slice.Brain Res. 1986;398:215–219.

16. Stasheff SF, Anderson WW, Clark S, Wilson WA. NMDA antagonistsdifferentiate epileptogenesis from seizure expression in an in vitro model.Science. 1989;245:648–651.

17. Sombati S, DeLorenzo RJ. Recurrent spontaneous seizure activity inhippocampal neuronal networks in culture.J Neurophysiol. 1995;73:1706–1711.

18. DeLorenzo RJ, Pal S, Sombati S. Prolonged activation of the N-methyl-D-aspartate receptor-Ca21transduction pathway causes spontaneousrecurrent epileptiform discharges in hippocampal neurons in culture.ProcNatl Acad Sci U S A. 1998;95:14482–14487.

19. Didier M, Heaulme M, Soubrie P, Bockaert J, Pin JP. Rapid, sensitive,and simple method for quantification of both neurotoxic and neurotrophiceffects of NMDA on cultured cerebellar granule cells.J Neurosci Res.1990;27:25–35.

20. McNamara JO. Cellular, and molecular basis of epilepsy.J Neurosci.1994;14:3413–3425.

21. Sombati S, Coulter DA, DeLorenzo RJ. Neurotoxic activation of glu-tamate receptors induces an extended neuronal depolarization in culturedhippocampal neurons.Brain Res. 1991;566:316–319.

22. Matsumoto H, Amjone-Marsan C. Cotrical cellular phenomena in exper-imental epilepsy: interictal manifestations.Exp Neurol. 1964;9:286–304.

23. Rogawski MA, Porter RJ. Antiepileptic drugs: pharmacological mech-anisms and clinical efficacy with consideration of promising develop-mental stage compounds.Pharmacol Rev. 1990;42:223–286.

24. Macdonald RL, Kelly KM. Antiepileptic drug mechanisms of action.Epilepsia. 1993;34(suppl 5):S1–S8.

25. Furshpan EJ, Potter DD. Seizure-like activity and cellular damage in rathippocampal neurons in cell culture.Neuron. 1989;3:199–207.

26. Koroshetz WJ, Furshpan EJ. Seizure-like activity and glutamate receptorsin hippocampal neurons in culture.Neurosci Res Suppl. 1990;13:S65–S74.

27. Hoffman H, Gremme T, Hatt H, Gottmann K. Synaptic activity-dependent developmental regulation of NMDA receptor subunitexpression in cultured neocortical neurons.J Neurochem. 2000;75:1590–1599.

28. Tecoma ES, Choi DW. GABAergic neocortical neurons are resistant toNMDA receptor-mediated injury.Neurology. 1989;39:676–682.

29. Dudek FE, Yasumura T, Rash JE. Non-synaptic mechanisms in seizuresand epileptogenesis.Cell Biol Int. 1998;22:793–805.

30. DeLorenzo RJ, Morris TA. Long-term modulation of genetic expressionin epilepsy and other neurological disorders.Scientist. 1999;5:86–99.

2350 Stroke October 2001

by guest on February 20, 2014http://stroke.ahajournals.org/Downloaded from