Receptor-Mediated Inhibition of the N-Methyl-O-Aspartate Component of Synaptic Transmission in the Rat Hippocampus

The Journal of Neuroscience, January 1991, 17(l): 203-209

GABA,-Receptor-Mediated Inhibition of the N-Methyl-o-Aspartate Component of Synaptic Transmission in the Rat Hippocampus

Richard A. Morrisett,1-5 David D. Mott,2 Darrell V. Lewis,3 H. Scott Swartzwelder,i,4.5 and Wilkie A. Wilson1-*v5 Departments of ‘Medicine (Neurology), *Pharmacology, 3Pediatrics (Neurology) and Neurobiology, and 4Psychology, Duke University Medical Center, Durham, North Carolina 27710. and 5Neurology Research Laboratory, Veterans Administration Medical Center, Durham, North Carolina 27705

GABA receptor regulation of NMDA-receptor-mediated syn- operated channelsare regulated in a voltage-dependentmanner aptic responses was studied in area CA, of the rat hippocam- by Mg2+ (Mayer and Westbrook, 1984; Nowak et al., 1984; pus using extracellular and intracellular recording tech- Herron et al., 1986). Becauseof the development of specific niques. Picrotoxin (PTX) was used to suppress GABA, non-NMDA receptor antagonistssuch as 6,7-dinitroquinoxa- inhibition and 6,7-dinitroquinoxaline-2,3-dione (DNQX) was line-2,3-dione (DNQX; Drejer and Honor& 1988) the dem- used to suppress non-NMDA receptor-mediated responses. onstration of NMDA-receptor-mediated synaptic potentialshas In this manner, we were able to avoid the complicating fac- been facilitated (Blake et al., 1988; Andreasenet al., 1989; Da- tors caused by potentials induced by other excitatory and vies and Collingridge, 1989; Iambert and Jones, 1989). When, inhibitory amino acid receptors. Under these conditions, large in the presenceof DNQX, synaptic inhibition is reduced or NMDA-receptor-mediated EPSPs were observed. When blocked, one can record largedepolarizing responsesto NMDA paired stimuli were given at interstimulus intervals from 100 receptor activation (Andreasenet al., 1989; Davies and Collin- to 400 msec, powerful inhibition of the second response was gridge, 1989;Lambert and Jones, 1989;Morrisett et al., 1990b). observed. This inhibition was reversed by the GABA, an- We have recently utilized excitatory amino acid antagonists tagonists phaclofen and P-hydroxy-saclofen; it was also de- in combination with the GABA, antagonist picrotoxin (PTX) pressed by removal of Mg2+ from the bath. Examination of to study NMDA-mediated synaptic potentials (NMDA EPSPs) non-NMDA receptor-mediated synaptic responses (deter- in area CA, of the rat hippocampus in the presenceof Mg2+ mined in the presence of D-2-amino-5-phosphonovalerate (Morrisett et al., 1990a,b). Very large, long-lasting NMDA-re- and PTX) showed no such inhibition, thereby supporting the ceptor-mediated synaptic potentials were recorded in the pres- hypothesis that GABA, inhibition of NMDA EPSPs is post- ence of thesedrugs. Indeed, the prominent effect of PTX upon synaptic. This difference in paired-pulse inhibition of NMDA the expression of NMDA EPSPs dramatically points up the and non-NMDA EPSPs leads us to conclude that there was apparent role of GABA,-mediated synaptic inhibition in reg- no evidence of GABA,-mediated presynaptic inhibition of ulating the Mg2+ block of NMDA-operated channels.We now excitatory transmitter release. Intracellular recordings in the report that NMDA-mediated synaptic potentials recorded in presence of DNQX and PTX revealed a phaclofen-sensitive dendritic and somatic layers of areaCA, of the rat hippocampus late IPSP that correlated in time with the period of inhibition exhibited marked paired-pulsedepression. Investigation of this of NMDA responses. Taken together, these data suggest phenomenonrevealed that GABA, receptors were responsible that paired-pulse-inhibition of NMDA responses is produced for paired-pulse inhibition of NMDA EPSPs. By this mecha- by a GABA,-receptor-mediated hyperpolarization of the nism, GABA, receptors could participate in the regulation of postsynaptic membrane, causing an enhanced block of the the activity of NMDA-receptor-operated channelsand, hence, NMDA channels by Mg2+. Regulation of NMDA-mediated syn- the development of neuronal plasticity. aptic responses by GABA, receptors constitutes a powerful A preliminary report of these findings has been submitted mechanism for control of a major excitatory system in hip- (Morrisett et al., 1990a). pocampal pyramidal cells. Materials and Methods Induction of neuronal plasticity in many systemsis dependent Male Sprague-Dawley rats (1 W-250 gm) were killed under CHCI, an- upon activation of the NMDA subtype of excitatory amino acid esthesia. The hippocampi were rapidly removed, and transverse slices receptor (Harris et al., 1984; Morris et al., 1986; Collingridge 625 wrn thick were prepared. After the slices were allowed to equilibrate for 1 hr in artificial cerebrospinal fluid (ACSF; containing, in mM, NaCl, and Bliss, 1987;Morrisett et al., 1989). However, in many cases, 120; KCl, 3.3; NaH,PO,, 1123; NaHCO,, 25; CaCl,, 1.8; MgSO,, 1.2; the involvement of NMDA receptorsin normal synaptic trans- dextrose, 10: continuallv gassed with 95% 0,:5% CO,: oH. 7.4: 32°C). mission is minimal becausethe inward currents via NMDA- a midtemporal slice was transferred to a su&fusion*~hamber’(which contained 1.0 mM MgSO, and 3.0 mM CaCl, in the ACSF unless oth- erwise indicated). Monophasic, constant current stimulus pulses (100 Received May 22, 1990; revised Aug. 10, 1990; accepted Sept. 5, 1990. psec, 100-1000 PA) were generated with a Dagan S-900 stimulator and This work was supported by Grants NS 17701 and AA 07207 and by the were applied through a tungsten stimulus electrode to the Schaffer col- Veterans Administration. R.A.M. is an NRSA fellow (MH 15 I77- 13). laterals in the stratum radiatum of area CA,. Correspondence should lx addressed to Richard Morris&t, Ph.D., Building 16, Extracellular population field potentials were recorded from the stra- Room 25, Veterans Administration Medical Center, Durham, NC 27705. tum radiatum and stratum pyramidale of area CA, using glass micro- Copyright 0 1991 Society for Neuroscience 0270-6474/91/010203-07$03.00/O electrodes filled with 150 mM NaCl (l-3 MQ; Fig. 1). Extracellular 204 Morrisett et al. * GABA,, Inhibition of NMDA-Mediated Synaptic Responses

DENDRITIC SOMAE SOMA,

ACSF

Figure 1. Isolation of NMDA-mediated responses in aTea CA,. Population responses recorded from the dendritic DNQX/ and somatic (extracellular) electrodes 1 T I’ (SOMA,) are shown in the first 2 col- PTX umns. The third column shows responses from the intracellular electrode (SOMA,). In the top row, responses re- J corded in the presence of normal ACSF L.f r"- ----IL are presented. In the bottom 2 rows, responses recorded in the presence of the non-NMDA antagonist DNQX ( 10 PM) and the GABA, antagonist PTX (10 DNQX/ FM) are presented. In the bottom row, 1 the effect of the NMDA receptor an- PTX/ tagonist D-APV (50 PM) upon the D-APV r NMDA EPSP induced by DNQX/PTX I 1.2 - ‘O is shown. Stimulus artifacts were omit- mV L I mV L ted for clarity. 20 mrsc 20 mmc

recording from the stratum radiatum in ACSF typically elicited responses (see Fig. 1) consisting of a negative-going presynaptic fiber Results volley, immediately followed by a negative-going population field potential, reflecting the dendritic excitatory postsynaptic potential. Si- The top row of Figure 1 depicts typical responsesrecorded from multaneous extracellular recording from the stratum pyramidale re- dendritic and somatic sitesin area CA,. Although not shown in sulted in a negative-going orthodromic population spike superimposed Figure 1, when the non-NMDA receptor antagonist DNQX (10 upon a positive-going EPSP. Slices were monitored until stable input- PM) was added to the bath, postsynaptic responseswere com- output curves were established. A slice was rejected if the maximal dendritic EPSP recorded from the stratum radiatum was of 1.0 mV pletely abolished. As shown in the middle row of Figure 1, when amplitude or less. the GABA, channelantagonist PTX ( 10 PM) wasadded in com- Intracellular recordings were made from the stratum pyramidale of bination with DNQX to the bath, excitatory post synaptic po- CA, using microelectrodes filled with 4 M K+ acetate (approximately tentials were recorded from the dendritic extracellular electrode. 100 Ma). Intracellular recording typically elicited positive-going EPSPs and action potentials, followed by early and late IPSPs. In every case, Thesedepolarizations (at maximal stimulusintensity) were about CA, pyramidal cells were electrophysiologically identified by stimula- the same amplitude as the control EPSP but of much greater tion of the alveus and intracellular recording of antidromic action po- duration; thus, the area under the NMDA EPSP was approxi- tentials. Neurons were considered acceptable for intracellular recording mately 3 times that of the control EPSP. if all of the following criteria were met: (1) The input resistance was The excitatory responsesrecorded in the combination of greater than 30 Ma, (2) the action-potential spike height was greater than 50 mV, and (3) the resting membrane potential was less than -60 DNQX and PTX were presumedto be initiated by activation mV. A total of 11 neurons was utilized for this study. of NMDA receptors. NMDA EPSPswere accompanied by re- To elicit the NMDA component of synaptic transmission (NMDA petitive cell firing in both individual cells and in the population EPSP) in the presence of Mg*+, 10 PM DNQX and 10 PM picrotoxin response.Addition of the NMDA receptor antagonist D-APV were added to the ACSF. The response typically became stable within 30-45 min, and input-output curves were obtained for the NMDA to the bath containing DNQX and PTX almostcompletely abol- EPSPs and intracellular responses. ished the EPSP recorded extracellularly and blocked cell firing. Extracellular dendritic waveforms were analyzed on a Compaq 386s In separateexperiments, theseresponses have been demonstrat- computer using Waveform Basic software (Blue Feather Software for ed to be almost completely blocked by the NMDA channel Nicolet Corp., Madison, WI). A program, written in this laboratory in antagonist MK-80 1 (10 MM; data not shown).For thesereasons, Waveform Basic language was used to calculate the area under the extracellular dendritic responses potentials. This value, in units of mV. the responserecorded from the dendritic electrode in the pres- msec, is assumed proportional to the charge entering the population of ence of DNQX/PTX was termed the NMDA EPSP. neurons detected by the recording electrode. In all extracellular exper- Non-NMDA EPSPs were also studied. These EPSPs were iments, every measurement was made at least 3 times and the mean obtained in the presenceof D-APV (50 PM) and PTX (10 PM). calculated (1 slice/animal). All drugs and compounds were from Sigma (St. Louis, MO), except DNQX, D-amino-5phosphonovalerate (D-APV), These non-NMDA responseswere not significantly different 2-hydroxy-saclofen, and phaclofen were from Tocris Neuramin (Essex, from control (normal ACSF) responses,except that they were England). prolonged about 30%, and that somatic responsesincluded re- The Journal of Neuroscience, January 1991, 1 I(1) 205

petitive cell firing, which was never seen in control responses. NON-NMDA EPSPs NMDA EPSPs Non-NMDA responses were blocked completely by DNQX (10 A -/-lr---- PM). 50 ms We first compared paired stimulation of NMDA and non- iili w-- In? NMDA EPSPs as demonstrated in Figure 2A. There were marked 1 100 ms differences seen when pairing NMDA EPSPs (Fig. 2A, left). At i 1 50 msec between stimuli, there was prominent paired-pulse w potentiation of the NMDA EPSPs. However, the second re- 200 ms :/----I IT--- sponsemarkedly decreasedin amplitude and duration at inter- i i stimulus intervals between 100 and 400 msec. Maximal paired- pulse inhibition was noted at 200 msec. This inhibitory effect was independent of stimulus intensity and could be produced from both distant and nearby stimulation siteswithin the stratum radiatum of CA, relative to the recording electrode in the stratum radiatum. --l/-- 17-- 1.0 ---r-- There was no apparent paired-pulseinhibition of non-NMDA l-- EPSPs at any interstimulus interval (Fig. 2A, right). In fact, at V intervals of 50 and 100 msec, the secondresponse was slightly (30%) potentiated. B The effect of varying the interstimulus interval on non-NMDA 200 - and NMDA EPSPs is compiled in Figure 2B. Paired-pulse potentiation of both NMDA and non-NMDA EPSPsis shown at intervals of 50 and 100 msec,while profound inhibition of only 150 - NMDA EPSPsoccurred at the 200-msecinterval. These2 curves d are significantly different by Kruskal-Wallis (p < 0.00 1). F The time courseof paired-pulse inhibition of NMDA EPSPs Too follows closely the known time courseof the late IPSP induced 0 by activation of the GABA, receptor subtype (Alger, 1984; Du- n tar and Nicoll, 1988a). We hypothesized that addition of the 50 - GABA, receptor antagonist phaclofen to the bath would reverse the 200-msecinhibition of the NMDA component of synaptic transmission. Figure 3 showsthe effect of phaclofen (667 PM) 0 I I I 1 1 on NMDA EPSPsand somatic responsesrecorded both intra- 0 50 100 200 400 800 1600 cellularly and extracellularly (left and middle columns). Inhi- INTERSTIMULUS INTERVAL (msec) bition of the NMDA EPSP recorded at 200 msec was almost completely abolished in the presenceof phaclofen, so that the Figure 2. A, Paired stimulation of NMDA and non-NMDA EPSPs. Responses are dendritic EPSPs induced by DNQX/PTX treatment size of each NMDA EPSP was equivalent. (NMDA EPSPs) on the leftpaneland APV/PTX treatment (non-NMDA Intracellular recordings in Figure 3 (right column) are shown EPSPs) on the right panel. Paired-pulse potentiation (50-100 msec) and at the reversal potential for the GABA, IPSP at the soma. This inhibition (100400 msec) of NMDA EPSPs was apparent, while non- allowed us to demonstrate in the clearestmanner the ability of NMDA EPSPs were only slightly enhanced by pairing. Note that the duration of NMDA EPSPs was much longer than non-NMDA EPSPs. phaclofen to prevent the 200-msec inhibition of the NMDA B, Time course of effect of pairing stimuli on NMDA and non-NMDA EPSPs.Note that, at this membranepotential, the GABA, IPSP, EPSPs. Inhibition of NMDA EPSPs was maximal at 200 msec, non- as recorded from the somatic site, is not hyperpolarizing; how- NMDA EPSPs displayed slight increases in response to pairing of stim- ever, the NMDA EPSP and its paired-pulse inhibition are still uli. Data are mean k SEM; n = 5-10 slices; curves significantly different seen.Presumably, the dendritic site of the NMDA EPSP is quite by Kruskal-Wallis; p -c 0.00 1. electrically remote from the cell body, and thus, hyperpolarization of the soma by current injection does not sufficiently hy- perpolarize the dendrites to suppressthese processes (see Dis- resistancechanges, if significant, would affect both responses. cussion).[Phaclofen also reversedthe inhibition of the NMDA Taken together, thesedata strongly suggestthat changesin input EPSPswhen recorded at resting membrane potentials (data not resistanceduring the GABA, IPSP are not sufficient to explain shown).] the paired-pulse inhibition of NMDA EPSPs. The input resistanceof several cells in DNQX/PTX was mea- The dose-responserelationship of the phaclofen block was sured by current injection at membrane potentials equivalent studied. Figure 4A depicts the effect of varying phaclofen con- to the peak of the late IPSP and during the late IPSP itself (data centrations (333-1000 PM) upon paired-pulse inhibition of not shown). The input resistanceduring the late IPSP was only NMDA EPSPsinduced at varying stimulusintensities. The abil- slightly less(20%) than the resistancetested at membrane po- ity of phaclofen to block inhibition of the NMDA EPSPs was tentials equivalent to the peak ofthe late IPSP. This conductance dose related. Also, in the presenceof phaclofen, there was a changealso occursfollowing non-NMDA EPSPs,yet no paired- significant relationship between stimulus intensity and the de- pulse inhibition of these EPSPs was demonstrated. Becauseit gree of paired-pulseinhibition of NMDA EPSPs(as opposedto is likely that non-NMDA- and NMDA-receptor-operated chan- the control response).The low concentration of phaclofentested nels occupy sites very close to one another (Monaghan and (333 PM) decreasedinhibition at low stimulus intensities by Cotman, 1985; Cotman et al., 1987) one would expect that about SO%, while at maximal stimulus intensities, inhibition 206 Morrisett et al. * GABA., Inhibition of NMDA-Mediated Synaptic Responses

DENDRITIC SOMAE SOMA,

CONTROL

v” +-T:,

PHACLOFEN Figure 3. Paired-pulseinhibition of NMDA EPSPsrecorded from dendritic and somatic(extracellular SOMA,; intracellular,SOMA,) layersof CA, was blockedby the GABA, receptorantag- onistphaclofen (667 PM). The intracellular data wererecorded at the reversal WASH potential for the GABA, IPSP (-92 mv) for clarity. Phaclofenalmost completely blockedpaired-pulse inhibition at 200 msecin a mannerthat wasre- versibleupon washout of the drug.An mV identical result was obtained in the lLyy-r-mV i!iY-pbmV 60 mrec presenceof 2-hydroxy-saclofen. 60 mrec 60 mrec

was completely blocked. Higher concentrations of phaclofen bition by digital subtraction of the first 2 recordings, thus dem- (667 PM, 1 mM) had more pronounced stimulus dependencies. onstrating that GABA,-mediated inhibition was present. At lower stimulus intensities (~600 MA), these concentrations Becausethe NMDA channel is increasingly blocked by Mg2+ of phaclofenproduced potentiation of the secondNMDA EPSP. asthe membraneis hyperpolarized (Mayer and Westbrook, 1984; However, at higher stimulus intensities (> 600 PA), there was Nowak et al., 1984), we hypothesized that the paired-pulsein- no potentiation or inhibition of paired responses(see Discus- hibition of the NMDA EPSPresulted from the hyperpolarizing sion). In all cases,the maximum areasof the first NMDA com- effectsof GABA, inhibition. If this were the case,then removal ponent measuredin the presenceof phaclofen did not differ of Mg2+ from the ACSF would suppressthe paired-pulse inhi- significantly from control. However, there wasa slight rightward bition of the NMDA EPSP, becausethe Mg2+-dependentblock shift (approximately 30%)in the input-output relationshipcaused would be prevented. Therefore, we studied the effect of varying by phaclofen. the concentration of Mg2+ upon NMDA EPSPs in comparison The effect of phaclofen was always reversible within 15 min to non-NMDA EPSPs. (Fig. 4,4, wash). Furthermore, the more potent GABA, antag- Figure 5, A and B, depicts the sensitivity of NMDA and non- onist 2-hydroxy-saclofen (200 PM) completely reversed inhibi- NMDA EPSPs to Mg*+ and showsthat the paired-pulse inhi- tion of the NMDA EPSPs(n = 4). In a separatebut related set bition of NMDA EPSPswas dependent upon Mg2+.As indicated ofexperiments, the effect of phaclofen(667 PM) upon non-NMDA in Figure 5A, the NMDA EPSPs were extremely sensitive to EPSPswas studied. In no casedid the GABA, antagonist alter Mg2+.As the concentration of Mg*+ was increasedfrom no added the amplitude or duration of non-NMDA EPSPs,either in single Mg2+ to 3.16 mM, NMDA EPSPs decreasedin area approxi- stimulation or in pairs over various interstimulus intervals (50- mately lo-fold (from 200 to 20 mV.msec). The apparent me- 1600 msec; n = 4). dian effective concentration (EC,,) of the NMDA component The block of paired-pulse inhibition by GABA, antagonists to Mg2+ was approximately 500 PM. Non-NMDA EPSPs were suggeststhat a late IPSP mediated by GABA, receptors might minimally altered by addition of Mg2+. be responsiblefor suppressionof the paired response.Therefore, Figure 5B shows an example of the effect of varying Mg2+ we felt that direct examination of the late IPSP following NMDA concentration upon the 200-msecinhibition. Mg2+appeared to EPSPswas necessary.Figure 4B depicts intracellular recordings inhibit the 200-msec responsemore readily than the first re- from CA, pyramidal cells in the presenceof DNQX/PTX. In sponse.This difference in Mg*+ sensitivity wasespecially evident the control responsefollowing the EPSP, a prolonged hyper- at 500 FM and 1 mM Mg*+, where the 200-msecNMDA EPSPs polarization of approximately 12 mV is present. Addition of were reducedalmost completely, while the initial responseswere phaclofen (667 PM) partially antagonized the late IPSP, leaving still quite large.These data are presentedcumulatively in Figure the EPSPand action potential unchangedexcept for a small shift SC, showing the Mg2+ dependenceof the 200-msecinhibition. in the input-output relationship. The effect of phaclofen on the The size of the second responserelative to the first was dose late IPSP was completely reversible upon washout. The final related to Mg*+, with an apparent EC,, of approximately 500 column of Figure 4B shows the phaclofen-sensitive late inhi- PM Mg2+. In the absenceof added Mg*+, there is apparently The Journal of Neuroscience, January 1991, If (1) 207

. 667 l 1000

A 0 I I I I 0.1 1.0 2.0 3.16 210 3.;6 200 400 600 600 1000 [Mg++l, mM STIMULUS INTENSITY IpA B B0.0 cI”--‘-- l.o-p-- O.’-v 2.0v---- *I i/+-‘ CONTROL PHACLOFEN WASH LATE IPSP

10 J mv 100 Figure 5. A, Mg*+ inhibits NMDA but not non-NMDA EPSPs. Dose msec response of the effect of Mg*+ (0.0-3.16 mM) upon NMDA and non- NMDA EPSPs in area CA,. Data are mean + SEM; n = 5-12 slices. B, Paired-pulse inhibition ofNMDA EPSPs is Mg*+ dependent (numbers Figure 4. A, Dose response of antagonism of paired-pulse inhibition at Ieji of tracings are [MgZ+lba,,,). All NMDA EPSPs were powerfully of NMDA EPSPs by phaclofen at different stimulus intensities. In con- inhibited by Mg2+, but the inhibition of the paired response at 200 msec trol media (DNQX/PTX), pairing of stimuli at various intensities elic- occurred at much lower concentrations of Mg*+. For example, note that ited inhibition of approximately 75%. Addition of phaclofen to the the responses in 0.1 mM Mg2+ are almost identical, while in 1 mM Mg*+, media resulted in a dose-dependent reduction (333-1000 PM) in paired- there was a large decrease in the paired response at 200 msec. C, MgZ+ pulse inhibition. Biphasic effects of phaclofen were observed to be de- dependence of paired-pulse inhibition of NMDA EPSPs. Inhibition of pendent upon stimulus intensity. At low intensity, phaclofen tended to NMDA EPSPs was approximately doubled in the presence of 1 mM induce paired-pulse potentiation, while at greater intensities, no effect Mg2+ relative to responses recorded in the absence of Mg*+. Increasing of pairing of stimuli was seen. In every case, the effects were reversible. the concentration of Mg*+ above 1 mM reduced the apparent inhibition Data are mean & SEM; n = 4-8 slices. B, Phaclofen-sensitive late IPSP of NMDA EPSPs due to the reduction of the size of the initial response. follows NMDA-mediated EPSP and action potentials in pyramidal cell Data are mean & SEM: n = 4-8 slices. of area CA,. Control response was recorded in DNQX/PTX; note the prominent IPSP following EPSP, with no early IPSP. Addition of phaclofen (667 PM) resulted in an almost complete block of late IPSP, with inhibition, (3) the mechanismof the paired-pulse inhibition of no effect upon excitatory response. Response to phaclofen was reversible NMDA EPSPs appearsto depend upon activation of GABA, upon washout. Late IPSP is the digital subtraction of the response recorded in phaclofen from the control response indicating the degree of receptorsbecause phaclofen and 2-hydroxy-saclofen reversedit antagonism of the late IPSP in this cell by phaclofen. Membrane po- in a dose-dependentmanner at pharmacologically effective con- tential, -67 mV. This response was consistent in 5 of 5 cells tested in centrations, and (4) the mechanismof paired-pulse inhibition 5 separate slices. appearsto be dependent upon Mg2+ block of the NMDA-receptor-channelcomplex, becausedependence of inhibition upon MgZ+ could be demonstrated. sufficient residual Mgz+ to block NMDA channelsto a significant extent. Above 1 mM Mg2+, no further inhibition could be in- Reversal of inhibition of NMDA EPSPs with GABA, duced, primarily becauseall NMDA EPSPswere largely blocked antagonists at those concentrations of Mg*+. Evidence indicates that late IPSPs are mediated by activation of GABA, receptors (Newberry and Nicoll, 1984; Dutar and Discussion Nicoll, 1988a,b; Soltesz et al., 1988; Lambert et al., 1989). The The primary resultsfrom theseexperiments include the follow- late IPSP is likely due to GABA, receptor regulation of a K+ ing: (1) NMDA EPSPs induced by DNQX/PTX exhibit sensi- channel in a G-protein-dependent mechanism(Andrade et al., tivity to competitive and noncompetitive NMDA antagonists, 1986; Dutar and Nicoll, 198813;Nicoll, 1988). Phaclofen and (2) NMDA EPSPs exhibit paired-pulse inhibition of a partic- 2-hydroxy-saclofen are specificantagonists of GABA, receptors ularly specific time frame (between 100 and 400 msec, peak at (Dutar and Nicoll, 1988; Hasuo and Gallagher, 1988; Kerr et 200 msec)while non-NMDA EPSPs displayed no paired-pulse al., 1989). 208 Morrisett et al. - GABA., Inhibition of NMDA-Mediated Synaptic Responses

GABA, antagonists had potent effects upon paired-pulse in- postsynaptic receptors were separately innervated, with presyn- hibition of NMDA EPSPs, while non-NMDA EPSPs were not aptic terminals of only the NMDA system containing GABA, changed. The relative potencies of these antagonists in blocking receptors, then paired-pulse inhibition of only the NMDA EPSPs paired-pulse inhibition of NMDA EPSPs agrees well with pre- would be explained. Nevertheless, it is well known that acti- viously published studies (Hasou and Gallagher, 1988; Kerr et vation of presynaptic GABA, receptors with exogenous agonists al., 1989). As seen in Figure 4A, it was readily apparent that potently inhibits responses due to release of endogenous excit- higher concentrations of phaclofen (667 and 1000 PM) fully atory amino acids (Hill and Bowery, 198 1; Dutar and Nicoll, blocked inhibition at higher stimulus intensities. However, at 1988a,b). Thus, while there may be evidence for GABA,-re- lower stimulus intensities, paired-pulse potentiation was receptor-mediated presynaptic regulation ofglutamate release, the vealed by the highest concentration of phaclofen. This stimulus process does not occur to a significant extent following a single intensity dependence may be due to several mechanisms: (1) At stimulation in this model. higher stimulus intensities, the increased release of endogenous Our intracellular experiments showed that, when the soma of GABA may compete more effectively with phaclofen, therefore the CA, cell was hyperpolarized from the resting potential to preventing the effect of phaclofen to fully reverse paired pulse the GABA, reversal potential (about -90 mV), NMDA EPSPs inhibition; (2) GABA, antagonists may regulate GABA and were still present, as was paired-pause inhibition of these EPSPs glutamate release by way of presynaptic GABA, receptors dif- (see Fig. 3). It is well known that marked inhibition of NMDA- ferently at different stimulus intensities; and (3) the paired-pulse mediated processes occurs when cells are hyperpolarized from potentiation of the NMDA response might exist only at low about -30 mV (Andreasen et al., 1989; Hestrin et al., 1990). stimulus intensities and thus be uncovered as inhibition is We assume that, in our hyperpolarized cells, there is a minimal blocked. contribution to the EPSPs from somatic or electrically nearby In most cases, the GABA, antagonists did not fully suppress (dendritic) NMDA channels. Therefore, our NMDA EPSPs al- the hyperpolarization after a stimulation. We suspect that the most certainly result from channels on electrically remote den- residual potential remaining in the presence of GABA, antag- drites that are not hyperpolarized by somatic current injection. onists is an afterhyperpolarization (AHP) due to the Ca*+-ac- Indeed, paired-pulse inhibition ofthese EPSPs is retained during tivated K+ conductance. There is good precedent for such AHPs, this somatic hyperpolarization, suggesting that the source of especially following repetitive cell firing, which may induce Ca2+ GABA, inhibition is also present locally at the dendrites. spikes and therefore activate Ca2+-dependent K+ conductances, therefore decreasing cell excitability (Dingledine, 1983). Physiologic signiJcance of GABA, inhibition of NMDA EPSPs The findings of this study indicate that pairing of stimuli 200 Mechanism of GABA,-mediated inhibition of NMDA EPSPs msec apart can suppress NMDA-mediated depolarizations. Dendritic GABA, receptors in area CA, have been demonstrat- However, Lynch and coworkers have shown that a single con- ed using both electrophysiologic and autoradiographic tech- ditioning stimulus given 200 msec before a weak stimulus train niques (Alger et al., 1984; Newbeny and Nicoll, 1984; Bowery will facilitate NMDA depolarizations and the induction of long- et al., 1987). GABA, receptors that exist on afferent terminals term potentiation (Larson and Lynch, 1986; Muller and Lynch, may inhibit transmitter release (Hill and Bowery, 198 1; Dutar 1988). They suggest that this occurs because the conditioning and Nicoll, 1988a,b). In the present study, it is conceivable that stimulus suppresses GABA,-mediated inhibition and thus per- GABA, receptors located on terminals of Schaffer collaterals mits depolarization and opening of NMDA channels by the are activated by inhibitory intemeurons and subsequently in- subsequent stimulus train. Our laboratory and others have shown hibit transmitter release by regulating Ca*+ levels in the terminal. that this disinhibition may be mediated by GABA, receptors On the other hand, it is also possible that the relevant GABA, acting to promote NMDA function by decreasing GABAergic receptors are located postsynaptically on the dendrites of CA, inhibition (Mott et al., 1989, 1990a,b, Davies et al., 1990). pyramidal cells. Activation of these receptors may increase the Nevertheless, in the present study, GABA, receptors had op- Mgz+ block of the NMDA channel (Nowak et al., 1984; Mayer posite effects to suppress NMDA function. and Westbrook, 1984; Mayer et al., 1984). This work indicates that GABA,-mediated responses can The data presented here suggest that there is a postsynaptic modulate the activity of the NMDA-receptor-channel complex localization of GABA, receptors that regulate NMDA EPSPs. under certain circumstances. This process appears to be related Inhibition of transmitter release does not appear to be the mech- to the proximity of the GABA, and NMDA channels on py- anism of the paired-pulse inhibition of NMDA EPSPs due to ramidal cell dendrites. It is entirely possible that, during normal the following observations: (1) that NMDA EPSPs and not non- (and/or excitable) states of hippocampal activity, GABA, re- NMDA EPSPs displayed paired-pulse inhibition, and (2) that ceptors may locally regulate currents through NMDA channels. paired-pulse inhibition of NMDA EPSPs was reversed by GABA, Through such a mechanism, it seems that the function of receptor antagonists, while GABA, antagonists had no effect NMDA-mediated processes may be regulated within these dis- upon non-NMDA EPSPs. Thus, the mechanism of inhibition crete dendritic regions. Therefore, GABA, modulation of NMDA of NMDA EPSPs most likely involves systems that are specific responses is potentially an important mechanism for the regu- to the NMDA receptor channel/complex. This observation sup- lation of neuronal plasticity and other NMDA-dependent pro- ports the hypothesis that the mechanism of inhibition of NMDA cesses. EPSPs is dependent upon GABA, modulation of the voltage- dependent block of NMDA-operated channels by Mgz+ because References it is well known that non-NMDA-operated channels display Alger BE (1984) Characteristics of a slow hyperpolarizing synaptic little Mg2+ sensitivity. potential in rat hippocampal pyramidal cells in vitro. J Neurophysiol A presynaptic locus of GABA,-mediated inhibition of NMDA 52~892-910. EPSPs is conceivable, but unlikely. If NMDA and non-NMDA Andrade R, Malenka RC, Nicoll RA (1986) A G-protein couples The Journal of Neuroscience. January 1991, 11(l) 209

serotonin and GABA, to the same channels in hippocampus. Science Lambert JDC, Jones RSG (1989) Activation of N-methyl-o-aspartate 234:1261-1265. receptors contribute to the EPSP at perforant path synapses in the rat Andreasen M, Lambert JDC, Jensen MS (1989) Effects of non-N- dentate gyrus in vitro. Neurosci Lett 97~323-328. methyl-o-aspartate antagonists on synaptic transmission in the in vitro Lambert Nk, Harrison NL, Kerr DIB, Ong J, Prager RH, Teyler TJ rat hippocampus. J Physiol (Lond) 4 14:3 17-336. (1989) Blockade of the late IPSP in rat CA, hiDDOCaDIDa1 neurons Blake JF, Brown MW, Collingridge GC (1988) CNQX blocks acidic by 2-hydroxy-saclofen. Neurosci Lett 107: 125-128. - amino acid induced depolarizations and synaptic components me- Larson J, Lynch G (1986) Induction of synaptic potentiation in hip- diated by non-NMDA receptors in rat hippocampal slices. Neurosci pocampus by patterned stimulation involves two events. Science 232: Lett 89: 182-186. 985-988. Bowery NG, Hudson AL, Price GW (1987) GABA, and GABA, re- Mayer ML, Westbrook GL (1984) Mixed-agonist action of excitatory ceptor site distribution in the rat central nervous system. Neurosci- amino acids on mouse spinal cord neurones under voltage clamp. J ence 20:365-383. Physiol (Lond) 354:29-53. Collingridge GL, Bliss TVP (1987) NMDA receptors-their role in Mayer ML, Westbrook GL, Guthrie PB (1984) Voltage-dependent lot&term potentiation. Trends Neurosci 10:288-293. block by Mg*+ of NMDA responses in spinal cord neurones. Nature Cotman CW, Monaghan DT, Ottersen OP, Storm-Mathisen J (1987) 309:261-263. Anatomical organization of excitatory amino acid receptors and their Monaghan DT, Cotman CW (1985) Distribution of NMDA-sensitive pathways. Trends Neurosci 10~273-280. t$H]-glutamate-binding sites in rat brain. J Neurosci 5:2909-2919. Davies CH. Davies SN. Collineridee GL (1990) Paired-oulse deores- Morris RGM, Anderson EA, Lynch GS, Baudry M (1986) Selective sion of monosynaptic GAB&mediated- inhibitory po&.ynap& re- impairment of learning and blockade of long term potentiation by an sponses in rat hippocampus. J Physiol (Land) 424:5 13-53 1. NMDA antagonist, AP5. Nature 319:774-776. Davies SN, Collingridge GL (1989) Role of excitatory amino acid Morrisett RA. Chow CC. Nadler JV. McNamara JO ( 1989) Biochem- receptors in synaptic transmission in area CA, of rat hippocampus. ical evidence for enhanced sensitivity to NMDA in‘the hippocampal Proc R Sot Lond [Biol] 236:373-384. formation of kindled rats. Brain Res 496:25-28. Dingledine R (1983) N-methyl aspartate activates voltage-dependent Morrisett RA, Mott DD, Swartzwelder HS, Lewis DV, Wilson WA calcium conductance in rat hippocampal pyramidal cells. J Physiol (1990a) GABA, receptor mediated inhibition of the NMDA com- (Lond) 343:385405. ponent of synapiic transmission. Sot Neurosci Abstr 16:463. Drejer J, Honor6 T (1988) New quinoxalinediones show potent an- Morrisett RA, Mott DD, Lewis DV, Wilson WA, Swartzwelder HS tagonism of quisqualate responses in cultured mouse cortical neurons. (1990b) Reduced sensitivity of the N-methyl-D-aspartate component Neurosci Lett 87: 104-108. of synaptic transmission to Mg*+ in hippocampal slices of immature Dutar P, Nicoll RA (1988a) A physiological role for GABA, receptors rats. Brain Res, in press. in the central nervous system. Nature 332: 156-l 58. Mott DD, Lewis DV (1990) Fading of recurrent inhibition in the rat Dutar P, Nicoll RA (1988b) Pre- and postsynaptic GABA, receptors dentate gyrus is mediated by GABA, receptors. Sot Neurosci Abstr in the hippocampus have different pharmacological profiles. Neuron 16:945. 1:585-591. Mott DD, Ferrari CM, Lewis DV, Wilson WA, Swartzwelder HS (1989) Harris EA, Ganong A, Cotman CW (1984) Long-term potentiation in Baclofen facilitates the induction of long term potentiation. Proc the hippocampus involves activation of N-methyl-D-aspartate. Brain GABA, Symp 1:47. Res 323: 132-l 37. Mott DD. Braadon AC. Lewis DV ( 1990) Phaclofen antagonizes post- Hasuo H, Gallagher JP (1988) Comparison of antagonism by phaclo- tetanic bisinlhibition in rat dentate gyrus. Neurosci Lett l-lo: 13 l-1 36. fen of baclofen induced hyperpolarizations and synaptically mediated Muller D, Lynch G (1988) Long-term potentiation differentially affects late hyperpolarizing potentials recorded intracellularly from rat dor- two components of synaptic responses in hippocampus. Proc Nat1 solateral septal neurons. Neurosci Lett 86:77-81. Acad Sci USA 85:9346-9350. Herron CE, Lester RAI, Coan EJ, Collingridge GC (1986) Frequency- Newberry NR, Nicoll RA (1984) A bicuculline-resistant inhibitory dependent involvement of NMDA receptors in the hippocampus: a post-synaptic potential in rat hippocampal pyramidal cells in vitro. J novel synaptic mechanism. Nature 322:265-268. Physiol (Lond) 348~239-254. Hestrin SR, Nicoll RA, Perkel DJ, Sah P (1990) Analysis of excitatory Nicoll RA (1988) The coupling of neurotransmitter receptors to ion synaptic action in pyramidal cells using whole-cell recording from rat channels in the brain. Science 24 1:545-55 1. hippocampal slices. J Physiol (Lond) 422:203-225. Nowak L, Bregestovski P, Ascher P, Herbet A, Prochiantz A (1984) Hill DR, Bowery NG (1981) [)H]baclofen and [)H]GABA bind to Magnesium gates glutamate-activated channels in mouse central neu- bicuculline-insensitive GABA, sites in rat brain. Nature 290: 19-l 52. rones. Nature 307:462-465. Kerr BIB, Ong J, Johnston GAR, Prager RH (1989) GABA.-receptor- Soltesz I, Haby M, Leresche N,‘Crunelli V (1988) The GABA, an- mediated actions of baclofen in rat isolated neocortical slice prepa- tagonist phaclofen inhibits the late K+-dependent IPSP in cat and rat rations: antagonism by phosphono-analogues of GABA. Brain Res thalamic and hippocampal neurones. Brain Res 448:35 l-354. 480:312-316.