component of the slow depolarization underlying burst firing is

J. Physiol. (1986), 380, pp. 175-189 175With 7 text-figures

Printed in Great Britain

INVOLVEMENT OF N-METHYL-D-ASPARTATE RECEPTORS INEPILEPTIFORM BURSTING IN THE RAT HIPPOCAMPAL SLICE

BY RAYMOND DINGLEDINE, MARY A. HYNES AND GREGORY L. KING*From the Department of Pharmacology and Neurobiology Curriculum, University of

North Carolina School of Medicine, Chapel Hill, NC 27514, U.S.A.

(Received 22 November 1985)

SUMMARY

1. The effects of the N-methyl-D-aspartate (NMDA) receptor antagonist, D-2-amino-5-phosphonovaleric acid (D-APV), and other excitatory amino acid antagon-ists, were studied on CAI pyramidal neurones treated with picrotoxin or bicucullineto reduce synaptic inhibition mediated by y-aminobutyric acid (GABA). Under theseconditions epileptiform burst firing is readily produced by orthodromic stimulationof the pyramidal cell population.

2. D-APV reduced the plateau amplitude and duration of the depolarizationunderlying evoked and spontaneous bursts without affecting membrane potential,input resistance or the ability of the cell to fire a Ca2+ spike or a short train of Na+spikes.

3. A late component of the subthreshold excitatory post-synaptic potential(e.p.s.p.) was voltage dependent, being reduced in amplitude on membrane hyper-polarization. D-APV selectively removed this component ofthe e.p.s.p. in disinhibitedslices. In contrast, in the absence of GABA antagonists, D-APV had no noticeableeffect on the e.p.s.p. as studied with field potential recordings.

4. The concentration-response relationship of the inhibitory effect of D-APV andL-APV on population spike bursts was studied. The action of APV was highlystereoselective; the EC50 of D-APV was approximately 700 nm, whereas a similarinhibition required 540 /iM-L-APV. A number of other excitatory amino acid antag-onists were tested at a fixed concentration (100 ,M). Among them, the quisqualateantagonist y-D-glutamylaminomethyl sulphonic acid was ineffective against epilep-tiform bursts.

5. In the low nanomolar concentration range both D- and L-APV potentiatedbursting.

6. These results suggest that in the absence of GABAergic inhibition, a significantcomponent of the slow depolarization underlying burst firing is voltage dependent,synaptic in origin and mediated by NMDA receptors. We propose that, under normal(non-epileptic) physiological conditions, the balance between synaptic inhibitionmediated by GABA receptors and synaptic excitation mediated by NMDA receptorsmay modulate the excitability of pyramidal cell dendrites.

* Present address: Department of Physiology, Armed Forces Radiobiology Research Institute,Bethesda, MD 20814, U.S.A.

R. DINGLEDINE, M. A. HYNES AND G. L. KING

INTRODUCTION

Hippocampal pyramidal cells strongly depolarize, generate Ca2+ spikes and fire inbursts in response to N-methyl-D-aspartate (NMDA) and its analogues. The apparentvoltage dependence of NMDA responses (Hablitz, 1982; MacDonald, Porietis &Wojtowicz, 1982; Dingledine, 1983; Flatman, Schwindt, Crill & Stafstrom, 1983)appears to be due to depolarization-induced removal of Mg2+ block from a conven-tional cation channel (Nowak, Bregestovski, Ascher, Herbet & Prochiantz, 1984;Mayer, Westbrook & Guthrie, 1984). Responses to NMDA can be antagonized bya variety of drugs, including the highly selective compound D-2-amino-5-phosphonovaleric acid (D-APV) (Davies, Francis, Jones & Watkins, 1981; Colling-ridge, Kehl & McLennan, 1983a; Jones, Smith & Watkins, 1984). Due to thefortunate existence of selective antagonists and their unusual blockade by Mg2+,NMDA receptors are currently the best characterized of the several excitatory aminoacid receptors.Although NMDA receptors do not appear to mediate conventional synaptic

excitation in the hippocampus (Koerner & Cotman, 1982; Collingridge, Kehl &McLennan, 1983b; Crunelli, Forda & Kelly, 1983), recent experiments have begunto reveal a possible synaptic function for these receptors. It has been suggested thatAPV can prevent long-term potentiation (Collingridge et al. 1983b) and reducesynaptic excitation in the absence of Mg2+ (Coan & Collingridge, 1985). Theparticipation ofNMDA receptors in the genesis of some seizures has been suggested,based on the anticonvulsant action of NMDA antagonists (Croucher, Collins &Meldrum, 1982; Ryan, Hackman & Davidoff, 1984).Thus the concept is emerging that NMDA receptors may serve roles in pathological

or plastic forms of synaptic transmission. Our work has focused on a possible rolefor NMDA receptors in epileptiform bursting. We show here that in the disinhibitedhippocampus a component of synaptic excitation is selectively and potently blockedbyNMDA receptor antagonists. It appears that part of the depolarization underlyingthe epileptiform burst is a synaptic potential mediated by NMDA receptors. Wesuggest that dendritic excitability may normally be finely tuned by the balancebetween GABAergic inhibition andNMDA receptor activation, and that epileptiformbursts could reflect one extreme in this balance. Preliminary accounts of this workhave appeared (Hynes & Dingledine, 1984; King & Dingledine, 1985 a, b).

METHODS

Procedures for preparing and maintaining hippocampal slices, for intracellular and extracellularrecording from CAI pyramidal neurones, for electrical stimulation of afferent fibres and forionophoresis ofN-methyl-DL-aspartate (NMA) were as described previously (Dingledine, 1983; King,Knox & Dingledine, 1985). The construction and interpretation of field potential input-outputcurves have been described (Valentino & Dingledine, 1982). Two types of recording chamber wereemployed. A conventional interface chamber was used for intracellular recording and forconstructing input-output curves (Fig. 4), whereas a submersion chamber (Nicoll & Alger, 1981),because of its faster fluid exchange time, was used for the pharmacological studies (Figs. 6 and 7)and some ofthe intracellular recordings. The perfusion rate was 0 5 ml/min in the interface chamberand 1P67 ml/min in the submersion chamber. In the submersion chamber drug effects appearedwithin 30 s of changing solution and reached a plateau by 2-3 min; the respective times for the

176

NMDA RECEPTORS AND INTERICTAL BURSTING

interface chamber were 4-6 and 12-20 min. Temperature was held at 32-350 in the interfacechamber and 29-30 0C in the submersion chamber. In some experiments drugs were applied bypressure ejection in a small (approximately 1 nl) droplet of bathing medium onto the slice surface.

Intracellular data were gathered from forty-eight pyramidal neurones in the CAI region. Theinput resistance of the sampled population was 38 + 1-5 MQ (mean+ S.E. of mean; n = 41) and themembrane potential was -69+16 mV (n = 41). Micropipettes were filled with 2 M-potassiummethylsulphate or 2 M-CsCl plus 05 M-tetraethylammonium bromide. In some experiments50 mM-QX-314, a quaternary lidocaine derivative (Dahlbom, Misiorny & Truant, 1965), wasincluded in the pipette solution to block Na+ spikes. Most experiments were performed with10-100 uM-bicuculline, 100,M-picrotoxin or, in a few cases, 2 mM-morphine, in the bathingmedium. Under these conditions GABAergic inhibition is greatly reduced and orthodromicstimulation elicits typical epileptiform burst firing.The intensity of evoked burst firing was quantified by measuring differences in the total linear

length of the population spike burst wave form, a method similar in principle to a coast-line lengthmeasurement. Two cursors were positioned visually to delineate the portion of the burst to bemeasured (see Fig. 6A), and a PDP-1 1/23 computer was used to calculate and display the 'coast-linebursting index' after each stimulus as a function of real time (e.g. Fig. 6B). The cursors werepositioned to measure all but the first two population spikes, since only the late part of the burstwas sensitive to antagonists. Other cursors were used simultaneously to measure the peak-to-peakamplitude of the first population spike and (from a separately positioned micropipette) the slopeof the field e.p.s.p. The effect of excitatory amino acid antagonists on population spike bursts wasexpressed as the ratio of coast-line burst indices in the presence and absence of drug. In theseexperiments a razor-blade chip was used to cut and remove the CA2-CA3 region of most slices.This prevented rebound excitation of CAI cells by antidromic activation ofCA3 cells; spontaneousinterictal-like bursts were absent in this preparation.The coast-line bursting index offers a measure of the intensity of an individual burst. The index

will increase if each neurone fires more spikes in a burst, if more neurones participate in a burst,or if cell synchrony is increased; these conditions result in more or larger population spikes. Anadvantage of this measure over a simple algebraic summation of population spike amplitudes isthat it eliminates subjective decisions about whether to include just-threshold spikes in theanalysis. Neither method, however, provides information concerning the precise mechanismunderlying a change in burst intensity.

Material8The following compounds were obtained from Tocris Chemicals (Buckhurst Hill, U.K.): D-APV,

L-APV, DL-APV and y-D-glutamylaminomethyl sulphonic acid (GAMS); from Cambridge ResearchBiochemicals (Harston, Cambs. U.K.): glutamic acid diethylester (GDEE), bicuculline metho-bromide, y-D-glutamylglycine (DGG) and DL-a, e-diaminopimelic acid, DL-APV; from Aldrich(Milwaukee, WI, U.S.A.): CsCl (gold label); from Sigma (St. Louis, MO, U.S.A.): tetrodotoxin,tetraethylammonium bromide (TEA), NMDA, picrotoxin and most other common chemicals.QX-314 was a generous gift from Dr Bertil Takman (Astra Pharmaceuticals).

RESULTS

D-APV attenuates burstingBlockade of GABAergic inhibition by perfusion of slices with GABA antagonists

resulted in characteristic burst firing in response to stratum radiation stimulation.Bursts in CAI pyramidal neurones consisted of a train of three to eight actionpotentials riding on a large (up to 40 mV) depolarizing wave, or paroxysmaldepolarizing shift (p.d.s.). Occasional spontaneous bursts, quite similar in wave formto synaptically evoked bursts, also occurred in disinhibited slices. Addition of8-5-85 ,sM-D-APV or -DL-APV to the tissue by either perfusion or droplet rapidlydecreased the plateau amplitude and duration of the p.d.s. (n= 13). Fig. 1 showsintracellular recordings of evoked and spontaneous epileptiform bursts before and

177


after application of 85 /LtM-D-APV in a small droplet onto the surface of the slice.The reduction in plateau amplitude could be so great that the spike inactivation inthe latter half of the burst was relieved (Fig. 1). Often, however, the number of spikesevoked by a constant stimulus was decreased, concomitant with a reduction in thedepolarization by D-APV or DL-APV (e.g. Fig. 2). The reduction in burst duration

A Evoked C Spontaneous

Controltro

B D

Fig. 1. Reduction in amplitude of p.d.s. wave form by D-APV. Bursts occurringspontaneously (C) or evoked by 0 1 Hz stimulation of Schaffer-commissural path (A) wererecorded in the presence of 100t /M-picrotoxin. After pressure ejection of a droplet ofD-APV (85 AM) onto the slice surface the amplitude of the depolarization underlying bothtypes of burst was decreased (B and D). The results from two different cells are shown,with resting membrane potentials of -61 mV (A and B) and -63 mV (C and D) and inputresistances of 40 MW (A and B) and 48 Mt (C and D).

and plateau amplitude by D-APV or DL-APV could be reversed by washing. Similarresults have recently been reported by Herron, Williamson & Collingridge (1985).D-APV and DL-APV reduced burst duration without affecting other indices of

pyramidal cell excitability. Input resistance, measured with small hyperpolarizingcurrent pulses, was 103 + 3% of control in the presence of 8-5-85 /,M-DL-APV or-D-APV (n = 23 cells). Likewise, membrane potential was not changed significantly(+ 1-6+09 mV from rest, n = 31). The reduction in burst duration did not appearto be due to a fall in soma excitability per se, since DL-APV abbreviated theepileptiform burst without impairing the ability of neurones (n = 6) to fire a shorttrain of Na+ spikes during a depolarizing current pulse (Fig. 2). The effect of DL-APVon Ca2+ spikes triggered alternately by strong depolarizing current pulses or briefionophoretic applications of NMA was studied after blocking Na+ spikes with1 ftM-tetrodotoxin. As shown in Fig. 3, perfusion with 8-5 ,SM-DL-APV blocked theCa2+ spike evoked by NMA (n = 9) but had no discernible effect on the slow spikeelicited directly by depolarizing current (n = 6). This suggests that DL-APV andD-APV are not acting simply as Ca2+ spike blockers to reduce the p.d.s. In ten

178


neurones, very prolonged depolarizations (up to 800 ms) were triggered by briefcurrent pulses after allowing the K+-channel blockers Cs+ and TEA to diffuse intothe cell from the recording pipette. D-APV (85 fSM in a droplet) did not affect theselarge, intrinsic depolarizations (not shown).The possibility that D-APV or DL-APV might reduce conventional excitatory

A1 Synaptic 2 Direct

81 DL-APV8 5 pM 2

\̂ ~~~~~50mrnVan

25 mns

Fig. 2. DL-APV abbreviates evoked bursts without enhancing accommodation. In 10 UM-bicuculline orthodromic stimuli were alternated with depolarizing transmembrane currentpulses to produce an epileptiform burst (Al) or a short train of action potentials (A2),respectively. Perfusion with 8-5 /SM-DL-APV reduced burst duration (Bi) but had no effecton the ability of the cell to fire a train of spikes (B2). Each panel shows traces from threesuperimposed trials.

transmission was explored with dendritic field-potential recordings. Separate micro-pipettes were positioned extracellularly in the cell layer to record population spikes,and in the apical dendritic layer to record the presynaptic fibre volley and fieldexcitatory post-synaptic potential (e.p.s.p.). Input-output curves were constructedby varying the stimulus intensity and plotting the slope of the field e.p.s.p. as afunction of fibre volley amplitude, and the population spike amplitude as a functionofthe field e.p.s.p. slope. These studies were done in the absence ofGABA antagonistsso that the input-output curves could be examined over their full range, withoutcontaminating population spike bursts. As illustrated in Fig. 4A, 85 /LM-DL-APV or-D-APV had no effect on the size of the field e.p.s.p. triggered by a given fibre volley(n = 6 slices). This confirms previous reports that concentrations of APV which areselective for NMDA receptors do not affect normal synaptic transmission in the CA1region (Collingridge et al. 1983b; Crunelli et al. 1983), although Hablitz & Langmoen(1986) recently reported that a higher concentration ofDL-APV (200 /LM) could cause

179

R. DINGLEDINE, M. A. HYNES AND 0. L. KING

a slight reduction in the field e.p.s.p. Fig. 4B demonstrates that the population spikeproduced by a given field e.p.s.p. was also unchanged by DL-APV.

It appears from these results that D-APV and DL-APV may have a selectiveinhibitory action on epileptiform bursts evoked in disinhibited slices, since passiveand active intrinsic membrane properties were unchanged and synaptic activation

Al Control 2

B 1 DL-APV 2

20 mV

100 ms

NMAFig. 3. Effect of DL-APV on Ca2+ spikes. In tetrodotoxin (1 pM) Ca2+ spikes were elicitedalternately by injection of depolarizing current pulses (AI) or brief ionophoretic pulsesof N-methyl-DL-aspartate (NMA; - 50 nA) near the apical dendrites (A2). Perfusionwith 8-5 /ZM-DL-APV reduced the underlying depolarization and blocked the spikes evokedby NMA (B2), but not the voltage-dependent spikes produced by direct depolarizationof the cell (Bi). Each panel shows the superimposed responses from five trials.

of pyramidal cells in the absence of GABA antagonists was unaffected. As shownbelow, however, e.p.s.p.s evoked in the presence ofGABA antagonists were diminishedby D-APV.

Dependence on membrane potentialUnder most conditions the depolarizing action of NMDA is strongly voltage

dependent, being increased by depolarization and reduced by hyperpolarizationaround the resting potential (MacDonald et al. 1982; Dingledine, 1983; Mayer et al.1984). The work of Nowak et al. (1984) indicates that this voltage dependence arisesfrom a channel blocking action of extracellular Mg2+. If the p.d.s. evoked in theabsence of GABAergic inhibition is partly due to the synaptic activation of NMDAreceptors, one might expect the synaptic potential underlying the p.d.s. itself, as wellas the action of D-APV, to exhibit voltage dependence in a similar fashion.

In these experiments 50 mM-QX-314 was included in the micropipette fillingsolution in order to block Na+ spikes and allow e.p.s.p.s to be studied over a widerange of membrane potentials. After reducing GABAergic inhibition with picrotoxin,application of 8-5 /ZM-D-APV by perfusion or droplet (n = 12) rapidly reduced the

180

NMDA RECEPTORS AND INTERICTAL BURSTING 181

amplitude of a subthreshold e.p.s.p. (arrow in Fig. 5B). A similar observation hasalso been reported by Hablitz & Langmoen (1986). The magnitude of the effect ofD-APV was quite variable and depended on the stimulus intensity, shape of thecontrol e.p.s.p., and membrane potential. Fig. 5A demonstrates that in some cellsa late component of the evoked synaptic potential was voltage dependent, being

A 10 B

2 oIT*0

E0 b~i °d'> i' e *0s°0L 8

E oc 08

0-~~~0o , , 0

O3 1 2 3 4 0 1"2

Fig.~ ~ 084.DLAPEfeco_____on_______ndsoatcfildpoenilsinthasnco__

0~~~~~~~~~

00 12 4 0 1 2

Volley (mV) Field e.p.s.p. (mV/ins)

Fig. 4. Effect Of DL-APV on dendritic and somatic field potentials in the absence ofGABAantagonists. A stimulating cathode in stratum radiatum was used to activate afferentfibres synapsing on the apical dendrites of CAI pyramidal cells. An extracellularmicropipette placed 'on-beam' in the apical dendritic field recorded the afferent fibrevolley and the subsequent slow negativity (field e.p.s.p.) taken to represent the magnitudeof the post-synaptic current. The peak-to-peak amplitude of the fibre volley, and theinitial slope of the field e.p.s.p. were measured. A separate pipette placed in stratumpyramidale recorded population spikes. By varying the stimulus current, input-outputcurves were constructed that related the slope of the field e.p.s.p. to the amplitude of thefibre volley (A) and the population spike to the field e.p.s.p. (B). Perfusion with85 /,M-DL-APV had no effect on either relationship.

reduced in amplitude as the cell was hyperpolarized by current injection. Theinhibition by D-APV was particularly prominent in these cells and was much lessapparent at hyperpolarized potentials; in effect D-APV selectively blocked thevoltage-sensitive component of the e.p.s.p. In some cells the e.p.s.p. half-width wasalso reduced by D-APV (e.g. Fig. 5A).The voltage dependence of D-APV action is summarized in Fig. 5C. For each of

eight cells a plot was made of e.p.s.p. amplitude as a function of membrane potential,before and after perfusion with 8'5 /SM-D-APV. The percentage reduction in e.p.s.p.amplitude by D-APV was interpolated at fixed (5 mV) intervals of membranepotential. The average change in e.p.s.p. amplitude for these eight neurones was thenplotted as a function of shift in membrane potential from rest (Fig. 5C). Theeffectiveness of D-APV was diminished at hyperpolarizing potentials (P < 002),

182 1. DINGLEDINE, M. A. HYNES AND G. L. KING

A C

100 815 /M-D-APV

-52 mV0

0

2575

58mV I rmv~~~~~~~~~~~~

ECU,5

-75mV/

B 5mV25 ms

25--40 -20 0 20

/ ~ ~~~~~~~~~~mV from resting potential

Fig. 5. Dependence on membrane potential of e.p.s.p. and D-APV effect. E.p.s.p.s wererecorded intracellularly in the presence of 100 IM-picrrotoxin to reduce i.p.s.p.s. In allcases the micropipette contained 50 mm-QX-314 to raise spike threshold. A, e.p.s.p.sevoked by weak orthodromic stimuli at three membrane potentials produced by passingsteady current through the recording pipette. Perfusion with 85 ,aM-D-APV reduced theamplitude and duration of e.p.s.p.s (arrows indicate responses in the presence of D-APV).The saw-toothed appearance of some responses is due to population spike bursts seen bythe intracellular recording pipette. B, D-APV (8-5 /uM) reduced the amplitude of an e.p.s.p.(arrow) evoked by an orthodromic stimulus subthreshold for a population spike. Thevertical calibration bar in B is 2 5 mV. C, reduction of e.p.s.p. amplitude by D-APV isweakly voltage dependent. In each of eight pyramidal cells a plot of e.p.s.p. amplitudeas a function of membrane potential was constructed before and after perfusion with85 JM-D-APV. The data were normalized by, interpolating e.p.s.p. values at discrete(5 mV) intervals of membrane potential above and below the resting potential. Then thee.p.s.p. amplitude in the presence of D-APVX expressed as a percentage of its controlamplitude, was plotted as a function of membrane potential shift from rest. The meanresting potential of these eight cells was - 69 mV. Each bar shows the mean + s.E. of meanat a given membrane potential. The regression of D-APV effect on membrane potentialwas significant (/' < 0 02). The apparent voltage-dependent action of n)-APV was probablydlue to the voltage dependence of the e.p.s.p. itself (A) in the presence of 100 /u.M-picrotoxin.

which reflects the loss of a voltage-sensitive component of the e.p.s.p. (Fig. 5A).Although the dependence of this action of D-APV on membrane potential was seenin every cell, the magnitude of the effect was quite variable and did not appear asmarked as that of NMT)A-evoked depolarizations (Dingledine, 1983). This ispresumably due to the composite nature of the e.p.s.p. in disinhibited slices. Thatis. a D-APV-sensitive e.p.s.p. appears superimposed on a (conventional e.p.s.p., andthe relative contribution of these two components of the e.p.s.p. was variable fromcell to cell.


A

v

B0*026 pM

I r-

Wash

2mV

10 ms

856Mm

I* Ar0 "SPS 1%000, 0 01- 0 0 '60 %Q-160, 0 4r,* 0 0 9 WV 0 0

as0 too16;z: 'o-C*0 .$ I of

69, 0SP0

li

10 20 30

Time (min)

Fig. 6. Concentration-dependent biphasic effect of D-APV on bursting. A, sample recordsof population spike bursts evoked by 01 Hz orthodromic stimulation in the presence of100 /zM-picrotoxin. This slice was perfused sequentially with control solution (A1),0-026 /LM-D-APV (A2), control solution (not shown), 85,uM-D-APV (A3) and controlsolution (A4). The amplitude of the last three population spikes within the burst wasincreased by 0'026 ,uM-D-APV but decreased by 85 /SM-D-APV. Each trace represents theaverage of ten responses. To quantify the intensity of evoked bursts the total length ofthe line representing the late part of the burst wave form was measured between thecursors shown in Al and 2. This procedure is similar to making a coast-line lengthmeasurement. B, the value of the coast-line bursting index for each evoked response,determined as in A, was plotted as a percentage of control base line over time during theexperiment. In this instance, base line was taken as the mean value before and afterperfusion with 0-026 ,/M-D-APV. D-APV was present in the perfusion solution for thetimes indicated by the open bars.

PharmacologyThe action of DL-APV on NMDA receptors is stereoselective, with the D-isomer

being much more potent that the L-isomer (Evans, Francis, Jones, Smith & Watkins,1982; Davies & Watkins, 1982). We compared the potency of the two isomers forattenuating evoked bursts by measuring the total length of the wave form in thelatter part of the burst, as described in the Methods. Only slices that generated atleast five population spikes were used in these studies. When tested in this mannerthe effect of D-APV was, surprisingly, biphasic. The coast-line bursting index was

enhanced at nanomolar concentrations but attenuated at micromolar concentrations

183

160x*0.' 2 1200 c.' o

w 80

400 60


(Fig. 6A). Typically the last two or three population spikes were affected by D-APV,with no discernible effect on the amplitude of the first spike until the concentrationwas raised to 1 mm (Fig. 7B; see also Fig. 4B). Simply changing control bathingmedium had no effect on the coast-line bursting index (101 + 2 % of control, n = 28).As shown in Fig. 6B, D-APV effects developed quickly and were entirely and rapidlyreversible when the concentration was less than 10-30 /M, but were only slowly andpartially reversible at concentrations of 85 fM or more.

A B

160 160

120 02

B0 6is+oi Kel

, 0 lie 4- - -4 -C)

.C ~~~~~~~~~~~~~0.80 V 80

Fg7.Conetainrsos uvsfrteefcCof -P n -P neietfo L-APV 0 L-APV* D-APV T *D-APV

0 0I-10 -8 -6 -4 -10 -8 -6 -4

log [APVJ (M) log [APV] (m)Fig. 7. Concentration-response curves for the effect of D-APV and L-APV on epileptiformbursts. Both coast-line bursting index (A) and the amplitude of the first population spikes(B) are expressed as percentage of control value in response to sequentially increasingconcentrations of drug. Plotted values represent the means+s.E. of means from six toeleven slices. The CA2-CA3 region was removed from these slices to prevent activationof this region by the Schaffer-commissural stimulus, which could in turn have re-excitedCAI cells. All studies were done in the presence of 100 /M-picrotoxin in a submersionchamber.

Full dose-response curves for D-APV and L-APV against evoked epileptiformbursts are shown in Fig. 7A. After an initial stabilization period, each slice wasexposed to sequentially increasing (up to ten) concentrations of an antagonist, withperiodic wash-outs during the experiment. Drug effects were expressed as percentageof the control coast-line bursting index, as illustrated in Fig. 6B. If it is assumed thatthe facilitation and depression of bursting are independent phenomena that summatealgebraically, the EC50 for D-APV as an inhibitor of bursting was found byinterpolation to be approximately 700 nM. L-APV was about 760 times less potent,with an equivalent effect occurring at 540 gLM. The EC50 of D-APV versus burstingshown here is similar to the KD of this compound for NMDA receptors determinedin frog spinal cord (1I4-2-4 #M; Davies et al. 1981; Evans et al. 1982; Jones et al. 1984).

In spite of its marked effects on bursting, D-APV had no effect on the amplitude

184


of the first population spike at concentrations up to 100 /M. L-APV, on the otherhand, produced a small (10 00) but consistent increase in the first population spike atall concentrations below 100 /M (Fig. 7B). At a concentration of 1 mm, both D-and L-APV had non-specific depressant effects on the first population spike andcoast-line bursting index (Fig. 7).A large number of compounds have been tested in the isolated frog spinal cord for

their ability to block responses to NMDA and other excitatory amino acids (Jones

TABLE 1. Attenuation of evoked epileptiform bursts by excitatory amino acid antagonistsCoast-line

bursting index 1st pop. spikeAntagonist (100 1M) (% control) (% control) n

D-2-Amino-5-phosphonovaleric acid (D-APV) 67-8 + 9-8* 100-9 + 2-0 9DL-2-Amino-5-phosphonovaleric acid (APV) 78-8 + 5-4t 99-9 + 1-6 7L-2-Amino-5-phosphonovaleric acid (L-APV) 1111 + 8-2 112 1+4-7t 7y-D-Glutamylglycine 92-0+ 6-3 93-4+3-2 7z,e-Diaminopimelic acid 100-9+9-4 94-7 +3-3 7

L-Glutamic acid diethyl ester (GDEE) 100 1 + 3-6 96-1 + 3-1 7y-D-Glutamylaminomethyl sulphonic acid (GAMS) 100-5 + 4-4 98-8 + 2-4 7

Surgically isolated CAI regions were perfused with 10-100 /tM-bicuculline or -picrotoxin toinduce bursting, and Shaffer-commissural afferents were stimulated at 0.1 Hz at an intensitysufficient to evoke at least five population spikes. After a period of stable recording, slices wereperfused with a single concentration (100 /M) of one antagonist. The changes in evoked burstintensity and amplitude of the first population spike were measured as described in Methods.Values represent the means+ 1 S.E. of means of the control bursting index, for n slices tested.

* P < 0-02 by two-tailed t test.t P < 0-01 by two-tailed t test.$ P < 0 05 by two-tailed t test.

et al. 1984). D-APV was reported to be the most selective antagonist of NMDAreceptors, and GAMS the least. We tested the ability of seven excitatory amino acidantagonists to attenuate evoked bursting in disinhibited hippocampal slices perfusedwith a constant concentration (100 /M) of an antagonist. The effect on coast-linebursting index was measured from plots similar to that in Fig. 6B. The results inTable 1 show that D-APV was the most active, whereas compounds less active atNMDA receptors, GAMS and GDEE, had no effect at 100 fM. These results areconsistent with the involvement of NMDA receptors in the expression of bursts (cf.Jones et al. 1984), although quantitative studies of antagonist potency are neededto substantiate this idea. Collingridge et al. (1983 a) also reported the relative potencyfor antagonism of NMDA responses in hippocampal slices to be DL-APV> DGG > GDEE, consistent with our results.

DISCUSSION

Role for NMDA receptor in burst firingFrom the known actions of NMDA on hippocampal neurones, one might predict

that a physiological role for NMDA receptors would involve a voltage-dependentsynaptic depolarization that is blocked by micromolar concentrations of the NMDA

185


receptor antagonist, D-APV. These criteria are met in disinhibited hippocampalslices. The high potency of D-APV for attenuating bursts, the stereospecificitydemonstrated, and the relative lack of effect of GAMS argue for NMDA receptormediation of a component of bursting. The wave form underlying the p.d.s. appearsto be a combination of an early depolarization that is relatively insensitive to D-APVand a later component that is voltage dependent and blocked by D-APV. The earlycomponent probably results simply from a combination of a large conventionale.p.s.p. and the explosive triggering of intrinsic voltage-dependent currents, bothsubsequent to i.p.s.p. blockade. The later, more labile depolarizing phase seems tobe due to the unmasking ofsynaptically activatedNMDA receptors. Both componentsare epileptiform in nature because they are associated with the synchronous dischargeof the CA1 pyramidal cell population.By this hypothesis, the p.d.s. consists of two synaptic potentials plus a depolari-

zation triggered intrinsically. The conventional e.p.s.p., unfettered by an i.p.s.p.,would depolarize the membrane above threshold for the intrinsic currents. The initialdepolarization that results moves the membrane potential into the range at whichthe blocking action of Mg2+ on NMDA channels (Nowak et al. 1984) is reduced. Thedepolarization resulting from NMDA receptor activation then adds to and prolongsthe p.d.s. A similar scheme has been suggested by Herron et al. (1985). Evidence fromother laboratories favours the idea that e.p.s.p.s contribute heavily to the wave formunderlying the p.d.s. (Johnston & Brown, 1981; Miles, Wong & Traub, 1984). Indeed,the observations that D-APV had no noticeable effect on a variety of active andpassive membrane properties, and that a p.d.s.-like potential cannot be triggered inhealthy CA1 cells by depolarization with injected current, argue against an exclusiverole for intrinsic conductances in the expression of the p.d.s. One argument againsta major synaptic component of the p.d.s. has been that its duration (approximately150-250 ms) is far longer than that of the normal e.p.s.p. (see Alger, 1984). However,if the late phase of the p.d.s. is a slow, NMDA-receptor-mediated e.p.s.p. that isvirtually suppressed under normal physiological conditions, this difficulty vanishes.Although the simplest explanation of our results, and those of others (Coan &

Collingridge, 1985; Herron et al. 1985), is based on a reduction of Mg2+ block ofpost-synaptic NMDA channels by unusually large dendritic depolarizations, alter-native explanations might be raised that involve a possible block of presynapticinhibition by GABA antagonists. One consequence of such a scheme could be therelease of a novel transmitter that selectively activates post-synaptic NMDAreceptors. Cuenod, Do & Herrling (1985) have recently reported that one such agonist,L-homocysteic acid, is present in rat hippocampus and can be released by elevatedK+. Another consequence could be increased release of the normal transmitter, whichmight recruit extrasynaptic NMDA receptors or increase the concentration oftransmitter in the synapse to the point at which NMDA receptors are activated.Given that presynaptic inhibition mediated by GABAA receptors has not beendemonstrated in the hippocampus, we prefer a mechanism based on the well-knownvoltage-dependent block of NMDA channels by Mg2+.

186


Role for NMDA receptors in normal synaptic transmission?Although most synaptic responses in the central nervous system appear resistant

to NMDA antagonists, or are blocked only at relatively high concentrations (e.g.Koerner & Cotman, 1982; Miles et al. 1984; Hablitz & Langmoen, 1985), synapticresponses in other preparations are selectively reduced by Mg2+ or low concentrationsof D-APV and are thus presumed to be mediated by NMDA receptors (Ikeda &Sheardown, 1982; Thomson, West & Lodge, 1985; Corradetti, King, Nistri, Rovira& Sivolotti, 1985). It has been estimated that approximately 800 of the glutamatebinding sites in CAI stratum radiatum represent NMDA receptors (Greenamyre,Olson, Penney & Young, 1985). An important issue raised by our studies is that, ifpyramidal neurones possess NMDA receptors that can be activated synaptically,what mechanism suppresses their expression under normal (non-epileptic) conditions?An answer may lie in the voltage dependence of NMDA potentials. We suggest thatthe hyperpolarization and conductance shunt accompanying the i.p.s.p. may normallyhinder the dendritic membrane from depolarizing into the range of membranepotentials at which the Mg2+ block of NMDA channels would begin to abate. Thetransmitters) released by Schaffer-commissural afferents could normally bind toboth NMDA and non-NMDA receptors, but the degree of inhibitory tone wouldregulate the expression of synaptic potentials mediated by NMDA receptors. Thecombination of dendritic GABAergic inhibition and synaptic excitation mediated byNMDA receptors could then balance each other to finely tune dendritic excitability,epileptiform bursts representing an extreme imbalance.

In the view outlined above, NMDA receptors may normally provide an amplifi-cation mechanism for the conventional e.p.s.p. Recent studies have demonstratedthat a voltage-dependen-t potential located on or near dendritic spines couldmodulate the efficacy of e.p.s.p.s (Miller, Rall & Rinzel, 1985; Perkel & Perkel, 1985).It is interesting in this regard that one form of synaptic plasticity in the hippocampus,long-term potentiation, is both facilitated by loss of inhibition (Wigstrom &Gustafsson, 1985) and prevented by NMDA antagonists (Collingridge et al. 1983b).Further, the tetanic stimulus trains used to evoke long-term potentiation canproduce a transient reduction of GABAergic inhibition (McCarren & Alger, 1985).Thus, it seems that the same process that contributes to epileptiform discharges may,in milder form, underlie long-term potentiation.

We thank Drs Catherine Rovira, Barry Pallotta and Stephen Korn for comments on themanuscript. This work was supported by NINCDS grants NS17771, NS22249 and NS06953, andby NIDA grant DA02360.

REFERENCES

ALGER, B. E. (1984). Hippocampus. Electrophysiological studies of epileptiform activity in vitro.In Brain Slices, ed. DINGLEDINE, R., pp. 155-199. New York: Plenum Press.

COAN, E. J. & COLLINGRIDGE, G. L. (1985). Magnesium ions block an N-methyl-D-aspartatereceptor-mediated component of synaptic transmission in rat hippocampus. Neuroscience Letters53, 21-26.

COLLINGRIDGE, G. L., KEHL, S. J. & McLENNAN, H. (1983 a). The antagonism ofamino acid-inducedexcitations of rat hippocampal CAI neurones in vitro. Journal of Physiology 334, 19-34.

187


COLLINGRIDGE, G. L., KEHL, S. J. & McLENNAN, H. (1983b). Excitatory amino acids in synaptictransmission in the Schaffer collateral-commissural pathway of the rat hippocampus. Journalof Physiology 334, 33-46.

CORRADETTI, R., KING, A. E., NISTRI, A., ROVIRA, C. & SIVILOTTI, L. (1985). Pharmacologicalcharacterization of D-aminophosphonovaleric acid antagonism of amino acid and synapticallyevoked excitations on frog motoneurones in vitro: an intracellular study. British Journal ofPharmacology 86, 19-25.

CROUCHER, M. J., COLLINS, J. F. & MELDRUM, B. S. (1982). Anticonvulsant action of excitatoryamino acid antagonists. Science 216, 899-901.

CRUNELLI, V., FORDA, S. & KELLY, J. S. (1983). Blockade of amino acid-induced depolarizationsand inhibition of excitatory post-synaptic potentials in rat dentate gyrus. Journal of Physiology341, 627-640.

CUENOD, M., Do, K. Q. & HERRLING, P. L. (1985). Release of sulfur-containing excitatory aminoacids related to synaptic transmission. Epilepsia 26, 504.

DAHLBOM, R., MISIORNY, A. & TRUANT, A. (1965). Quaternary derivatives of aminoacylanilines.Acta pharmacologica suecica 2, 213-218.

DAVIES, J., FRANCIS, A. A., JONES, A. W. & WATKINS, J. C. (1981). 2-Amino-5-phosphonovalerate(2APV), a potent and selective antagonist of amino acid-induced and synaptic excitation.Neuroscience Letters 21, 77-81.

DAVIES, J. & WATKINS, J. C. (1982). Actions of D and L forms of 2-amino-5-phosphonovalerate and2-amino-4-phosphonobutyrate in the cat spinal cord. Brain Research 235, 378-386.

DINGLEDINE, R. (1983). N-methyl aspartate activates voltage-dependent calcium conductance inrat hippocampal pyramidal cells. Journal of Physiology 343, 385-405.

EVANS, R. H., FRANCIS, A. A., JONES, A. W., SMITH, D. A. & WATKINS, J. C. (1982). The effectsof a series of omega-phosphonic alpha-carboxylic amino acids on electrically evoked and excitantamino acid-induced responses in isolated spinal cord preparations. British Journal ofPharmacology75, 65-75.

FLATMAN, J. A., SCHWINDT, P. C., CRILL, W. E. & STAFSTROM, C. E. (1983). Multiple actions ofN-methyl-D-aspartate on cat neocortical neurons in vitro. Brain Research 266, 169-173.

GREENAMYRE, J. T., OLSON, J. M. M., PENNEY, J. B. & YOUNG, A. B. (1985). Autoradiographiccharacterization of N-methyl-D-aspartate-, quisquilate- and kainate-sensitive glutamate bindingsites. The Journal of Pharmacology and Experimental Therapeutics 233, 254-263.

HABLITZ, J. J. (1982). Conductance changes induced by DL-homocysteic acid and N-methyl-DL-aspartic acid in hippocampal neurons. Brain Research 247, 149-153.

HABLITZ, J. J. & LANGMOEN, I. A. (1986). NMDA receptor antagonists reduce synaptic excitationin the hippocampus. Journal of Neuroscience 6, 102-106.

HERRON, C. E., WILLIAMSON, R. & COLLINGRIDGE, G. L. (1985). A selective N-methyl-D-aspartateantagonist depresses epileptiform activity in rat hippocampal slices. Neuroscience Letters 61,255-260.

HYNES, M. A. & DINGLEDINE, R. (1984). Attenuation of epileptiform burst-firing in the rathippocampal slice by antagonists of N-methyl-D-aspartate receptors. Neuroscience Abstracts 10,229.

IKEDA, H. & SHEARDOWN, M. J. (1982). Aspartate may be an excitatory transmitter mediatingvisual excitation of 'sustained' but not 'transient' cells in the cat retina: ionophoretic studiesin vivo. Neuroscience 7, 25-36.

JOHNSTON, D. & BROWN, T. H. (1981). Giant synaptic potential hypothesis for epileptiformactivity. Science 211, 294-297.

JONES, A. W., SMITH, D. A. S. & WATKINS, J. C. (1984). Structure-activity relations of dipeptideantagonists of excitatory amino acids. Neuroscience 13, 573-581.

KING, G. L. & DINGLEDINE, R. (1985a). Micromolar and nanomolar concentrations of D-2-amino-5-phosphonovalerate have opposite effects on epileptiform burst-firing in the rat hippo-campal slice. Neuroscience Abstracts 11, 103.

KING, G. L. & DINGLEDINE, R. (1985b). Role for N-methyl-D-aspartate receptors in epileptiformbursting? Epilepsia 26, 509.

KING, G. L., KNOX, J. J. & DINGLEDINE, R. (1985). Reduction of inhibition by a benzodiazepineantagonist, Rol5-1788, in the rat hippocampal slice. Neuroscience 15, 371-378.

KOERNER, J. F. & COTMAN, C. W. (1982). Response of Schaffer collateral-CAl pyramidal cellsynapses of the hippocampus to analogues of acidic amino acids. Brain Research 251, 105-115.

188


MCCARREN, M. & ALGER, B. E. (1985). Use-dependent depression of i.p.s.p.s in rat hippocampalpyramidal cells in vitro. Journal of Neurophysiology 53, 557-571.

MACDONALD, J. F., PORIETIS, A. V. & WOJTOWICZ, J. M. (1982). L-aspartic acid induces a regionof negative slope conductance in the current-voltage relationship of cultured spinal cordneurons. Brain Research 237, 248-253.

MAYER, M. L., WESTBROOK, G. L. & GUTHRIE, P. B. (1984). Voltage-dependent block by Mg2+ ofNMDA responses in spinal cord neurones. Nature 309, 261-263.

MILES, R., WONG, R. K. S. & TRAUB, R. D. (1984). Synchronized after-discharges in the hippo-campus: contribution of local synaptic interaction. Neuroscience 12, 1179-1189.

MILLER, J. P., RALL, W. & RINZEL, J. (1985). Synaptic amplification by active membrane indendritic spines. Brain Research 325, 325-330.

NICOLL, R. A. & ALGER, B. E. (1981). A simple chamber for recording from submerged brain slices.Journal for Neuroscience Methods 4, 153-156.

NOWAK, L., BREGOSTOVSKI, P., ASCHER, P., HERBET, A. & PROCHIANTZ, A. (1984). Magnesiumgates glutamate-activated channels in mouse central neurones. Nature 307, 462-465.

PERKEL, D. H. & PERKEL, D. J. (1985). Dendritic spines: role of active membrane in modulatingsynaptic efficacy. Brain Research 325, 331-335.

RYAN, G. P., HACKMAN, J. C. & DAVIDOFF, R. A. (1984). Spinal seizures and excitatory aminoacid-mediated synaptic transmission. Neuroscience Letters44, 161-166.

THOMSON, A. M., WEST, D. C. & LODGE, D. (1985). An N-methylaspartate-mediated synapse inrat cerebral cortex: a site of action of ketamine? Nature 313, 479-481.

VALENTINO, R. J. & DINGLEDINE, R. (1982). Pharmacological characterization of opioid effects inthe rat hippocampal slice. Journal of Pharmacology and Experimental Therapeutics 223, 502-509.

WIGSTR6M, H. & GuSTAFSSON, B. (1985). Facilitation of hippocampal long-lasting potentiation byGABA antagonists. Acta physiologica scandinavica 125, 159-172.

189

component of the slow depolarization underlying burst firing is

Documents