B R A I N R E S E A R C H 1 1 5 0 ( 2 0 0 7 ) 6 9 – 7 3

ava i l ab l e a t www.sc i enced i rec t . com

www.e l sev i e r. com/ l oca te /b ra in res

Research Report

In vivo temporal property of GABAergic neuraltransmission in collateral feed-forward inhibitionsystem of hippocampal–prefrontal pathway

Masatoshi Takitaa,⁎, Masahito Kuramochia,b, Yoshinori Izakic, Michiko Ohtomib

aCognition and Action Group, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Ibaraki 305-8566, JapanbDepartment of Biomolecular Science, Faculty of Science, Toho University, Funabashi, JapancDepartment of Physiology, St. Marianna University School of Medicine, Kawasaki, Japan

A R T I C L E I N F O

⁎ Corresponding author. Fax: +81 29 861 6632.E-mail address: takita.m@aist.go.jp (M. TaAbbreviations: ACSF, artificial cerebrospin

ylmethyl)-phosphinic acid; GABA, gamma-ammuscimol

0006-8993/$ – see front matter © 2007 Elsevidoi:10.1016/j.brainres.2007.02.063

A B S T R A C T

Article history:Accepted 23 February 2007Available online 2 March 2007

Anatomical evidence suggests that rat CA1 hippocampal afferents collaterally innervateexcitatory projecting pyramidal neurons and inhibitory interneurons, creating a disynaptic,feed-forward inhibitionmicrocircuit in themedial prefrontal cortex (mPFC).We investigatedthe temporal relationship between the frequency of paired synaptic transmission andgamma-aminobutyric acid (GABA)ergic receptor-mediatedmodulation of themicrocircuit invivo under urethane anesthesia. Local perfusions of a GABAa antagonist (−)-bicuculline intothe mPFC via microdialysis resulted in a statistically significant disinhibitory effect onintrinsic GABA action, increasing the first and second mPFC responses followinghippocampal paired stimulation at interstimulus intervals of 100–200 ms, but not those at25–50 ms. This (−)-bicuculline-induced disinhibition was compensated by the GABAaagonist muscimol, which itself did not attenuate the intrinsic oscillation of the local fieldpotentials. The perfusion of a sub-minimal concentration of GABAb agonist (R)-baclofenslightly enhanced the synaptic transmission, regardless of the interstimulus interval. Inaddition to the tonic control by spontaneous fast-spiking GABAergic neurons, it is clear thesequential transmission of the hippocampal-mPFC pathway can phasically drive thecollateral feed-forward inhibition system through activation of a GABAa receptor, bringingan active signal filter to the various types of impulse trains that enter the mPFC from thehippocampus in vivo.

© 2007 Elsevier B.V. All rights reserved.

Keywords:Paired-pulse facilitationMicrodialysisPrefrontal cortex(R)-baclofen(−)-Bicuculline(RS)-saclofenMuscimolCGP46381 ((3-aminopropyl)(cyclohexylmethyl)-phosphinic acid)

1. Introduction

Dysfunctions in the hippocampus and prefrontal cortex (PFC)have been recognized in aging (Raz et al., 1997) and schizo-phrenia (Harrison, 1999). Functional changes in gamma-

kita).al fluid; BAC, (R)-baclofeinobutyric acid; ISI, inte

er B.V. All rights reserved

aminobutyric acid (GABA) have also been recently reportedin interneurons in the PFC (Lewis et al., 2005). Anatomically,the rat hippocampus sends monosynaptic glutamatergicinputs to the medial PFC (mPFC, mainly prelimbic, andinfralimbic areas), collaterally innervating both pyramidal

n; BIC, (−)-bicuculline; CGP CGP46381, (3-aminopropyl)(cyclohex-rstimulus interval; SAC, (RS)-saclofen; PFC, prefrontal cortex; MUS,

.

70 B R A I N R E S E A R C H 1 1 5 0 ( 2 0 0 7 ) 6 9 – 7 3

cells and parvalbumin-positive, GABA-like interneurons inthe mPFC (Gabbott et al., 2002). Electrophysiologically, bothtypes of synaptic connection have been identified in thehippocampal-mPFCpathwayby intracellular recording studies(Degenetais et al., 2003; Tierney et al., 2004). In fact, thelatencies of the monosynaptic responses in both pyramidalcells and interneurons are about 15 ms.

The cortical, parvalbumin-positive, GABA-like neuronsexhibit fast spiking, providing the potential for tonic inhibitorycontrol (Kawaguchi andKubota, 1993, 1997). On the otherhand,the collateral GABAergic system seems to provide phasicdisynaptic feed-forward inhibition for the firing frequency ofpyramidalneurons that respond tohippocampal-mPFC inputs.

When initial stimulation of the hippocampal-mPFC path-way collaterally evokes GABA release, the glutamatergicsynaptic response to the second stimulation should includethe effects of evoked GABA release on GABAergic receptors.The major receptor subtypes “a” and “b”, should also beinfluenced by extracellular GABA concentrations. Postsynap-tic GABAa receptors reportedly produce fast synaptic inhibi-tion (Stephenson, 1988), while presynaptic GABAb receptorsare highly sensitive in controlling endogenous transmitterrelease in the rat brain (Bowery, 1993; Waldmeier et al., 1988).In the present study, receptor agonists and antagonists ofGABAa and GABAb receptors were applied to themPFC using areverse dialysis technique (Gurden et al., 2000). The hippo-campal-mPFC pathway (Gabbott et al., 2002) was tested bypaired stimulation, as reported previously (Mulder et al., 1997;Izaki et al., 2002). By focusing attention on the receptorproperties, we were able to investigate the temporal involve-ment of GABAergic modulation in the hippocampal-mPFCglutamatergic synaptic transmission in vivo.

Fig. 1 – Representation of the hippocampal-mPFC evoked andlocal field potentials (four single-trace waves from oneanimal, arrowheads indicate stimulation timing for evokedresponse) under perfusion with ACSF for 10 min before and100 min after a 10-min perfusion of a high dose (1 mM) of theGABAa agonist muscimol (MUS, top panel). Typicalwaveforms of the synaptic responses following pairedstimulation (arrowheads, identical animal data) at ISIs of 25,50, 100, and 200 ms, under perfusions of ACSF, MUS, theGABAa antagonist (−)-bicuculline (BIC), and a combination ofMUS and BIC (Comba) (middle panel) are shown, togetherwith the relative values of the first response under perfusionsof ACSF (lower panel) (n=8 in total). A post hoc test withtwo-wayANOVAwith repeatedmeasures is used to test drugeffects independent of stimulus version, and these effects arecompared with the respective control ACSF for each stimulusversion (*p<0.05, NSp>0.1, Fisher's PLSD).

2. Results

Under the steady-state control condition, the first postsynap-tic maximal slope response was −37.7±3.6 mV/s; the latenciesof onset, maximal slope point, and peak in the hippocampal-mPFC pathway were 13.4±0.3, 18.1±0.1, and 23.1±0.4 ms,respectively (n=16). Initially, we carefully pre-tested theconcentration of GABAa agonist muscimol (MUS), which isknown to be a major inactivation drug for local brain activity.As demonstrated by single-trace waves (Fig. 1, top panel), thereverse dialysis of 1000 μM MUS drastically attenuated boththe evoked and local field potentials compared with that ofartificial cerebrospinal fluid (ACSF), even 100 min afterwashout. The effects of GABAa receptor-related drugs onpaired responses were then tested under four perfusionconditions: ACSF alone; MUS (200 μM, a washable concentra-tion); the GABAa antagonist (−)-bicuculline (BIC, 10 μM); and acombination of MUS and BIC (Comba) (n=8 in total, Fig. 1).Two-way ANOVA with repeated measures showed significanteffects across the drugs and stimulus versions (F3,21=9.391,p=0.0004 and F4,28=40.025, p<0.0001, respectively) and theirinteraction (F12,84=4.487, p<0.0001). Post hoc analysis revealedsignificant increases under the BIC and Comba conditionscompared with the control ACSF (p<0.001 and 0.05, respec-tively, Fisher's PLSD, Fig. 1). Under BIC perfusion conditions,the first and second responses at ISIs of 100 and 200 ms

increased significantly, but those at 25 and 50 ms did not. Inthe Comba perfusion condition, these BIC increases disap-pearedwithout any change in the BIC-unaffected responses atISIs of 25 and 50 ms, whereas the MUS condition did not showa significant decrease for any of the stimulus versions (Fig. 1);

Fig. 2 – Typical waveforms of a hippocampal-mPFC synapticresponse followingpaired stimulation (identical animaldata),under perfusions of ACSF, BAC, SAC, and a combination ofBAC and SAC (Combb), together with their relative values forthe first response under perfusions of ACSF (n=8 in total). Apost hoc test is used to test the drug effects independent ofstimulus timing. No significant effects were found for any ofthe stimulus versions.

Fig. 3 – Hippocampal stimulation and prefrontal cortex (PFC)recording sites are represented in the atlas and numbered inan anteroposterior direction from the bregma (mm).

Table 1 – BAC-like effect of CGP46381 (CGP, 0.3μM) and thecumulative effect of BAC+CGP on the paired pulsedresponse in hippocampal-mPFC pathway

ACSF(mV/s)

BAC (%) CGP (%) BAC+CGP(%)⁎#

1st −39.01±7.47 105.16±5.62 114.07±6.96 107.22±9.27ISI25 ms −29.52±0.65 89.89±11.78 90.03±12.69 80.36±8.1850 ms −55.24±0.96 123.48±17.13 112.72±7.84 148.76±25.96100 ms −102.65±2.43 128.89±12.99 139.29±10.21 163.99±14.42#

200 ms −89.96±2.15 126.74±20.04 130.09±18.35 150.08±26.44#

The changes in the first or second response (1st, ISI of 25, 50, 100, or200 ms) were evaluated as percent changes of the respective ACSFresponses (mean±SEM, n=8 [same group as used for GABAa expe-riment]). ⁎p<0.05 compared to BAC perfusion condition. #p<0.05compared with ACSF (post hoc test).

71B R A I N R E S E A R C H 1 1 5 0 ( 2 0 0 7 ) 6 9 – 7 3

suppression of Comba/BIC was 73.6±5.7% in the first responseand 91.4±16.0% or 79.5±14.1% in the second response at ISIs of100 or 200 ms, respectively. Similarly, GABAb-related drugswere tested under four perfusion conditions: ACSF alone; theGABAb agonist (R)-baclofen (BAC, 1 μM); the antagonist (RS)-saclofen (SAC, 30 μM); and a combination of BAC and SAC(Combb) (n=8 in total, Fig. 2). Two-way ANOVA with repeatedmeasures showed significant effects across drugs and stimu-lus versions (F3,21=5.203, p=0.0076 and F4,28=21.070, p<0.0001)and their interaction (F12,48=1.887, p=0.0488). Post hoc analysisrevealed significant increases in the BAC and Combb condi-tions regardless of stimulus version (p<0.03 for both, Fig. 2).We also tested a low concentration of CGP46381 (CGP, 0.3 μM),which is generally categorized as a GABAb antagonist, andobserved its BAC-like agonistic effect (Table 1), as previouslyreported (Green et al., 2000) when tested in a GTPgS-bindingassay using membrane fractions (Dr. A Green, personalcommunication). We found a significant cumulative effectfor the combination of BAC and CGP (p=0.015, Fisher's PLSD)following two-way ANOVA with repeated measures (signifi-

cant effects of stimulus versions [F4,28=4.458, p=0.0066] andinteraction between drugs and stimulus versions [F8,56=2.264,p=0.0358]). None of the drugs showed statistically significantchanges in the latency of onset, maximal slope, or peak. Thestimulated and recorded sites were confirmed in the hippo-campus CA1 (the posterior dorsal subregion of our previousreport; Izaki et al., 2002) and the mPFC (Fig. 3).

3. Discussion

The present study confirms that even with perfusion of ACSFvia a microdialysis probe placed in the PFC, the dual-phasetendency fromdepression to facilitation appears in the paired-pulsed hippocampal-mPFC pathway. This corresponds to alengthening of the ISI when the posterior dorsal subregion ofhippocampus CA1 is stimulated, as seen in previous studies(Christoffersen et al., 2003; Izaki et al., 2002). Perfusion with arelatively high dose (400–1000 μM) of the GABAa agonist MUSshowed a specific secondary effect on the mPFC network,attenuating both single-pulsed responses and intrinsic oscilla-tions of local field potentials far from thepaired-pulse test. Thedual attenuation persisted after a sufficient washout of 400–1000 μM, but not of 200 μM, atwhich concentrationMUSdidnotattenuate the hippocampal-mPFC synaptic responses of any of

72 B R A I N R E S E A R C H 1 1 5 0 ( 2 0 0 7 ) 6 9 – 7 3

the stimulus versions under our experimental conditions (Fig.1). However, MUS selectively compensated the antagonist BIC-mediated disinhibition of the first and second responses at 100and 200 ms ISIs but did not interfere with the BIC-unaffectedresponses at ISIs of 25 and 50 ms. These data imply thatendogenous extracellular GABAhas tonic andphasic effects onGABAa receptors in the collateral microcircuit formed by theprojection from the hippocampus CA1 subregion that westimulated (Fig. 3), as well as on GABA-like neurons in themPFC (Gabbott et al., 2002). In short, beyond the GABAergictonic inhibition through the GABAa receptor, the first stimula-tion phasically evokes release of GABA, which acts on theGABAa receptor for at least 50 ms to oppose the antagonisticaction of BIC in the microcircuit (Fig. 1). This time window ofGABAa-related feed-forward inhibition corresponds to animportant frequency theta rhythm (frequency range, 4–10 Hz)that regulates the hippocampal-mPFC pathway in rats (Siapaset al., 2005). TheGABAasystemshould tightly control themPFCnetwork under not only tonic but also phasic conditions.Therefore, the higher dose of MUS should easily attenuate theintrinsic oscillation of local field potentials and single-pulsedresponses in parallel, as mentioned previously.

Although the sub-minimal concentration of GABAb agonistBAC did not interfere in a statistically meaningful way withthe single-pulsed hippocampal-mPFC responses (as did MUS),it facilitated to a small extent synaptic transmission, regard-less of the ISI used (Fig. 2). Thus, BAC appears to facilitateGABAergic presynaptic/autoreceptor control in reducing theextracellular concentration of GABA. However, the antagonistSAC was not sufficiently efficient to act on the GABAbreceptor. We also observed that CGP46381, which is generallyclassed as an antagonist, showed a BAC-like agonistic effecton the mPFC, as reported previously (Green et al., 2000). In theprefrontal microcircuit for feed-forward inhibition, the role ofthe GABAb system was tonic but not phasic. These resultssuggest that the hippocampal-mPFC transmission driving theglutamatergic postsynaptic responses collaterally evokedGABA release from the GABA interneuron, which acts onGABAa receptors for at least 25–50 ms, but not for more than100ms. Regarding the local field potential in this pathway, thetheta rhythm plays an important role, as mentioned above(Siapas et al., 2005), as does the gamma band activity (40–100 Hz) for neuroplastic events (Izaki et al., 2001), with 10–40 Hz corresponding to the intervening ISIs of 25–100 ms. Thistime window of collateral feed-forward inhibition may under-lie the temporal control in the prefrontal neural network thatallows timing-dependent cooperation/competition betweensynaptic plasticities (Kawashima et al., 2006).

4. Experimental procedures

Male Sprague–Dawley rats (300–400 g) were anesthetized withurethane (1.5 g/kg, i.p.) and placed in a stereotaxic frame withbody temperature maintained at 37 °C. The experimentalprocedures and treatment of the laboratory animals wereapproved by the Animal Ethical Committee of the NationalInstitute of Advanced Industrial Science and Technology,Japan, and were in accordance with the National Institutesof Health Guide for the Care and Use of Laboratory Animals,

revised 1996 (NIH Publications No. 80-23). Care was taken tominimize the number of animals and their suffering.

A concentric, bipolar, stainless steel, stimulating electrodewas lowered into the CA1 region of the hippocampus 6.3 mmposterior and 5.4 mm lateral to bregma, and 3.5 mm below thedura (Paxinos andWatson, 1998). A recording electrode (64 μmin diameter, nichrome wire, impedance ∼0.5 MΩ) fixed to amicrodialysis probe was lowered into the mPFC 3.2 mmanterior and 0.8 mm lateral to bregma, and 3.2–4.0 mmbelow the dura (Gurden et al., 2000). The I-shaped micro-dialysis probes were prepared as previously described (mem-brane length, 2 mm; outer diameter, 0.24 mm; donated byAsahi Kasei, Tokyo, Japan) and were connected via Teflontubing to a microinfusion pump (CMA102; Carnegie Medicine,Sweden) equipped with a 40-μL loop-injector (model 9125;Rheodyne, Rohnert Park, CA, USA) (Takita et al., 1997).

Electrophysiological recordingswere carried out via reversemicrodialysis under steady-state conditions 2–3 h after theinsertion of the electrodes with the probe perfused by ACSF(145 mM NaCl, 2.7 mM KCl, 1 mM MgCl2, 1.2 mM CaCl2) at aflow rate of 2 μL/min. This procedure allowed recordings ofintact cortical tissue and simultaneous control of the infusionand washout of drugs (Gurden et al., 2000). Stimulation of theCA1 region evokes a characteristic monosynaptic, negative-going, field potential in the mPFC, with stable latencies ofonset, maximal slope, and peak. Constant-current pairedstimulations of 100-μs duration at ISIs of 25, 50, 100, and200 ms (Izaki et al., 2002) were delivered every 30 s at anintensity that evoked a 65%-of-maximum response to a singlestimulation (110–260 μA). Under pharmacological conditions,the ACSF-containing GABA drugswere infused one at a time inrandom order into the probe for 20 min via a loop-injector atan inter-infusion interval of 30–60 min for washout. Since thebinding ability of the presynaptic receptor is known to bemoresensitive than that of the postsynaptic receptor, the doses ofeach drug were kept as low as possible, to minimize the directeffect of the agonist on the first response to paired-pulsestimulation or to antagonize the effect of the agonist on thesecond response i.e., 200 μM MUS, 10 μM BIC (Harte andO'Connor, 2005), 1 μM BAC (interference without serotoninrelease, Abellan et al., 2000), 30 μM SAC (Ishizuka et al., 2000),and 0.3 μM CGP46381. The effective extracellular concentra-tions of the drugs, extrapolated from the relative recovery ofGABA drugs through the microdialysis probe during inter-infusion intervals, were estimated to be tenfold lower than theconcentrations within the probe. The data were recordeddigitally at a frequency of 0.08–1000 Hz, and analyzed aspreviously described (Takita et al., 1999). The maximal slopesof the postsynaptic field potentials were expressed aspercentage change (mean±SEM) of the first response underperfusion of ACSF, and analyzed using ANOVA. The positionsof the electrodes were marked by passing a DC current andverified histologically in serial sections stained with thionin.

Acknowledgments

We thank Dr. A. Green for information regarding the assay forthe agonistic action of CGP46381. This work was supported inpart by a Research Grant for long-range research initiative

73B R A I N R E S E A R C H 1 1 5 0 ( 2 0 0 7 ) 6 9 – 7 3

from JCIA and by a Grant-in-Aid for Science Research from theMinistry of Education, Culture, Sports, Science and Technol-ogy of Japan (16530484).

R E F E R E N C E S

Abellan, M.T., Jolas, T., Aghajanian, G.K., Artigas, F., 2000. Dualcontrol of dorsal raphe serotonergic neurons by GABA(B)receptors. Electrophysiological and microdialysis studies.Synapse 36, 21–34.

Bowery, N.G., 1993. GABAB receptor pharmacology. Annu. Rev.Pharmacol. Toxicol. 33, 109–147.

Christoffersen, G.R., Petersen, S., daCosta, N.M., 2003. Potentiationof prelimbic field potentials during and seconds after trains ofexcitations in the rat hippocampo-prefrontal pathway.Neurosci. Lett. 341, 143–146.

Degenetais, E., Thierry, A.M., Glowinski, J., Gioanni, Y., 2003.Synaptic influence of hippocampus on pyramidal cells of therat prefrontal cortex: an in vivo intracellular recording study.Cereb. Cortex 13, 782–792.

Gabbott, P., Headlam, A., Busby, S., 2002. Morphological evidencethat CA1 hippocampal afferents monosynaptically innervatePV-containing neurons and NADPH-diaphorase reactive cellsin the medial prefrontal cortex (Areas 25/32) of the rat. BrainRes. 946, 314–322.

Green, A., Walls, S., Wise, A., Green, R.H., Martin, A.K., Marshall,F.H., 2000. Characterization of [(3)H]-CGP54626A binding toheterodimeric GABA(B) receptors stably expressed inmammalian cells. Br. J. Pharmacol. 131, 1766–1774.

Gurden, H., Takita, M., Jay, T.M., 2000. Essential role of D1 but notD2 receptors in the NMDA receptor-dependent long-termpotentiation at hippocampal–prefrontal cortex synapsesin vivo. J. Neurosci. 20, RC106.

Harrison, P.J., 1999. The neuropathology of schizophrenia—acritical review of the data and their interpretation. Brain 122,593–624.

Harte, M., O'Connor, W.T., 2005. Evidence for a selective prefrontalcortical GABA(B) receptor-mediated inhibition of glutamaterelease in the ventral tegmental area: a dual probemicrodialysis study in the awake rat. Neuroscience 130,215–222.

Ishizuka, Y., Ishida, Y., Jin, Q.H., Mitsuyama, Y., Kannan, H., 2000.GABA(A) and GABA(B) receptors modulating basal andfootshock-induced nitric oxide releases in rat prefrontal cortex.Brain Res. 872, 266–270.

Izaki, Y., Takita, M., Jay, T.M., Kaneko, H., Suzuki, S.S., Nomura, M.,2001. Effect of long-term potentiation induction on

gamma-band electroencephalograms in prefrontal cortexfollowing stimulation of rat hippocampus in vivo. Neurosci.Lett. 305, 57–60.

Izaki, Y., Takita, M., Nomura, M., 2002. Local properties of CA1region in hippocampo-prefrontal synaptic plasticity in rats.NeuroReport 13, 469–472.

Kawaguchi, Y., Kubota, Y., 1993. Correlation of physiologicalsubgroupings of nonpyramidal cells withparvalbumin-immunoreactive andcalbindin(d28k)-immunoreactive neurons in layer V of ratfrontal cortex. J. Neurophysiol. 70, 387–396.

Kawaguchi, Y., Kubota, Y., 1997. GABAergic cell subtypes and theirsynaptic connections in rat frontal cortex. Cereb. Cortex 7,476–486.

Kawashima, H., Izaki, Y., Grace, A.A., Takita, M., 2006.Cooperativity between hippocampal–prefrontal short-termplasticity through associative long-term potentiation. BrainRes. 1109, 37–44.

Lewis, D.A., Hashimoto, T., Volk, D.W., 2005. Cortical inhibitoryneurons and schizophrenia. Nat. Rev., Neurosci. 6, 312–324.

Mulder, A.B., Arts, M.P., Lopes da Silva, F.H., 1997. Short- and long-term plasticity of the hippocampus to nucleus accumbens andprefrontal cortex pathways in the rat, in vivo. Eur. J. Neurosci. 9,1603–1611.

Paxinos, G., Watson, C., 1998. The Rat Brain in StereotaxicCoordinates, Vol. Academic Press, Sydney.

Raz, N., Gunning, F.M., Head, D., Dupuis, J.H., McQuain, J., Briggs,S.D., Loken, W.J., Thornton, A.E., Acker, J.D., 1997. Selectiveaging of the human cerebral cortex observed in vivo:differential vulnerability of the prefrontal gray matter. Cereb.Cortex 7, 268–282.

Siapas, A.G., Lubenov, E.V., Wilson, M.A., 2005. Prefrontal phaselocking to hippocampal theta oscillations. Neuron 46, 141–151.

Stephenson, F.A., 1988. Understanding the GABAA receptor: achemically gated ion channel. Biochem. J. 249, 21–32.

Takita, M., Kaneko, H., Suzuki, S.S., Akamatsu, M., 1997. Lastingeffect of NO on glutamate release in rat striatum revealed bycontinuous brain dialysis. NeuroReport 8, 567–570.

Takita, M., Izaki, Y., Jay, T.M., Kaneko, H., Suzuki, S.S., 1999.Induction of stable long-term depression in vivo in thehippocampal–prefrontal cortex pathway. Eur. J. Neurosci. 11,4145–4148.

Tierney, P.L., Degenetais, E., Thierry, A.M., Glowinski, J., Gioanni,Y., 2004. Influence of the hippocampus on interneurons of therat prefrontal cortex. Eur. J. Neurosci. 20, 514–524.

Waldmeier, P.C., Wicki, P., Feldtrauer, J.J., Baumann, P.A., 1988.Potential involvement of a baclofen-sensitive autoreceptor inthe modulation of the release of endogenous GABA from ratbrain slices in vitro. Naunyn-Schmiedeberg's Arch. Pharmacol.337, 289–295.