Activation and desensitization of glutamate-activated channels mediating fast excitatory synaptic currents in the visual cortex

Neuron,

Vol. 9, 991-999, November,

1992, Copyright

0 1992 by Cell Press

Activation and Desensitization of G lutamate-Activated Channels Mediating Fast Excitatory Synaptic Currents in the Visual Cortex Shaul Hestrin Department of Physiology School of Medicine University of California at San Francisco San Francisco, California 94143

Summary Brief glutamate applications to membrane patches, excised from neurons in the rat visual cortex, were used to assess the role of desensitization in determining the AMPAfkainate receptor-mediated excitatory postsynaptic current (EPSC) time course. A brief (1 ms) application of glutamate (l-10 mM) produced a response that mimicked the time course of miniature EPSCs fmEPSCs). Direct evidence is presented that the rate of onset of desensitization is much slower than the decay rate of the response to a brief application of glutamate, implying that the decay of mEPSCs reflects channel closure into a state readily available for reactivation. Rapid application of glutamate combined with nonstationary variance analysis provided an estimate of the single-channel conductance and open probability, allowing an approximation of the number of available channels at a single synaptic site. tntroduction Glutamate receptors are thought to mediate synaptic transmission in the majority of fast excitatory CNS synapses. There are two broad classes of glutamate receptors: N-methyl-u-aspartate (NMDA) and AMPAI kainate (Nicoll et al., 1990). AMPA/kainate receptors play important roles in mediating synaptic transmission, and their regulation has been suggested to underlie a wide range of processes, including synapse formation, long-term potentiation, and long-term depression (Collingridge and Lester, 1989; McDonald and Johnston, 1990; Miller et al., 1990; ConstantinePaton et al., 1990; Ito, 1989). It has been shown that the time course of NMDA receptor-mediated excitatory postsynaptic currents (EPSCs) is determined by the NMDA channel properties rather than removal of transmitter from the synaptic cleft (Lester et al., 1990; Hestrin et al., 1990). The mechanisms underlying the time course of the AMPA/ kainate receptor-mediated EPSC, however, are only poorly understood. AMPAIkainate receptors are unusual in exhibiting fast desensitization (Kiskin et al., 1986; Trussell and Fischbach, 1989), defined as a reduced responsivity induced by previous activation. Several lines of evidence suggest that desensitization could be underlying the EPSC time course. Trussell and Fischbach (1989) found that, in chick spinal neurons, the response to a step application of glutamate had a fast decay component of 2.2 ms, which was

similar to the decay of the response to a 1 ms pulse of glutamate and to the decay of the EPSC. Further evidence for the role of desensitization has been derived using aniracetam (ho et al., 1990), which reduces desensitization in response to exogenous glutamate application and increases the amplitude and slows the decay rate of hippocampal EPSCs (Tang et al., 1991; lsaacson and Nicoll, 1991; Vyklicky et al., 1991). The effects of aniracetam on the time course of the response to brief glutamate application have not been studied so far. Therefore, interpretation of the action of aniracetam is inconclusive, given that it might produce some or all of its effects by slowing the closing rate of AMPAlkainate channels rather than reducing desensitization (Tang et al., 1991; Vyklicky et al., 1991; lsaacson and Nicoll, 1991). Rapid desensitization presents a technical challenge in determining the affinity and efficacy of glutamate to the AMPAlkainate receptors. Under equilibrium conditions, more than 90% of available receptors are desensitized, and their affinity to glutamate is much higher than that measured prior to desensitization (Vyklicky et al., 1991). Furthermore, Tang et al. (1989) and Trussell and Fischbach (1989) found evidence suggesting that single-channel currents observed with the rapid application of agonist may have conductances of 18-35 pS, which are larger than those observed using equilibrium measurements (reviewed by Cull-Candy et al., 1992). In the visual cortex, AMPA/kainate receptors, along with NMDA receptors, mediate EPSCs (Jones and Baughman, 1988; Artola and Singer, 1990; Stern et al., 1992). The time course of the AMPA/kainate receptormediated evoked EPSC measured in nonpyramidal neurons in the visual cortex was 2.4 ms (Stern et al., 1992). The present report is based on whole-cell recording of miniature EPSCs (mEPSCs) and rapid glutamate application to outside-out patches obtained from cells in layers II-IV of rat visual cortex. Several properties of AMPAlkainate receptors relevant to the generation of EPSCs have emerged. These properties are used to assess the role of desensitization in determining the time course of the EPSC and to estimate the number of channels that mediate mEPSCs.

Results mEPSCs in the Visual Cortex It is thought that mEPSCs that occur in the absence of presynaptic action potentials originate at individual synaptic contacts and therefore represent the unit of “quantal” transmission (Redman, 1990). mEPSCs are not distorted by nonsynchronized activation and can be used to characterize synaptic currents. Patchclamp and brain slice techniques were used to record from layer II-IV neurons of rat visual cortex. mEPSCs

were recorded at a holding potential of -80 mV in the absence of stimulation and in the presence of the Na+ channel blocker tetrodotoxin (Figure la). These mEPSCs were detected and aligned using a threshold crossing method (see Experimental Procedures). mEPSCs presumably arise at different electrotonic locations and may undergo different degrees of attenuation. To reduce the differences in attenuation, only mEPSCs with rise times shorter than 0.6 ms were selected for analysis. The mEPSCs recorded under these conditions exhibited variable peak amplitudes and highly skewed amplitude histograms (Figure lb). The average amplitude of mEPSCs recorded in 5 experiments was 13.0 + 2.9 pA (mean + SD), and the average decay time constant was 2.6 * 0.5 ms. The coefficient of variation (CV = SD/mean) measured in 5 experiments was 0.51 + 0.08. This high degree of variability in mEPSC amplitude was not related to the different attenuation by dendritic filtering because there was only a poor correlation between the decay times and the amplitudes (Figure Ic). Large variations in mEPSC amplitudes have been previously reported at central synapses (Bekkers et al., 1990; Manabe et al., 1991; Livsey and Vicini, 1992; Silver et al., 1992; Malgaroli and Tsien, 1992). It has been argued that the large CV arisesfrom thevariabilityoftransmitterconcentration (Bekkers et al., 1990), but postsynaptic mechanisms have not been ruled out.

Response of Outside-Out Membrane to Brief Glutamate Application

to membrane patches produces a response mimicking the NMDA receptor-mediated EPSC in cultured hippocampal neurons and superior colliculus slices (Lester et al., 1990; Hestrin, 1992). Outside-out patch recording configurations were obtained by withdrawing the pipette from the slice after establishing whole-cell recording. Patches of membrane exhibiting high resistance were used in rapid perfusion experiments. In these experiments, patch pipettes were positioned close to the interface of the solution flowing out of a 100 urn glass tube. Rapid application of glutamate was achieved by displacement of the flow pipe using a piezo-electric device (see Experimental Procedures). The average current elicited by a step application of glutamate (l-10 mM) was 127.1 f 71 pA (range, 14-254 pA; n = 30; holding potential, -80 mV). In most experiments the maximally evoked current amplitude decreased with time (“run down”); after IO-20 min the size of the maximally evoked current was typically about half of the initial response. Figure 2a illustrates an average of aligned mEPSCs selected for their fast rise times. The decay of the ensemble average was fitted by a single exponential function. The time course of the mEPSC was compared with the response of an outside-out patch to a brief application of glutamate (Figure2b). Application of I-IO mM glutamate for a duration of about 1 ms activated an inward current with a time constant of less than 500 us. The decay of the response to a brief pulse could be fitted with a single exponential function (Figure 2b). The average decay time constant of responses to a brief pulse of glutamate, obtained in 28 experiments, was 2.15 & 1.2 ms (range, 0.8-3.3). The similarity of the mEPSCs and the response to a brief pulse of glutamate (Figures 2a and 2b) suggest that the time course of the AMPAlkainate receptor-mediated

Patches

Rapid agonist application techniques provide a powerful tool for the study of the regulation of synaptic currents. It is thought that transmitter concentration rises only briefly in the cleft of excitatory CNS synapses (Eccles and Jaeger, 1958). Moreover, activation of NMDA channels by brief application of glutamate

a

Figure 1. Properties

Y’

v I

10 pA

,

50 ms

b

C 20

6r

+ ++

0.05.

O100 0

Amplitude

v1

(PA)

of mEPSCs

(a) Superposition of six traces. Whole-cell recording holding potential of -80 mV in the presence of tetrodotoxin. (b) Amplitude histogram of 265 mEPSCs (same cell as in [a]) that had a rise time (20%-80%) of less than 0.6 ms. (c)A plot of the amplitude against the decay time constant of the mEPSCs shown in (b). The correlation coefficient between these two parameters was

Glutamate 993

Channels

Mediating

Fast Cortical

EPSCs

a

--I

5 PA

5 ms

5 ms

b

r -

5 ms

1

50 pA

Figure2. Comparison of mEPSCs and the Response cised Patch with a Brief Application of Glutamate

OoW of an Ex-

(a) Average of 143 mEPSCs that had a rise time of less than 600 us (see Experimental Procedures for details). Theaverage mEPSC was fitted withasingle exponential function (marked bythedots) with a 2.3 ms time constant and amplitude of 12.7 pA. (b) Upper trace of open pipette current: the current had a rise time (20%80%) of 200 ps and duration of 1 ms. Lower trace, average of four responses from an excised patch (holding potential, -80 mV) to an application of 10 m M glutamate. The rise time was 400 us. The average response was fitted with a single exponential function (marked by the dots) with a time constant of 2.1 ms. The two “glitches”observed before the responses are electrical “artifacts” resulting from the command voltage (30-60 V).

EPSCs in the visual cortex reflects channel kinetics. In addition, the similar time course of mEPSCs and the response of excised patches to glutamate application suggest that somatic and synaptic channels are generally comparable, as has been found in chick spinal neurons (TrusselI and Fischbach, 1989). It appears, therefore, that brief glutamateapplications can mimic both NMDA and AMPAlkainate components of the EPSC. The Response to a Step Application of Glutamate Trussell and Fischbach (1989) found that the response of excised patches to a step of 1 m M glutamate was biphasic and that the fast component was similar to the decay time constant of the EPSCs. In the visual cortex the step response of outside-out patches was also biphasic, but the fast component was significantly slower than that of the EPSC or the brief pulse response. In the experiment illustrated in Figure 3a the response to a step application of IO m M glutamate was fitted bytwo exponential functions. The fast component had a time constant of 6.3 ms; this was much

2

3

Pulse Tau (ms)

4

I0

2

3

4

Pulse Tau (ms)

Figure 3. The Response of an Excisecl Patch to 1.1 ms, 6.6 ms, or a Step Application of IO m M Glutamate (a) Upper trace illustrates the open pipette current recorded in response to 1.1 ms, 6.6 ms, and a step glutamate application. The response of the patch held at -80 mV to the 1.1 ms pulse was fitted with a single exponential function (marked by the dots) with a time constant of 1.31 ms. The step response was fitted with a biexponential function: 6.3 ms (relative amplitude 63%) and 36.0 ms (37%). The response to a 6.6 ms application was fitted with a single exponential of 1.9 ms beginning at the glutamate offsettime.(b)Thetimeconstantof the responses toa brief pulse application (pulse T) plotted against the offset response (offset T). Each point represents different experiments. The pulse T was correlated to the offset r; correlation coefficient, 0.84. (c)The fast component of the step response to glutamate application (step T) plotted against the time constant of the response to a brief pulse application of glutamate (pulse T). The pulse 7 was not correlated to the step T; correlation coefficient, -0.04.

slower than the decay time constant of the brief pulse response, which was 1.3 ms in this experiment. The difference in the decay time constant of the step and brief pulse responses indicates that these responses reflect different kinetic steps, suggesting that the decay of the step response reflects desensitization and the decay of the brief pulse response reflects channel closure. Figure 3c illustrates a comparison of the fast component of the step decay with the decay in response to a brief application. The average fast component of the step application in 18 experiments was 7.8 f 2.7 ms, and the average decay of the brief application response, in these experiments, was 2.03 f 0.67 ms. There was no correlation between the fast component of the step application responses and thedecaytimeconstant induced bya brief application

Neuron 994

(r = -0.04). In one exception the fast component of the step response was 1.7 ms, which was similar to the brief pulse response with a time constant of 1.6 ms (Figure 3~). The average slow component of the step decay was 38.9 + 23 ms (n = I@, and the average relative component of the slow component was 24.6%. Application of glutamate for an intermediate duration produced a response that initially resembled the step response. However, the sudden removal of agonist after several milliseconds resulted in much faster decay (“offset response”; Figure 3a). The offset response was faster than the decay of the step response and was similarto the brief pulse response (Figure 3a). The average decay of the offset response was 2.77 + 1.2 ms (n = 20). Moreover, the offset decay time constant was correlated with the brief pulse decay measured in the same patch of membrane (r = 0.84; Figure 3b). Desensitization Induced by Brief Application of Agonist The poor correlation between the fast component of the step response and the response to a brief application of glutamate may be interpreted as indicating that desensitization is not involved in determining the time course of the latter. Moreover, given the similarity of the brief application response and the mEPSCs, these data suggest that the time course of EPSCs in the visual cortex reflects channel closure and not desensitization. However, desensitization induced by a step application of glutamate could reflect kinetic steps that are different from desensitization induced by a brief glutamate application. This suggestion can be tested directly by measuring the extent and rate of onset of the desensitization induced by a brief application of glutamate. Figures 4a-4d illustrate the response of an excised patch to two 1 ms pulses of 5 m M glutamate, which were applied at various time intervals. When the time interval was 3 ms the peak amplitude in response to the second pulse was similar to that reached at the peak of the first pulse. However, at longer time intervals the response to the second pulse was markedly reduced. The degree of desensitization was defined as [A(t) - B(t + At)]l[A(t) - A(t + At)] (equation I), where At is the time interval, A(t) and B(t + At) are the peak amplitudes of the first and second responses, and A(t + At) is the amplitude of the response to a single application of glutamate at the time when the second response peaked. Using this definition, a plot of desensitization time course was constructed (Figure 4e). In the example illustrated in Figure 4, desensitization peaked at an interval of 20 ms and attained a magnitude of 0.6. In 17 experiments desensitization peaked on average at 10.3 ms (range, 5-25 ms); the average maximal desensitization magnitude was 0.52 (range, 0.17-0.93). As illustrated in Figure 4 the rate of desensitization onset was significantly slower than the decay rate of the response to a brief glutamate application (2.7 ms; Figure 4e). The desensitization onset rate was, however,

av---bw-L 10ms

e

o.oL, 0

100

200

I 300

/ 400

Interval (ms) Figure 4. Desensitization tamate

Induced

by a Brief Application

of Glu-

(a-d) The response of an excised patch to dual 1 ms applications of 5 m M glutamate. The interval between the 1 ms application was 3.0, 5.0, 10.0, and 20 ms for (a), (b), Cc), and Cd), respectively. Pulse T was 2.7 ms; step T was 8.2 ms (78%) and 32.6 ms (22%). (e) Desensitization in relation to the time interval between dual brief pulses. The maximal desensitization was 60% and occurred at 20 ms intervals. Inset illustrates the step and pulse responses measured from the same patch superimposed on the initial desensitization plot.

similar to the initial decay rate of the step response (Figure 4e), suggesting that the fast decay component of the step response is related to the desensitization induced by a brief glutamate application. Effect of Aniracetam on Glutamate-Activated Currents Pharmacological tools may help in elucidating the role of desensitization in generating the AMPA/kainate receptor-mediated EPSC. Aniracetam increases the equilibrium response to glutamate application (Ito et al., 1990; Vyclicky et al., 1991; lsaacson and Nicoll, 1991; Tang et al., 1991). In addition, aniracetam increases the peak amplitude of AMPA/kainate mEPSCs and increasestheirdecaytimeconstant(Vyclickyetal.,1991; lsaacson and Nicoll, 1991; Tang et al., 1991). However, it is not known whether these effects of aniracetam result from its effects on the closing rate of AMPA/ kainate channels or on their desensitization. Figure 5 illustrates an attempt to study the specific effects of aniracetam on channel kinetics. The decay time con-

Glutamate

Channels

Mediating

Fast Cortical

EPSCs

995

a-

Figure 5. The Effect of Aniracetam on Brief Pulse Response and Desensitization ’

Ani 110 10 ms w

d Control

-I

10 pA

10 ms

T

of the response to brief glutamate application stant (2 mM) was 1.9 ms in control conditions (Figure 5a). During the application of 5 m M aniracetam, the brief pulse response decay time increased to 4.8 ms. These data suggest that aniracetam affects the closing rate of AMPAlkainate channels. However, aniracetam also affected the decay of the response to a step application of glutamate (Figure 5b). In the experiment illustrated in Figure 5 the step response decay time constants were 5.3 ms (relative amplitude, 51%) and 12.8 ms (49%) in control conditions. In the presence of aniracetam the step response decay time constant increased to 18.9 ms (69%) and 86.4 ms (31%). In addition, aniracetam also affected the degree of desensitization induced by a brief pulse of glutamate application. Figure 5c illustrates the reduction of the response to a brief glutamate application, which was preceded by a brief pulse. At an interval of 20 ms

PA

(a) The response to a 1 ms application of 2 m M glutamate in theabsence or presence of 5 m M aniracetam (Ani). The pulse T was 1.9 ms in control conditions and 4.8 ms in the presence of aniracetam. (b) The patch response to a step application of 2 m M glutamate. The step decay times were 5.3 ms (relativeamplitude, 51%) and 12.8 ms (49%) in control conditions and 18.9 ms (relative amplitude 69%) and 86.4 ms (31%) in the presence of 5 m M aniracetam. (c) Desensitization induced by a brief application of 2 m M glutamate. Interval between the 1 ms application was 20 ms. The degree of desensitization was 60.7% in control conditions (c) and 31.6% in the presence of 5 m M aniracetam (d). Glutamate application is shown diagrammatically above the current traces (a and b)

desensitization was 0.61 in the absence of aniracetam. However, in the presence of aniracetam desensitization at the same time interval was only0.32 (Figure 5d). These results suggest that aniracetam affects both the channel closing rate and desensitization, excluding the simple interpretation of its effects on EPSCs. The Efficacy of Glutamate at the AMPAlKainate Receptor To understand the mechanisms generating the EPSC it is important to estimate the efficacy of glutamate in opening AMPA/kainate channels. Figure 6a illustrates the responses of an excised patch to step applications of glutamate at different concentrations. In general the response to a given concentration of glutamate exhibited “rundown,” and therefore the peak responses illustrated in Figure6bwere normalized compared with the 10 m M response obtained by changing

a

Figure 6. Dose-Response ary Analysis

I

G E

a 20 1 ,, ./CT01

*' ...'.:'

0.1

/

/ .:

(a) Responseof a patch toa stepapplication of different concentrations of glutamate (millimolarconcentration indicated nextto the traces). Each trace represents average of four step responses. (b) Dose-response plot of data illustrated in part (a). (c) Six superimposed responses to identical step applications of 5 m M glutamate. (d) Variance-amplitude plot computed for 20 step responses; same patch as that illustrated in part (c). The amplitude and variance were computed for a period of 100 ms starting at the peak of the average response (see Experimental Procedures). The dots mark the fitted model: o2 = i. I - P/N (see text), where i = 0.75 pA and N = 302.

I,,ll

i'

1

Glutamate

10

(mM)

100

Amplitude

and Nonstation-

(PA)

the concentration of glutamate in the flow pipe. The average ECso was obtained by fitting the logistic equation VI,,, = I/[1 + (EC,,JC)~] (equation 2) to the peak response (Figure 6b), where I and I,,, are the peak current and the maximal peak current induced by step glutamate application, respectively, EC0 is the concentration producing 50% maximal response, c is glutamate concentration, and h is the Hill coefficient. The average Hill coefficient obtained in 5 experiments was 1.28 + 0.13 (mean k SD, n = 5), and the average EC0 was 1.05 + 0.17 mM. The concentration response analysis indicated that application of several millimolar glutamate would lead to near saturation of theAMPA/kainate receptors. However, it is not known whether these concentrations of agonist would also lead to simultaneous opening of all the available channels, because the maximum open probability (PO,,,) of the AMPA/kainate channels is not known. To estimate theopen probability of the AMPA/kainate channels, a series of responses to identical 100 ms steps of 5 mM glutamate (Figure 6c) were recorded. Nonstationary variance analysis (Sigworth, 1980; see Experimental Procedures) was used to obtain variance-amplitude plots (Figure 6d). Assuming that the patch current is generated by superposition of unitary currents originating from identical and independentchannels, the relation of the variance [o*(t)] to the current I(t) is given (Sigworth, 1980; Hille, 1992) by o*(t) = i. l(t) - l(t)*/N (equation 3), where i is thecurrent through a single channel and N is the total number of channels available for activation in the patch. The data obtained for the patch illustrated in Figure 6d were best fitted with this parabolic expression using a single-channel current, i = 0.75 pA, and available channels (N = 302). The average single-channel current was 0.61 k 0.1 pA, and the range of the number of available channels in 6 experiments was 120-655. The single current estimateswereobtained in solution containing lower permeating ions compared with the usual extracellular solution; therefore, the average single-channel conductance corrected for this effect was calculated to be about 10 pS (see Experimental Procedures). For the estimate of the single-channel current it is sufficient to obtain the slope of the asymptote near zero current. However, a reliable estimation of the number of available channels can be obtained only if the open probability is high at some time during the step application (Dilger and Brett, 1990). It is expected that if Pornax is much less than about 0.5 the varianceamplitude plots would be linear. Indeed, using 1 mM glutamate, which is expected to produce low Pornax, the variance-amplitude plot could be fitted with a linear function. As expected, the single-channel current estimated from 2 experiments (0.48 pA and 0.51 pA) was similar to that obtained using 5 mM glutamate. The maximal current obtained for the patch illustrated in Figure 6d was 205 pA, which was 0.9 of the maximal predicted current if all channels would open.

Therefore, under these conditions the Pornah was 0.9. The average PO,,, obtained in patcheswith step application of 5 mM glutamate was 0.64 + 0.2 (mean & SD, n = 6). The logistic equation indicates that at 5 mM glutamate the current would be only 0.88 of the maximal current (Figure 6b). Therefore the average value of Pornax at saturation would be about 0.7. Discussion The Role of Desensitization in Determining the Decay of the Response to a Brief Activation by Glutamate In patches from the visual cortex the response to brief glutamate application decayed with an average time constant of 2.15 ms, which was about 5-fold faster than the decay of the response to a step application of glutamate (Figure 3a). The rate of desensitization onset induced by brief glutamate application was much slower than the decay rate in response to a brief agonist application. These findings provide direct evidence that the decay rate following brief glutamate application reflects channel closure into a state readily available for reactivation and that desensitization develops slowly, secondary to channel closure. In chick spinal neurons the onset of desensitization in response to step application exhibited a rapid component of decay (2.2 ms), which was similar to the brief application response (TrusselI and Fischbach, 1989). In thevisual cortex (this study) and in hippocampal neurons (Colquhoun et al., 1992, Proc. Physiol. Sot., abstract) step response desensitization is slower than the decay in response to a brief application, suggesting that the decay of a brief activation does not reflect desensitization. These findings suggest that glutamate channels in chick spinal neurons differ in their gating properties from glutamate channels found in brain slices of the hippocampus and visual cortex. Differences in the properties of glutamate channels are also indicated by the single-channel current measurements. Tang et al. (1989) and Trussell and Fischbach (1989) reported that single-channel conductance measured with rapid application of agonist is 35 pS and 18 pS, respectively. The variance analysis of rapidly activated channels in the visual cortex indicated that single-channel conductance is about 10 pS. These differences suggest that heterogeneity of glutamate channel subtypes could produce different gating properties. Molecular biology studies reveal a family of AMPA/ kainate receptors that is specifically expressed in different cortical layers (Bettler et al., 1990; Monyer et al., 1991). The role played by these AMPAlkainate receptors in determiningthe EPSCtimecourse is not understood. However, it is plausible that the range of kinetic parameters that has been found in both membrane patches and mEPSCs in the visual cortex reflects the heterogeneity of the underlying receptors. It is possible that channel modulation occurred in the excised patches. However, the general similarity of the brief application response and the mEPSC time course sug-

Glutamate 997

Channels

Mediating

Fast Cortical

gests that channel modulation main kinetic features.

Aniracetam

EPSCs

did

not

affect

their

Slows Closing Rate and Desensitization

Aniracetam, an agent that reduces desensitization, slows the decay of mEPSCs and increases their amplitude (Isaacson and Nicoll, 1991; Vyklicky at al., 1991; Tang et al., 1991). The data obtained using dual brief applications of glutamate indicated that aniracetam slows the rates of channel closure and slows the onset rate of desensitization. If hippocampal glutamate channels are similar to cortical channels it is plausible that the effect of aniracetam on mEPSC amplitude and decay rate results primarily from its effects on the closing rate. On the other hand the lectins concanavalin A and wheat germ agglutinin, which also reduce desensitization, did not significantly slow the decay of EPSCs and did not increase the peak glutamate response (Vyklicky et al., 1991). The difference in the effects of lectins and aniracetam could be explained if the lectins have a specific effect on desensitization alone compared with aniracetam, which affects both desensitization and closing rate.

Open Probability

Is High

Glutamate has relatively low affinity to the AMPA/kainate receptor sites. Previously, reported E&s are 1.1 m M and 0.482 m M using whole-cell recording (Kiskin et al., 1986; Patneau and Mayer, 1990) and 0.51 m M using outside-out patches (Trussell and Fischbach, 1989). The EC*,, in the visual cortex, 1 mM, is similar to that reported in other preparations. The open probability of glutamate-activated channels in their native state has not been reported so far. However, Huettner (1990) reported that in dorsal root ganglion neurons, treated with concanavalin A to reduce desensitization, the open probability of glutamate-activated channels was about 0.5. In the present study, nonstationary analysis indicated that open probability could be high and that the maximal open probability at saturation is about 0.7. Determination of the maximal open probability was difficult because the rapid desensitization could have reduced the peak response. In addition, the rundown that was observed in the patch recording could also reduce the estimate of the open probability. It has been estimated recently that the maximal open probability at the NMDA receptor site (Jahr, 1992) is 0.3. The affinity of glutamate to the NMDA receptors is high (Patneau and Mayer, 1990), and its affinity to the AMPAlkainate receptors is low; however, it appears that glutamate is an efficacious agonist at both NMDA and AMPAIkainate sites.

Implication

for the Mechanisms Generating EPSCs

Several lines of evidence suggest that in the visual cortex, desensitization does not play a major role in determining the decay rate of EPSCs. Brief application of glutamate produces only partial desensitization, and moreover the rate of desensitization onset is slower than the decay rate of brief glutamate applica-

tion. Thus it appears that the decay rate of the response to a brief application of glutamate reflects the channel closing rate rather than desensitization. Given that brief 1 m s activation by glutamate produces responses that mimic both evoked EPSCs (Stern et al., 1992) and mEPSCs in their time course, it is likely that the time course of EPSCs also reflects the channel closing rate rather than desensitization. To understand howthe EPSC isgenerated it is necessary to have a kinetic model of AMPA/kainate channels in addition to an estimate of the time-dependent concentration changes of glutamate at these receptors. Accurate and complete data concerning these processes are not available but speculations using estimates may be useful. An mEPSC amplitude of IO15 pA requires opening about IO-20 channels with conductances of 10 pS. The number of available AMPAlkainate receptors can be derived from the magnitude of glutamate concentration at the cleft, estimated to be about 2 m M (Clements et al., 1991, Sot. Neurosci., abstract). The time course of glutamate concentration in the cleft is not known, but assuming that it is brief (Eccles and Jaeger, 1958) and using an ECso of AMPA/kainate receptors of 1 m M and a Pornax of about 0.7, it can be roughly estimated that only about one-third of available channels at a single postsynaptic site open to produce the peak mEPSC. Assuming that the mean mEPSC was produced byopening about IO-20 channels, the number of available channels would be about 30-60 at a single synaptic site. If the peak response of EPSCs at a given synaptic contact is on average generated by activating onethird of the availablechannels then the peak response would fluctuate from trial to trial. Assuming that EPSCs are generated by identical independent channels then the peak response would follow a binomial distribution, m = N.p.i and o2 = N.p,(l - p).P, where m is the mean current, N is the number of available channels, p is the open probability at the peak, i is single-channel current, and o2 is the variance. The CV, arising from channel activation, would be [(I - p)/(N . p)]‘“, and using the estimated parameters the CV would be about 0.2. The CV of mEPSCs measured in the visual cortex was 0.51, similar to that measured at CA1 neurons in hippocampal slices (CV = 0.42; Bekkers et al., 1990). Thus, channel activation parameters would be expected to contribute to variation of mEPSC amplitude but other factors are probably also important. Bekkers et al. (1990) suggested that the large CV of mEPSCs derives from fluctuation in released transmitter. In their model linear relation between glutamate released and channel activation is assumed. However, if AMPAlkainate receptors are about half saturated, their response to variation in transmitter concentration would be sublinear. Another source of variation of mEPSC amplitude would arise from variation in the number of available postsynaptic receptors. Electron microscopic measurements of postsynaptic density (PSD), which are

Neuron 998

thought to correspond to junctional membrane, indicate that PSD area in CA1 neurons is 0.069 f 0.08 pm* (Harris and Stevens; 1989). The number of glutamate receptors in a PSD is not known. However, Harris and Landis (1986) using freeze-fractured hippocampal synapses found particle aggregates at postsynaptic sites. The density of particles within aggregates was about 2800 particles per pm*, translating into about 200 particles per PSD (assuming that PSD area is similar to aggregate area). Harris and Landis raise the hypothesis that the particles within postsynaptic aggregates represent glutamate channels (a fraction of which would be AMPA/kainate receptors). An estimate of particle numbers in the visual cortex is not available. However, both their number and variation (Harris and Landis, 1986) are similar to the estimated number and variation of postsynaptic channels. Taken together, the results of this study suggest that the observed variability in mEPSC amplitude can originate from the fluctuation of transmitter peakconcentrations, channel activation kinetics, and variation of channel numbers among different synapses. Furthermore, modulation of any of these parameters would change synaptic strength, which underlies many forms of plasticity in the CNS. Experimental

Procedures

Slices Parasagittal, 300 urn thick slices from the visual cortex area were obtained using a vibratome (Lancer) from rats IO-25 days postnatal. Solutions The slices were maintained in a solution containing the following: 120 m M NaCI, 2.5 m M KCI, 1.3 m M MgSO+ 2.5 m M CaC&, 1 m M NaHP04,26 m M NaHC03,and 20 m M glucose. The internal solution contained 112 m M cesium gluconate, 17.5 m M CsCI, 10 m M CsOH-HEPES, 10 m M ECTA, 2 m M Mg-ATP, 0.2 m M GTP, and 8 m M NaCI. The solution used for rapid perfusion contained 100 m M NaCI, 2 m M CaCI,, 1 m M MgS04, 10 m M HEPES (NaOH), and 80 m M sucrose. Aniracetam was added to the experimental solutions from a dimethyl sulfoxide stock (final dimethyl sulfoxide, 1%).

brane patch can be estimated from the rate of rise of the response to a stepapplication of subsaturating agonist (Dilger and Brett, 1990). When the concentration of glutamate was l-3 m M the rising phase of the response was fitted with an exponential function with a time constant of about 0.5 ms. Measurements of the patch response to different concentrations of glutamate were achieved by changing the solution in the glutamatecontaining stream. Full exchange required about 30 s. mEPSC Detection Whole-cell currents were filtered at 1 or 2 kHr and stored on a video recorder (Vetter). Detection of mEPSCs was obtained offline by setting a threshold of -4 pA, which was compared with a baseline value computed by averaging the amplitude over an interval of 10 ms. The rise time (20%-80%), peak amplitude, and decay time constant were computed. The mEPSCs were aligned at the 50% crossing of the rising phase and stored for later averaging. A file containing the rise time, peak amplitude, and time constant of each mEPSC was used for histogram construction. Nonstationary Variance Analysis Transient responses to step application of 5 m M glutamate were filtered at l-2 kHz and stored on a computer disk. The ensemble average and variance for each sample point between the peak of the average response and the end of the trace (about 100 ms) were computed across 16-64 traces (Sigworth, 1980). A plot of the variance versus amplitude was constructed. A fitting routine minimized the sum of squares of differences between equation 3and thevariance-amplitude plots. Single-channel conductance was estimated assuming a linear current-voltage relationship and linear dependence on ion concentration. Single-channel current estimations were obtained in solution containing only 110 m M sodium compared with 146 m M sodium in the usual bathing media used for mEPSC recording. The single-channel conductance was calculated using -80 mV as E - E,. Singlechannel conductances were”corrected” using a ratio of 146/110. Acknowledgments I would like to thank Sascha Du Lac, Jeff Isaacson, Dimitri Kullmann, Roger Nicoll, and David Perkel for comments on the manuscript. This work was supported by National Institutes of Health grant EY09120. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertiseme&’ in accordance with 18 USC Section 1734 solely to indicate this fact. Received

July 16, 1992; revised

August

30, 1992

References Recording Slices were positioned on an upright microscope stage allowing visual inspection of cells within the slices. Wholecell recordings were obtained using the “blind” technique (Blanton et al., 1989). Currenttracesobtained with a patch-clampamplifier (List, EPC-7) were filtered at l-2 kHz (Frequency Devices), digitized (Axon Instruments, TLI), and stored using a PC. Rapid Perfusion Rapid application of agonists was achieved using a piezo-electric element that displaced adoublebarrel pipette (8 tubing; Trussell and Fischbach, 1989). The Ringer’s solution in the double-barrel pipette contained 80 m M sucrose, allowing clear demarcation of the outflow stream. Rapid exchange at the patch of membrane was obtained by positioning the patch in the control stream near the interface to the agonist containing stream. In some experiments a singleglutamate-containing stream was used. Under these conditions care was taken to prevent accumulation of agonist in the bath by rapid suction. The rate of solution exchange was measured as an offset current using an open patch pipette. The solution exchanged with a rising/falling time (20%80%) of 200 vs. The rate of solution exchange at the intact mem-

Artola, A., and Singer, W. (1990). The involvement of N-methyl-Daspartate receptors in induction and maintenance of long-term potentiation in rat visual cortex. Eur. J. Neurosci. 2, 254-269. Bekkers, J. M., Richerson, C. B., and Stevens, C. F. (1990). Origin of variability in quanta1 size in cultured hippocampal neurons and hippocampal slices. Proc. Natl. Acad. Sci. USA 87,5359-5362. Bettler, B., Boulter, J., Hermans-Borgmeyer, O’Shea-Greenfield, A., Deneris, E. S., Mall, C., Borgmeyer, U., Hollmann, M., and Heinemann, S. (1990). Cloning of a novel glutamate receptor subunit, GluR5: expression in the nervous system during development. Neuron 5, 583-595. Blanton, M. G., LoTurco, J. J., and Kriegstein,A. R. (1989). Wholecell recording from neurons in slices of reptilian and mammalian cerebral cortex. J. Neurosci. Meth. 30, 203-210. Collingridge, C. L., and Lester, A. J. (1989). Excitatory amino acid receptors in thevertebrate central nervous system. Physiol. Rev. 40, 145-210. Constantine-Paton, M., Cline, H. T., and Debski, E. (1990). Patterned activity, synaptic convergence, and the NMDA receptor in developingvisual pathways.Annu. Rev. Neurosci. 73,129-154.

Glutamate 999

Channels

Mediating

Fast Cortical

EPSCs

The ontogeny of excitatory amino acid receptors in the rat forebrain-11 kainic acid receptors. Neuroscience 35, 45-51.

Cull-Candy, S. G., Traynelis, 5. F., and Wyllie, D. J. A. (1992). Excitatory amino acid-gated channel types in mammalian neurones and glia. In Excitatory Amino Acids and SynapticTransmission, H. V. Wheal and A. M. Thomson, eds. (London: Academic Press), pp. 69-91.

Monyer, H., Seeburg, P. H., and Wisden, W. (1991). Glutamateoperated channels: developmentally early and mature forms arise by alternative splicing. Neuron 6, 799-810.

Dilger, J. M., and Brett, R. S. (1990). Direct measurement of the concentration-and time-dependent open probability of the nicotinic acetylcholine receptor channel. Biophys. J. 57, 723-731.

Nicoll, R. A., Malenka, R. C., and Kauer, J. A. (1990). Functional comparison of neurotransmitter receptorsubtypein mammalian central nervous system. Physiol. Rev. 70, 514-565.

Eccles, J. C., and Jaeger, J. C. (1958). The relationship between the mode of operation and the dimensions of the junctional regions at synapses and motor-end-organs. Proc. R. Sot. Lond. (B) 748, 38-56.

Patneau, D. K., and Mayer, M. L. (1990). Structure-activity relationships foraminoacid transmittercandidatesactingat N-methyl-oaspartate and quisqualate receptors. J. Neurosci. 10, 2385-2399.

Harris, K. M., and Landis, D. M. D. (1986). Membrane structure at synaptic junctions in area CA1 of the rat hippocampus. Neuroscience 19, 857-872. Harris, K. M., and Stevens, J. K. (1989). Dendritic spines of CA1 pyramidal cells in the rat hippocampus: serial electron microscopy with reference to their biophysical characteristics. J. Neurosci. 9, 2982-2997. Hestrin, S. (1992). Developmental regulation of NMDA receptormediated synaptic currents at acentral synapse. Nature357,686689. Hestrin, S., Sah, P., and Nicoll, R. A. (1990). Mechanisms generating the time course of dual component excitatory synaptic currents recorded in hippocampal slices. Neuron 5, 247-253. Hille, B. (1992). Ionic Channels of Excitable Membranes, 2nd ed. (Sunderland, Massachusetts: Sinauer Associates, Inc.), p. 321. Huettner, J. E. (1990). Glutamate receptor channels in rat DRG neurons: activation by kainate and quisqualate and blockade of desensitization by Con A. Neuron 5, 255-266. Isaacson, J. S., and Nicoll, R. A. (1991). Aniracetam reduces glutamate receptor desensitization and slows the decay of fast excitatory synaptic currents in the hippocampus. Proc. Natl. Acad. Sci. USA 88, 10936-10940. ho, M. (1989). Long-term 85-102.

depression.

Annu.

Rev. Neurosci.

12,

ho, I., Tanabe, S., Kohda, A., and Sugiyama, H. (1990). Allosteric potentiation of quisqualate receptors by nootropic drug aniracetam. J. Physiol. 424, 533-543. Jahr, C. E. (1992). High probability of NMDA by L-glutamate. Science 255, 470-472.

receptor

channels

Jones, K. A., and Baughman, R. W. (1988). NMDA- and nonNMDA-receptor components of excitatory synaptic potentials recorded from cells in layer V of rat visual cortex. J. Neurosci. 8, 3522-3534. Kiskin, N. I., Krishtal, 0. A., and Tsyndrenko, A. Y. (1986). Excitatoryaminoacid receptors in hippocampal neurons: kainatefails to desensitize them. Neurosci. Lett. 63, 225-230. Lester, R. A. J., Clements, J. D., Westbrook, G. L., and Jahr, C. E. (1990). Channel kinetics determine the time course of NMDA receptor-mediated synaptic currents. Nature 346, 565-567. Livsey, C. T., and Vicini, S. (1992). Slower spontaneous excitatory postsynaptic currents in spiny versus aspiny hilar neurons. Neuron 8, 745-755. LoTurco, J. J., Mody, I., and Kriegstein, A. R. (1990). Differential activation of glutamate receptors by spontaneously released transmitter in slices of neocortex. Neurosci. Lett. 774, 265-271. Malgaroli, A., and Tsien, R. W. (1992). Glutamate-induced longterm potentiation of the frequency of miniature synaptic currents in cultured hippocampal neurons. Nature 357, 134-139. Manabe, T., Renner, P., and Nicoll, R. A. (1991). Postsynaptic contribution to long-term potentiation revealed by the analysis of miniature synaptic currents. Nature 355, 50-55. McDonald, J. W., and Johnston, M. V. (1990). Physiological pathophysiological rolesof excitatoryaminoacidsduringcentral nervous system development. Brain Res. Rev. 75, 41-70. Miller,

L. P., Johnson,

A. E., Celhard,

and

R. E., and Insel, T. R. (1990).

Redman, S. (1990). Quanta1 analysis of synaptic potentials in neurons of the central nervous system. Physiol. Rev. 70, 165-198. Sigworth, F. J. (1980).Thevarianceof at the node of ranvier. J. Physiol.

sodium currentfluctuations 307, 97-129.

Silver, R. A., Traynelis, S. F., and Cull-Candy, S. G. (1992). Rapidtime-course miniatureand evoked excitatorycurrentsat cerebelJar synapses in situ. Nature 355, 163-166. Stern, P., Edwards, F. A., and Sakmann, B. (1992). Fast and slow components of unitary EPSCs on stellate cells elicited by focal stimulation in slices of rat visual cortex. J. Physiol. 449, 247-278. Tang, C-M., Dichter, M., and Morad, M. (1989). Quisqualate activates a rapidly inactivating high conductance ionic channel in hippocampal neurons. Science 243, 1474-1477. Tang, C.-M., Shi, Q.-Y., Katchman,A., and Lynch, G. (1991). Modulation of the time course of fast EPSCs and glutamate channel kinetics by aniracetam. Science 254, 288-290. Trussell, L. O., and Fischbach, G. D. (1989). Glutamate receptor desensitization and its role in synaptic transmission. Neuron 3, 209-218. Vyklicky, L., Jr., Patneau, D. K., and Mayer, M. L. (1991). Modulation of excitatory synaptic transmission by drugs that reduce desensitization at AMPA/kainate receptors. Neuron 7, 971-984.