Possibility of multiquantal transmission at single inhibitory synapse in cultured rat hippocampal neurons

Pergamon PII:

Neuroscience Vol. 92, No. 4, pp. 1217–1230, 1999 Copyright q 1999 IBRO. Published by Elsevier Science Ltd Printed in Great Britain. All rights reserved 0306-4522/99 $20.00+0.00 S0306-4522(99)00084-6

POSSIBILITY OF MULTIQUANTAL TRANSMISSION AT SINGLE INHIBITORY SYNAPSE IN CULTURED RAT HIPPOCAMPAL NEURONS S. A. FEDULOVA,*† D. V. VASILYEV,* E. V. ISAEVA,‡ S. G. ROMANYUK‡ and N. S. VESELOVSKY* *Center of Molecular Physiology, National Academy of Science, Bogomoletz Street 4, Kiev-24, Ukraine ‡Bogomoletz Institute of Physiology, National Academy of Science Bogomoletz Street 4, Kiev-24, Ukraine

Abstract—Miniature, spontaneous and evoked inhibitory postsynaptic currents were studied using the whole-cell patch-clamp technique on synaptically connected cultured hippocampal neurons, at a holding potential of 275 mV. All experiments were done in tetrodotoxin-containing solution to exclude an action potential generation. Spontaneous miniature inhibitory postsynaptic currents were observed in Ca 21-free solution. The distribution of miniature inhibitory postsynaptic currents was skewed to larger current amplitudes and could be fitted reliably by one Gaussian with the mean at 10.0 ^ 1.2 pA (n ˆ 7). Spontaneously occurring whole-cell spontaneous inhibitory postsynaptic currents were recorded in physiological solution (Ca 21 2 mM). The average amplitude of spontaneously occurring currents depended on membrane potential and reversed at 2 18 ^ 5 mV (n ˆ 5). The amplitude distribution of spontaneous inhibitory postsynaptic currents had one peak clearly detectable with the mean of 20.0 ^ 2.0 pA (n ˆ 6) or 10.0 ^ 2.0 pA (n ˆ 2). Inhibitory postsynaptic stimulus-evoked currents arose in responses to gradual activation of neurotransmitter release by direct extracellular electrical stimulation of a single presynaptic bouton by short depolarizing pulses. The current–voltage relation of the averaged amplitudes of evoked inhibitory postsynaptic currents was linear and reversed at potential predicted by the Nernst equation for corresponding intra- and extracellular Cl 2 concentrations. The time-course of decay of miniature, spontaneous and evoked inhibitory postsynaptic currents was fitted by a sum of two exponents and their timeconstants were the same in the range of standard deviation. The stimulus-evoked inhibitory postsynaptic currents fluctuated with regard to the discrete aliquot values of their peak amplitudes in all the investigated synapses from a measurable minimum of about 8 pA to 200 pA. The evoked inhibitory postsynaptic currents were assumed as superimposition of statistically independent quantal events. Fitting amplitude histograms of evoked inhibitory postsynaptic currents with several Gaussian curves resulted in peaks that were equidistant with the mean space of 20 ^ 3 pA (n ˆ 10), which was assumed as one quantum (quantum size) to construct the Poisson’s distribution. A possibility of simultaneous multiquantal release at single inhibitory synapses of rat hippocampal neurons was discussed. q 1999 IBRO. Published by Elsevier Science Ltd. Key words: single presynaptic bouton, IPSC, Poisson’s statistics.

Inhibitory postsynaptic currents (IPSC) mediated by GABA presynaptic release in inhibitory synapses of cultured hippocampal neurons were studied in this paper. IPSC arises from the opening of Cl 2 —selective ionic channels activated by GABAA receptors. 4 GABAA receptors are major inhibitory receptors in the CNS and their functional properties have been extensively studied in the somatic membrane of hippocampal neurons in culture and in slices by patch-clamp technique. 4,6,32,33 However, the precise †To whom correspondence should be addressed. Abbreviations: DL-AP5, dl-2-amino-5-phosphonovaleric acid; DNQX, 6,7-dinitroquinoxaline-2,3-dione; EGTA, ethyleneglycolbis(aminoethylether)tetra-acetate; eISPC, evoked inhibitory postsynaptic currents; HEPES, N-2hydroxyethylpiperazine-N 0 -2-ethanesulphonic acid; ISPC, inhibitory postsynaptic currents; mIPSC, miniature inhibitory postsynaptic currents; sISPC, spontaneous inhibitory postsynaptic currents; TTX, tetrodotoxin.

statistical analyses of quantal GABA release in single central synapses have been complicated by technical limitations. Some difficulties specific for central neurons, such as numerously distributed inputs of neurons, difficulties with identification of pre- and postsynaptic cells, noises of various origins, a very high probability of transmitter release, etc. 2,10,14,15,31 precluded reliable recording and statistical analyses of data. To adequately describe probabilistic process, which is the principal feature of spontaneous and evoked transmitter release, the Poisson’s statistics were successfully applied. 2,3,10 The Poisson’s statistics describe the casual number of the appearance of a given event at a given time at a given place. The probability P(x) that a given event was observed exactly x times in a package of N observations or the probability P(x) that a given response is made of 1, 2, 3,...x quantal

1217

1218

S. A. Fedulova et al.

Fig. 1. Local extracellular stimulation of single presynaptic terminals. (A) Equivalent circuit of experiments. Rpip, Cpip, resistance and capacitance of stimulating pipette; Rsh, shunt resistance of bath solution; Rb, Cb, input resistance and capacitance of the bouton and Rpre, input resistance of presynaptic neuron, Rp, Cp, resistance and capacitance of localized surface of postsynaptic membrane involved in potential perturbation. Point X indicates pipette mouth. (B) Linear dependence of local short-time potential perturbation amplitude on output voltage of stimulating device at a minimal fixed distance between the two pipettes. Linear increase in longitudinal (C) and transversal (D) distances between stimulating and recording pipettes resulted in exponential decrease of field potential perturbation. Scales of axes are the same for C and D. (C, D) Inset—the glass pipettes (tip openings 1 mm) were filled with bath solution and connected to the input of the microelectrode amplifier (right) and the output of stimulation device (left), which generated 3 ms stimuli of 15 V. Recording pipette was moved by a micromanipulator (shown in insets) to measure produced perturbation of bath potential by stimuli of 15 V.

components is given by P(x) ˆ e 2mm x/x!, x ˆ 0, 1, 2,.....The mathematical expectation of Poisson’s statistics has to be a constant value m. The m is the average number of units responding to one impulse, or the mean quantal content. This description implies a low probability of the event P(x) (less than 0.1) and a large number of N observations in a package. 2,10,14,15 The main goal of this study was to analyse the quantal nature of evoked IPSC (eIPSC) as a response to neurotransmitter release from single synaptic boutons. Our experimental approach combined local extracellular electrical stimulation of a single presynaptic bouton and whole-cell patch-clamp recording of IPSCs. These techniques allowed us to obtain a reliable simultaneous control of

potentials over identified pre- and postsynaptic membranes. Under these conditions we could provide a low probability of release that was necessary for statistical analyses of amplitude distribution of IPSCs predicted by Poisson’s statistics. We also compared the time-course of spontaneous and eIPSCs, analysed the quantal properties of their amplitudes and estimated the value of one quantum.

EXPERIMENTAL PROCEDURES

Culture with low density of neurons The culture of rat hippocampal neurons was prepared as described previously. 18 The animals were decapitated by guillotine. The hippocampi of newborn Wistar rats (Animal Resources Unit, Bogomoletz Institute of Physiology, Kiev)

Multiquantal transmission at inhibitory synapses

1219

Fig. 2. Effectiveness of local extracellular stimulation of individual boutons. (A) In the top—phase-contrast image of a hippocampal neuron with putative axo-dendritic synapses. In the bottom—the enlarged framed part of the top. Positions of the stimulating pipette are shown by arrows. (B) In the top—evoked IPSCs with position of the stimulating pipette directly at the center of investigated bouton (R0). The amplitude of stimulating current 0.5 mA, voltage output 5 V. In the bottom—shift of the stimulating pipette (arrows) in any direction (indicated by circle) with 3 mm (R) away from the centre of synaptic bouton resulted in disappearance of postsynaptic response; the amplitude of stimulating current 1 mA, voltage output 10 V. Here and thereafter a trace of averaged failures (n ˆ 3) was subtracted to avoid stimulating artifacts.

were mechanically dissociated, the cells were plated onto poly-l-ornithine-laminine-coated Petri dishes at a low density of 30,000 to 35,000/cm 2. The neurons were cultured in Eagle’s modified medium containing 10% horse serum, 1% glucose, 0.6 ml glutamine, 2.3 g/l NaHCO3, 6 mg/ml insulin at 358C in a 95% air and 5% CO2 humidified incubator. Three days after plating 1 mM cytosine-A-D-arabinofuranoside was added for 24 h to the culture medium to reduce glia proliferation. Solutions Extracellular bath solution contained (in mM): NaCl 140, KCl 3, CaCl2 2, MgCl2 2, Glucose 30, HEPES 20, pH 7.4. Tetrodotoxin (TTX) 0.25 mM, dl-2-amino-5-phosphonovaleric acid (DL-AP5) 20 mM, 6,7-dinitroquinoxaline-2,3dione (DNQX) 20 mM were always added to the bath solution. Potassium current blockers (4-aminopyridine, tetraethylammonium) and solutions with different Ca 21 concentration were applied by fast local superfusion 34 close to the nearby zone of the investigated neuron. The intracellular solution for patch pipette contained (in mM):

K-gluconate, 100, KCl 50, MgCl2 5, EGTA 5, HEPES 20, pH 7.4 (all from Sigma). Recording of postsynaptic currents IPSCs were recorded from a postsynaptic neuron in a whole-cell configuration using home-made amplifier for continuous voltage-clamp with series resistance compensation. 16 Patch pipettes (opening diameter 1 mm) were pulled out from borosilicate glass (WPI, U.S.A.) in two steps. 8 The resistance of patch pipettes ranged between 3–4 MV. To obtain the best control of intracellular potential and to reduce “dendritic filtering”, we recorded only from the bouton situated on or close to the soma of postsynaptic neuron (not farther than 10–20 mm). Series resistance compensation was in the range of 50–70%. Currents were filtered by a Bessel bandpass filter of the 2nd order with cut off frequency of 2–5 kHz and digitized at 5–10 kHz using LabMaster TL-1 and a pClamp v. 5.0 software (Axon Instruments). All experiments were carried out on seven to 18 day cultured neurons at 18–208C.

1220

S. A. Fedulova et al.

Fig. 3. Amplitude distributions of sIPSCs in external physiological (Ca 21, 2 mM) and Ca 21-free solutions for three different neurons. (A) Histogram of sIPSCs amplitude obtained in physiological solution. Here and thereafter continuous lines represent fitting of the histogram with Gaussian curves. Histogram was fitted by two Gaussian distributions with maxima at 217.5 and 238 pA. (B) Examples of continuous recording of sIPSCs in physiological solution. The distance between dashed lines is equal to 20 pA. (C) Amplitude distribution of sIPSCs for the second neuron with position of histogram maxima at 29.7 and 221.3 pA in physiological solution. (D) Amplitude distribution of mIPSC for the third neuron in Ca 21-free solution. Selected traces of mIPSC are shown in inset. Histogram was fitted by two Gaussian curves with maxima at 29.5 pA and 222 pA.

Multiquantal transmission at inhibitory synapses

Fig. 4. Effect of postsynaptic holding potential and extracellular presynaptic stimulating current on stimulusevoked IPSCs. (A) Inset—averaged traces of 40 eIPSCs at different membrane potentials of postsynaptic neuron; Vh is indicated near the corresponding curve. IPSCs were evoked by 3 ms stimuli of fixed amplitude. The Nernst equilibrium potential for given extra-/intracellular concentrations of Cl 2 was 2 19 mV. (A) Current–voltage (I– V) dependence of the averaged eIPSC on the level of postsynaptic holding membrane potential. (B) Top traces— the averaged traces of IPSCs evoked by 3 ms stimuli pulses of different intensity are shown. Each trace represents the average of 40 eIPSCs at fixed amplitudes of extracellular pulse. Stimulus amplitude was linearly increased (traces 1–6), the seventh trace is control experiment of stable recording. Bottom graph— dependence of the mean amplitude of eIPSCs on stimulus intensity.

1221

1222

S. A. Fedulova et al.

Fig. 5. Comparison of time-course of miniature, spontaneous and stimulus-evoked IPSCs. (A–C) Selected individual traces of spontaneous (A), miniature (B) and evoked (C) IPSCs with small and large amplitudes. The scale bars refer to all traces. (D) Normalized traces of individual evoked and miniature IPSCs. The decay phases of IPSCs were fitted by two exponents (thin lines). The exponents had time-constants of tf ˆ 7.1 ms, (Af ˆ 55%) and ts ˆ 48.4 ms, (As ˆ 45%) for mIPSC and tf ˆ 8.2 ms (Af ˆ 40%) and ts ˆ 46.9 ms (As ˆ 60%) for eIPSC in physiological solution. (E) Left traces—rising phase for two selected individual eIPSCs with amplitudes of 40 pA and 120 pA. Right traces—normalized traces of those individual eIPSCs. Smooth line represents eIPSC with small (40 pA) amplitude, dotted line represents eIPSC with large (120 pA) amplitude. Points indicate the experimental sampling interval.

Extracellular local stimulation of single boutons A stimulating pipette with inner diameter of 1 mm was placed close to the bouton at a distance of 0.5–1 mm (Fig. 2A). The pipette was filled with bath solution (resistance 8– 10 MV) and connected to a self-made stimulus isolator unit. Shape and amplitude control of stimulating current through

the pipette was monitored by a special current amplifier in the output circuit. The current–voltage relationships of these pipettes were linear in the range of applied voltages 0–25 V. Voltage applied to glass micropipette filled with electrolyte usually results in a dual effect on extracellular stimulation. This voltage produces total current flowing from the

Multiquantal transmission at inhibitory synapses

output of stimulating device via pipette resistance (Rpip ˆ 8– 10 MV) and small value of shunting resistance of bath solution, 60 V × cm, 9 to the grounded reference electrode (Fig. 1A). The first effect is revealed when a part of this current (proportional to the ratio Rsh/Rm) flows via membranes of neurons, depolarizes them and produces an effective voltage for activation potential-dependent conductances. Small amplitudes of stimulating voltage (0–25 V) in our experiments produced total current not larger than 2.5 mA. A part of total current flowing through the neuronal membrane was too small to exert any reasonable effect of the current stimulation itself. In this case the stimulation could be performed in the other way—by local change of potential in bath solution. The stimulating current flowing through the divider Rp/ Rsh produced a local change in bath potential at point X (Fig. 1A), on the border between the mouth of the pipette and bath solution. This potential decreased gradually in all directions as a function of distance from the pipette tip according to equation:

w ˆ wo exp…2x=xo †; where wo is potential at x ˆ 0 (opening of the pipette) and xo is a length constant. The relationship between the output voltage of stimulating device (0–25 V) and the potential close to the mouth of stimulating pipette was linear (Fig. 1B). The dependence of potential on the distance from stimulating pipette could be measured experimentally (Fig. 1C, D). Two pipettes were filled with bath solution and connected correspondingly to the output of stimulating device and the input of the microelectrode amplifier (MZ-4, Nihon Kohden). The potential at a distance of 5 mm from the stimulation pipette in longitudinal (Fig. 1C) and transverse (Fig. 1D) directions decreased to the values less than 5 mV when stimulating voltage was 15 V. Usually stimulating pulses used for single bouton activation were in the range of 6.5 ^ 3.5 V (n ˆ 143) and a zone of effective voltage shift (0–100 mV) was not more than 2–3 mm in diameter. These pulses altered the extracellular potential in a small area that included presynaptic bouton and a small patch of underlying postsynaptic membrane. By applying negativegoing pulses to stimulating pipette, a bouton could be depolarized for definite periods that resulted in the appearance of typical fluctuating postsynaptic currents. The shift of stimulating pipettes even for 1.5–2 mm (Fig. 2A) from the bouton resulted in a complete disappearance of eIPSCs (Fig. 2B). The area of bouton membrane that becomes discharged is approximately 1 mm 2. The corresponding area of underlying postsynaptic membrane is of the same order. The membrane potential change followed the shift of extracellular potential with a very little time lag. Corresponding time-constants (about 10–20 ms) could be calculated approximately from the capacitance of small membrane area (about 0.01 pF) and high input resistance of the bouton and postsynaptic membrane under the stimulating pipette. 13 Identification of single boutons After establishing the whole-cell configuration between recording pipette and postsynaptic neuron a single bouton on this neuron could be defined (Fig. 2A). Visual control of investigated neurons and nerve terminals was performed at total magnification of × 1000 (objective Plun-NEOFLUAR × 100, aperture 1.3, oil immersion, phase-contrast pH 3, Zeiss). Optical resolution of this system for the wavelength of visual spectrum (0.65 mm) is 0.35 mm, that is usually enough to identify different types of nerve terminals. Usually only two to three neurons could be seen in the field of vision (diameter ˆ 200 mm) and visual identification of single synaptic connections was not difficult (Fig. 1A). In some special cases we identified boutons with fluorescent dye FM1-43 that demonstrated complete correlation with

1223

visual control under phase contrast as was shown earlier. 6,7 These conditions and mechanical stability of micromanipulators were the criteria for proper local stimulation. Data analysis Families of original traces of eIPSCs were usually analysed by one of the following softwares: WinTida v. 2.64 (3.0) (HEKA, Germany), AutesP v. 6.0 (Dr H. Zuker, MPI for Psychiatry, Martinsried, Germany), Clampfit v.6.0 and AxoScope v. 1.1 (Axon, U.S.A.). Amplitude and time parameters of each spontaneous and eIPSC were collected manually or automatically into a worksheet for analyses in Origin v.3.5 (4.1) (Microcal, U.S.A.). The eIPSCs amplitude histograms with peaks clearly detectable by eye were fitted by Gaussian distribution using Simplex-based least square algorithm. The variance of instrumental noise sN ˆ 1.9–2.8 pA was always 10–15% of the peak of the first distribution. The S.D.(n) of Gaussian fitting to peak n in the√IPSCs amplitude histogram were calculated as S.D.(n) ˆ (sN2 1 nsq2)., where sq2 is S.D. arising from biological sources 4 and in our experiments sq ˆ 3.6–4.2 pA. Autocorrelation analysis was performed for histograms with an equidistant peaks as was described earlier. 4 Unless indicated otherwise, experimental data specified in the text are represented as means ^ S.D. RESULTS

Spontaneous inhibitory postsynaptic currents Whole-cell IPSCs were studied using the patchclamp technique on synaptically connected cultured hippocampal neurons. Analogously to earlier observations, 4,6,7,22 spontaneously occurring whole-cell IPSCs were recorded in TTX containing physiological solution at a holding potential of 275 mV. (Figs 3B, 5A). The average amplitude of spontaneously occurring currents depended on membrane potential and reversed at 2 18 ^ 5 mV (n ˆ 5), calculated by the Nernst equation equilibrium potential for given in- and outside concentrations of Cl 2, was 219 mV. Bath application of 2 mm bicuculline resulted in a complete reversible block of any detectable activity immediately after application during recording period of about 30 min (n ˆ 4). We defined these currents as GABA-mediated Cl 2 currents—IPSCs. Amplitude distributions of spontaneous IPSC (sIPSC) were constructed for eight neurons in physiological solution with the number of realizations not less than 130, collected over 10 min from each neuron. Continuous recording demonstrated occurrence of sIPSC at a rate of about 8–17/min. In all the investigated neurons (including the two shown in Fig. 3A, C) the amplitude distribution of sIPSC had one peak clearly detectable by eye which could be fitted by a Gaussian curve. In some cases, the skewed amplitude distribution could be fitted by a sum of two Gaussian curves (Fig. 3A, C). The autocorrelation analysis performed for all neurons as criterion for the existence of peaks, indicated the second peak as non-significant. Mean distance between the peaks calculated from the means of the fitted distributions coincided with the position of the

1224

S. A. Fedulova et al. Table 1. Comparison of time-courses for spontaneous and evoked inhibitory postsynaptic currents IPSC

Time to peak, ms tf ms

mini 3 ^ 1.2 9.7 ^ 4.2 Spontaneous 1.4 ^ 0.7 9.1 ^ 4.5 Evoked 1.1 ^ 0.4 8.4 ^ 3.9

Time of decay Af%

ts ms

As%

n

43 41 49

54 ^ 15 62 ^ 18 52 ^ 16

57 59 51

18 35 32

Time-to-peak for individual current traces was estimated from 10% to 90% of the maximal amplitude. tf, time-constant and Af, the amplitude of the fast exponent in currents decay; ts, time-constant and As, the amplitude of the slow exponent in currents decay. n, the number of individual traces for calculating the mean value. Data are means ^ S.D.

about 2–4/min. The amplitude distributions of mIPSC revealed one peak clearly detectable by eye that could be described by a Gaussian distribution (Fig. 3D) with the mean at 10.0 ^ 1.2 pA. In four of seven neurons (including one shown in Fig. 3D) additional peaks formed by larger current amplitudes could be distinguished and fitted by Gaussian curves, 2 but the autocorrelation analysis showed their non-significance. All the investigated neurons had spontaneously occurring mIPSCs with the amplitude more than 80 pA (Fig. 3D, in the inset). Evoked inhibitory postsynaptic currents

Fig. 6. Quantal origin of eIPSCs amplitudes distribution. (A) Amplitude histogram of stimulus evoked IPSCs (n ˆ 320). Dotted line represents the sum of five underlying Gaussian distributions (smooth line) assuming independent superposition of quantal events, failures being included. Inset—first 50 traces out of 320 evoked IPSCs. (B) Criterion for equidistant peaks is represented by calculation of autocorrelation function (squares) for the experimental data shown in panel A. Smoothed autocorrelation function is shown by circles. The original autocorrelation function intersects the smoothed function at regular intervals of about 20 pA (n ˆ 8), that indicates three successive equally spaced peaks and dips.

first peak in Gaussian distribution and was equal to 20.0 ^ 2.0 pA in six neurons (Fig. 3A), in two investigated neurons this value was 10.0 ^ 2.0 pA (Fig. 3C). Miniature IPSCs (mIPSC) were recorded in Ca 21free solution in the presence of 0.25 mm TTX (Fig. 5B). mIPSCs were collected from seven neurons with the number of realizations for each of them not less than 120. Continuous recording (over 25 min) indicated a rate of mIPSCs occurrence of

Evoked IPSCs were revealed in the TTX-containing physiological solution as a response to short extracellular depolarizing voltage pulses in close vicinity of a single presynaptic bouton (Figs 2B, 5C, 6A, 8A, C). The stimulus-evoked IPSCs were similar to spontaneously occurring IPSCs and reversed at the potential of 216 ^ 4 mV (n ˆ 4), the equilibrium potential calculated by the Nernst equation for the corresponding Cl 2 concentrations was 219 mV (Fig. 4A). The linear increase of stimulating pulse amplitude caused a nonlinear bell-like changes of the averaged eIPSC amplitude (Fig. 4B). During small increase of stimulating pulse the averaged eIPSC enhanced until some maximal value. Further linear increase of stimulating pulse intensity resulted in the reversible decrease of the averaged eIPSC amplitude (Fig. 4B, on the top). Time-course of individual eIPSCs did not depend on the value of stimulus (not shown). The time-course of the stimulus-evoked IPSCs was identical to that of spontaneous currents, as shown for two different amplitudes of individual mini, spontaneous and evoked current traces in Fig. 5A, B, C, respectively. The average parameters of time-course were estimated for selected traces of equal amplitudes in each from 10 investigated synapses (eIPSC) and five neurons (mIPSC and sIPSC). Time to peak was estimated for the rising part of the current from 10% to 90% of maximal

Multiquantal transmission at inhibitory synapses

1225

Amplitude distribution of the stimulus-evoked inhibitory postsynaptic currents

Fig. 7. Stimulus-evoked transmitter release from single central synaptic bouton follows Poisson’s statistics. (A) Experimentally obtained numbers of failures, singles, doubles, etc. of quantal units (columns) and predicted corresponding values from Poisson’s statistics (asterisks) for the value of one quantum 20 pA (the same data as shown in Fig. 6A). Here and thereafter N ˆ total number of traces, the mean quantal content m ˆ mean amplitude of IPSC/mean value of one quanta. (B) The mean quantal content estimated by the method of counting failures, assuming Poisson’s release statistics corresponds to the value calculated by dividing the average eIPSC amplitude by one quantum (20 pA). Plot represents data for 12 individual boutons. The line indicates the linear regression with a correlation coefficient of 0.95.

amplitude. The decay of individual evoked, spontaneous and mini IPSCs was best fitted by two exponents (Fig. 5D). The time to peak was the same for eIPSC with small and large amplitudes, as shown in Fig. 5E for two individual current amplitudes of 40 pA and 120 pA. Numerical values for time to peak and time-constants of current’s decay of mini, spontaneous and evoked IPSC could be considered as equal in the range of S.D. (Table 1).

The eIPSC amplitudes varied from a measurable minimum of about 7–8 pA to more than 200 pA, at holding potential 275 mV (Fig. 6A, in the inset). The stimulus-evoked IPSCs fluctuated with regard to the discrete aliquot values of peak amplitude in all the investigated synapses (n ˆ 32) (Figs 6A; 8A, C, in the inset). In the majority of neurons the reliable stimulation of single bouton and recording of eIPSC was not long-lasting enough to collect an adequate number of eIPSC for detailed quantal analysis. The frequency of stimulation was one 1/ 5 s, hence to collect at least 300 traces we needed 25 min of an absolutely fixed position of stimulating pipette, because its drift even at 1.5–2 mm caused a complete disappearance of eIPSC (Fig. 2B). The amplitude distributions were constructed for eight different synapses which met these conditions. They revealed clearly distinguished regularly spaced multiple peaks as shown in Figs 6A and 8A, C for two selected boutons. Fitting histograms with several Gaussian curves assuming independent superposition of quantal events resulted in equally spaced (quantal) successive peaks. In histograms with equidistant peaks the space between the peaks was on the average 20 pA irrespective of the amplitude of stimulating pulse (Figs 6A; 8A, C). Assuming that space between the peaks was one quantum (or quantal unit) of an integral IPSCs, 4 the mean value of one quantum was equal to 20 pA at the holding potential of 275 mV. The value of one quantum 20 ^ 3 pA was obtained from the amplitude distributions for eight single synaptic boutons with number of realizations in each experiment not less than 270. A superposition of autocorrelation analyses performed for the histogram in Fig. 6A and smoothed autocorrelation function showed successive equally spaced peaks and dips (Fig. 6B), thus the IPSCs amplitudes were significantly quantal. Counting quanta and description by Poisson’s statistics To use Poisson’s statistics for the analysis of our experimental data, we should transfer continuous amplitude distribution into discrete distribution of quantal units. To construct the distribution of the number of quantal units in each response, a single quantal response (quantal unit) has to be distinguished from multiquantal events. The distribution of eIPSCs amplitudes gave the value of one quantum 20 pA. The IPSCs amplitudes falling into the range of ^ 8 pA were not included in the analysis as independent currents, and the corresponding traces were assumed as failures. To construct Poisson’s distribution we took the bin size equal to the value of one quantum (20 pA). The range for an event consisting of one quantum was between 28 pA and 228 pA,

1226

S. A. Fedulova et al.

Fig. 8. Effect of different intensities of stimulation on eIPSC amplitude distributions at the same hippocampal bouton. (A) Experimental histogram of eIPSCs amplitudes was fitted by three Gaussians (smooth line) assuming independent superposition of quantal events, failures were included. IPSCs evoked by brief (5 ms) depolarizing pulses, stimulating current 0.5 mA. Inset—first 40 traces of eIPSC out of total number n ˆ 201. (B) Experimentally obtained numbers of failures, singles, doubles, etc. of quantal units (columns, the same data as shown in panel A) and predicted corresponding values from Poisson’s statistics (asterisks) for the value of one quantum 20 pA. (C) Amplitude distribution of eIPSCs obtained for the same bouton by stimuli of 5 ms, stimulating current 0.8 mA. Inset—first 40 traces of eIPSC out of total number n ˆ 310. (D) Experimental diagram of the number of quantal units in each response (column, the same data as shown in panel C) and predicted values from Poisson’s statistics (asterisks).

Multiquantal transmission at inhibitory synapses

two quanta ranged between 2 28 pA and 2 48 pA and so on. The dependence of the number of quantal units on the number of their realizations in each multiquantal event is shown in Fig. 7A and Fig. 8B, D. In all the investigated synapses, the eIPSCs with the amplitude fluctuating around 9 pA were measured. For example, in the described synapse (Fig. 6A) eIPSCs with amplitude of about 9 pA were recorded eight times from the total number of traces N ˆ 320. We did not make special analysis of small currents because of the insignificant number of such realizations in each synapse. To test the applicability of Poisson’s law, we calculated the m value in two ways. 3,10 The first: m ˆ mean amplitude of IPSC= mean value of one quantum: The mean amplitude was obtained as the algebraic sum of all amplitudes divided by a total number of traces. The second way was obtained from Poisson’s law for x ˆ 0 or failure of response m ˆ ln …number of observations N= number of failures in the package N0 †; where the number of N observations is the total number of traces, so mean amplitude of IPSC= mean value of one quanta ˆ ln …N=N0 †: The graphic representation of this equation (Fig. 7B) was satisfactorily fitted by linear function with a slope of 458. This correlation between the two determinations of m for 12 cells with different amplitudes of extracellular stimulating current pulses supported the original hypothesis about the value of one quantum and applicability of Poisson’s statistics to our experimental data. The quantal content m for all the investigated synapses was enhanced with the increase of stimulating pulse amplitude in the range of linear dependence of the averaged eIPSC amplitudes on stimulating pulse amplitude (Fig. 4B). In this range the mean value of one quantum was constant and approximately equal to 20 pA (Fig. 8) and the enhancement of the quantal content m was based on the increase probability of realizations of currents with large amplitudes. DISCUSSION

The application of local extracellular electrical stimulation to a single presynaptic bouton and

1227

patch-clamp recording of eIPSCs in postsynaptic neuron broadened the possibilities to study quantal nature of eIPSC in single inhibitory synapses. In our experimental conditions, Poisson’s statistics adequately described the amplitude distributions of eIPSCs without artificial lowering the probability of release by pharmacological agents. 10 The value of quanta did not depend on experimental procedures applied to the same synapse, and the mean value of one quantal unit was 20 pA for various depolarizing stimuli in all the investigated synapses. Statistical description of our experimental data suggested the possibility of simultaneous multiquantal release of neurotransmitter from presynaptic bouton in a single synapse. Thus, some features of IPSCs recorded in response to stimulation of a single presynaptic bouton demonstrated that transmission in these experiments has specific properties which could not be observed under other experimental conditions. Spontaneous and mini inhibitory postsynaptic currents The amplitude distributions of spontaneous IPSC were skewed, although in half of the experiments the second peak formed by larger current amplitudes could be distinguished and fitted by Gaussian distribution. The sIPSC (Fig. 3B) demonstrated amplitudes which fluctuated with regard to the value of 20 pA and double value of 40 pA. The distance between the peaks coincided with the position of the first peak for such distributions that is consistent with the assumption that sIPSC amplitudes were quantally distributed in physiological and Ca 21free solutions. However, such modification of sIPSC amplitude could be explained by the fact that currents may originate from several synapses of different presynaptic neurons or by the different number of available postsynaptic receptors opposite a different single synaptic bouton. It should be mentioned that the main feature of sIPSCs was the essential difference between the currents recorded in normal physiological saline and Ca 21-free solution. The first distinction is that the values of the quantal event in Ca 21-free and physiological solutions measured at the same holding potential were 10 pA and 20 pA, respectively. The increased value of one quantum in normal solution may result from either an increase in open probability of postsynaptic GABAA-receptor Cl 2 channels in Ca 21-containing solution or a change in single-channel conductance. The increase in open probability to its maximal possible value (Po ˆ 1) could cause no more than 25% of the increase in one quantum. 22 From our experiments it is not possible to determine the reason for such a change of one quantum value but most likely the reason might be the existence of two common conductance states (14 and 23 pS), probably representing two different

1228

S. A. Fedulova et al.

subtypes of GABAA-receptor channels. 4 Presumably, in Ca 21-containing solution, an opening probability of channels with larger conductance is much higher than that of channels with lower conductance, whereas in Ca 21-free solution the lower conductance channels had higher opening probability. 4 The differences in times to peak measured for sIPSC in Ca 21-free and in physiological solutions (Table 1) are corroborate the letter assumption, i.e. the channels with different conductances may mediate sIPSC in Ca 21-free and Ca 21-containing solutions. In TTX-containing non-buffered external Ca 21free solution the generation of action potential was impossible but spontaneous minis with large amplitudes could be caused by synchronizing release at few synapses. On the other hand, calculated probability of a few uniquantal minis if uniquantal release is assumed was less than experimentally obtained probability of appearance spontaneous minis with the amplitude larger than 30 pA (which coincide with three independent events) during the recording period of 40 min. However, the difference was not significant enough to draw an explicit conclusion about the possibility of spontaneous multiquantal minis.

Evoked inhibitory postsynaptic currents The stimulus-evoked IPSCs reversed at the equilibrium potential calculated by the Nernst equation for the corresponding Cl 2 concentration and voltage current dependence of IPSCs was linear. This means that controlled voltage on the postsynaptic membrane permitted to avoid dendritic filtration resulting in sufficient voltage gradient on postsynaptic membrane and to describe the kinetics of eIPSC. The quantity of neurotransmitter released from each presynaptic nerve terminal is directly proportional to the increase in the intraterminal Ca 21 concentration evoked by each invading action potential. 12 Previous studies performed on neuromuscular junction under voltage-clamp of presynaptic membrane 17,37 indicated that the voltagedependent calcium current triggered the release of synaptic transmitter, so the averaged EPSC amplitude increased and decreased directly proportional to calcium current amplitude and duration. In our experiments the dependence of averaged eIPSC on stimulating current was similar to that for the relationship between EPSCs and calcium current amplitude. 17,37 It could mean that (i) local extracellular stimulation did not evoke any action potential-like generation on presynaptic membrane otherwise we would get dependence of the averaged eIPSCs with saturation, and (ii) 3–5 ms of stimulus duration are enough to get activation of calcium currents that evoked neurotransmitter release. Obviously, such nonlinear dependence proved that the increase of stimulating current intensity did not increase the

field experienced by recruiting responses from other remote boutons. Possibility of simultaneous multivesicular release The basic idea of the quantal model is that neurotransmitter is released from presynaptic terminal in discrete units thought to be equal to the amount of neurotransmitter packaged within a single presynaptic vesicle. 12 Each vesicle contains one quantum (several thousand molecules) of transmitter. The vesicles are thought to fuse to the inside surface of the presynaptic terminal at specific release sites. The membrane opens transiently so as to allow the vesicle to extrude its entire contents, in an all-ornone fashion, into the extracellular space of the synaptic cleft. The number of release sites associated with an axon has been identified by the total number of releasable vesicles, the number of boutons, and the number of active zones. 11,14,15,21,23,24,26–29,35,36,38 The number of morphologically distinguishable active zones in one bouton of central neurons is defined as one, sometimes two 5,29 and rarely three 29 as contrasted to junction with high quantal content and correspondingly large number of presynaptic active zones, e.g., 200–300 at the frog neuromuscular junction and about 10,000 at the squid giant synapse. 20,25 In our experimental configuration it means that visible morphological structure, i.e. the bouton of nerve terminal, has to be identical to one active zone. The number of release sites is generally accepted as corresponding to the number of active zones, although without definite evidence. In our experimental condition, this could mean that the bouton is identical to one active zone with one release site for only one vesicle. The mean diameter of a vesicle measured for brain synapses was 35.2 ^ 3.5 nm. 29 This variability may provide a significant difference in vesicular entire content or in one quantum that could result in fluctuation of IPSCs caused by the release of each vesicle. However, such fluctuations have to vary randomly from release to release around some mean value and such probabilistic process is described by normal distribution (Gaussian) with one peak. In our experiments, we obtained histograms for the amplitude distribution of IPSCs with regularly spaced peaks (Figs 6A; 8A, C). This allows us to suggest that transmitter release occurs by discrete, approximately equal portions which could sum up and result in multipeak distributions of eIPSCs, because the amplitude of the postsynaptic response is proportional to the quantity of released neurotransmitter. The number of peaks changed for different boutons from two to six excluding first peak around zero. For the bouton with six peaks in our experimental condition this could mean that one bouton had (i) one active zone with six release sites or (ii) two active zones with three release sites, or (iii) three active zones with two release sites,

Multiquantal transmission at inhibitory synapses

because the presence of the sixth peak in the histogram means only simultaneous release of six vesicles. These assumptions were confirmed by morphological study which showed the presence of a few active zones as well as multiple docked vesicles per one active zone in single boutons of mouse hippocampus. 29 The possibility of multivesicular release was assumed for excitatory synapses of cultured hippocampal neurons in control conditions and when baclofen, 4-aminopyridine and adenosine were applied, 31 and for inhibitory synapses in cerebellar stellate and basket cells. 1 Nevertheless, individual hippocampal synapses differ greatly from each other with respect to their release probability, the value for one quantum being approximately equal. The Poisson’s statistics were successfully applied to 20 pA as one quantum assuming the possibility of simultaneous release of several vesicles in single bouton. The value of IPSC for GABAA receptors could be determined as IPSC ˆ N × GCl2 × E, where N is the number of opened GABAA channels, GCl 2 is conductance of single GABAA channel, E is electromotive force E ˆ VH 2 ECl, VH ˆ 275 mV; ECl ˆ 219 mV—the Nernst potential for Cl 2 concentrations in physiological solutions. In previous papers the value of one quantum for hippocampal neurons at the same electromotive force was 7–12 pA for small quantal size and 17–21 pA for a larger quantal size. 4 This was explained by the existence of two GABAAreceptor channel subtypes with a low and large conductances. 19,30 In our experiments the value of 18–22 pA could be reliably defined as quantum size using data of spontaneous activity and statistical analyses, although the eIPSC of small amplitudes of about 7–10 pA could also be measured. The value of one quantum calculated for eIPSCs

1229

coincided with that obtained from distribution of spontaneously occurring currents for the same bouton (Figs 3A, 6A). One may suggest that direct electrical stimulation of a single bouton is an adequate method to further study the mechanisms of synaptic transmission. The Poisson’s statistics described probabilistic processes with a low probability of events or a low mean quantal content of the release. Our data show that the mean quantal content is directly correlated with the amplitude of the stimulus (Fig. 8) for small amplitudes of stimulating current. In each set of observations the amplitude of stimulating current was kept constant to maintain the release probability unchanged. In our experiments the mean quantal content m did not exceed the value of 3 (Fig. 7B) that allowed us to describe our data by Poisson’s statistics, despite the parameters of distribution have been modified (the mean quantal content, probability of release, the number of observations). The maximal value of the mean quantal content (m ˆ 3) was obtained for 5 mM external Ca 21 concentration. The time to peak was identical for the currents with small and large amplitudes (Fig. 5E) that give evidence for simultaneous activation of postsynaptic receptors by large and small portions of neurotransmitter. This fact indirectly proves the simultaneous release of any neurotransmitter portion from the same morphologically distinguishable active zone on the presynaptic membrane, and indicates the presence multiple release sites on the active zone. Acknowledgements—The research described in this publication was possible in part by Award No UB2-323 of the Government of Ukraine and U. S. Civilian Research and Development Foundation for the Independent States of the Former Soviet Union (CRDF).

REFERENCES

1. Auger C., Kondo S. and Marty A. (1998) Multivesicular release at single functional synaptic sites in cerebellar stellate and basket cells. J. Neurosci. 18, 4532–4547. 2. Bekkers J. M. (1994) Quantal analysis of synaptic transmission in the central nervous system. Curr. Opin. Neurobiol. 4, 360–365. 3. dell Castillo J. and Katz B. (1954) Quantal component of the end-plate potential. J. Physiol. 124, 560–573. 4. Edwards F. A., Konnerth A. and Sakmann B. (1990) Quantal analysis of inhibitory synaptic transmission in the dentate gyrus of rat hippocampal slices: a patch-clamp study. J. Physiol. 430, 213–249. 5. Forti L., Bossi M., Bergamaschi A., Villa A. and Malgaroli A. (1997) Loose-patch recordings of single quanta at individual hippocampal synapses. Nature 388, 874–878. 6. Grantyn R. and Veselovsky N. S. (1997) Quantal analysis of “evoked” and “spontaneous” GABA release from individual synaptic bouton. Soc. Neurosci. Abstr. 23, 579. 7. Grantyn R. and Veselovsky N. S. (1998) Release capacity of single inhibitory synapses in collicular cultures. HFSP Workshop, 146–153 (Abstract). 8. Hamill O. P., Marty A., Neher E., Sakmann B. and Sigworth F. J. (1981) Improved patch-clamp techniques for high-resolution current recording from cells and cell-free membrane patches. Eur. J. Physiol. 391, 85–100. 9. Hille B. (1984) Ionic Channels of Excitable Membranes. Sunderland, Massachusetts. 10. Isaacson J. S. and Walmsley B. (1995) Counting quanta: direct measurements of transmitter release at a central synapse. Neuron 15, 875–884. 11. Jack J. J., Redman S. J. and Wong K. (1981) The components of synaptic potentials evoked in cat spinal motoneurones by impulses in single group Ia afferents. J. Physiol., Lond. 321, 65–96. 12. Katz B. (1969) The Release of Neural Transmitter Substances. Liverpool University Press, Liverpool. 13. Katz B. and Miledi R. (1967) A study of synaptic transmission in the absence of nerve impulses. J. Physiol. 192, 407–436. 14. Korn H. and Faber D. S. (1987) Regulation and significance of probabilistic release mechanisms at central synapses. In Synaptic Function. (eds Edelman G., Gall E. and Cowan M.), pp. 57–108, Chichester. 15. Korn H. and Faber D. S. (1991) Quantal analysis and synaptic efficacy in the CNS. Trends Neurosci. 14, 439–445.

1230

S. A. Fedulova et al.

16. Kostyuk P. G., Veselovsky N. S. and Tsyndrenko A. Y. (1981) Ionic currents in the somatic membrane of rat dorsal root ganglion neurons. I. Sodium currents. Neuroscience 6, 2423–2430. 17. Llinas R., Steinberg I. Z. and Walton K. (1981) Relationship between presynaptic calcium current and postsynaptic potential in squid giant synapse. Biophys. J. 33, 323–351. 18. Lux H. D. and Veselovsky N. S. (1994) Glutamate-produced long-term potentiation by selective challenge of presynaptic neurons in rat hippocampal cultures. Neurosci. Lett. 178, 231–234. 19. MacDonald R. L., Rogers C. J. and Twyman R. E. (1989) Kinetic properties of the GABAA receptor main conductance state of mouse spinal cord neurons in culture. J. Physiol. 410, 479–499. 20. McLachlan E. M. (1978) The statistics of transmitter release at chemical synapses. Int. Rev. Physiol. Neurophysiol. 17, 49–117. 21. Neale E. A., Nelson P. G., Macdonald R. L., Christian C. N. and Bowers L. M. (1983) Synaptic interactions between mammalian central neurons in cell culture. III. Morphophysiological correlates of quantal synaptic transmission. J. Neurophysiol. 49, 1459–1468. 22. Nusser Z., Hajos N., Somogyi P. and Mody I. (1998) Increased number of synaptic GABAA receptors underlies potentiation at hippocampal inhibitory synapses. Nature 395, 172–177. 23. Pierce J. P. and Lewin G. R. (1994) An ultrastructural size principle. Neuroscience 58, 441–446. 24. Pierce J. P. and Mendell L. M. (1993) Quantitative ultrastructure of Ia boutons in the ventral horn: scaling and positional relationships. J. Neurosci. 13, 4748–4763. 25. Propst J. W. and Ko C. P. (1987) Correlations between active zone ultrastructure and synaptic function studied with freezefracture of physiologically identified neuromuscular junctions. J. Neurosci. 7, 3654–3664. 26. Pun R. Y., Neale E. A., Guthrie P. B. and Nelson P. G. (1986) Active and inactive central synapses in cell culture. J. Neurophysiol. 56, 1242–1256. 27. Redman S. (1990) Quantal analysis of synaptic potentials in neurons of the central nervous system. Physiol. Rev. 70, 165–198. 28. Redman S. and Walmsley B. (1983) Amplitude fluctuations in synaptic potentials evoked in cat spinal motoneurones at identified group Ia synapses. J. Physiol., Lond. 343, 135–145. 29. Schikorski T. and Stevens C. F. (1997) Quantitative ultrastructural analysis of hippocampal excitatory synapses. J. Neurosci. 17, 5858–5867. 30. Smith S. M., Zorec R. and McBurney R. N. (1989) Conductance states activated by glycine and GABA in rat cultured spinal neurons. J. Membr. Biol. 108, 45–52. 31. Tong G. and Jahr C. E. (1994) Multivesicular release from excitatory synapses of cultured hippocampal neurons. Neuron 12, 51–59. 32. Vautrin J., Schaffner A. E. and Barker J. L. (1993) Quantal and subquantal GABAergic transmissions in cultured rat hippocampal neurons. Hippocampus 3, 93–101. 33. Vautrin J., Schaffner A. E., Fontas B. and Barker J. L. (1993) Frequency modulation of transmitter release. J. Physiol., Paris 87, 51–73. 34. Veselovsky N. S., Engert F. and Lux H. D. (1996) Fast local superfusion technique. Pflu¨gers Arch. 432, 351–354. 35. Walmsley B. (1991) Central synaptic transmission: studies at the connection between primary afferent fibers and dorsal spinocerebellar tract (DSCT) neurons in Clarke’s column of the spinal cord. Prog. Neurobiol. 36, 391–423. 36. Walmsley B., Wieniawa-Narkiewicz E. and Nicol M. J. (1985) The ultrastructural basis for synaptic transmission between primary muscle afferents and neurons in Clarke’s column of the cat. J. Neurosci. 5, 2095–2106. 37. Yazejian B., DiGregorio D. A., Vergara J. L., Poage R. E., Meriney S. D. and Grinnell A. D. (1997) Direct measurements of presynaptic calcium and calcium-activated potassium currents regulating neurotransmitter release at cultured Xenopus nerve– muscle synapses. J. Neurosci. 17, 2990–3001. 38. Zucker R. S. (1973) Changes in the statistics of transmitter release during facilitation. J. Physiol., Lond. 229, 787–810. (Accepted 1 February 1999)