depolarization a

Journal of Physiology (1988), 404, pp. 365-384 365With 11 text-figuresPrinted in Great Britain

LONG-LASTING MODIFICATION OF THE SYNAPTIC PROPERTIES OFRAT CA3 HIPPOCAMPAL NEURONES INDUCED BY KAINIC ACID

BY YEHEZKEL BEN-ARI AND MICHEL GHOFrom the Unite 29, INSERM, 123 Boulevard Port Royal, 75014 Paris, France

(Received 11 February 1988)

SUMMARY

1. The action of a short bath application of kainic acid (KA, 200-250 nM, 3-5 min)on the CA3 region of rat hippocampal slices has been studied with intracellular andextracellular recording techniques.

2. KA evoked bursts which persisted for 10-15 min. In addition, after KA,electrical stimulation of various inputs to CA3 which elicited an EPSP-IPSPsequence in control conditions evoked an EPSP followed by a burst. This evokedresponse persisted for several hours after removal of KA suggesting the occurrenceof a long-lasting modification of the synaptic properties of CA3 neurones.

3. Intracellular recordings showed the spontaneous and evoked bursts to consistof five to ten action potentials riding on a depolarizing shift 10-25 mV in amplitudeand 40-100 ms in duration. Both spontaneous and evoked bursts were followed bya long-lasting hyperpolarization 15-25 mV in amplitude and 1-15 s in duration.

4. We propose that both spontaneous and evoked synchronized bursts aregenerated by a polysynaptic network since: (a) intracellularly recorded bursts weresynchronized with the bursts in extracellular field recording; (b) bursts disappearedwhen synaptic transmission or Na+ action potential were blocked by cobalt (1 mm)or TTX (1 ,UM) respectively; (c) bursts were suppressed by elevated divalent cationconcentration; (d) burst occurrence was independent of the membrane potential ofthe cell; (e) the depolarization shift that underlies the bursts was a linear functionof the membrane potential and reversed in polarity at 0 mV. In addition, the evokedbursts were all-or-none events with a variable latency.

5. Laminar profile analysis of the spontaneous and evoked bursts suggests thatthey were generated by synapses located on the distal apical segments of thedendrites of CA3 pyramidal cells.

7. The persistence of the evoked bursts was neither due to a persistent change incell excitability nor to a long-lasting reduction in GABAergic synaptic inhibition.

8. Bath application of a high concentration of potassium (7 mm) also inducedspontaneous and evoked bursts; the latter also persisted several hours after returnto control medium.

9. The N-methyl-D-aspartate (NMDA) antagonist, D-APV (D(-)-2-amino-5-phosphonovaleric acid) (30 /tM), did not block the spontaneous discharges induced byKA or high potassium, but prevented the long-lasting effects on the synapticresponses.

10. In conclusion, we suggest that the long-lasting change of synaptic responses is

Y. BEN-ARI AND M. GHO

generated by the spontaneous synchronized discharges present during and shortlyafter the application of KA or high potassium. D-APV experiments suggest thatNMDA receptors are involved in this change as in other forms of long-term synapticplasticity.

INTRODUCTION

The mammalian hippocampus is thought to be the brain area with the lowestseizure threshold (Green, 1964). Within the hippocampus the CA3 region has beenreferred to as the pacemaker region due to its unique ability to generate synchronizedbursts following the application of various convulsive agents or procedures. AlthoughCA3 pyramidal neurones have intrinsic mechanisms that can generate endogenousbursts (Hablitz & Johnston, 1981), the pacemaker properties of this region are likelyto be due to the presence of recurrent excitatory collateral connections betweenpyramidal neurones (MacVicar & Dudek, 1980; Miles & Wong, 1986).

Kainic acid (KA), a rigid analogue of glutamic acid (Olney, Rhee & Ho, 1974), hasbeen extensively used to study the mechanisms of epileptogenesis. Parenteral orintracerebral administration of KA into animals produces a seizure and a braindamage syndrome in which the hippocampus, and in particular the CA3 subfield,plays a central role (Ben-Ari, 1986). In keeping with this, the CA3 region isparticularly rich in high-affinity specific KA receptors in both human (Represa,Tremblay, Schoevart & Ben-Ari, 1986) and rat brains (Berger & Ben-Ari, 1983;Unnerstall & Wamsley, 1983). In slice preparations, KA (Robinson & Deadwyler,1981; Westbrook & Lothman, 1983), like other convulsive agents (Wong & Traub,1983; Johnston & Brown, 1984; Rutecki, Lebeda & Johnston, 1985; Bingmann &Speckmann, 1986; Walther, Lambert, Jones, Heinemann & Hamon, 1986; Neuman,Cherubini & Ben-Ari, 1987; Cherubini, Neuman, Rovira & Ben-Ari, 1988), readilygenerates spontaneous bursts in the CA3 region, probably because of an increase inexcitatory and a decrease in inhibitory processes. Thus in the CA3 region KA hasbeen shown to produce multiple effects on pyramidal cells including: (i) adepolarization (Robinson & Deadwyler, 1981; Collingridge, Kehl, Loo & McLennan,1983; Cherubini, Rovira, Gho & Ben-Ari, 1986); (ii) a reduction of GABAA- (Fisher& Alger, 1984) and GABAB- (Ben-Ari, Gho & Rovira, 1988) mediated inhibitions;(iii) a reduction of various inhibitory conductances including IQ (Halliwell & Adams,1982), which is responsible for the anomalous rectification, and the slow calcium-dependent potassium current Ic (Brown & Griffith, 1983), which is responsible for theafter-hyperpolarization following a train of spikes (Gho, King, Ben-Ari & Cherubini,1986).In this paper, using intracellular and extracellular recording techniques, we

describe in detail the action of KA on CA3 with particular reference to the long-lasting changes in synaptic responses which occur following a brief application of thedrug. Since long-term alterations in synaptic responses in the hippocampus are welldocumented, notably after a train of high-frequency electrical stimulation (Bliss &L0mo, 1973; Bliss & Gardner-Medwin 1973; Schwartzkroin & Wester, 1975), we haveexamined the possibility that brief epileptiform activity may lead to similar long-lasting modifications. A preliminary communication of this report has been presented(Ben-Ari, Cherubini & Gho, 1987).

366

SU"STAINED EFFE'CTS OF KAINIC ACID IN CA3

METHODS

Male Wistar rats (I120-200 g, 30-40 days old) were anaesthetized with ether and killed by a heavyblow to the chest or cervical dislocation. One hippocampus was quickly removed and transverseslices of 500 ,um nominal thickness were cut using a Mcllwain tissue chopper. Slices were incubatedat room temperature in artificial cerebrospinal fluid (ACSF) for at least 60 min; when required, oneslice was transferred to a recording chamber (Williams. Henderson & North, 1984). A completelysubmerged slice was laid on a nylon mesh and was superfused (3 ml/min, dead space 3 ml) withACSF at 34+1 °C. ACSF had the following composition (in mM): NaCl, 126; KCl, 3-5; NaH2PO4,1 2; MgCl2, 1-3; CaCl2, 2; glucose, 10; NaHCO3, 25; pH 7 3, when equilibrated with 95% O2-5% CO2. High-divalent-cation ACSF was made by increasing the concentrations of Mg2+ andCa2+ to 6 and 4 mm respectively. High-potassium ACSF was made by increasing the K+concentration to 7 m M.

Pyramidal cells in the CA3b (Lorente de N6, 1934) region were impaled with microelectrodesfilled with either 4 M-potassium acetate, 2 M-potassium methylsulphate or 2 M-CS2SO4 which had afinal resistance of 50-120 MQ before use. A conventional bridge circuit allowed injection of currentthrough the recording electrode and measurement of the potential difference between the electrodetip and a Ag/AgCl pellet bath electrode. Membrane potentials were estimated from the potentialobserved upon withdrawal of the electrode from the cell. Field potentials were recorded with glassmicroelectrodes filled with 2 M-NaCI with a resistance of 1-5 MQ and usually positioned about200(,um from the intracellular recording microelectrode. Electrical stimulation (10-50 /ts durationand 10-50 V intensity) was performed with bipolar, electrolytically etched tungsten electrodesplaced in the granular layer or in the hilar zone. Signals were displayed on a digital oscilloscope andrecorded on a computer-driven pen recorder.

D)rugs and other solutions were bath applied via a three-way tap system. Drugs used were kainicacid (KA. 200 nm, Sigma), tetrodotoxin (TTX, 1 /tM, Sigma) and D(-)-2-amino-5-phosphonovalericacid (D-APV, 30 /tM, (ambridge Research Biochemicals).

RESULTS

This report is based on intracellular recordings from 101 neurones in the CA3bregion of rat hippocampal slices. Intracellular records were considered acceptable ifa stable resting membrane potential greater than -60 mV was maintained for atleast 20 min. The neurones had a resting membrane potential of 67+7 mV(mean+ S.D.). In twenty-six neurones we have measured the input resistance andmembrane time constant and found values comparable to earlier studies (56 + 8 MQand 25+6 ms respectively; e.g. Brown, Fricke & Perkel, 1981; Johnston, 1981).

Kainic acid induces synchronized burstsElectrical stimulation of the granular layer of the fascia dentata or the mossy fibre

zone evoked in control conditions a negative field potential in the stratum lucidumof the CA3 region. With intracellular recordings, the synaptic response consisted ofan EPSP-IPSP sequence (Fig. IA). After 2-3 min bath application of KA (200-250 nM) synchronized bursts at a frequency of about 0 5 Hz were conspicuous ineighty-five out of ninety-five slices (including field potential data, Fig. 1 B). Withextracellular recordings in the pyramidal layer, the discharges consisted of a burst ofpopulation spikes riding on a positive potential wave. With intracellular recordings,it consisted of a brief high-frequency burst of action potentials (five to ten innumber), arising from a large depolarizing shift (Matsumoto & Ajmone-Marsan,1964), which was followed by a large hyperpolarization or post-activationhyperpolarization (Fig. 1 B). The depolarizing shift was 10-25 mV in amplitude and40-100 ms in duration; the amplitude of the post-activation hyperpolarization was

367

Y. BEN-ARI AND M. GHO

oL Eo oco N

:1.00LO

000

1"EI

_

E

0

CN

4)

0co

Ar _

.~~~~~~~~~~~~~~~~~~~~~~~~~~~~

_=

II

I~~~~~~~~~~~~~~~~~~~~~~~~~~

N

6w .~~~~~~~

v4

I

E

Ii

IiIII

1-1 .4

4

368

I

4

.4

SUSTAINED EFFECTS OF KAINIC ACID IN CA3

15-25 mV and its duration 1-1.5 s. KA also produced a reduction (about 15 %) of theinput membrane resistance associated with a small initial depolarization (2-6 mV,n = 45). Similar depolarizations induced by current injection through the recordingelectrode did not produce synchronized bursts, ruling out the possibility that theseare causally related to the depolarization. Furthermore, the threshold of burstgeneration was frequently more negative than the resting membrane potential (seeFig. 1). After returning to control medium, the spontaneous bursts lasted for about10-15 min (Fig. 1). In contrast, electrical stimulation of the fascia dentata or themossy fibre zone now evoked an EPSP followed by a burst of action potentials (Fig.1C). A similar evoked burst could be elicited for up to 4 h after return to controlmedium (longest duration of stable intracellular recordings performed).

Nature of spontaneous and evoked burstsUsing the criteria proposed by Johnston & Brown (1984), a series of experiments

were performed in order to determine whether the bursts are generated by amechanism intrinsic to a single neurone (endogenous pacemaker) or by a network ofneurones (circuit generation).To determine the degree of dependence of the occurrence of the bursts on the

membrane potential of the cell recorded, the frequency of the intracellularspontaneous bursts was monitored at various membrane potentials (between -100and -50 mV). A change in the membrane potential did not modify the burstfrequency (n = 5, not shown). A similar experimental paradigm was performed onthe evoked burst comparing the probability of eliciting a burst at differentmembrane potentials. Again, the probability of evoking a burst was not related tothe membrane potential of the cell recorded (n = 5, not shown).The relationship between the membrane potential of the cell recorded and the

amplitude of the depolarizing shift underlying the bursts was studied using Cs'-filledmicroelectrodes. This procedure enables a wider range of membrane potentials to betested, possibly by a blockade of potassium channels (Brown & Johnston, 1983). Theamplitude of the spontaneous and evoked depolarizing shift was a linear function ofthe membrane potential (Fig. 2). The mean reversal potential of the spontaneousburst was 1+ 2 mV (n = 6) and that of the evoked burst 0+ 2 mV (n = 6).

Other observations also suggest that the bursts were generated by a polysynapticnetwork. First, extracellular and intracellular recordings showed both spontaneousand evoked bursts always to be synchronized (see Fig. 1). Second, both spontaneousand evoked bursts were blocked by agents such as Co2+ (3 mM) which blocks synaptictransmission or TTX (1 sM) which blocks action potential propagation (n = 5, notshown). Third, spontaneous and evoked bursts were reversibly blocked by high-divalent-cation ACSF, a procedure which is known to preferentially blockpolysynaptic transmission (Berry & Pentreath, 1976; Miles & Wong, 1986). As shownin Fig. 3, spontaneous bursts were first induced by bath application ofKA in controlACSF; a brief (3 min) application of KA in a high-divalent-cation ACSF blocked thebursts (Fig. 3A) which reappeared a few minutes after return to KA in control ACSF.Thirty minutes later (Fig. 3 B) the evoked bursts elicited by stimulation of the mossyfibres were reversibly blocked by superfusion with high-divalent-cation ACSF,leaving only the monosynaptic mossy fibre EPSP.

369

370 Y. BEN-ARI AND M. GHO

A B 30 CSpontaneous 5

bursts E25

a) 20Evoked bursts2 5 *\,20 1

15

15 * E15 A E N10 11

0 . \ m

-10Membrane potential (mV)* 0-60 -40 -20 .20 -10

-28 * Spontaneous -10 -28bursts

A--- AEvokedbursts

-64 20 mV -45 J 20 mV

40 ms 40 ms

Fig. 2. Relationship between the amplitude of the depolarizing shift underlying thespontaneous and evoked bursts and the membrane potential. A and C show representativerecords obtained with Cs2SO4-containing electrodes of spontaneous and mossy-fibre-evoked bursts from the same neurone at the membrane potentials indicated (mV). In Bthe amplitudes of the depolarizing shifts are plotted against the membrane potential. Theamplitude of the depolarizing shift was measured at a fixed latency indicated by thearrows. The reversal potential of the spontaneous and evoked bursts was about 0 mV.

AKA (250 nM)

High Mg2+ + Ca2+

-69 mV25 mV

High Mg2+ + Ca2+ 1 minB

-69 mV +A A A AA AA AA AA AAAA A. A A A- AAAAA AA A AA AAA AA AAAAAAAA

J_ 25 mV1 min

20 mV

100 ms

A A A

Fig. 3. High-divalent-cation ACSF (6 mM-Mg2" and 4 mM-Ca2") blocks both spontaneousand evoked bursts. A, a brief application of high-divalent-cation ACSF reversibly blockedthe spontaneous bursts induced by KA. The asterisk shows where the manual restorationof the resting potential failed to prevent the suppression of the bursts. B shows, in thesame neurone, that more than 1 h after KA wash-out stimulation of the mossy fibresevoked bursts which were also reversibly blocked by high-divalent-cation ACSF. In high-divalent-cation ACSF stimulation only evoked a mossy fibre EPSP.

SUSTAINED EFFECTS OF KAINIC ACID IN CA3 371

The following additional features of the bursts following the transient applicationof KA also suggest that they were generated by a polysynaptic circuitry. Thus, thelatency of the bursts was variable in contrast with the fixed latency of the mossy fibremonosynaptic EPSP (Fig. 4A). Furthermore, the evoked bursts had an all-or-nonethreshold which was not modified by a large sustained depolarization of the neurone(Fig. 4B).

AEPSP Burst

4

-69 mV - n-\

*Vr * 20 mV 10 20 30

Latency (ms)80 ms

B

il 1 500 !! I0AV500 pV

40 mV 'I !1 n 20 mV

20 in \43 43 mV N80 msI ~~ ~ ~ ~ ~ ~ -~~~~~C~-43mVI ~~AAAAA A

_71 mV v~q,>? 7 9 11 V -I -,,,71 mV

A AAA A A A A7 9 11 v

Fig. 4. Variable onset latency and all-or-none characteristic of the evoked bursts. A,bursts evoked by stimulation of the mossy fibres 40 min after KA application (for 3 min).Note the variable latency of burst onset. The mossy fibre EPSP is conspicuous in the thirdtrace. The histograms on the right show the range of latencies of the mossy fibre EPSPand subsequent bursts observed in eleven trials. Latency was measured as the timeinterval between the stimulus artifact and the peak of the EPSP or the first actionpotential of the. burst. B, the evoked burst had an all-or-none threshold which wasindependent of the membrane potential. Note that at the resting membrane potential(-71 mV) a burst was elicited in both the extracellular (upper) and intracellular (lower;Cs'-containing microelectrode) records with a stimulation intensity of 10 V or more (butnot 9 V). When the cell was depolarized to -43 mV by current injection the threshold wasunchanged. The right-hand panel illustrates, using a faster time base and twosuperimposed traces, how an increase in the intensity of stimulation from 9 to 10 Vevoked a full burst in an all-or-none manner.

Topographical features of the burst generatorTo determine the site(s) of generation of the bursts along the CA3 dendrites, a

laminar profile analysis was performed with extracellular field recording. In tenexperiments, an extracellular electrode was moved along the apical and basaldendrite of pyramidal neurones and the field potentials recorded. As shown in Fig.5, spontaneous and evoked bursts (Fig. 5A and B respectively) had similar laminarprofiles. By plotting the peak amplitude of both field potentials against the distancefrom the pyramidal layer, both profiles were shown to reverse approximately 100 jtmabove the pyramidal layer, between the stratum radiatum and the mossy fibre

372 Y. BEN-A-RI AND M. GHO

A Spontaneous B Evokedbursts bursts

SR

500 pV SL

100 ms

SO

C Burst amplitude (pV)-300 -100 100 300 500

300

X/- * Spontaneous bursts

/A* 200 A Evoked bursts

A A

-100 A

-200

Fig. 5. Laminar profile analysis of extracellular field potentials for the spontaneous bursts(during the KA application, first column) and mossy fibre-evoked bursts (1 h after wash-out of the drug, second column). In A and B the traces show the field potentials recordedfrom the dendritic regions of the CA3 pyramidal neurones (see diagram on the right). Anextracellular electrode was placed in the pyramidal layer and another microelectrode wasmoved at approximately 50,um steps along the apical and basal dendrites; the burstrecorded from the former electrode served as a control for burst amplitude and to triggerthe digital oscilloscope, thus synchronizing the events observed with the displacedelectrode. In C, the peak amplitude of each of the bursts was plotted against the distancefrom the pyramidal layer. To measure the amplitude of the bursts, the signal was filtered(100 Hz, low pass) in order to eliminate the fast population spikes. Note that the maximalnegative field potential for spontaneous and evoked bursts was recorded in the stratumradiatum and the reversal occurred at the boundary between stratum radiatum (SR) andstratum lucidum (SL). The drawing of the cell has the same scale as the ordinate in C.Additional abbreviations: SP, stratum pyramida; SO, stratum oriens.

SUSTAINED EFFECTS OF KAINIC ACID IN CA3

A a bA b c0

-70mV f----| 0

20OmVA 0Q

B a b c 80ms

J 1O mV15 ms

-70 mVj10 mV

0 A 0AA 0 80 ms

Fig. 6. Convergence of afferent pathways on the polysynaptic circuitry generating theevoked burst. A, superimposed responses produced by mossy fibre (A, A a), commissuralpathway (O, A b) and perforant pathway (*, Ac) stimulation before (arrows) and afterKA application. The diagram indicates the location of the stimulating and recordingelectrodes. Note that similar bursts were evoked by all three stimuli. B shows how pairsof subthreshold stimuli applied to commissural and mossy fibre pathways summed toevoke a burst (Ba). Bb and Bc show how pairs of subthreshold stimuli of the mossy fibre(Bb) or commissural pathway (Bc) failed to evoke a burst. The insets show the sameresponses at a faster sweep speed. The interpulse delay was 1 ms.

terminal region. Also, the positive and negative maximal values of the spontaneousand evoked bursts occurred at identical points (Fig. 5C).

Heterosynaptic convergence of various afferents on the network generating thebursts was also conspicuous. Thus, as shown in Fig. 6, electrical stimulation of thegranular layer (Fig. 6A a), the vicinity of the hippocampal fissure (Fig. 6A b) or thefimbria (Fig. 6A c) to stimulate the mossy fibres, the perforant pathways and thecommissural pathways respectively, elicited typical evoked bursts in the same sliceafter KA. Furthermore, there was spatial summation between the stimuli, such thata pair of subthreshold stimuli applied at an interval of 1 ms to the mossy fibres andthe commissural pathway evoked a burst (Fig. 6Ba). In contrast, a pair of similarsubthreshold stimuli, applied at an interval of 1 ms to the mossy fibres (Fig. 6Bb) orthe commissural pathways (Fig. 6Bc), failed to evoke a burst.

373

Y. BEN-ARI AND M. GHO

B

No!

: \M\!~9 \

0 20. . .

8. .

0 20 40 60 80 100

Current step (pA)

-400 -200

,,#/ //

a'

Time (ms)

C

-61

I 20 mV

20 ms

Th 10 mV

500 ms

Fig. 7. Long-lasting changes in the evoked response are not due to a sustained change inthe intrinsic properties of CA3 pyramidal neurones. The same cell as in Fig. 1. A and B,the membrane time constant and the voltage-current relationship before (EO ) and 30 minafter (*) KA application. C, superimposed traces of the response evoked by anintracellular depolarizing current injection (15 ms, 700 pA) before and after KAapplication show the after-hyperpolarization following the pulse, the threshold of spikegeneration and the spike accomodation (inset) not to be modified by KA.

TABLE 1. Effects of kainic acid on intrinsic properties of CA3 neurones

ControlAfter KAPercentage ofcontrol values

n

Membranepotential(mV)66+565+598+2

83

Membranetime constant

(ms)

25+424+493+5

26

The parameters were measured before and 30 min after KA application, in neurones in whichspontaneous bursts were induced by the drug. The after-hyperpolarization (AHP) was elicited bya depolarizing current pulse that evoked three spikes.

374

A

3-0

2-5mxE

c 2-0

1-5

-20

-40

AV (mV)

Inputresistance(MQ)56+955+596+5

26

AHP(mV)11+311+298+5

7

w w

SUSTAINED EFFECTS OF KAINIC ACID IN CA3

Mechanisms of the long-lasting synaptic modification observed after kainic acidSeveral experiments were performed to examine the mechanisms of the long-

lasting induction of evoked bursts by KA.We first examined whether the procedure produced a long-lasting change in the

intrinsic properties of CA3 neurones by measuring the following parameters incontrol conditions and 30 min after application of KA (for 3 min): (i) the membranepotential (n = 83); (ii) the input resistance estimated from the voltage deflectioncaused by hyperpolarizing constant-current steps, 200 ms in duration and 300 pA inamplitude (n = 26); (iii) the membrane time constant measured from the off-onphase of the voltage deflection after careful bridge balancing and assuming theresponse could be fitted by a single-exponential decay (n = 26; Johnston, 1981); and(iv) the after-hyperpolarization which followed a discharge of spikes produced by anintracellular injection of depolarizing current (n = 9). As shown in Fig. 7 and Table1 there were no significant changes in these parameters. Therefore, although KAproduced significant modifications in these parameters during and shortly after drugapplication (i.e. Robinson & Deadwyler, 1981; Westbrook & Lothman, 1983;Cherubini et al. 1986), the long-lasting effects are not explicable by a sustainedchange in these intrinsic properties.Two sets of experiments were done to test the possibility that the long-lasting

effects are due to a failure or a reduction in inhibition. We first used a procedureapplied by Alger (1984). In control conditions (Fig. 8) mossy fibre stimulation evokedan EPSP followed by a fast and a slow IPSP, presumably mediated by GABAA andGABAB receptors respectively (Alger & Nicoll, 1982; Newberry & Nicoll, 1985;Hablitz & Thalmann, 1987). Forty-five minutes after briefKA application, a similarstimulation now evoked a burst (clearly recorded at the extracellular level) followedby a post-activation hyperpolarization which prevented the possibility of measuringthe IPSP. However, a hyperpolarizing current step (1-5 nA, 100 ms), applied 50 msbefore the electrical stimulation of the mossy fibres, blocked most of the voltage-dependent hyperpolarization components enabling the measurement of the slowIPSP evoked by the stimulation. As shown in Fig. 8 the slow IPSP was similar incontrol and 45 min after wash-out of KA. Identical observations were obtained in sixother cells.

In the second set of experiments, we used high-divalent-cation ACSF to block thebursts elicited after KA application. As shown in Fig. 9, the fast and slow IPSPs werefirst measured in a high-divalent-cation ACSF. Then, while perfusing with controlACSF, KA was applied for 3 min inducing typical spontaneous bursts; 1 h later,while recording from the same neurone, the slice was again superfused with high-divalent-cation ACSF which blocked the evoked burst enabling the measurement ofboth fast and slow IPSPs. As shown in Fig. 9, the fast and slow IPSPs were similarin control and after KA. Similar observations were performed in two otherneurones.The long-lasting effects of KA are also not due to either a slow removal or a long-

lasting action of the drug since its effects on membrane properties and GABA-mediated inhibitions disappear shortly after the return to control solution (Figs 7, 8and 9). This is confirmed by the experiment described in Fig. 10. In medium

375

Y. BEN-ARI AND M. GHO

A Control

-61 mV,

A

8 45 min after KA

-1

A

C

0

n -E

'D-5

'a-E

cn- -100

c,,

-~ ~ ~ ~ ~ --

MO,

° C Control

%.i a 45 min after KA

0 200 400Time (ms)

600

Fig. 8. The long-lasting change in synaptic response is not due to a reduction in slowGABAergic inhibition. A and B, extracellular (upper trace) and intracellular (lower trace)recordings show mossy fibre stimulation to evoke a fast and a slow hyperpolarization incontrol conditions (A) and a burst 45 min after KA application (conspicuous in theextracellular record, B). To measure the slow GABAergic inhibition and to differentiateit from the voltage-dependent hyperpolarization which follows the burst, a hyper-polarizing current pulse (1-5 nA, 90 ms) was applied before the stimulation, preventingthe occurrence of the voltage-dependent hyperpolarization. The remaining hyper-polarization, presumably a slow IPSP, was not different from that seen in controlconditions. C, plot of the slow inhibition elicited by the stimulation in control conditions(LI) and 45 min after KA application (U).

1250uV

I | 10 mV

200 ms

A^

376

-ftb

SUSTAINED EFFECTS OF KAINIC ACID IN CA3

containing high-divalent-cation ACSF, a brief application of KA depolarized theneurone and reduced its input resistance but did not produce spontaneous bursts.After the return to control media, stimulation of the mossy fibres failed to evoke aburst (Fig. IOA b). However, a second application of KA, in control ACSF, elicitedspontaneous bursts and the usual long-lasting effects (Fig. lOBa and Bb). Similarobservations were made in two other neurones. These observations also suggest thatspontaneous events are of importance in sustaining the hyperactivity.

High Mg2+ + Ca2+ KA (250 nM)

10 min

-62 mV 1 mi J

\~~~~~~~~~~~~~I 1

| r210mV_~rA

X/ ,f u ~~~~~~~~~10mVA>f A 200ms

High Mg2+ +Ca2+ /

h--- -m-.. M. ft-

I 40 mVA 500 ms A 1 min

Fig. 9. The long-lasting change in synaptic response is not due to a reduction in fast andslow GABAergic inhibitions. Upper and lower records from the same cell. First (uppertrace), high-divalent-cation ACSF was applied and the fast and the slow GABAergicinhibitions evoked by mossy fibre stimulation measured. After return to normal ACSF,KA was applied inducing the typical spontaneous synchronized bursts and (1 h later)stimulation-evoked bursts. When the slice was again superfused with high-divalent-cation ACSF the burst was blocked enabling the fast and slow inhibitions elicited bymossy fibre stimulation to be measured. Clearly, there was no change in both types ofGABAergic inhibitions before and after KA.

Long-lasting effects induced by high-potassium ACSFTo evaluate whether the long-lasting effects are exclusively induced by KA, we

have used high-K+ ACSF, another epileptogenic procedure known to inducespontaneous synchronized discharges in CA3 (Rutecki, Lebeda & Johnston, 1985;Korn, Giacchino, Chamberlin & Dingledine, 1987).

Elevating the K+ concentration in ACSF (from 3-5 to 7 mm, 15 min) producedspontaneous bursts at a frequency of about 03 Hz which rapidly disappeared afterreturn to control medium (about 5 min). Moreover, like KA, high-K+ ACSF inducedlong-lasting changes in the synaptic response since stimulation of mossy fibres,commissural pathways or perforant pathways evoked bursts for several hours afterreturn to control medium (n = 18, see Fig. 1 B).

377

Y. BEN-ARI AND M. GHO

The spontaneous and evoked bursts induced by high-K+ ACSF had similarcharacteristics to those induced by KA. Thus (i) the occurrence of the bursts wasindependent of the membrane potential (n = 3), (ii) the amplitude of thedepolarization underlying the burst was a linear function of the membrane potential(n = 2), (iii) the bursts were blocked by high-divalent-cation solution (n = 6) and thebursts had a similar laminar profile as KA-induced bursts (n = 5). These observationssuggest that the bursts induced by K+ or KA are generated by the same polysynapticcircuitry.

A a Control KA Wash b(200 nM, 3 min) (45 min) Wash

Mg2+: Ca2+,6: 4 mm 11 -60 mV A

Control

10mV b j10mV5 s 40 ms

B a , Wash

Control

Mg2+: Ca2+'-60 mV A

13:2mM

Fig. 10. Relationship between spontaneous and evoked bursts. Intracellular recordingsfrom the same cell in A and B. In high-divalent-cation ACSF, a brief application of KA(A) depolarized the neurone and reduced its input resistance but did not producespontaneous bursts. After KA was removed, mossy fibre stimulation in control ACSFfailed to evoke a burst (A b). Thirty minutes later (B), a second application of KA incontrol ACSF induced spontaneous and evoked bursts which persisted for over 1 h (Bb;45 min after wash of KA).

Long-lasting effect of kainic acid or high K+ involves NMDA receptorsThe involvement of N-methyl-D-aspartate (NMDA) receptors, a subclass of

glutamate receptors (Watkins & Evans, 1981), in neural plasticity, notably in thehippocampus, is well documented (see Collingridge, 1987). We have used D(-)-2-amino-5-phosphonovaleric acid (D-APV), a specific NMDA antagonist, to test thehypothesis that the ability of KA or high K+ to evoke a burst involves NMDAreceptors. As shown in Fig. I1 A, superfusion with high-K+ ACSF in the presence ofD-APV (30 ,UM) produced spontaneous synchronized bursts which rapidly dis-appeared 1-2 min after the D-APV-containing ACSF was restored. Stimulation of themossy fibres 20 min later evoked a slightly augmented EPSP but no burst. Thus infour experiments, D-APV blocked the long-lasting effect without preventing thespontaneous synchronized bursts. Furthermore, 15 min after washing with normalACSF (i.e. D-APV free) and while recording from the same neurone, a secondapplication of high-K+ ACSF induced typical spontaneous bursts. Stimulation of themossy fibres 30 min later now evoked a typical burst confirming that high-K+ ACSF

378

SUSTAINED EFFECTS OF KAINIC ACID IN CA3 379

A K (7 mM)7 min

20 min

-61 mVV 44 9 'A0j2

1 min

- -. - E - .. . 20 mV

- - lOOms

B K (7 mM)10 min

-1 - < Ij20 mV

A

A

100 ms

Fig. 11. High-K+ ACSF induced long-lasting synaptic changes which involve NMDAreceptors. A and B, records from the same cell. A, superfused with D-APV- (30,M)containing ACSF; stimulation of the mossy fibres is first shown to evoke a typical EPSP(left). The addition of high-K+ ACSF (7 mV) for 15 min (in the presence of D-APV) theninduced spontaneous bursts which disappeared a few minutes after return to the controlD-APV-containing ACSF solution. However, 20 min later in a control ACSF solution (i.e.without D-APV) stimulation of the mossy fibres evoked an EPSP but not a burst. B,when, 15 min after return to control medium (without D-APV), the K+ concentration wasagain increased (to 7 mm, for 15 min), spontaneous bursts were induced. Thirty minuteslater mossy fibre stimulation also evoked typical bursts indicating that in the sameneurone high-K+ ACSF had the ability to induce a long-lasting change in the evokedresponse. Note that the spontaneous burst was shorter in duration in the presence of D-APV.

has the ability to induce long-lasting change in the same neurone. Similarobservations were made with KA (n = 6). It perhaps should be stressed that once thelong-lasting changes had been elicited by KA, D-APV (30 gM) failed to block theevoked burst (n = 6, not shown) confirming previous studies using variousepileptogenic procedures which indicate that CA3 bursts once induced are notblocked by high concentration of NMDA antagonists (Anderson, Swartzwelder &Wilson, 1987; Cherubini et at. 1988). These observations therefore support the ideathat NMDA receptors are involved in the initiation of the long-lasting changes butnot in their maintenance.

DISCUSSION

The principal conclusion of the present work is that in addition to the spontaneousrecurrent bursts which are present during and shortly after KA or high-K+

Y. BEN-ARI AND M. GHO

application, there is a long-lasting modification of synaptic activity in CA3,characterized by the ability of electrical stimulation to evoke bursts several hoursafter returning to normal solution.

Principal features and mechanisms of bursts generation in CA3Spontaneous (and evoked) bursts are not endogenous (pacemaker) events but are

generated by a polysynaptic network. They fulfil all the criteria for network-drivenevents suggested by Johnston & Brown (1984), including the lack of correlationbetween their frequency (or probability) of occurrence and the membrane potential,the linear relationship between the amplitude of the depolarizing shift and themembrane potential (Fig. 2) and their susceptibility to procedures which blocksynaptic transmission (Fig. 3). The network-driven bursts induced by KA aretherefore reminiscent of those induced in CA3 by a variety of other convulsive agentsincluding pentylenetetrazol (Bingmann & Speckmann, 1986), (+)-tubocurarine(Johnston & Brown, 1984), the mast cell-degranulating peptide (Cherubini et al.1988), penicillin, bicuculline and picrotoxin (Wong & Traub, 1983), reduced Mg2+concentration (Walther et al. 1986; Neuman et al. 1987), application of NMDA(Walther et al. 1986; Neuman et al. 1987) or elevated K+ concentration (Rutecki etal. 1985; Korn et al. 1987, and the present study).The network in which the bursts are generated is clearly intrinsic to a relatively

small population of CA3 neurones. This is in keeping with the heterosynapticconvergence on the circuit (Fig. 6) and with the observation that KA, like picrotoxin(Wong & Traub, 1983), evokes spontaneous bursts in small islands of CA3 neuronesseparated by knife cuts (Y. Ben-Ari, unpublished observations). The laminar profileanalysis, like a number of previous studies using penicillin (Swann, Brady, Friedman& Smith, 1986), mast cell-degranulating peptide (Cherubini et al. 1988), NMDA orMg2+-free solutions (R. Neuman, E. Cherubini & Y. Ben-ari, personal communica-tion), indicates that the negative deflections, which presumably correspond to thelocation of current sinks (Mitzdorf, 1985), occur in the stratum radiatum (Fig. 5). Inthis region the recurrent collaterals between pyramidal neurones (Ramon y Cajal,1911) are known to synapse with the apical dendrites of neighbouring pyramidal cells(Miles & Wong, 1986). The present observations are therefore in keeping with earlierstudies which suggest that the pacemaker properties ofCA3 neurones are largely dueto the presence of recurrent excitatory collaterals between adjacent neurones (Traub& Wong, 1982).

Intracellular stimulation of a single CA3 neurone can evoke a network burst inslices in which excitability has been enhanced either by blockade of GABAA-mediated inhibition (Miles & Wong, 1983) or by activation of the NMDA receptor(Neuman et al. 1987). Furthermore, using intracellular recordings from pairs of CA3neurones, Miles & Wong (1987) have shown that blockade of GABAA-mediatedinhibition or tetanic stimulation of the mossy fibres can reveal polysynapticexcitatory pathways between two previously non-interacting cells. This type ofplasticity leads to the emergence of synchronous discharges in groups of CA3neurones by means of the recurrent excitatory collaterals. Since KA strongly reducesGABAA- and GABAB-mediated inhibitions (Fisher & Alger, 1984; Ben-Ari et al.1988) it is likely that KA generated the bursts by a similar mechanism.

380

SUSTAINED EFFECTS OF KAINIC ACID IN CA3

Mechanisms of long-lasting modification in synaptic responsesThe persistent ability to evoke a burst after a brief epileptic episode reflects a

change in the integrative properties of the circuitry which has, as it were, lost itscapacity to modulate the response to stimulation of a number of inputs. We considerthis an important point which deserves further investigation as it is this long-lastinghyperreactivity which may be responsible for epileptogenic activity.Our observations provide direct evidence that this long-lasting change is neither

due to a long-lasting change in intrinsic membrane properties of CA3 pyramidal cells(Fig. 7) nor to a persistent reduction in GABAergic inhibitions (Figs 8 and 9). It isalso unlikely to be due to a slow removal of the toxin causing a persistenthyperexcitable state nor to be specific to KA since other treatments includingelevated K+ (present study), (+ )-tubocurarine (Johnston & Brown, 1984; M. Gho &Y. Ben-Ari, unpublished observations), mast cell-degranulating peptide (Cherubiniet al. 1988) or trains of electrical stimulations in the stratum radiatum (Stasheff,Bragdon & Wilson, 1985) also induce a similar long-lasting effect. Severalobservations suggest that the long-lasting effects of KA or high K+ are related to thepresence of spontaneous bursts (e.g. Fig. 10). However, in a parallel study (Y. Ben-Ari, E. Cherubini & R. Neuman, personal communication) have shown exposure toMg2+-free ACSF or NMDA to trigger spontaneous bursts similar to those generatedby KA or high K+, but not the long-lasting hyperactivity after return to controlmedium. Therefore the spontaneous bursts are necessary but probably not sufficientto initiate the long-lasting change, i.e. other mechanisms are involved in the initiationprocess, perhaps a -block of repolarizing K+ conductances or GABA-mediatedinhibitions.Whatever the exact mechanism of the long-lasting effect, this is clearly reminiscent

of the long-term potentiation produced in CAI or fascia dentata by trains of high-frequency electrical stimuli (Bliss & Gardner-Medwin, 1973; Bliss & L0mo, 1973).Thus, in both processes there is no persistent change in intrinsic cell properties or inGABAergic inhibition (Griffith, Brown & Johnston, 1986; Abraham, Gustafsson &Wingstrom, 1987). Also, as in long-term potentiation, the long-lasting changeinduced by KA involves NMDA receptors. Thus D-APV blocks long-termpotentiation without reducing the control EPSP (Collingridge, Kehl & McLennan,1983; Harris, Ganong, & Cotman, 1984) and in the present experiments prevents thelong-lasting effects ofKA without blocking the spontaneous bursts or the mossy fibreEPSP.We are indebted to Dr E. Cherubini for his participation in some experiments and his suggestions

and criticism. We also thank Drs R. Neuman and K. Krnjevic for critical comments on themanuscript and S. Guidasci for photographic assistance. M. Gho was a fellow of Fondation Fyssen.Financial support from D.R.E.T. is acknowledged.

REFERENCES

ABRAHAM, W. C., GuSTAFSSON, B. & WINGSTRdM, H. (1987). Long-term potentiation involvesenhanced synaptic excitation relative to synaptic inhibition in guinea-pig hippocampus. Journalof Physiology 394, 367-380.

ALGER, B. E. (1984). Characteristics of a slow hyperpolarizing synaptic potential in rathippocampal pyramidal cells in vitro. Journal of Neurophysiology 52, 892-910.

381

382 Y. BEN-ARI AND M. GHO

ALGER, B. E. & NiCOLL, R. A. (1982). Feed-forward dendritic inhibition in rat hippocampalpyramidal cells studied in vitro. Journal of Physiology 328,105-123.

ANDERSON, W. W., SWARTZWELDER, H. S. & WILSON, W. A. (1987). The NMDA receptorantagonist 2-amino-5-phosphonovalerate blocks stimulus train-induced epileptogenesis but no

epileptiform bursting in the rat hippocampal slice. Journal of Neurophysiology 57,1-21.BEN-ARI, Y. (1986). Limbic seizure and brain damage produced by kainic acid: mechanisms and

relevance to human temporal lobe epilepsy. Neuroscience 14, 375-403.BEN-ART, Y., CHERUBINI, E. & GHo, M. (1987). Long-lasting induction of an evoked network burst

in rat hippocampal CA3 neurones in vitro. Journal of Physiology 388, 50P.BEN-ARI, Y., GHO, M. & ROVIRA, C. (1988). Kainic acid reduces both the baclofen induced outward

current and inhibitory postsynaptic current in rat hippocampal CA3 neurones in vitro. Journalof Physiology 398, 20P.

BERGER, M. & BEN-ART, Y. (1983). Autoradiographic visualization of[3H]kainic acid receptor

subtypes in the rat hippocampus. Neuroscience Letters 39, 337-342.BERRY, M. S. & PENTREATH, V. W. (1976). Criteria for distinguishing between monosynaptic and

polysynaptic transmission. Brain Research 105,1-20.BINGMANN, D. & SPECKMANN, E.-J. (1986). Actions of pentylenetetrazol (PTZ) on CA3 neurones

in hippocampal slices of guinea pigs. Experimental Brain Research 64, 94-104.BLISS, T. V. P. & GARDNER-MEDWIN, A. R. (1973). Long-lasting potentiation of synaptic

transmission in the dentate area of the unanaesthetized rabbit following stimulation of theperforant path. Journal of Physiology 232, 357-374.

BLISS, T. V. P. & L0Mo, T. (1973). Long-lasting potentiation of synaptic transmission in thedentate area of the anaesthetized rabbit following stimulation of the perforant path. Journal ofPhysiology 232, 331-356.

BROWN, D. A. & GRIFFITH, W. H. (1983). Calcium-activated outward current in voltage-clampedhippocampal neurones of the guinea-pig. Journal of Physiology 337, 287-301.

BROWN, T. H., FRICKE, R. A. & PERKEL, D. H. (1981). Passive electrical constants in three clasesof hippocampal neurones. Journal of Neurophysiology 46, 812-827.

BROWN, T. H. & JOHNSTON, D. (1983). Voltage-clamp analysis of mossy fibre synaptic input to

hippocampal neurones. Journal of Neurophysiology 50, 487-507.CHERUBINI, E., NEUMAN, R., ROVIRA, C. & BEN-ARI, Y. (1988). Epileptogenic properties of themast cell degranulating peptide in CA3 hippocampal neurones. Brain Research 445, 91-100.

CHERUBINI, E., ROVIRA, C., GHO, M. & BEN-ARI, Y. (1986). Effects of kainate on CAl hippocampalneurones recorded in vitro. In Excitatory Amino Acids and Seizure Disorders, ed. BEN-ARI, Y. &

SCHWARCZ, R., pp. 475-484. New York: Plenum Press.COLLINGRIDGE, G. (1987). The role of NMDA receptors in learning and memory. Nature 330,

604-605.COLLINGRIDGE, G. L., KEHL, R., Loo, R. & McLENNAN, H. (1983). Effects of kainic acid and otheramino acids on synaptic excitation in rat hippocampal slices. I. Extracellular analysis.Experimental Brain Research 52, 170-178.

COLLINGRIDGE, G. L., KEHL, S. J. & McLENNAN, H. (1983). Excitatory amino acids in synaptic,transmission in the Schaffer-commissural pathway of the rat hippocampus. Journal ofPhysiology334, 33-46.

FISHER, R. S. & ALGER, B. E. (1984). Electrophysiological mechanisms of kainic acid-inducedepileptiform activity in the rat hippocampal slice. Journal of Neuroscience 4, 1312-1323.

GHO, M., KING, A. E., BEN-ARi, Y. & CHERUBINI, E. (1986). Kainate reduces two voltage-dependent potassium conductances in rat hippocampal neurons in vitro. Brain Research 385,411-414.

GREEN, J. D. (1964). The hippocampus. Physiological Review 44, 561-608.GRIFFITH, W. H., BROWN, T. H. & JOHNSTON, D. (1986). Voltage-clamp analysis of synaptic

inhibition during long-term potentiation in hippocampus. Journal of Neurophysiology 55, 757-775.

HABLITZ, J. J. & JOHNSTON, D. (1981). Endogenous nature of spontaneous bursting in hippocampalpyramidal neurons. Cellular and Molecular Neurobiology 1, 325-334.

HABLITZ, J. J. & THALMANN, R. H. (1987). Conductance changes underlying a late synaptic

hyperpolarization in hippocampal neurons. Journal of Neurophysiology 58, 160-179.

382

SUSTAINED EFFECTS OF KAINIC ACID IN CA3 383

HALLIWELL, J. V. & ADAMS, P. R. (1982). Voltage-clamp analysis of muscarinic excitation inhippocampal neurons. Brain Research 250, 71-92.

HARRIS, E. W., GANONG, A. H. & COTMAN, C. W. (1984). Long-term potentiation in thehippocampus involves activation of N-methyl-D-aspartate receptors. Brain Research 323, 132-137.

,JOHNSTON, D. (1981). Passive cable properties of hippocampal CA3 pyramidal neurons. Cellularand Molecular Neurobiology 1, 41-55.

JOHNSTON, D. & BROWN, T. H. (1984). Mechanisms of neuronal burst generation. In Electro-physiology of Epilepsy, ed. SCHWARTZKROIN, P. A. & WHEAL, H. V., pp. 227-301. New York:Academic.

KORN, S. J., GIACCHINO, J. L., CHAMBERLIN, N. L. & DINGLEDINE, R. (1987). Epileptiform burstactivity induced by potassium in the hippocampus and its regulation by GABA-mediatedinhibition. Journal of Neurophysiology 57, 325-340.

LORENTE DE N6, R. (1934). Studies on the structure of the cerebral cortex. II. Continuation of thestudy of the ammonic system. Journal psychologie und neurologie 46, 113-177.

MACVICAR, B. A. & DUDEK, F. E. (1980). Local synaptic circuits in rat hippocampus: interactionsbetween pyramidal cells. Brain Research 184, 220-223.

MATSUMOTO, H. & AJMONE-MARSAN, C. (1964). Cortical cellular phenomena in experimentalepilepsy: Interictal manifestations. Experimental Neurology 9, 286-304.

MILES, R. & WONG, R. K. S. (1983). Single neurones can initiate synchronized populationdischarges in the hippocampus. Nature 306, 371-373.

MILES, R. & WONG, R. K. S. (1986). Excitatory synaptic interactions between CA3 neurones in theguinea-pig hippocampus. Journal of Physiology 373, 397-418.

MILES, R. & WONG, R. K. S. (1987). Latent synaptic pathways revealed after tetanic stimulationin the hippocampus. Nature 329, 724-726.

MITZDORF, U. (1985). Current source-density method and application in cat cerebral cortex:investigation of evoked potentials and EEG phenomena. Physiological Reviews 65, 37-100.

NEUJMAN, R., CHERUBINI, E. & BEN-ARI, Y. (1987). Network bursts triggered by single cellstimulation in the rat hippocampus. Society for Neuroscience Abstracts 13, 765.

NEWBERRY, N. R. & NICOLL, R. A. (1985). Comparison of the action of baclofen with y-aminobutyric acid in rat hippocampal pyramidal cells in vitro. Journal of Physiology 360, 161-185.

OLNEY, J. W., RHEE, X. & Ho, 0. L. (1974). Kainic acid: a powerful neurotoxic analogue ofglutamate. Brain Research 77, 507-512.

RAM6N Y CAJAL, S. (1911). Histologie du systeme nerveux de l'homme et des verte'bre's, vol. II. Paris:MIaloine.

REPRESA, A., TREMBLAY, E., SCHIOEVART, D. & BEN-ARI, Y. (1986). Development of high affinitykainate binding sites in human and rat hippocampi. Brain Research 384, 170-174.

ROBINSON, J. H. & DEADWYLER, S. A. (1981). Kainic acid produces depolarisation of CA3pyramidal cells in the in vitro hippocampal slice. Brain Research 221, 117-127.

RUTECKI, P. A., LEBEDA, F. J. & JOHNSTON, D. (1985). Epileptiform activity induced by changesin extracellular potassium in hippocampus. Journal of Neurophysiology 54, 1363-1374.

SCHWARTZKROIN, P. A. & WESTER, K. (1975). Long-lasting facilitation of a synaptic potentialfollowing tetanization in the in vitro hippocampal slice. Brain Research 89, 107-119.

STASHEFF, S. F., BRAGDON, A. C. & WILSON, W. A. (1985). Induction of epileptiform activity inhippocampal slices by trains of electrical stimuli. Brain Research 344, 296-302.

SWANN, J. W., BRADY, R. J., FRIEDMAN, R. ,J. & SMITH, E. J. (1986). The dendritic origins ofpenicillin-induced epileptogenesis in CA3 hippocampal pyramidal cells. Journal of Neuro-physiology 56, 1718-1738.

TRAUB, R. D. & WONG, R. K. S. (1982). Cellular mechanism of neuronal synchronization inepilepsy. Science 216, 745-747.

UNNERSTALL, J. R. & WAMSLEY, J. K. (1983). Autoradiographic localisation of high affinity(3H)kainic acid binding sites in rat forebrain. European Journal of Pharmacology 86, 361-371.

WA1,THER, H., LAMBERT, J. 1). (C., TJONES, R. S. G., HEINEMANN, U. & HAMON, B. (1986).Epileptiform activity in combined slices of the hippocampus, subiculum and entorhinal cortexduring perfusion with low magnesium medium. Neuroscience Letters 69, 156-161.

384 Y. BEN-ARI AND M. GHO

WATKINS, J. & EVANS, R. H. (1981). Excitatory amino acid neurotransmitters. Annual Review ofPharmacology and Toxicology 21, 165-204.

WESTBROOK, G. L. & LOTHMAN, E. W. (1983). Cellular and synaptic basis of kainic acid-inducedhippocampal epileptiform activity. Brain Re8earch 273, 97-109.

WILLIAMS, J. T., HENDERSON, G. & NORTH, R. A. (1984). Locus coeruleus. In Brain Slice, ed.DINGLEDINE, R., pp. 297-312. New York: Plenum Press.

WONG, R. K. S. & TRAUB, R. D. (1983). Synchronized burst discharge in disinhibited hippocampalslice. I. Initiation in the CA2-CA3 region. Journal of Neurophysiology 49, 442-458.