Effect of thiocyanate on ampa receptor mediated responses in excised patches and hippocampal slices

Neuroscience Vol. 66, No. 4, pp. 815 827, 1995

~

Pergamon

0306-4522(94)00616-4

Elsevier ScienceLtd Copyright © t995 IBRO Printed in Great Britain. All rights reserved 0306-4522/95 $9.50 + 0.00

EFFECT OF THIOCYANATE ON AMPA RECEPTOR MEDIATED RESPONSES IN EXCISED PATCHES AND HIPPOCAMPAL SLICES A. A R A I , * J. S I L B E R G , M. K E S S L E R and G. L Y N C H Center for the Neurobiology of Learning and Memory, University of California, Irvine, CA 92717-3800, U.S.A.

Abstract--The binding affinity of alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors for [3H]AMPA is increased 10-30-fold by the chaotropic anion thiocyanate. The present experiments tested if thiocyanate alters AMPA receptor mediated current fluxes and if any such effects are reflected in the waveform of synaptic responses. Currents were measured after a step application of glutamate or AMPA to patches excised from pyramidal cells of hippocampal slice cultures. Application of I mM AMPA produced responses with an average peak amplitude of 86 pA at - 50 mV and a 10-90% rise time of 1.7 + 0.1 ms; the responses desensitized to a steady-state level below 10% of the peak current with a time constant of 11.1 + 0.7 ms. Glutamate in presence of D-amino-phosphonopentanoate produced similar responses which were inhibited by 6-cyano-7-nitro-quinoxaline-dione and enhanced by aniracetam or cyclothiazide and thus are characteristic for AMPA receptors. Thiocyanate accelerated the decay of AMPA responses two-fold and reduced the peak current by 30-50% with an ECs0 of 3.2 mM which is comparable to its ECs0for enhancing binding. Effects on the desensitization of glutamate induced responses were much smaller and only evident at the highest thiocyanate concentration; no effect was seen on response amplitude. Binding and physiological effects can be adequately explained by assuming that thiocyanate enhances conversion from the sensitive to the desensitized state of the receptor and reduces ligand dissociation from the desensitized state. Synaptic responses were measured in disinhibited hippocampal slices. Perfusion with 20 mM sodium thiocyanate increased the slope of the field excitatory postsynaptic potential by 44.9 + 4.2% and reduced its decay time by 10.4 + 4.3%. The former effect appears to result at least in part from an increase in transmitter release since it was accompanied by a decrease in paired-pulse facilitation and was reduced in magnitude after enhancing transmitter release. The decrease in the decay time constant points to an effect of thiocyanate on AMPA receptors in situ which is similar to that seen in excised patches. These results demonstrate that an increase in binding affinity may be indicative of reduced rather than enhanced current flow through AMPA receptors. In addition, the results provide further evidence that the kinetics of the AMPA receptor channel contribute significantly to at least the decay phase of fast excitatory synaptic responses.

There is a significant body of evidence indicating that factors present in the synaptic environment exert a pronounced effect on the binding properties of the A M P A - t y p e glutamate receptors which mediate fast excitatory transmission at many sites in the brain. Solubilization of the receptors from synaptic membrane fractions causes a more than 10-fold increase in their affinity for [3H]AMPA; 8 the binding properties of the solubilized receptors resemble those of a small pool of high-affinity sites found in brain homogenates thought to arise from extra-synaptic regions. These observations led to the idea that insertion .of A M P A *To whom correspondence should be addressed. Abbreviations: ACSF, artificial cerebrospinal fluid; AMPA,

alpha-amino- 3-hydroxy- 5-methyl-4-isoxazolepropionic acid; o-AP5, D-2-amino-5-phosphonopentanoic acid; CNQX, 6-cyano-7-nitro-quinoxaline-2,3-dione); EPSP, excitatory postsynaptic potential; HEPES, N-2-hydroxyethyl piperazine-N'-2-ethanesulphonic acid; LTP, longterm potentiation; NMDA, N-methyl-D-aspartate; PTX, picrotoxin; TTX, tetrodotoxin.

receptors into the synaptic zone brings them into contact with regulatory processes that reduce their affinity. In accord with this, radiation inactivation experiments have provided evidence that membrane bound receptors are associated with a protein that keeps the receptor in a state of reduced affinity, m14 Regulatory factors of this type are plausible candidates for the mechanisms responsible for long-term potentiation (LTP), a form of synaptic plasticity which has been linked to changes in A M P A receptor functioning. 1'2'35 Although the nature of the changes is poorly understood, it is not unreasonable to assume that a simple increase in receptor affinity could be the cause of potentiation. Arguments of this kind have led to biochemical explorations into mechanisms that increase binding affinity (e.g., treatment of brain membranes with phospholipases or phospholipids 4m) and attempts at linking these with LTP. 22 Conversely, several studies have examined whether [3H]AMPA binding properties are changed after LTP.m m33 815

A. Arai et al.

816

A n implicit a s s u m p t i o n b e h i n d m o s t studies o f this kind has been t h a t the affinity m e a s u r e d in binding tests correlates with the size of synaptic responses. However, a direct c o n n e c t i o n between these parameters, while intuitively attractive, c a n n o t be t a k e n as given. In the c o n t i n u e d presence o f agonist, A M P A receptors desensitize a n d evidence from physiological a n d biochemical studies suggest t h a t the desensitized state is of higher a p p a r e n t affinity t h a n the initial receptor state a n d c o n t r i b u t e s significantly to the equilibrium binding properties of the receptor. T M Moreover, binding tests are routinely done in the presence of the c h a o t r o p i c a n i o n thiocyanate which substantially increases the affinity of A M P A receptors; 6'9'14'15'2527 binding in the absence of Ihiocyanate, in fact, is likely to detect only the small s u b p o p u l a t i o n o f high affinity sitesJ ° Little is k n o w n a b o u t the effect o f t h i o c y a n a t e o n the physiological properties o f A M P A receptors a l t h o u g h Bowie a n d S m a r t 5 f o u n d t h a t it reduces steady-state currents p r o d u c e d by p r o l o n g e d applications of A M P A to cultured cerebellar neurons. When applied ionophoretically to the h i p p o c a m p u s , thiocyanate is reported to produce a sizeable facilitation o f the field excitatory postsynaptic potential ( E P S P ) ] ° an effect t h a t a p p e a r e d to be selective in as m u c h as N M D A receptor-mediated responses were unchanged. The present experiments used excised patches from cultured slices to examine h o w thiocyanate affects the response of h i p p o c a m p a l A M P A receptors to brief pulses of agonists. Infusions into acute slices were then employed to test if the effects of t h i o c y a n a t e o n patches are predictive o f its effects on the waveform of the field EPSP. EXPERIMENTAL PROCEDURES

Hippocampal slice cultures were prepared from 10 12day-old rat neonates (Sprague-Dawley, Charles River) according to the technique introduced by Stoppini et al. 3~ Transverse slices (400#m thick) were placed on to the membrane of a 25-mm diameter insert (Millipore CM cellulose) and maintained in an incubator with 5% CO2 at 36°C for three to four weeks. For the excised patch experiment, cultured slices were immersed in a modified medium (at room temperature) containing (in mM): NaC1 125, KC1 2.5, CaC12 2, MgCI 2 1, NaHCO 3 25, glucose 25 and HEPES 20 (pH 7.2). Patch pipettes were pulled from borosilicate glass tubing (1.5 mm outer diameter, 1.12 mm inner diameter, 0.19mm wall thickness). The electrodes typically showed a resistance of 5 8 MYL The intracellular solution contained (in mM): CsC1 140, EGTA I0, MgC12 2, ATP disodium salt 2, and HEPES 10 mM; the pH was adjusted with CsOH to 7.3. The osmolarity of all solutions was adjusted at around 360 mOsm. Outside-out patches were obtained from CA3 or CA1 pyramidal neurones exposed at the surface of the tissue and were transferred via a solution bridge to a spacially separated recording chamber containing the following recording medium (in mM): NaCI 135, KC1 4.5, CaC12 1.8, MgC12 I, H E P E S 20, MK-801 0.01 and o-amino-phosphonopentanoate (D-AP5) 0.02. The patches were positioned near the septum of a 0-shaped bifurcated glass pipette which provided at any one time either a background flow of recording medium through one of the lines or recording

medium plus agonist through the other; the flow was changed from one line to the other by fast-switching valves. The position of the patch was adjusted until a sharp pulse onset was obtained. The solution exchange time was 0.61 _+0.04ms (10-90%; n = 20) according to junction potential measurements. The flow rate was regulated by gravity and adjusted to approximately 0.3 ml/min. Both the background flow line and the pulse flow line were connected to four supply lines, only one of which was active at each phase of the experiment; solutions in the flow lines could be exchanged by a 1 2-s purge. Thus, four different experimental conditions could be tested for each patch. The agonist-containing medium contained 50/~M instead of 20/~M o-AP5 and further additions such as thiocyanate or aniracetam as required by the experiment. Cyclothiazide and 6-cyano-7-nitro-quinoxaline-2,3-dione (CNQX) were only partially effective when applied together with the agonists and were, therefore, applied to the background flow line prior to the agonist pulse. Data were collected with a patch-clamp amplifier (AxoPatch-lD) and digitized at 10kHz with PClamp (Axon Instrument) on line. Pulse applications were alternated between the four different experimental conditions for as long as stable recordings could be maintained. For data analysis, comparisons between two conditions, A and B, were made by comparing response B with the average of the responses from A taken immediately before and after response B. Data from 68 patches were submitted to analysis. Currents obtained from each patch were normalized to the mean value of all the experiments at - 5 0 m V and the currents at different holding potentials were multiplied with a constant deduced from the normalization. The decay phase of the currents activated by AMPA or L-glutamate was fitted with a single exponential curve since twoexponential functions in general did not give a significantly better fit. To distinguish the decay time constant of the patch current from that of EPSPs, it is referred to in the following as desensitization time-constant. Hippocampal slices were prepared from six- to sevenweek-old Sprague-Dawley rats (Charles River) using conventional methods. Slices were maintained in a linearflow chamber with low volume (0.2 ml) at an interface between humidified oxygen-rich atmosphere and artificial cerebrospinal fluid (ACSF) containing (in mM); NaCI 124, KC1 3, KH2PO 4 1.25, MgSO 4 2.5, CaCI: 3.4, NaHCO 3 26 and D-glucose 10. In most experiments, an incision was made between CA3 and CA1. Extracellular recordings were made from stratum radiatum of the field CA1 and CA3 in response to Schaffer-commissural (S-C) fiber stimulation. Data were digitized at 5-10 kHz and recorded on line. The amplitude of the maximum EPSP (without spike generation) was measured and the stimulation intensity was adjusted so as to obtain 30-40% of the maximum amplitude for baseline. In order to isolate the EPSP component, a disinhibited slice preparation was established by applying picrotoxin (PTX), tetrodotoxin (TTX) and 2-hydroxy-saclofen into the ACSF as follows (see also Ambros-Ingerson et a l l ) . After slices had recovered from the cutting process, the perfusion medium was changed to ACSF plus 50/~M PTX. When spike activity appeared upon stimulation of S-C fibers, TTX at a final concentration of 0.5 # M was introduced into the chamber for approximately 1 2 min until the recurrent action potential generation was suppressed. 2-Hydroxysaclofen (100 ,uM) was then added to the perfusion (ACSF plus 50 # M PTX) for the rest of the experiment to block the GABA~ component in the EPSPs. Upon completion of this preparation, the recording electrodes were re-positioned so as to be located near the current sink in stratum radiatum where the decay phase of EPSPs does not produce an overshoot. The decay time constant was determined by fitting a single exponential function between 90-50% and 90-10% of the maximum amplitude and the corresponding regression coefficient was calculated. Decay time-constants

Thiocyanate effect on AMPA receptor mediated responses with a correlation coefficient of higher than 0.990 were employed for data analysis. When 90 50% and 90-10% values differed by more than 2 ms, the data were not used. After establishing a 10-20-min stable baseline, l0 or 20mM sodium thiocyanate was introduced into the perfusion line substituting for the equivalent concentration of sodium chloride. Comparisons were made between records taken before and during thiocyanate application, as well as between those taken during thiocyanate perfusion and after wash-out. The results were expressed as per cent change from the corresponding control group (before or after thiocyanate perfusion). In some experiments, two pulses separated by 40 80 ms were applied to measure paired-pulse facilitation in the presence and the absence of thiocyanate. In an other group of experiments, hippocampal slices were first maintained in a low Ca2+-ACSF (2raM Ca2+/4mM Mg 2+) and later in a high CaZ+-ACSF (5 mM Ca2+/1 mM Mg 2+ ). The magnitude of paired-pulse facilitation and the effect of sodium thiocyanate on EPSP slope were compared in these two media. [3H]AMPA binding was measured in a rat whole brain membrane preparation obtained by subjecting a lysed and washed P: fraction to a mild Triton X-100 treatment which does not solubilize AMPA receptors (0.5% for 10min at 0"C) and extensively washing the membranes in HEPES (100 mM)/Tris buffer (pH 7.4)] 7 Details of the binding assay are described in the legend to Fig. 4. Student's t-test and matched paired t-test were employed for statistical analysis. Unless indicated otherwise, data are presented as means and S.E.M.

Materials Sodium thiocyanate was obtained from Mallinckrodt, potassium thiocyanate from Fisher Scientific. CsOH and CsC1 were purchased from Aldrich. D-AP5 (Sigma), MK-801 (Research Biochemicals, Inc.) and CNQX (Tocris) were dissolved in distilled water to prepare a stock solution and were diluted for every experiment. Aniracetam and cyclothiazide were kindly provided by Dr G. Rogers, U.C. Santa Barbara, and were dissolved in DMSO to a concentration of 1 M and 100 mM, respectively. RESULTS

Characteristics of AMPA and L-glutamate responses in patches excised from cultured hippocampal slices The top two panels of Fig. la show responses to four different concentrations of L-glutamate or A M P A given consecutively to the same patch. The responses reached a peak at about 2 - 4 ms and then declined to a steady-state level which in general was about one-tenth or less of the peak current. Currents produced by 1 m M glutamate and A M P A at a holding potential of - 50 mV were characterized by peak amplitudes of 86.3_+ 13.0pA and 60.9_+7.1 pA (mean _+ S.E.) and by desensitization time constants of l l.l _+ 0.7 ms and 10.0-+0.8ms, respectively (averages from 27 patches); the decay phase of the response was fitted with single exponential functions to determine desensitization time-constants. Rise times ( 1 0 - 9 0 % ) were 1.7 _+ 0.1 ms for both agonists. Kainate produced non-decremental responses which were larger than the steady-state current but smaller than the peak current produced by A M P A at the same concentration (see Fig. 2 below); the rise time was similar to that of the other agonists (1.7 _+ 0.4 ms; n = 10). The ultra-fast transient current reported for

817

kainate 29 could not be resolved with the technique employed here. The dose-response relations for the peak responses to L-glutamate and A M P A are shown in the lower row of Fig. IA. Peak currents measured in the same patch at different agonist concentrations were normalized relative to that at 1 m M agonist. The curves that best fit these data predict half-maximal responses at 1.05 m M with a Hill coefficient of 1.2 for glutamate and 640 ~ M with a Hill coefficient of 1.38 for A M P A . The voltage-current relation of the peak current induced by 1 m M L-glutamate, A M P A and kainate is illustrated in Fig. lB. The inward currents showed a nearly linear correlation (r > 0.990) with a slight rectification in the case of A M P A and reversal potentials around 0 mV. Steady-state currents also reversed at 0 mV and changed their amplitude linearly with holding potentials (not shown). Recordings were routinely done in presence of D-AP5 and MK-801 to eliminate N M D A receptor currents. The glutamate responses recorded under these conditions were examined for their pharmacological properties by testing their sensitivity to the n o n - N M D A receptor antagonist C N Q X and the specific A M P A receptor modulators aniracetam and cyclothiazide (Fig. 1C). Inclusion of 20/~ M C N Q X in the background flow produced a nearly complete blockade of the glutamate peak current; the small delayed current presumably developed because C N Q X was not included in the glutamate solution and therefore may have partially dissociated from the receptor during the glutamate pulse. Aniracetam which has been reported to reduce receptor desensitization 32'36 (but see Ref. 13) prolonged the response and increased the peak current; the desensitization time constant increased on the average by 48_+ 14.2% (n = 3). Cyclothiazide at 10/~M completely abolished desensitization; in its presence, both A M P A and glutamate produced a stable plateau current as previously reported. 37'39 Cyclothiazide had to be applied in the background flow for several seconds before it became fully active and wash-out times on the order of a minute were required before responses returned to the pre-drug profile. Taken together, these results indicate that patches excised from cultured hippocampal slices produce responses to glutamate and A M P A that are characteristic for A M P A receptors recorded in other preparations.

Effect of thiocyanate on AMPA receptor currents in excised patches Sodium thiocyanate reliably produced a reduction in the amplitude and the width of A M P A induced currents; an example of this is shown in Fig. 2 (bottom left). Thiocyanate was effective when applied together with the agonist, indicating that the onset of its action is rapid. Effects were evident at concentrations as low as 1 2 m M and reached a maximum at 10 to 2 0 m M (Fig. 2, top). The desensitization time constant was reduced from an average of

818

A. Arai et al.

"i-

_9o -~ o E

E o

(5 o

o

nn

.-2

LZ

5 tl. <

E t2.

"E:

o

uin (9 ~ ~ - r E tj

<

,.z o

O

Thiocyanate effect on AMPA receptor mediated responses 14.1 + 0.7 ms in control responses to 8.6 + 1.0ms in the presence of 20 m M thiocyanate; the Ecs0 determined from regression analysis was 3.2 mM. The peak current was reduced by 30-50% with a similar concentration profile. Changes in the rise time were insignificant (not shown). The effects of thiocyanate on currents induced by 1 m M L-glutamate were less evident than those obtained with A M P A . Peak currents remained constant in magnitude at all thiocyanate concentrations. A consistent effect was observed, however, in the desensitization rate. At a thiocyanate concentration of 2 0 m M , the desensitization time constant was significantly reduced to 70.9 + 12.4% of that measured in corresponding control responses. Responses produced by kainate were not affected by thiocyanate. An example for this is shown in the bottom row of Fig. 2; the average change in the steady-state level of kainate responses was + 5.7 _+ 4.5% (n = 10, n.s.). In agreement with observations from other laboratories, 13 responses gradually decreased after excision until they reached a relatively stable level. The cause for this run-down effect is not clear, but loss of A T P is unlikely to be the reason since high concentrations of A T P were included in the electrode solution. The decrease in response amplitude did not appear to be accompanied by a change in any of the other response properties. It should be noted that multiple responses were recorded in each patch with alternating application of agonist alone versus agonist plus thiocyanate (see Experimental Procedures); if the effect of thiocyanate on A M P A responses were to occur only in the run-down state then it would be expected to become larger with time. This was not the case as the reduction in amplitude and desensitization time constant was of comparable magnitude for the first responses taken after excision and for the later responses (not shown).

Effect of thiocyanate on synaptic responses Field EPSP studies were conducted in slices in which inhibitory pathways had been blocked with

819

5 0 # M picrotoxin and 1 0 0 # M 2-hydroxy-saclofen (see Experimental Procedures). The EPSP waveform in these disinhibited slices has been shown elsewhere to be sensitive to manipulations which influence the kinetic behavior of A M P A receptor channels, u9 Thiocyanate perfusion was carried out by substituting sodium thiocyanate for the equimolar concentration of sodium chloride. Perfusion of thiocyanate produced two effects. The first was an increase in the amplitude and slope of synaptic EPSPs in response to stimulation of the S-C fibers. Effects were observed 10-13 min after the initiation of the perfusion and reached a plateau after 20-30 min. The increase in the slope was 10-20% at 1 0 m M thiocyanate and 3 0 - 4 5 % at 20 mM; the effect was barely detectable at concentrations below 10 raM. Responses returned to pre-drug levels after about 30 min of wash-out. Most experiments were conducted in field CA i but comparable effects were seen in six experiments on associational responses in field CA3. The effects of thiocyanate on the EPSP amplitude were similar in magnitude in disinhibited slices and in conventional slice preparations, i.e. without G A B A receptor antagonists. The second effect of thiocyanate was an acceleration in the decay of the field EPSP. The time-constant for this decay was determined by fitting a single exponential function to the falling phase of the EPSP over the range from 90% to 50% of the maximum amplitude (see Experimental Procedures). The average decay time-constant for the field EPSPs in CA1 was 9.15 +0.31 ms. In the presence of thiocyanate, the decay time-constant was significantly reduced by 10.4__+ 4.3% (P < 0.025, seven pairs) or - 0 . 9 5 _+ 0.39 ms (Fig. 3A). This effect reversed after 30-40 rain of washing. Two observations indicate that the reduction in the time constant is not a secondary consequence of the increase in the response amplitude. First, an additional measure of the decay time constant was taken after thiocyanate had washed out and after the amplitude had been increased so that it matched that measured in the presence of thiocyanate. The decay time-constant under these

Fig. 1. Glutamate and AMPA-induced inward currents in outside out patches. (A) Top: current activated by step application of different concentrations (0.5, 1, 2 and 5 mM) of glutamate (left) or AMPA (right) at a holding potential of - 5 0 mV. Calibration: 50 pA/50 ms. Bottom: dose-response relation for peak currents produced by glutamate and AMPA. For each patch, currents obtained at different concentrations were normalized to the response obtained at 1 mM ligand. Each point represents the mean and S.E.M. of seven to 12 determinations; data were collected from 16 patches (glutamate) and 10 patches (AMPA). Curves show the best fit obtained with a four-parameter logistic equation; ECs0values and Hill coefficients are shown as inserts. (B) Current-voltage relations for 1 mM L-glutamate (n = 7), 1 mM AMPA (n = 8) and 5 mM kainate (n = 4). Currents obtained from each patch were normalized to the mean value of all the experiments at - 50 mV and the currents at different holding potentials were multiplied with the factor derived from the normalization process. Correlation coefficients in all cases are >0.99. (C) Top left: current induced by 1 mM glutamate before, during and after application of 20 # M CNQX. The medium in the background flow line was switched approximately 5 s before the glutamate pulse to one containing CNQX. Below: effect of 1 mM aniracetam on glutamate-induced current. Aniracetam was applied together with glutamate. The trace on the right is a superposition of those with and without aniracetam. Right: effect of 20 #M cyclothiazide on current induced by 1 mM glutamate; cyclothiazide was applied to the background flow solution about 2 min before the glutamate pulse. A superposition of the traces is shown on the right. Calibration: 20 pA/50 ms. Holding potential: - 5 0 mV.

820

A. Arai et al.

conditions was 8.82 + 0.65 ms; this was again significantly different in pairwise comparisons from the decay time-constant measured in the presence of thiocyanate (second bar in Fig. 3A; P < 0.05, seven pairs). Second, paired-pulse stimulation with an interval of 80 ms produced a facilitation of the second response of about 33% which is comparable in magnitude to the increase produced by thiocyanate. In this case, however, the difference in the decay time-constant between the first and the second pulse was only - 1 . 4 - + 1 . 1 % (Fig. 3A, third bar); this confirms previous reports that changes in response size either through increasing stimulation intensity or via modulation of transmitter release have minimal effects on the decay time-constant of EPSPs. In the course of these experiments, it was found that the amount of paired-pulse facilitation was reduced during the perfusion of thiocyanate. Facilitation decreased from 37.0-+ 2.3% under control conditions to 30.0-+ 2.4% in the presence of thiocyanate (Fig. 3B); in pairwise comparisons, facilitation was reduced by 15.9 -+ 3.0% (P < 0.005, eight

Peak current

160

pairs). The effect was almost completely reversed after washing out thiocyanate. To exclude a ceiling effect, paired-pulse facilitation was measured again after the washout and after the stimulation intensity had been increased so as to obtain the same slope as in the presence of thiocyanate (0.5+0.1 vs 0.5 + 0.1 mV/ms). The resultant paired-pulse facilitation was 35.9+2.2%, which was significantly larger than that measured in presence of thiocyanate (P < 0.005, eight pairs). The changes in paired-pulse facilitation raised the possibility that thiocyanate increases synaptic responses by enhancing transmitter release, This was tested further by combining thiocyanate perfusion with a known presynaptic modulation. It has previously been demonstrated that manipulations which enhance release produce non-multiplicative effects when applied together whereas manipulations acting separately at pre- and postsynaptic sites do not show an interaction of their effects. 38 In this experiment, slices were perfused with 20 mM thiocyanate twice, first while being maintained in a low CaZ+-ACSF

11111

Desensitization constant time

140

(16) 120

120 1

100

.

.

.

.

.

80 *

(le)

60 40

(13) 20

20 0

0

5

AMPA

10 [SCN] (mM)

:~ SCN

15

20

Glutamate :1: SCN

0

5

10 [SCN] (mM)

15

Kainate :1: SCN

Fig. 2. Effect of thiocyanate on AMPA and glutamate-induced currents in outside-out patches. Top left: peak current produced by 1 mM AMPA (closed circles) and l mM glutamate (open circles) in the presence of the thiocyanate concentration indicated on the x-axis (in mM). The data are shown as means and S.E.M. of the number of determinations shown in parentheses; the current in each case was normalized to that measured in the absence of thiocyanate. Data were collected from a total of 40 patches. Values significantlydifferent from 100% are indicated by asterisks with *P < 0.05, **P < 0.01, and ***P < 0.001. Top right: desensitization time constants for the same set of recordings. Bottom: examples of traces obtained with agonist alone and with agonist plus 20 mM thiocyanate for 1 mM AMPA (left), 1 mM L-glutamate (middle) and 5 mM kainate (right). Desensitization time-constants were determined from the fitting of single-exponentialdecay curves as two-exponential curve fits in most cases did not prove to be superior. Calibration: 50 pA/50 ms (AMPA), 25 pA/50 ms (glutamate), 25 pA/100 ms (kainate). SCN, sodium thiocyanate.

20

Thiocyanate effect on AMPA receptor mediated responses (2 mM Ca 2+/4 mM Mg 2+ ) and the second time after changing the medium to a high calcium concentration (5raM Ca2+/l mM M f +) to increase transmitter release. In agreement with previous reports, 24 the degree of paired-pulse facilitation was significantly reduced after changing to the high Ca 2+ medium (42.1 + 9.4% after vs 59.3 + 13.6% before, nine pairs, P < 0.005). The response enhancement produced by sodium thiocyanate showed a reduction in the high calcium medium which was comparable to that seen in the paired-pulse facilitation; the increase in the EPSP slope was 58.9+16.1% in the low-Ca 2+ medium but only 42.0_+ 12.4% in the high-Ca 2+ A C S F (Fig. 3C). The difference was significant at P < 0.05 (nine pairs). The fact that the effect produced by thiocyanate in both experiments is smaller when combined with a manipulation which enhances transmitter release suggests that the effect of thiocyanate on field EPSPs is at least in part mediated by an action at a presynaptic site.

Effect of thiocyanate on [3H]AMPA binding and dissociation To facilitate comparison with physiological studies, the effects of thiocyanate on [3H]AMPA binding were re-assessed at 25°C. Under these conditions, thiocyanate increased equilibrium [3H]AMPA binding 7 . 7 + 0 . 8 times with an ECs0 of 8 . 7 + 0 . 1 m M (n = 3), a value which is slightly larger than that obtained for its effect on desensitization in the excised patch experiments of Fig. 2. Dissociation from the A M P A receptor was biphasic under all conditions tested, i.e. in the presence and absence of thiocyanate and at both 0 and 25°C (Fig. 4), presumably because of the presence of high- and low-affinity binding sites in these membranes. At 0°C, thiocyanate reduced the dissociation rate of both the fast and the slow component by a factor of about 10 (see T~/2 values in figure insets). This accords with the previous findings that thiocyanate produces a comparable shift in affinity for both populations of sites? ° The respective values (i.e. plus thiocyanate vs minus thiocyanate) at 25°C differed by a factor of 6; this most likely represents an underestimation since dissociation in the absence of thiocyanate was too fast to be reliably assessed and since dilution presumably was not instantaneous and may thus have imposed limitations during the early phase of dissociation.

DISCUSSION

Effect of thiocyanate on A M P A receptor currents in excised patches and on excitatory postsynaptic potentials Patches could be reliably excised from the pyramidal cells of cultured hippocampal slices. Responses to brief application of A M P A and glutamate were similar to those reported in the literature m6,23 and exhibited the expected sensitivities to CNQX, aniracetam,

821

and cyclothiazide. Current-voltage relations were linear as they are in patches excised from fresh hippocampal tissue. The most prominent effect of thiocyanate in these patches was a reduction by about half in the amplitude of A M P A elicited currents and an equivalent increase in the rate at which the responses desensitized. These effects occurred over a concentration range of 1-10 mM which is comparable to the range effective in enhancing binding. Thiocyanate also increased the rate of desensitization of glutamate induced currents but only at the highest ion concentration; it had no evident effects on the peak current. The lesser effect of thiocyanate on glutamate vs A M P A induced currents is predicted from binding experiments. Earlier reports had suggested that glutamate binding is not affected by thiocyanate, 26 however, more recent studies measuring the displacement of [3H]CNQX by A M P A and glutamate found that thiocyanate does increase the affinity for glutamate but that the shift in affinity is about four times smaller than that for AMPA. 1° The interaction of kainate with the AMPA receptor appears to be insensitive to thiocyanate according to converging evidence from binding H and physiological experiments (Fig. 2). The results with excised patches indicate that thiocyanate is a first tool for exploring the effects on synaptic responses of accelerating the decay phase of A M P A receptor-mediated currents. Experiments with disinhibited slices showed that the compound produces a small but reliable decrease in the decay time constant of the field EPSP. The magnitude of the effect is not unexpected given that glutamate is thought to be the transmitter released at hippocampal synapses and thiocyanate, as discussed, has modest effects on A M P A receptor currents elicited by glutamate. Previous studies have shown that cyclothiazide, which slows desensitization or accelerates resensitization 29'37'39 and reduces binding, 9 slows the decay phase of the field EPSP. 19 Thus, drugs with opposite effects on desensitization/resensitization of the A M P A receptor are found to have opposite effects on synaptic responses in slices. These observations strongly suggest that channel kinetic behavior plays a prominent role in shaping the waveform of synaptic currents in adult hippocampus. Thiocyanate also caused an increase in the slope and amplitude of the field EPSP, effects which did not correspond with those observed with excised patches. While it is possible that the compound has more complex actions on synaptic A M P A receptors than on the presumably non-synaptic receptors studied in excised patches, the amplitude increase obtained with thiocyanate may reflect, at least in part, enhanced transmitter release. This is suggested by the observed interactions between the response increase caused by thiocyanate and those produced by manipulations which are known to act on transmitter release. Moreover, the magnitude of the decrease in the effects of thiocyanate produced by switching from low to high

A. Arai et al.

822

Decay time constant

A 4] ¢-

I

CA1 vs before

0 SCN

0 0

CA3

vs after

SCN

C

PPF

vs b e f o r e

SCN

PPF

SCN

Low Hi

Low Hi

80

vs after

SCN

60

E -4 o

.__.

40

-8

(11 e-

o

20

-12

-16

B

Paired pulse facilitation 50

before SCN

SCN

after SCN

~. 40 0 .=_ tO .m

+sc

z////~, 20

LL -

S

C

N

-

z/_f//~, r-.zffh, ~.¢ff/'t //////

Fig. 3. Effect of sodium thiocyanate on synaptic responses in field CA1 and CA3 of hippocampal slices. (A) Effect of thiocyanate on the 90-50% decay time-constant. The two columns on the left compare the decay time constant in presence of 20 mM thiocyanate with that before thiocyanate application and after wash-out of thiocyanate (seven pairs each). For the latter comparison, the EPSP amplitude after wash-out was adjusted to the same size as that during thiocyanate perfusion. The columns on the right show the change in decay time constant for similar experiments recorded in field CA3 (six pairs). The column labeled PPF shows a comparison in paired-pulse measurements with an interpulse interval of 80 ms; the decay time-constants of the first and second pulse were compared and expressed as per cent change (four pairs). Paired t-test was used for statistical analysis, *P < 0.05. (B) Effect of thiocyanate on paired pulse facilitation of CA1 EPSPs. Two pulses were given separated by 40 ms. Per cent facilitation of the EPSP slope was compared in responses taken before and during thiocyanate application and in responses taken during and after application (11 pairs each). In the latter case, the amplitude of the EPSP was adjusted to approximately the same size as that measured in the presence of thiocyanate. Differences in both comparisons were significant at P < 0.005. Representative traces from one of the paired-pulse facilitation experiments are shown on the right side, responses to the first and the second pulse were superimposed. Top: in the presence of 20 mM thiocyanate. Bottom: after wash-out of thiocyanate and after increasing the amplitude to match that measured in the presence of thiocyanate. Calibration: 0.5 mV/10 ms. (C) Paired pulse facilitation (PPF; 40 ms interval) and increase in EPSP slope induced by 20 mM sodium thiocyanate (SCN) were measured in the same slice in two different media characterized by either a low calcium concentration (2mM Ca2+/4mM Mg z+, Low) or a high calcium concentration (5mM Ca2+/l mM Mg 2+, Hi). Each column represents the mean and S.E.M. of nine experiments. *P < 0.05; ***P < 0.005 (paired t-test).

calcium c o n c e n t r a t i o n s in the perfusion m e d i u m was a b o u t the same as t h a t observed for paired-pulse facilitation. E n h a n c e m e n t of synaptic responses by drugs k n o w n to act o n A M P A receptors has elsewhere b e e n - s h o w n to r e m a i n u n c h a n g e d when combined with m a n i p u l a t i o n s acting o n t r a n s m i t t e r release. 38 T a k e n together, these o b s e r v a t i o n s suggest t h a t the increases in slope a n d amplitude o b t a i n e d

with t h i o c y a n a t e can be fully accounted for by a n increase in t r a n s m i t t e r release. The increase in EPSP amplitude is in agreement with the observations reported by Shahi and Baudry. 3° The a u t h o r s o f that study, however, did not find any change in the N M D A receptor mediated responses or in paired-pulse facilitation a n d thus a t t r i b u t e d the EPSP e n h a n c e m e n t to the effect of

Thiocyanate effect on AMPA receptor mediated responses thiocyanate on AMPA receptor affinity. The reasons for the experimental differences between these earlier results and the findings reported here are not clear. In the present study, thiocyanate was applied through the perfusion medium rather than by ionophoresis and thus the effect was of longer duration; non-receptor effects may accordingly have been more prominent. As thiocyanate, based on the excised patch results, would be expected to cause minimal changes in the amplitude of synaptic responses, the facilitation reported by Shahi and Baudry may indicate that the chaotropic ion produces somewhat different effects on receptors embedded in synaptic membranes. However, as noted, there was little evidence in the present experiments that thiocyanate's effects on slope and amplitude were mediated via an action on synaptic receptors once changes in release were factored out. It thus seems necessary to postulate that the brief and local applications of the chaotropic ion produce receptor changes of a type not seen with infusions. Steady-state currents could not be reliably assessed in the present study and a direct comparison with the results reported by Bowie and Smart 5 was not attempted. It can be predicted, however, that the enhancement of desensitization seen here would manifest itself in a reduction in the steady-state current as reported in their study (see Fig. 5B below).

gradients tend to expel the anion from the cell interior and as the effects of thiocyanate reversed within 30 min of wash-out; the ion could be applied repeatedly without evident deterioration of slice physiology. Chaotropic ions at near molar concentrations may produce general disturbances in protein structure and in the way proteins interact with their environment,7 but the fact that thiocyanate acts on AMPA receptors at millimolar concentrations (see above, and Refs 6, 10) and has no comparable effects on the binding properties of kainate and NMDA receptors (unpublished observations) suggests a more specific mode of action. The target site on the receptor remains unknown, but the available data permit to draw certain conclusions as to which aspects of receptor kinetics are modulated by thiocyanate. A receptor model similar to that described by Vyklicky et al. 36 and by Ambros-Ingerson and Lynch 3 (Fig. 5) has been adopted in an attempt to reconcile the effects of thiocyanate on physiological responses with those on binding. Models such as this one can with appropriately selected rate constants reproduce many of the observed features of AMPA receptors including the kinetics of desensitization, open and closed time characteristics, and binding properties (see, e.g., Ref. 3 for a discussion). The dissociation constant Kd of equilibrium binding is described by the following equation: 3

Mode q f action o f thiocyanate

=

The delocalization of its charge allows thiocyanate to penetrate membranes and it is not unlikely that this produces some perturbations in membrane potentials resulting, among other things, in changes in transmitter release. Such disturbances may be small and transient, however, as prevailing potential

823

4

+~2+

Z ) -1.

The Kd is approximately 15/~M for the set of constants defined in Fig. 5 which corresponds to affinity values typically measured in the absence of thiocyanate at 25°C. The Kd is evidently determined not only by the rate constants around the sensitive states i¸

o

OoC

25°C

iI ~\ 4

< 2~ <

3i!,

+ SCN '~" ~'~

T'/2 22 s, 242 s

~ - SCN

/

~ ..... 4

0

10

+ SCN

~ - - ~ _

TV2 1.8 s, 17 s

------e~

20

30

(1)

~'~

40

- SCN

10

20

Dissociation time (seconds)

Fig. 4. Thiocyanate effect on [3H]AMPA dissociation at 0°C and 25°C. Membranes were allowed to equilibrate for 1 h at 0°C in 1000nM [3H]AMPA and 50 mM KSCN; 4-pl aliquots were then diluted 250-fold into buffer with or without 50 mM potassium thiocyanate maintained at 0°C or 25°C. After the duration shown on the x-axis, 4 ml ice-cold stop solution (50 mM Tris/HCl plus 50 mM KSCN, pH 7.4) was injected into the sample which was then rapidly collected on a Whatman GF/C filter and washed three times with the same buffer (duration of procedure about 10 s). Non-specific .bindingfor each time point was determined by subjecting aliquots from a parallel incubation containing 5 mM e-glutamate to the same dilution regimen. Exponential decay curves were fitted to the complete set of data points which included additional time points extending up to 40 s (25°C) and 90 min (0°C). Under all four conditions, an adequate fit was obtained only by assuming at least two components. T~/2values for the dissociation at 0°C are shown in the figure. SCN, potassium thiocyanate.

824

A. Arai et al.

B

A L*k 1

a k.

\

k.1

ks

k_s

Con

" RLO

/

klk2

k4

Rd~

RL

a, a* /

L*k s k_3

k -1,1, k_3 '["

RdL i

-

b,b*

/ k i t

C

k _3,[.

c+d /~Jf

I

f~

/

/

/f

~ ~ 7 -

C, C*

k2,1, k _3,[. d, d*

k2'l' k _3,1, Fig. 5. Receptor model; changes in responses after 10-fold changes in selected rate-constants. (A) Receptor model. R denotes the sensitized, R d the desensitized state, and RLO the open state of the receptor; L, ligand. The responses shown in B and C were calculated with the following basic set of rate constants, adopted with minor modifications from Ref. 3: k~ 1 #M ~ s ~; k ~ 1000s-~; k 2 800s-~; k 2 8s-~; k3 10#M l s ~; k 3 50s-~; k4 1 s - t ; k 4 2s-I; k5900s ~; k s 500s-=- Estimates for these rate-constants were derived from published values of open time, mean burst time, and shut time distributions of AMPA receptor currents, from its desensitization characteristics, and from its binding affinity for glutamate (for further details, see Ref. 3). (B) The trace in the top panel and the dotted lines in panels 2 5 show the response to a 40-ms application of 1 mM agonist produced by a receptor which is governed by the basic set of rate constants shown above; solid lines in panels 2-5 show the response after changing by a factor of I0 the pair of rate constants indicated by the letters a~l and a*~t* in Table 1. (C) A combination of the conditions c and d with equal weighting produces a response which is characterized by both a decrease in the amplitude and a reduction in the desensitization rate constant similar to that found during AMPA application to excised patches.

R and RL, but also by those connected to the desensitized state o f the receptor R d and RdL. Table 1 lists how an assumed change in one o f the rateconstants by a factor o f 10 would affect the dissociation c o n s t a n t Kd, the arrows indicate whether the respective rate-constant is increased (1") or decreased (~). Interconversions within the loop (subscripts _ 1 to + 4 ) have to satisfy microscopic reversibility and their rate-constants can accordingly

be changed only in pairs; all permissible combinations are s h o w n with the first constant listed in the top row and the paired c o n s t a n t in the second row. Identical changes are obtained in all cases if an n-fold change in a forward rate c o n s t a n t is substituted with a 1/n fold change o f the reverse rate constant. Changes between constants belonging to the same interconversion do not affect the dissociation constant as is evident from Eqn 1.

Table 1. Changes in the dissociation constant K a produced by a 10-fold increase (T) or decrease (~) of selected rate-constants First constant Second constant k d decrease

k I $ ( o r k I T) and k_2 T k_3 + a
k_ 2 $ (or k 21") and k 3 $ c(d) k 4 $ or or k3 Y c, (d*) k41" 9.5 × 2.5 x

k 3 + ( o r k 3T) k 51" k51" and or k 5 + or k 5 k 4T or k4 1.4 x ~ 1x 1.33 x

Combinations which are characterized by producing a large decrease in Ka are marked by the letters a~l and a*~t*, letters in parenthesis refer to the combinations in which the first constant is selected as k~ I" or k 2 1"-

Thiocyanate effect on AMPA receptor mediated responses It is clear that changes in many of these combinations and in the opening (ks) and closing rates (k 5) have little impact on binding affinity. The only combinations which produce a large decrease in K d are the correlated changes between k+l and k± 3 and between k + 2 and k ± 3 and it is therefore likely that the changes produced by thiocyanate are to be found among these combinations which have been marked with the superscripts a - d and a*~t*. Combinations with and without asterisks are indistinguishable in their Kd and in the response profiles shown in Fig. 5B (see below), the only evident difference exists in the rate at which the bound ligand dissociates from the equilibrium state of the receptor. Simulations of the dissociation process produced time-constants with values of 13 ms for the standard set of rateconstants, 12-14 ms for combinations a*, b*, c* and d* (characterized by a 10-fold increase in k3), and 74-127 ms for combinations a, b, c and d (10-fold decrease in k 3). Since dissociation of bound [3H]AMPA is strongly reduced in presence of thiocyanate (Fig. 4), combinations with asterisks are not likely to make a significant contribution to the effect of thiocyanate. Figure 5B depicts how a 10-fold change in each of the pairs of rate constants a~d would alter the response produced by a 40 ms application of 1 mM agonist: calculations are based on the model of Fig. 5A. Dotted lines show the response obtained with the basic set of rate constants, solid lines those obtained after changing two of the rate-constants as indicated in Table 1. Actual receptor modifications probably concern more than two rate-constants and may accordingly have to be seen as superpositions of these sets; the traces shown in Fig. 5B should then be taken as elements which can be assessed for their possible contribution to actual receptor responses. All responses except c are associated with a decrease in the time-constant of the desensitization, the effect being largest for d ( ~ 5 0 % ) and less than 25% for a and b. Response c shows little change in response amplitude, whereas a and b are associated with a small increase; combination d stands apart in that it produces a large reduction in the amplitude. As thiocyanate produces a sizeable reduction in both amplitude and desensitization rate-constant of the actual A M P A response, it can be adequately reproduced only by a combination of d and one of the other sets; one possible combination with equal weighting of d and c is shown in Fig. 5C. This suggests that thiocyanate primarily changes the rateconstants leading to and away from the desensitized state RdL, and that a large reduction in the dissociation rate k 3 and an increase in the desensitization rate k 2 are likely to be among its main effects. To summarize, the main effect of thiocyanate may be to stabilize the desensitized state of the receptor. The changes evidently lead to a substantial increase in equilibrium binding affinity but have only minor effects on initial receptor responses. The most consist-

825

ent change across different ligands and preparations was the acceleration in the decay of the response. The model calculations indeed suggest that most combinations of rate-constants likely to be affected by thiocyanate would produce such an effect. Effects on the peak amplitude in excised patches were more variable in that responses to A M P A were strongly reduced while glutamate responses exhibited little change in the presence of thiocyanate. Theoretical considerations again suggest that effects on amplitude may be more variable depending on the particular set of rate-constants connected with a certain ligand. For example, thiocyanate induced changes in receptor operation when A M P A is bound may be predominantly of the type shown under d, whereas changes after binding of glutamate may have been produced by a combination of d with either a or b in such a manner that effects on amplitude cancelled out. A dominant contribution from the latter might even result in an increase in response amplitude, but whether thiocyanate indeed had such effects on EPSPs could not be reliably assessed in our study due to its probable side effects on transmitter release. The relationship between binding and response changes explored in Table 1 and Fig. 5 is not confined to the particular model used above. Similar results were obtained with sets of rate constants such as those used by Vyklicky e t a/., 28'36 except that changes in some combinations involving k±4 were also associated with a considerable increase in binding affinity; those combinations did not, however, produce the reduction in the ligand dissociation characteristic for thiocyanate (not shown). The results presented above suggest that changes in binding caused by a given treatment cannot be reliably used to predict either the direction or the magnitude of the change which the same treatment may produce on synaptic responses, particularly if the binding changes are assessed in the presence of thiocyanate as is usually the case. Conversely, conditions leading to an increase in A M P A receptor currents such as long-term potentiation may or may not be positively correlated with any binding changes. Ambros-Ingerson e t al. 2 found that the EPSP waveform changes in a subtle way after long-term potentiation and they were able to adequately reproduce these changes and the increase in the synaptic response in their model by increasing the opening and closing rates k 5 and k 5. It is evident from Table 1 that changes in these parameters have minimal effect on binding affinity. This may explain why attempts in this laboratory to find binding changes after inducing LTP in hippocampal slices have not been successful/* CONCLUSION

The increase in binding affinity produced by thiocyanate is likely to be the result of an enhanced receptor desensitization. The observation that thiocyanate accelerates the decay of synaptically evoked

826

A. Arai et al.

responses in a manner which is similar to its effect on responses in excised patches provides further evidence that the kinetics of A M P A receptors contribute significantly to the waveform of fast excitatory synaptic responses. The negative correlation between binding affinity and receptor efficacy calls in question hypotheses linking biochemical processes that lead to increased binding to the receptor enhancement presumed to subserve long-term potentiation. The nature of the relationship between biochemical and functional receptor changes may have to be established separately for each such process, but it would be greatly facilitated by the further development

of models which properly take into account the constraints from experimental studies. Acknowledgements--This research was supported by the

AFOSR grant 92-0307. The authors wish to express their sincere thanks to Dr C.-M. Tang (Philadelphia/Baltimore) for his technical advice concerning the construction of the fast-switch system and for making available various components of the setup. We are further indebted to Dr Jose Ambros-Ingerson for many helpful discussions and for providing initial information about the connection between receptor binding affinity and receptor responses. J.S. is the recipient of an Instructional Development Award of the University of California, Irvine, for undergraduate students. The authors also wish to thank Jackie Porter and Maria Lay for their excellent secretarial assistance.

REFERENCES

1. Ambros-Ingerson J., Larson J., Xiao P. and Lynch G. (1991) LTP changes the waveform of synaptic responses. Synapse 9, 314-316. 2. Ambros-Ingerson J., Xiao P., Larson J. and Lynch G. (1993) Waveform analysis suggests that LTP alters the kinetics of synaptic receptor channels. Brain Res. 620, 237 244. 3. Ambros-Ingerson J. and Lynch G. (1993) Channel gating kinetics and synaptic efficacy: a hypothesis of LTP expression. Proc. natn. Acad. Sci. U.S.A. 90, 7903-7907. 4. Baudry M., Massicotte G. and Hauge S. (1991) Phosphatidylserine increases the affinity of the AMPA-quisqualate receptor in rat brain membranes. Behav. neural Biol. 55, 137-140. 5. Bowie D. and Smart T. G. (1993) Thiocyanate ions selectively antagonize AMPA-evoked responses in Xenopus laevis oocytes microinjected with rat brain mRNA. Br. J. Pharmac. 109, 779 787. 6. Cha J.-H, J., Makowiec R. L., Penney J. B. and Young A. B. (1992) Multiple states of rat brain (RS)
Thiocyanate effect on AMPA receptor mediated responses

827

25. Murphy D. E., Snowhill E. W. and Williams M. (1987) Characterization of quisqualate recognition sites in rat brain tissue using DL-[3H]amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA) and a filtration assay. Neurochem. Res. 12, 775-782. 26. Nielsen E. O., Cha J.-H. J., Honore T., Penney J. B. and Young A. B. (1988) Thiocyanate stabilizes AMPA binding to the quisqualate receptor. Eur. J. Pharmac. 157, 197-203. 27. Olsen R. W., Szamraj O. and Houser C. R. (1987) [3H]AMPA binding to glutamate receptor subpopulations in rat brain. Brain Res. 402, 243-254. 28. Patneau D. K. and Mayer M. L. (1991) Kinetic analysis of interactions between kainate and AMPA: evidence for activation of a single receptor in mouse hippocampal neurons. Neuron 6, 785-798. 29. Patneau D. K., Vyklicky L. Jr and Mayer M. L. (1993) Hippocampal neurons exhibit cyclothiazide-sensitive rapidly desensitizing responses to kainate. J. Neurosci. 13, 3496-3509. 30. Shahi K. and Baudry M. (1992) Increasing binding affinity of agonists to glutamate receptors increases synaptic responses at glutamatergic synapses. Proe. natn. Aead. Sci, U.S.A. 89, 6881-6885. 31. Stoppini L., Buchs P.-A. and Muller D. (1991) A simple method for organotypic cultures of nervous tissue. J. Neurosci. Meth. 37, 173-182. 32. 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, 288590. 33. Tocco G., Maren S., Shors T. J., Baudry M. and Thompson R. F. (1992) Long-term potentiation is associated with increased [3H]AMPA binding in rat hippocampus. Brain Res. 573, 228-234. 34. Trussell L. O. and Fischbach G. D. (1989) Glutamate receptor desensitization and its role in synaptic transmission. Neuron 3, 209-218. 35. Vanderklish P., Neve R., Bahr B. A., Arai A., Hennegriff M., Larson J. and Lynch G. (1992) Translational suppression of a glutamate receptor subunit impairs long-term potentiation. Synapse 12, 333-337. 36. 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. 37. Yamada K. A. and Tang C.-M. (1993) Benzothiadiazides inhibit rapid glutamate receptor desensitization and enhance glutamatergic synaptic currents. J. Neurosci. 13, 3904-3915. 38. Xiao P., Staubli U., Kessler M. and Lynch G. (1991) Selective effects of aniracetam across receptor types and forms of synaptic facilitation in hippocampus. Hippocampus 1, 373-380. 39. Zorumski C. F., Yamada K. A., Price M. T. and Olney J. W. (1993) A benzodiazepine recognition site associated with the non-NMDA glutamate receptor. Neuron 10, 61-67. (Accepted 18 November 1994)