669
Journal of Physiology (1991), 443, pp. 669-682 With 7 figures Printed in Great Britain
TETANIC STIMULI INDUCE A SHORT-TERM ENHANCEMENT OF RECURRENT INHIBITION IN THE CA3 REGION OF GUINEA-PIG HIPPOCAMPUS IN VITRO
BY RICHARD MILES From the Laboratoire de Neurobiologie Cellulaire, Institut Pasteur, 25 rue de Dr Roux, 75264 Paris, Cedex 15, France
(Received 14 January 1991) SUMMARY
1. Recordings were made from pre- and postsynaptic cells at synapses involved in generating synaptic inhibition in the CA3 region of slices from guinea-pig hippocampus to examine the short-term effects of tetanic stimuli on recurrent inhibitory circuits. 2. Disynaptic inhibitory interactions were facilitated for a period of 2-5 min after tetanic stimulation of afferent fibres. The facilitation was due to an increase in the probability that IPSPs were transmitted. The amplitude of IPSPs that were evoked did not change. 3. IPSPs elicited by single action potentials in inhibitory cells were not changed immediately after tetanic stimulation. 4. There was no short-term change in the amplitude of unitary or spontaneously occurring EPSPs recorded from inhibitory cells. 5. The frequency of spontaneous IPSPs recorded in CA3 pyramidal cells increased transiently after tetanic stimuli delivered in the presence of excitatory amino acid receptor antagonists. 6. Tetanic stimuli caused a voltage-dependent increase in excitability of CA3 inhibitory cells. Little effect was apparent at hyperpolarized potentials. Near resting potential, tetanic stimuli enhanced inhibitory cells firing frequency partly by increasing the rate of membrane repolarization after an action potential. 7. These findings suggest tetanic stimulation releases a transmitter or other factor that alters intrinsic currents of CA3 inhibitory cells for several minutes, increasing their excitability and resulting in a transient facilitation of recurrent synaptic inhibition. INTRODUCTION
Excitatory synaptic responses in hippocampal pyramidal cells are persistently enhanced after high-frequency stimulation of afferent fibres (Bliss & L0mo, 1973). It is less clear whether tetanic stimuli induce changes in inhibitory synaptic circuits. Several indirect observations suggest short-term and long-term changes in synaptic inhibition may occur. First, postsynaptic firing may be enhanced more than MS 9075
670
R. MILES
expected from EPSP potentiation (Bliss & L0mo, 1973). This spike potentiation might result from a reduction in efficacy of inhibitory synaptic circuits (Abraham, Gustafsson & Wigstrom, 1987; Chavez-Noriega, Halliwell & Bliss, 1990). Second, tetanic stimuli may induce epileptiform discharges similar to those induced by suppressing inhibition pharmacologically (MacVicar & Dudek, 1979; Stasheff, Bragdon & Wilson, 1985; Stelzer, Slater & ten Bruggencate, 1987; Miles & Wong. 1987b; Higashima, 1988). A third observation suggests inhibitory circuits may exhibit a short-term facilitation. While EPSPs are enhanced immediately when inhibition is suppressed, potentiation may be delayed or preceded by a short-term depression when synaptic inhibition is functional (Alger, Megela & Teyler, 1978; Yamamoto & Chujo, 1978; Andersen, Sundberg, Sveen, Swann & Wigstrom, 1980). However, studies using afferent stimuli to examine plasticity in inhibitory circuits (Misgeld, Sarvey & Klee, 1979; Haas & Rose, 1982; Abraham et al. 1987) have been compromised for several reasons. First, the time course of excitatory and inhibitory synaptic events overlaps (Griffith, Brown & Johnston, 1986). Changes in either excitation or inhibition could then affect postsynaptic responses. Second, afferent stimuli activate inhibitory cells indirectly via feedforward or feedback excitatory synapses. So changes in afferent-evoked inhibition might reflect alterations at several sites. These include the synapses which excite inhibitory cells, the inhibitory cells themselves and inhibitory synapses. Analysis of unitary events at inhibitory synapses and excitatory synapses onto inhibitory cells might overcome some of these problems. This paper therefore presents a unitary analysis of plasticity in inhibitory circuits in the CA3 region of the hippocampus. As a first step, short-term changes were examined. Previous work has shown recurrent excitatory pathways between CA3 pyramidal cells may become functional after tetanic stimuli (Miles & Wong, 1987 a). These changes in polysynaptic excitatory circuits occurred with a delay of several minutes after stimulation. Since recurrent inhibition controls transmission in excitatory circuits (Miles & Wong, 1987 b), the delay might result from a short-term facilitation of synaptic inhibition. Now I report recurrent inhibition is enhanced for 2-5 min after tetanic stimuli. The enhancement seems to result not from changes at inhibitory synapses, nor in recurrent excitation of inhibitory cells but from an increase in inhibitory cell excitability. METHODS
Experiments were done on transverse hippocampal slices, of thickness 400 ,um, prepared with a vibratome from guinea-pigs (weight 150-300 g) which were killed by cervical dislocation. Slices were put in a recording chamber, maintained at 37 °C and at a pH close to 7 4, where they were supported on nylon mesh, at the interface between a 5 % C02 in 02 atmosphere, and a physiological saline of normal composition (mM); NaCl, 124; KCl, 4; CaCl2, 2; MgCl2, 2; NaHCO3, 26; and Dglucose, 10. Intracellular recordings were made from pyramidal cells and from inhibitory cells located close to the stratum pyramidale in the CA3 region. Glass recording electrodes were usually filled with 3 M-potassium acetate and bevelled to a final resistance of 30-50 MQ. In some experiments, electrodes were filled with 3 M-KCl (resistance 15-30 MQ) to increase intracellular Cl- so that IPSPs could be examined with improved voltage resolution. Excitatory amino acid receptor antagonists were used in these experiments to ensure that depolarizing events, recorded at potentials of -80 to -I00 mV, represented IPSPs (Miles, 1990 b). The antagonists used were D,L-2-amino-5phosphonovaleric acid (APV, 100 ,tM) and 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX,
PLASTICITY IN HIPPOCAMPAL INHIBITORY CIRCUITS
671
10-40 4uM), both obtained from Tocris Neuramin. The GABA, channel antagonist picrotoxin (50 /SM, Sigma) was used in some experiments to suppress Cl--mediated synaptic inhibition. Divalent cation concentrations were sometimes increased (4 mM-Ca2+ and 4 mM-Mg2+) for recordings of spontaneous synaptic events. In elevated Ca2+ and Mg2+, neuronal firing threshold was increased, possibly due to an effect on membrane surface charges (Hille, 1968). This reduced the frequency of spontaneous inhibitory and excitatory synaptic events and facilitated measurements. To examine unitary synaptic events, presynaptic cells were made to fire single action potentials at 0 5 Hz by depolarizing current injections. Hyperpolarizing holding current was applied if needed to prevent firing between pulses. After an interaction between two cells was established, at least 10 min was allowed to elapse before tetanic stimulation of afferent fibres. Responses of the postsynaptic cell to 0 5 nA, 100 ms current pulses were examined during this period and again 10 min after tetanic stimulation. Results were not accepted if input resistance changed by more than 10 % or if postsynaptic resting potential deviated by more than 3 mV from its initial value. In addition results were not accepted if subsequent analysis showed a persistent increase or decrease in amplitude of synaptic events before stimulation. Electrical pulses of duration 100 /is applied to bipolar tungsten electrodes, normally placed near the trajectory of the mossy fibres in the CA4 region, were used to stimulate afferent fibres. In experiments carried out in the presence of CNQX and APV the distance between the stimulating and recording electrodes was between 1 and 2 mm. Tetanic stimulation consisted of ten to fifty impulses at 50-100 Hz repeated 3-5 times within 30 s. Procedures for recording, storage and analysis of synaptic events were as described previously (Miles, 1990a, b).
RESULTS
Short-term enhancement of recurrent inhibition The purpose of this study was to detect whether tetanic stimulation of afferent fibres induced short-term changes in inhibitory synaptic circuits. First, disynaptic inhibitory interactions between CA3 pyramidal cells (Miles, 1990a) were used to examine changes in the efficacy of recurrent inhibition. Interactions studied were limited to those where single pyramidal action potentials evoked IPSPs of latency greater than 3 ms and with mean amplitude greater than 1 mV so that transmission failures could be clearly identified (Fig. 1). At inhibitory synapses which generate IPSPs of this amplitude, transmission does not fail (Miles, 1990 b). Failure of disynaptic transmission may then reflect instances when an EPSP in an intercalated inhibitory cell did not cause it to fire. IPSP amplitude depends on the efficacy of the inhibitory synapse. So disynaptic interactions provide information both on inhibitory synaptic strength and on the effectiveness of recurrent excitation of inhibitory cells. After tetanic stimulation the amplitude of averaged disynaptic IPSPs increased by more than 20 % in eleven of twelve interactions examined. Enhancement was apparent within several seconds after stimulation, reached a maximal value within 2 min and subsequently declined. In five of twelve interactions, averaged IPSPs declined to an amplitude less than control at times greater than 10 min after stimulation (Miles & Wong, 1987 a). When tetanic stimulation was delivered twice at intervals longer than 3 min disynaptic IPSPs were enhanced after both stimuli (n = 4 cell pairs). The transient enhancement of recurrent inhibition might reflect changes at inhibitory synapses or in pyramidal cell excitation of inhibitory cells. Amplitude distributions constructed for disynaptic IPSPs suggested the number of transmission failures was reduced (Fig. 1B). From twelve disynaptic interactions, the probability
R. MILES
672
that IPSPs were transmitted during a 5 min control period before stimulation was 0-56 + 0'20 (mean + S.D., n = 12 disynaptic interactions). The probability increased to 0-86 + 0'17 in the 2 min period after stimulation and declined to 0-59 + 0-23 for the period from 2 to 5 min after tetanization. Transmission probability was enhanced by more that 20 % of control in ten of twelve interactions. A
B
I ,
>-'1"A-
0.2
Noise
(a-c)rlIL
p 2
L
Control
20V
0.0.
mV ~~~~~~~~~~120 0.2- Control
(b-a)
*
2
-----
2.o0.,01.. , A , 0-2
0-3 min
-0-3 min
0-2
6-9 min
6-9 min
2 mV
.20 ms cab
-5
-4
-3 -2 -1 Amplitude (mV)
0
1
Fig. 1. A short-term facilitation of disynaptic inhibition. A, single action potentials in cell 1 elicited disynaptic IPSPs in cell 2 with latency 2-6 ms and frequent transmission failures. The frequency of failure was reduced at 0-3 min after tetanic stimulation and then increased again. B, IPSPs could be differentiated from transmission failures. Postsynaptic responses were measured between a time before the spike, a, and the peak of the averaged IPSP, b, and noise was measured with a similar time difference (a-c). The peak in the control panel centred about 0 mV corresponds to transmission failures. In a control period of 5 min before stimulation the probability of failure was 0-52 and mean IPSP amplitude was 2-4 mV. Tetanic stimulation (100 Hz, twenty impulses, three trains) was then applied at a site close to the trajectory of the mossy fibres. From 0 to 3 min after tetanus the probability of failure was reduced to 0-08 and the mean IPSP amplitude was 2-7 mV. At 6-9 min the probability of failure increased to 0-72 and the mean IPSP amplitude was 3-1 mV.
Changes in amplitude of IPSPs evoked disynaptically were less consistent. Mean IPSP amplitude in the control period was 1-5 +0-4 mV (mean+ S.D., n = 12). It increased to 1-6 + 0-7 mV in the 2 min immediately after the tetanus and declined to 1-4 + 0-6 mV in the period from 2 to 5 min after tetanization. A change of more than 20% of control occurred at five connections immediately after stimulation. In two cases the change was a facilitation and in the other three cases a depression occurred.
PLASTICITY IN HIPPOCAMPAL INHIBITORY CIRCUITS 673 Figure 2 shows the time course of the tetanus-induced changes in the probability of IPSP transmission and in IPSP amplitude from disynaptic inhibitory interactions. Averages of both parameters for 20 s periods were made from twelve interactions for 5 min before and after tetanic stimulation. The increase in transmission probability A 1.0-
.0 o
0-5-
E C,,
3 B
-
Journal of Physiology (1991), 443, pp. 669-682 With 7 figures Printed in Great Britain
TETANIC STIMULI INDUCE A SHORT-TERM ENHANCEMENT OF RECURRENT INHIBITION IN THE CA3 REGION OF GUINEA-PIG HIPPOCAMPUS IN VITRO
BY RICHARD MILES From the Laboratoire de Neurobiologie Cellulaire, Institut Pasteur, 25 rue de Dr Roux, 75264 Paris, Cedex 15, France
(Received 14 January 1991) SUMMARY
1. Recordings were made from pre- and postsynaptic cells at synapses involved in generating synaptic inhibition in the CA3 region of slices from guinea-pig hippocampus to examine the short-term effects of tetanic stimuli on recurrent inhibitory circuits. 2. Disynaptic inhibitory interactions were facilitated for a period of 2-5 min after tetanic stimulation of afferent fibres. The facilitation was due to an increase in the probability that IPSPs were transmitted. The amplitude of IPSPs that were evoked did not change. 3. IPSPs elicited by single action potentials in inhibitory cells were not changed immediately after tetanic stimulation. 4. There was no short-term change in the amplitude of unitary or spontaneously occurring EPSPs recorded from inhibitory cells. 5. The frequency of spontaneous IPSPs recorded in CA3 pyramidal cells increased transiently after tetanic stimuli delivered in the presence of excitatory amino acid receptor antagonists. 6. Tetanic stimuli caused a voltage-dependent increase in excitability of CA3 inhibitory cells. Little effect was apparent at hyperpolarized potentials. Near resting potential, tetanic stimuli enhanced inhibitory cells firing frequency partly by increasing the rate of membrane repolarization after an action potential. 7. These findings suggest tetanic stimulation releases a transmitter or other factor that alters intrinsic currents of CA3 inhibitory cells for several minutes, increasing their excitability and resulting in a transient facilitation of recurrent synaptic inhibition. INTRODUCTION
Excitatory synaptic responses in hippocampal pyramidal cells are persistently enhanced after high-frequency stimulation of afferent fibres (Bliss & L0mo, 1973). It is less clear whether tetanic stimuli induce changes in inhibitory synaptic circuits. Several indirect observations suggest short-term and long-term changes in synaptic inhibition may occur. First, postsynaptic firing may be enhanced more than MS 9075
670
R. MILES
expected from EPSP potentiation (Bliss & L0mo, 1973). This spike potentiation might result from a reduction in efficacy of inhibitory synaptic circuits (Abraham, Gustafsson & Wigstrom, 1987; Chavez-Noriega, Halliwell & Bliss, 1990). Second, tetanic stimuli may induce epileptiform discharges similar to those induced by suppressing inhibition pharmacologically (MacVicar & Dudek, 1979; Stasheff, Bragdon & Wilson, 1985; Stelzer, Slater & ten Bruggencate, 1987; Miles & Wong. 1987b; Higashima, 1988). A third observation suggests inhibitory circuits may exhibit a short-term facilitation. While EPSPs are enhanced immediately when inhibition is suppressed, potentiation may be delayed or preceded by a short-term depression when synaptic inhibition is functional (Alger, Megela & Teyler, 1978; Yamamoto & Chujo, 1978; Andersen, Sundberg, Sveen, Swann & Wigstrom, 1980). However, studies using afferent stimuli to examine plasticity in inhibitory circuits (Misgeld, Sarvey & Klee, 1979; Haas & Rose, 1982; Abraham et al. 1987) have been compromised for several reasons. First, the time course of excitatory and inhibitory synaptic events overlaps (Griffith, Brown & Johnston, 1986). Changes in either excitation or inhibition could then affect postsynaptic responses. Second, afferent stimuli activate inhibitory cells indirectly via feedforward or feedback excitatory synapses. So changes in afferent-evoked inhibition might reflect alterations at several sites. These include the synapses which excite inhibitory cells, the inhibitory cells themselves and inhibitory synapses. Analysis of unitary events at inhibitory synapses and excitatory synapses onto inhibitory cells might overcome some of these problems. This paper therefore presents a unitary analysis of plasticity in inhibitory circuits in the CA3 region of the hippocampus. As a first step, short-term changes were examined. Previous work has shown recurrent excitatory pathways between CA3 pyramidal cells may become functional after tetanic stimuli (Miles & Wong, 1987 a). These changes in polysynaptic excitatory circuits occurred with a delay of several minutes after stimulation. Since recurrent inhibition controls transmission in excitatory circuits (Miles & Wong, 1987 b), the delay might result from a short-term facilitation of synaptic inhibition. Now I report recurrent inhibition is enhanced for 2-5 min after tetanic stimuli. The enhancement seems to result not from changes at inhibitory synapses, nor in recurrent excitation of inhibitory cells but from an increase in inhibitory cell excitability. METHODS
Experiments were done on transverse hippocampal slices, of thickness 400 ,um, prepared with a vibratome from guinea-pigs (weight 150-300 g) which were killed by cervical dislocation. Slices were put in a recording chamber, maintained at 37 °C and at a pH close to 7 4, where they were supported on nylon mesh, at the interface between a 5 % C02 in 02 atmosphere, and a physiological saline of normal composition (mM); NaCl, 124; KCl, 4; CaCl2, 2; MgCl2, 2; NaHCO3, 26; and Dglucose, 10. Intracellular recordings were made from pyramidal cells and from inhibitory cells located close to the stratum pyramidale in the CA3 region. Glass recording electrodes were usually filled with 3 M-potassium acetate and bevelled to a final resistance of 30-50 MQ. In some experiments, electrodes were filled with 3 M-KCl (resistance 15-30 MQ) to increase intracellular Cl- so that IPSPs could be examined with improved voltage resolution. Excitatory amino acid receptor antagonists were used in these experiments to ensure that depolarizing events, recorded at potentials of -80 to -I00 mV, represented IPSPs (Miles, 1990 b). The antagonists used were D,L-2-amino-5phosphonovaleric acid (APV, 100 ,tM) and 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX,
PLASTICITY IN HIPPOCAMPAL INHIBITORY CIRCUITS
671
10-40 4uM), both obtained from Tocris Neuramin. The GABA, channel antagonist picrotoxin (50 /SM, Sigma) was used in some experiments to suppress Cl--mediated synaptic inhibition. Divalent cation concentrations were sometimes increased (4 mM-Ca2+ and 4 mM-Mg2+) for recordings of spontaneous synaptic events. In elevated Ca2+ and Mg2+, neuronal firing threshold was increased, possibly due to an effect on membrane surface charges (Hille, 1968). This reduced the frequency of spontaneous inhibitory and excitatory synaptic events and facilitated measurements. To examine unitary synaptic events, presynaptic cells were made to fire single action potentials at 0 5 Hz by depolarizing current injections. Hyperpolarizing holding current was applied if needed to prevent firing between pulses. After an interaction between two cells was established, at least 10 min was allowed to elapse before tetanic stimulation of afferent fibres. Responses of the postsynaptic cell to 0 5 nA, 100 ms current pulses were examined during this period and again 10 min after tetanic stimulation. Results were not accepted if input resistance changed by more than 10 % or if postsynaptic resting potential deviated by more than 3 mV from its initial value. In addition results were not accepted if subsequent analysis showed a persistent increase or decrease in amplitude of synaptic events before stimulation. Electrical pulses of duration 100 /is applied to bipolar tungsten electrodes, normally placed near the trajectory of the mossy fibres in the CA4 region, were used to stimulate afferent fibres. In experiments carried out in the presence of CNQX and APV the distance between the stimulating and recording electrodes was between 1 and 2 mm. Tetanic stimulation consisted of ten to fifty impulses at 50-100 Hz repeated 3-5 times within 30 s. Procedures for recording, storage and analysis of synaptic events were as described previously (Miles, 1990a, b).
RESULTS
Short-term enhancement of recurrent inhibition The purpose of this study was to detect whether tetanic stimulation of afferent fibres induced short-term changes in inhibitory synaptic circuits. First, disynaptic inhibitory interactions between CA3 pyramidal cells (Miles, 1990a) were used to examine changes in the efficacy of recurrent inhibition. Interactions studied were limited to those where single pyramidal action potentials evoked IPSPs of latency greater than 3 ms and with mean amplitude greater than 1 mV so that transmission failures could be clearly identified (Fig. 1). At inhibitory synapses which generate IPSPs of this amplitude, transmission does not fail (Miles, 1990 b). Failure of disynaptic transmission may then reflect instances when an EPSP in an intercalated inhibitory cell did not cause it to fire. IPSP amplitude depends on the efficacy of the inhibitory synapse. So disynaptic interactions provide information both on inhibitory synaptic strength and on the effectiveness of recurrent excitation of inhibitory cells. After tetanic stimulation the amplitude of averaged disynaptic IPSPs increased by more than 20 % in eleven of twelve interactions examined. Enhancement was apparent within several seconds after stimulation, reached a maximal value within 2 min and subsequently declined. In five of twelve interactions, averaged IPSPs declined to an amplitude less than control at times greater than 10 min after stimulation (Miles & Wong, 1987 a). When tetanic stimulation was delivered twice at intervals longer than 3 min disynaptic IPSPs were enhanced after both stimuli (n = 4 cell pairs). The transient enhancement of recurrent inhibition might reflect changes at inhibitory synapses or in pyramidal cell excitation of inhibitory cells. Amplitude distributions constructed for disynaptic IPSPs suggested the number of transmission failures was reduced (Fig. 1B). From twelve disynaptic interactions, the probability
R. MILES
672
that IPSPs were transmitted during a 5 min control period before stimulation was 0-56 + 0'20 (mean + S.D., n = 12 disynaptic interactions). The probability increased to 0-86 + 0'17 in the 2 min period after stimulation and declined to 0-59 + 0-23 for the period from 2 to 5 min after tetanization. Transmission probability was enhanced by more that 20 % of control in ten of twelve interactions. A
B
I ,
>-'1"A-
0.2
Noise
(a-c)rlIL
p 2
L
Control
20V
0.0.
mV ~~~~~~~~~~120 0.2- Control
(b-a)
*
2
-----
2.o0.,01.. , A , 0-2
0-3 min
-0-3 min
0-2
6-9 min
6-9 min
2 mV
.20 ms cab
-5
-4
-3 -2 -1 Amplitude (mV)
0
1
Fig. 1. A short-term facilitation of disynaptic inhibition. A, single action potentials in cell 1 elicited disynaptic IPSPs in cell 2 with latency 2-6 ms and frequent transmission failures. The frequency of failure was reduced at 0-3 min after tetanic stimulation and then increased again. B, IPSPs could be differentiated from transmission failures. Postsynaptic responses were measured between a time before the spike, a, and the peak of the averaged IPSP, b, and noise was measured with a similar time difference (a-c). The peak in the control panel centred about 0 mV corresponds to transmission failures. In a control period of 5 min before stimulation the probability of failure was 0-52 and mean IPSP amplitude was 2-4 mV. Tetanic stimulation (100 Hz, twenty impulses, three trains) was then applied at a site close to the trajectory of the mossy fibres. From 0 to 3 min after tetanus the probability of failure was reduced to 0-08 and the mean IPSP amplitude was 2-7 mV. At 6-9 min the probability of failure increased to 0-72 and the mean IPSP amplitude was 3-1 mV.
Changes in amplitude of IPSPs evoked disynaptically were less consistent. Mean IPSP amplitude in the control period was 1-5 +0-4 mV (mean+ S.D., n = 12). It increased to 1-6 + 0-7 mV in the 2 min immediately after the tetanus and declined to 1-4 + 0-6 mV in the period from 2 to 5 min after tetanization. A change of more than 20% of control occurred at five connections immediately after stimulation. In two cases the change was a facilitation and in the other three cases a depression occurred.
PLASTICITY IN HIPPOCAMPAL INHIBITORY CIRCUITS 673 Figure 2 shows the time course of the tetanus-induced changes in the probability of IPSP transmission and in IPSP amplitude from disynaptic inhibitory interactions. Averages of both parameters for 20 s periods were made from twelve interactions for 5 min before and after tetanic stimulation. The increase in transmission probability A 1.0-
.0 o
0-5-
E C,,
3 B
-
