JOURNALOF NEUROPHYSIOLOGY Vol. 68, No. 1, July 1992. Printed in U.S.A.

Dual-Component Miniature Excitatory Synaptic Currents in Rat Hippocampal CA3 Pyramidal Neurons CHRIS

McBAIN

AND

RAYMOND

Department

of Pharmacology,

SUMMARY

AND

DINGLEDINE

University of North Carolina, Chapel Hill, North Carolina 27599

CONCLUSIONS

1. Spontaneous miniature synaptic events were studied with tight-seal whole-cell recordings from CA3 neurons maintained in the hippocampal slice from immature rats ( 3- 15 days). CA3 neurons suffer a constant, high-frequency barrage of inhibitory synaptic input. When inhibitory postsynaptic currents were suppressed by bicuculline, a smaller contribution from excitatory synapses was revealed. 2. Addition of tetrodotoxin (TTX) removed a persistent inward current and substantially reduced the baseline noise facilitating the detection of “miniature” excitatory currents. Addition of hyperosmotic media increased the frequency of spontaneous excitatory postsynaptic currents ( EPSCs). 3. Under both physiological and elevated potassium conditions, individual spontaneous miniature EPSCs( 1O-30 pA amplitude) were composed of components mediated by N-methyl-Daspartate (NMDA) and non-NMDA receptors as determined by their voltage dependence, time course, and sensitivity to selective antagonists. 6-Cyano-7-nitro-quinoxaline-2,3-dione (CNQX) or D-Zamino-5-phosphonovaleric acid ( D-APV) shifted the amplitude distribution of miniature EPSCsto a smaller mode at both +40 mV and -40 mV. Similar to EPSCsrecorded in CA1 neurons, the rise and decay times of the NMDA receptor component were slower than those of the non-NMDA component. The time course of the non-NMDA component was voltage independent. 4. In 13 of 2 1 neurons, no correlation existed between individual EPSC rise times and their corresponding halfwidth, peak amplitude, or decay time constant. This suggeststhat the large range of EPSC kinetics observed in each individual neuron was not due solely to cable attenuation of EPSCswidely distributed over the dendritic tree. Plots of the mean EPSC rise time against mean halfwidth for each cell, however, revealed a striking correlation, suggesting that in neonates, active synapsesmay be grouped in a restricted region of the dendritic tree and as such are subject to similar amounts of dendritic filtering. 5. The electrotonic length of CA3 neurons (L = 0.52) predicted that at this maturity the electrotonic compactness of the neuron facilitated voltage control over all but the most distal synapses.The reversal potential of the fast component of spontaneous events was close to 0 mV, whereas the reversal potential of exogenously applied kainate and NMDA was more positive. This discrepancy likely reflects a compromise of the voltage clamp by the activation of conductances distributed over the entire cell. 6. Elevation of [ K+ 1, from 3.5 to 8.5 mM produced a moderate inward current (80- 100 PA) and elevated the frequency of spontaneous EPSCs before the occurrence of the first interictal burst. The inward current persisted in TTX and CNQX, suggesting that neither synaptic nor glial cell releaseof excitatory transmitter was responsible. The barrage of EPSCsproduced by high potassium was eliminated in TTX, however, indicating that they are synaptically driven by spiking in nearby neurons. 7. These results indicate that NMDA and non-NMDA receptors are colocalized at individual excitatory synapsesin the hippo16

campal CA3 region. The role of recurrent excitation in high-potassium epileptogenesis is emphasized. INTRODUCTION

CA3 pyramidal neurons in the hippocampus receive excitatory inputs from dentate granule cells and from other CA3 cells in a recurrent circuit. The axon collaterals of adult CA3 pyramidal cells ramify and diverge extensively within the CA3 subfield (Finch et al. 1983; Tamamaki et al. 1984) and synapse onto other CA3 pyramids (MacVicar and Dudek 1980; Miles and Wong 1986, 1987). It is thought that these recurrent synapses sustain firing in the cell population and contribute to the generation of synchronous burst firing (Dichter and Spencer 1969; Miles and Wong 1987; Traub and Dingledine 1990; Traub and Wong 1982). Epileptiform bursts mediated by the recurrent pathway can be blocked by 6-cyano-7-nitroquinoxaline-2,3dione (CNQX; Chamberlin et al. 1990; Lee and Hablitz 1989; Neuman et al. 1988a), indicating a strong non-Nmethyl-D-aspartate (NMDA) receptor component to these synapses. Autoradiographic studies have identified NMDA binding sites in the stratum radiatum layer of CA3 (Monaghan and Cotman 1985; for review see Cotman et al. 1987). Although D-2-amino-5phosphonovaleric acid ( D-APV) slightly reduces the amplitude of the paroxysmal shift underlying interictal bursts in both immature (King et al. 1989) and mature CA3 neurons (Chamberlin et al. 1990; Lee and Hablitz 1989), the role of NMDA receptors in recurrent excitatory synaptic transmission is unexplored. Moreover, NMDA receptors appear to be involved in longterm potentiation ( LTP) at the association / commissural input to CA3 (Zalutsky and Nicoll 1990) but not at mossy fiber synapse on CA3 neurons (Harris and Cotman 1986; Zalutsky and Nicoll 1990). LTP at the mossy fiber input is thought to require non-NMDA glutamate receptors exclusively (Andreasen et al. 1989; Neuman et al. 1988b). It is well known that both NMDA and non-NMDA receptors contribute to the postsynaptic response of CA1 pyramidal cells to CA3 input (Collingridge et al. 1983; Dingledine et al. 1986; Kauer et al. 1988) and play a critical role in LTP in area CA 1 (Collingridge et al. 1983 ) . Individual excitatory synapses made between hippocampal neurons in culture have been shown to use both NMDA and non-NMDA postsynaptic receptors (Bekkers and Stevens 1989; Forsythe and Westbrook 1988 ) . Similar studies have not been carried out on spontaneous excitatory synaptic currents received by pyramidal neurons of CA3.

0022-3077192$2.00 Copyright 0 1992 The AmericanPhysiologicalSociety Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (130.056.064.029) on July 31, 2018. Copyright © 1992 American Physiological Society. All rights reserved.

EXCITATORY

SYNAPTIC

CURRENTS

We have used whole-cell patch recording techniques (Blanton et al. 1989;Edwards et al. 1989) to study spontaneous excitatory postsynaptic currents (EPSCs) in CA3 neurons of hippocampal slices prepared from neonate rats. Of particular interest were the observations that spontaneous miniature EPSCs contain both NMDA and nonNMDA receptor-mediated components, and that EPSC frequency increases during the transition period to interictal burst firing after introduction of high-potassium medium. Chamberlin et al. ( 1990) have previously demonstrated that the probability of occurrence of spontaneous EPSPs in CA3 neurons increased immediately before the onset of each burst, which suggests that a “wind-up” of excitatory activity may contribute to the initiation of spontaneous interictal activity. We show that this increased EPSP activity is spike triggered rather than due to the spontaneous release of transmitter from potassium-depolarized presynaptic nerve terminals. METHODS

Rat hippocampal slices (300 pm thick) were prepared with a Vibratome (Oxford, Series 1000) from 3- to 15day-old SpragueDawley rats. Sliceswere allowed a recovery period of 45 min before use, during which they were held in oxygenated media at room temperature. Sliceswere held in a submerged chamber and perfused at a rate of - 1 ml/min with the following medium (in mM): 130 NaCl, 24 NaHCO,, 3.5 KCl, 1.25 NaH,PO,, 1.5 CaCl, . 2H20, 1.5 MgS04 7H20, and 10 glucose, saturated with 95% 02-5%COZ (pH 7.4,307 mOsm) at 22-25°C. [K+], was elevated from 3.5 to 8.5 mM by addition of 1M KC1 to the normal medium. All drugs were bath-applied in known concentrations by direct addition to the perfusate via a three-way tap. The dead time of the perfusion system was - 1 min. Unless stated otherwise, all experiments were performed in the presence of 5 PM bicuculline to eliminate most inhibitory synaptic currents. Tight-seal (> 1 G8) whole-cell recordings (Edwards et al. 1990; Hamill et al. 198 1) were obtained from the cell bodies of 57 neurons in stratum pyramidale of CA3. Patch electrodes were fabricated from borosilicate glass and were typically not fire polished. They had resistancesof 1-5 Mti when filled with (in mM): 140 Cs-methanesulphonate, 10 N-2-hydroxyethylpiperazine-N ’ -2ethanesulfonic acid (HEPES), and 2 MgC& (buffered to pH 7.3 with CsOH, 275 mOsm). Cell sealswere attained with the technique originally described by Blanton et al. ( 1989). The electrode was positioned under visual control within the stratum pyramidale of CA3; usually, individual cells were not visualized. Positive pressure was applied to the recording electrode before insertion into the slice. The close apposition of the electrode with a neuron resulted in a small increase in the voltage deflection to a 100.pA current pulse. Release of the positive pressure and gentle suction facilitated attainment of a gigaohm seal. Breakthrough to the whole-cell mode was performed under current clamp to provide an initial evaluation of the passivemembrane and action potential properties of the neuron. Cells possessingan initial resting potential more negative than -50 mV (-59 t 2.2 mV, mean t SE; YI= 39 ) and overshooting action potentials were accepted. The dialysis of the neuron by Cs’ usually resulted in stabilization of the cell at a more positive potential. In addition, neurons that fired action potentials on the anode break were discarded. Neurons were then voltage clamped at -60 mV, unless stated otherwise in the text. CA3 neurons had input resistancesranging from 100 to 700 Mfi (411 t 42 MS2; n = 39) at this holding potential. Occasionally, glial cellswere encountered, identified as such by their high resting potential (greater than -70 mV), lower input resistance ( -200 MQ), and absence of spikes. l

IN CA3 NEURONS

17

All currents were recorded with an Axopatch- 1D amplifier (Axon Instruments); records were filtered at 2 kHz ( -3 dB) and digitized at 3- 10 kHz on an 80386 computer or stored on FM tape (5 kHz bandwidth) for off-line analysis. Individual EPSCs were identified and accepted if the mean current associated with the event was significantly different from that of an equivalent length of pre-event baseline (mean standard deviation of noise 0.32 pA; n = 14 neurons). The lo-90% rise time was calculated by linear regression over the appropriate data points. Measurements of the decay time constant of synaptic currents represent the mean of > 15 individual events from each neuron and were fit by the sum of one or two exponential components (simplex algorithm, leastsquares criteria). When a more accurate estimation of the slow decay time constant was required and the events were not adequately described by a single time constant, the fast time constant was assignedthat value obtained at a holding potential of -70 mV where contamination from NMDA receptor activation was minimal, allowing the extraction of a second decay time constant. Amplitude histograms were constructed from individually measured EPSC amplitudes and binned in 2-pA intervals. For measurements of membrane and dendritic time constants, electrotonic pulses ( 100 pA, duration 100-250 ms) were delivered at 0.5- 1 Hz in current-clamp mode. The voltage deflection was digitized at 10 kHz. The series resistance was determined by balancing of the “bridge” in current clamp. The seriesresistance typically ranged from 10 to 30 Ma (24 t 2.8 MQ; from a sample of 10 neurons). The amplitude of the excitatory synaptic currents was never ~70 pA; thus a series resistance of 20 M8 would result in a series resistance error of no greater than 1.5 mV. Membrane potentials were not corrected for these errors. The current-voltage relationships of exogenously applied agonists were obtained after achievement of the steady-state response to each drug. Steps of 10-mV increments were given every 5 or 10 s, and the current measurement was made at the end of each step. All data are represented as the mean * SE. Where appropriate, a Student’s t test was performed. Drugs used were tetrodotoxin (TTX, Sigma), bicuculline methobromide (Sigma), D-APV (Sigma), and CNQX (Tocris Neuramin) , NMDA (Cambridge Research Biochemicals), and kainic acid ( Sigma). RESULTS

In initial experiments, field potential recordings were made in an attempt to address the limitations of permissible slice thickness for satisfactory synchronized neuronal burst firing and electrographic seizure activity. “Thin slices” ( 180 pm thick) did not possess adequate circuitry to elicit consistently a reliable field potential. Slices 300 pm thick, however, contained enough circuitry to engage both interictal and electrographic seizure activity while permitting adequate visualization of the neurons. Cells in the CA3 pyramidal cell layer that were filled with Lucifer yellow typically possessed the classic pyramidal morphology.

Immature CA3 neuronsreceiveintensetonic inhibitory input Initial recordings where CsCl-filled microelectrodes were used revealed that CA3 pyramidal neurons possess a higher frequency of spontaneous synaptic activity than could be predicted from previous intracellular recordings (Fig. 1) . The frequency of spontaneous inward currents ranged from 1 to 80 Hz in different cells (22 t 5.7 Hz; n = 10 neurons).

Downloaded from www.physiology.org/journal/jn by ${individualUser.givenNames} ${individualUser.surname} (130.056.064.029) on July 31, 2018. Copyright © 1992 American Physiological Society. All rights reserved.

18

C. McBAIN

AND R. DINGLEDINE

B Control

50pA 2s 35ms

C 5pbl

Bicuculline

D Bicuculline

Amplitude

5pM

of

events

(PA)

FIG. 1. A : typical recording of a CA3b neuron voltage clamped at -80 mV. Downward deflections represent inward currents. The intracellular solution in this experiment contained CsCl ( 140 mM). A high frequency of spontaneous synaptic activity was observed ( 8.1 Hz), concomitant with the firing of spontaneous single action currents ( *, events are truncated). A and C bottom: a portion of the top trace displayed at a faster time base. B: amplitude histogram (bin size 2 PA) shows the broad range of postsynaptic current amplitudes. Data taken from 43 s of continuous recording. Average amplitude -80 & 4.6 pA ( n = 134 events). C: in the presence of the GABA* receptor antagonist bicuculline ( 5 PM), the frequency of events is markedly reduced to 1.5 Hz. D: amplitude histogram constructed from an equivalent time period demonstrates that in the presence of bicuculline, only small-amplitude events persist (average -20 & 2.1 pA; n = 2 1 events) and are presumably EPSCs.

The amplitude distribution was typically quite broad, ranging from - 10 pA to >200 pA in the cell shown in Fig. 1B. Addition of the y-aminobutyric acid (GABA), receptor antagonist bicuculline (5 PM) reduced the frequency of events to 1.7 t 0.5 Hz (n = 1 1 ), with only small amplitude events persisting ( 18 t 2.3 pA; n = 11, Fig. 1, C and D) . These events are presumably EPSCs, because they reversed at ~0 mV and were blocked by glutamate receptor antagonists ( see below).

Tetrodotoxininducesan outwardcurrent Addition of TTX to the bathing medium invariably resulted in a reversible outward current of +30 t 5.1 pA at a holding potential of -60 mV ( n = 10 neurons, Fig. 2). This outward current was associated with a large decrease in the baseline noise, which was probably not due to tonic synaptic activity because neither bicuculline, CNQX, nor D-APV reproduced the effect of TTX. The outward current induced by TTX is likely caused by the removal of a persistently activated inward sodium current (French et al. 1990). Under these conditions, spontaneous “miniature” EPSCs were typically of small amplitude (see below) and low frequency (