HIPPOCAMPUS, VOL. I , NO. 1, PAGES 67-78, JANUARY 1991
Local Circuit Synaptic Interactions Between CAl Pyramidal Cells and Interneurons in the Kainate-Lesioned Hyperexcitable Hippocampus S. Nakajima,*$ J. E. Franck,* D. Bilkey,§ and P. A. Schwartzkroin,*T D e p a r t m e n t s o f "Neurological S u r g e r y , and +Physiology and Biophysics, University of W a s h i n g t o n , Seattle, WA 98195 U.S.A., $ D e p a r t m e n t o f N e u r o p s y c h i a t r y , University o f K y o t o Prefectural University of Medicine, K y o t o , J a p a n , and $ D e p a r t m e n t of Psychology, University of O t a g o , D u n e d i n ,
New Zealand
ABSTRACT Following kainate (KA)-induced lesions of subfield CA3-a lesion relevant to human temporal lobe epilepsy-remaining pyramidal cells in CAI display synchronous hyperexcitability associated with a loss of synaptic inhibition. Despite this loss, inhibitory interneurons in CAI remain viable, and the density and function of GABAergic receptors on the CAI pyramidal cells are maintained at approximately normal levels. To further evaluate inhibition in this system, the authors examined interactions between pyramidal cells and inhibitory interneurons in paired intracellular recordings. Recordings were carried out in rat hippocampal slices 2-4 weeks following bilateral intraventricular KA injections. The frequency of synaptic interactions between CAI basket cells and pyramidal cells was lower in hyperexcitable slices than in controls: both synapses in the recurrent inhibitory circuit appeared to be involved. No recurrent excitatory interactions were seen between pyramidal cell pairs in lesioned or normal slices. The weakened interconnections between pyramidal cells and interneurons are consistent with the decreased inhibition previously found in this model. Unexpectedly, strong stimulation, which may directly activate local inhibitory circuitry, was effective in reducing hyperexcitability in KA-lesioned slices. These data suggest that development of recurrent excitatory connections among CAI hippocampal pyramidal cells contribute little to tissue excitability. and support the hypothesis that a functional uncoupling between inhibitory interneurons and CAI pyramidal cells is responsible for the seizure-like activity typical of KA-lesioned hippocampus. The data are also consistent with the hypothesis that in the KA model, the structural circuitry needed for inhibition in CAI is maintained, and can be functionally activated by appropriate stimuli. Key words: hippocampus, kainic acid, epilepsy, paired intracellular recording, inhibitory interneuron
Kainic acid (KA) selectively destroys hippocampal CA3 pyramidal cells when injected into the ventricles in low doses (Nadler et al., 1978a; Lancaster and Wheal, 1982). This KAlesioned hippocampus bears structural similarities to the pathologic findings in human temporal lobe epilepsy (Margerison and Corsellis, 1966; Corsellis and Meldrum, 1976). Significantly, neurons that remain viable following this lesion-CAl pyramidal and dentate gyrus granule cells-develop long-term hyperexcitability and synchrony, and may ~
~
Correspondence and reprint requests to P. A . Schwartzkroin, Department of Neurological Surgery, RI-20, University of Washington, Seattle, WA 98195 U.S.A.
provide a model for the study of hippocampal epileptogenesis, which accompanies cell loss (Franck and Schwartzkroin, 1985; Ashwood et al., 1986a). Several electrophysiological studies have demonstrated that the primary feature of these hyperexcitable CAI pyramidal cells is the presence of both synaptically elicited and spontaneous synchronous bursting and a reduction o r loss of GABA-mediated inhibitory postsynaptic potentials (IPSPs) (Ashwood et al., 1983, 1986a; Franck and Schwartzkroin, 1985, Franck et al., 1988). Curiously, however, while functional inhibition is compromised, the inhibitory cells and their terminal distribution in area CAI survive this lesion, much as they do in human epileptic hippocampus (Babb et al.,
67
68
HIPPOCAMPUS VOL. 1, NO. 1, JANUARY 1991
1989): they are electrophysiologically viable, synthesize GABA, and nearby pyramidal cells maintain their responsivity to exogenous GABA application (Davies et al., 1985: Ashwood et al., 1986b: Franck et al., 1988). In addition to a loss of functional inhibition, the CAI region of these lesioned hippocampi may develop significant recurrent excitation. Detailed histologic studies show that the excitatory synaptic density in stratum radiatum and stratum oriens of area CAI is reduced by 80% 7 days after KA injection, mainly due to a loss of Schaffer collaterals and commissural fibers arising from CA3 pyramidal cells. The synaptic density returns to a near-normal level by 40-SO days post-lesion (Nadler et al., 1978b, 1980a,b). The source of these reinnervatingfibers is still unknown (Nadler et al., 1980b). However. in view of the abnormal hyperexcitability and synchrony, these newly formed synapses might originate from the axon collaterals of nearby CAI pyramidal cells (Nadler et al., 1980a; Franck et al., 1988). These observations suggest that the hyperexcitability and synchrony that develop in the CAI region of the chronic KA model of epileptogenesis may involve alterations in both the manner in which pyramidal cells interact with interneurons and the way in which they interact with each other. We studied simultaneously impaled pyramidal cell pairs and pyramidal cell-interneuron pairs in the CAI region of KA-lesioned hyperexcitable hippocampi to determine the following: ( I ) Does the loss of functional inhibition in this tissue result from a loss of one component of the bidirectional interaction between pyramidal cells and interneurons and (2) Is there an increase in recurrent excitatory collateralization in this hyperexcitable CA 1 region? The data presented here suggest that the loss of inhibition is correlated with decreased efficacy of both inhibitory synapses from interneurons onto pyramidal cells and excitatory synapses from pyramidal cells onto interneurons. Furthermore, the hyperexcitability and synchrony that develops in this tissue does not depend on aberrant recurrent excitation in CAI.
extra.
FW \I
Jio
mV
10 msec
Pyramids Cell
lnhbilory
Fig. 1 , Diagrammatic representation of the experimental paradigm. An extracellular recording electrode (extra) was positioned in the CAI pyramidal cell layer, and the stimulating electrode was placed into CAI stratum radiatum (stim). Inset shows a typical abnormal multiple population response to stratum radiatum stimulation (.2 mA, .3 Hz) in a KA-A13 slice. After determining whether the electrophysiologic response from such stimulation consisted of a normal (single population) response or an abnormal (multiple population) response, intracellular microelectrodes (intra) were introduced into the pyramidal cell region. close to the site at which the field potential response had been monitored. DG, dentate granule cells.
MATERIALS A N D METHODS Male Long Evans rats (50-60 days old) were anesthetized with a ketarnineixylazine mixture ( 3 : I , 0.15 ml/lOO g, i.m.) and KA was slowly administered into each ventricle (.6 pg in I p1 of salineihemisphere, pH 7.2-7.8) from a 10 pl Hamilton syringe positioned stereotaxically according to the atlas of Paxinos and Watson (1986); coordinates for injections were 0.6 mm posterior to bregma, 2.0 mm lateral to the midline, and 3.5 mm deep to the dura. Control animals underwent the same procedure, but received injections of buffered saline. The animals were sutured, treated with antibiotics, and allowed to recover from anesthesia for 2-4 weeks prior to electrophysiological studies. Rats were decapitated, and a section of brain containing the hippocampus was quickly removed. Four to eight transverse slices (450 p m thick) were cut (Vibroslicer, Frederick. Haer) from each dorsal hippocampus. Slices were placed in an interface-type recording chamber, warmed to 35°C. humidified with 95% 0,/50/0 C02,and perfused with modified Krebs-Ringer buffer (126 mM NaCI, 5 mM KCI, 2 mM CaCI,, 1.2 mM MgCI,, 1.2 mM NaH,P04, 26 mM NaHCO?, and 10 mM dextrose) (pH 7.4, flow rate I.O
mlimin). All physiologic recordings were performed by the same group of investigators. At the start of each electrophysiological recording session. extracellular field potential recordings were made with electrodes containing 2 M NaCl(1-5 M i l ) to determine the degree of abnormal excitability present in the CA1 region of each slice. With the recording electrode placed in the pyramidal cell layer, a bipolar stimulating electrode (insulated tungsten wire; .5 mm tip separation) was placed in stratum radiatum (or the alveus) to orthodrornically (or antidromically) activate CAI pyramidal cells (.05 msec duration; .01-.3 mA; .3-3.0 Hzj. In normal slices, this stimulation elicits a single population spike; the presence of 2 o r more population spikes is a reliable indicator of bursting activity at the single cell level (Wheal et al., 1984). Slices were categorized a s abnormal if stimulation elicited multiple ( 2 2 ) population spikes (Fig. 1). In each slice, paired intracellular recordings were made in the CAI region. Glass microelectrodes ( 4 M potassium acetate, 40-1 10 M i l ) were inserted into o r near stratum pyramidale. In those slices with abnormal burst discharge in the field, the intracellular electrodes were introduced in the re-
LOCAL CIRCUIT INTERACTIONS IN KA-LESIONED HIPPOCAMPUS / Nakajima et al. gion from which the field burst had been recorded (Fig. I). Two intracellular electrodes were placed within about 200 p n of each other, and the activity of impaled cells was manipulated to determine whether the cells were synaptically connected (see below). The impaled neuron was classified as either a pyramidal cell or interneuron on the basis of its physiological properties (Schwartzkroin, 1975; Schwartzkroin and Mathers, 1978). In attempts to penetrate interneurons, the microelectrode was generally introduced at the border of the stratum pyramidale and stratum oriens, where the “basket cell” interneuron type is most often located (Schwartzkroin and Mathers. 1978; Ashwood et al., 1984). Synaptic interactions between pairs of impaled neurons were determined by the following process. One neuron was maintained at resting potential, while the other was depolarized by injecting current pulses (100-200 msec duration, .21.0 nA amplitude, .3 Hz) that evoked spike trains. The potential “postsynaptic” cell was monitored for the occurrence of depolarizing or hyperpolarizing potentials that were timelocked to action potential discharge in the stimulated neuron. In order to uncover small synaptic potentials, the membrane potential of the “postsynaptic” cell was manipulated in depolarizing and hyperpolarizing directions (injection of DC current of up to 1 nA). This procedure was then repeated,
69
reversing the stimulated and “postsynaptic” cell to test connections in the opposite direction. In all pairs, hylperpolarizing current pulses (100 msec duration pulses, .5 nA) were also introduced in order to determine the input resistance of these neurons and to assess electrotonic communication between cells. In a small number of experiments, the effect of stimulus proximity on field potentials was evaluated. Abnormal, multiple population spikes were first obtained in response to stratum radiatum stimulation. The stimulating electrode was then moved closer to, and then more distant from, the irecording electrode. PopuIation spike responses to stimulation, for all locations, were monitored. As for the intracellular data, all field potential data were digitized and stored on VCR tape. At the end of each electrophysiological experiment, the monitored slices were fixed in 10% buffered formalin, and subsequently stained with a Nissl stain in order to determine the degree of cell damage in each slice.
RESULTS In the present study, intraventricular K A injections destroyed subfield CA3 in almost all experimental anirnals, leaving the CAI and dentate regions intact (Fig. 2). N o lesion was found in the hippocampus in 4% of the KA-injected animals
Fig. 2. Histology of a typical kainate lesion. Pyramidal cells in CA3 are gone. Nissl stain shows cell bodies of surviving CAI and dentate gyrus (DG) cell populations. Cresyl Violet x 38.
70
HIPPOCAMPUS VOL. 1, NO. 1, JANUARY 1991
(2 of 53); in both of these cases, only normal field potentials were evoked in hippocampal slices. Somewhat surprisingly, abnormal, multiple population responses were evoked in slices of only 45% of KA-injected animals (24 of 53) with verified lesions. Such responses were seen in at least 1 slice from each of these animals, but were generally not seen in every slice. In the remaining animals (27 of 53), none of the slices generated abnormal population spikes in response to radiatum or alveus stirnulation. The likelihood of abnormal electrophysiological responsiveness was not clearly associated with the degree of CA3 damage. Because a significant proportion of lesioned animals appeared normal electrophysiologically, we divided our slice population into 3 groups: ( I ) KA-induced structural lesion with abnormal electrophysiological excitability (KA-AB); (2) KA lesion with normal electrophysiological responsiveness
(Le., single population spike in response to stimulation; KANOR); and (3) non-KA-treated and electrophysiologically normal (NOR).
Single cell properties and local interactions between pyramidal cells In 17 slices from KA-AB rats, we obtained 39 pairs of intracellular recordings, which included 67 pyramidal cells. The cell properties of these neurons are compared with 34 pyramidal cells monitored in control tissue (Table 1). Since the properties of cells in unlesioned animals (NOR) was identical to that of cells in KA-lesioned, but electrophysiologically normal, slices (KA-NOR), the control group for statistical comparisons was composed of the KA-NOR and NOR groups. The mean resting membrane potential (RMP) of cells in slices from KA-AB animals was not different from that of cells in
Table 1. Comparison of Cell Properties and Local Circuit interactions in Kainate-lesioned and Normal Slices . _ _ _ ~ ~ ~ ~ _ _ _ _ _ ~ _ _ . _ _ ~ _ _ _ _ _ _ _ _ _ _ _
_
~
_
_
_
KA-NOR + NOR
-
~
-
~
~
~
_
KA-AB _
KA-NOR
____~.__
NOR
(Control)
013
0125
~
A
Interactions P+P Cell properties Pyramidal cells
0139
RMP (mV) R,, (MR)
0122
-59.5 t 5.4 37.2 t 10.8** ( n = 67)
-60.4 27.6
(n
* 4.9 s 7.5
=
30)
-58.0 2 5.2 20.4 2 4.4 (n = 4)
-60.1 26.7
* 5.0 * 7.6
(n = 34)
B
Interactions P+I I+P Total interacting pairs Cell properties Pyramidal cells
119 019 119*
217 317 517
6/13 3113 6113
8120 6/20 11120
RMP (mV) Rin (Mn)
-58.9 '' 4.4 36.6 5 7.8**
RMP Rin
-59.3 t 3.0 41.7 7.2 (n = 6)
-58.3 t 3.6 27.9 s 6.1 (n = 7) -55.0 t 3.5 41.3 s 8.7 (n = 4)
-58.4 Z 5.0 24.6 2 7.6 (n = 13) -59.2 2 2.8 42.1 5 12.8 (n = 4)
-58.4 ? 4.5 25.8 t 7.1 (n = 20) -57.1 5 3.7 41.7 5 10.1 (n = 8)
(n
lnterneurons
(mV) (Ma)
_
_
=
_
9)
_
~
.
~
_
_
_
~
* P = 0.03, Fisher Exact Probability relative to control. Three normal pairs had bidirectional interactions, and the totals count these pairs
only once. **P < 0.01, z-test relative to control
Table 1. (A) Pyramidal cell properties and pyramidal cell-pyramidal cell interactions in KA-treated, electrophysiologically abnormal slices (KA-AB) (column l ) , KA-treated, electrophysiologically normal slices (KA-NOR) (column 2), and normal slices (no KA treatment; NOR) (column 3). No differences in input resident and resting membrane potential were found between the KA-NOR and NOR slices, and thus the groups were combined to provide the control population (column 4). A comparison between KA-AB and control populations showed a significantly higher input resistance in pyramidal cells of the KA-AB tissue. In neither KA-AB nor control slices were any synaptic interactions between pyramidal cells (P -+ P) uncovered (number of pairs showing synaptic interactioninumber of pairs surveyed). (B) Interactions and cell properties of pyramidal cell-interneuron (basket cell) pairs in KA-AB and control populations. Input resistance for this population of pyramidal cells was higher than Ri, in controls. However, interneuron properties did not differ between electrophysiologically abnormal and normal slices. In control slices, 8 of 20 pairs showed an excitatory drive from the pyramidal cell onto the interneuron; in contrast, only 1 of 9 such pairs in the KA-AB slices showed such an interaction. Six of 20 pairs showed inhibition from the interneuron onto the pyramidal cell in control slices, whereas no pairs showed such interaction in the KA-AB slices. Resting membrane potential and input resistance measurements are given as means -C standard deviation. Statistical comparisons were made between the KA-AB and control populations.
_
_
_
~
LOCAL CIRCUIT INTERACTIONS IN KA-LESIONED HIPPOCAMPUS / Nakajima et al. slices from controls (59.5 ? 5.4 mV vs. 60.1 & 5.0 mV). However, as reported previously (Franck and Schwartzkroin, 1985), cells from KA-AB slices had significantly higher input resistance than did control cells (37.2 10.8 M a vs. 26.7 2 7.6 M i l , P < .01, t-test). Cellular responsiveness to stratum radiatum stimulation in control animals consisted of an initial excitatory postsynaptic potential (EPSP), often triggering a single action potential, followed by an inhibitory postsynaptic potential (IPSP). Alveus stimulation elicited an antidromic spike, followed by an IPSP. In contrast, in the majority of cells (69%, 46 of 67) recorded from KA-AB slices, stimulation at either site failed
*
B
I l l I
71
to elicit an IPSP. These cells responded to stimulation with multiple action potentials. It is of interest that in the remaining 21 cells recorded in KA-AB slices, stimulation at slow rates elicited only an EPSP or EPSP with single action potential. However, in all these cells, burst discharge was easily elicited when the stimulation rate was increased (from 0.3 to 3.0 Hz). Thus, even this latter cell population appeared to be marginally “epileptogenic”. In examining the synaptic connectivity between pairs of pyramidal cells, we could find neither excitatory nor inhibitory interactions in KA-AB slices. Figure 3 shows an example of the recordings from a pyramidal cell pair. Each cell,
/
L
L
P1
P2 I
20 mV
2.0 nA 100 m8ec Fig. 3 . (A) Diagrammatic representation of the tested circuitry between simultaneously impaled pairs of pyramidal cells (P). (B) Typical result of a pyramidal cell-pyramidal cell interaction test. Cell PI was first injected with depolarizing current (current monitor above recording trace), and the effect of the evoked spike train monitored in cell P2 (left). The current injection electrode was then switched, so that P2 was activated and PI monitored (right). The small spike-like potentials (*) in the recording traces represent artifactual capacitative coupling in the postsynaptic cell. P1 RMP = - 58 mV; P2 RMP = -62 mV. These cells were recorded from a KA-AB slice, but are typical, too, of pyramidal cell-pyramidal cell interactions in control slices; no interactions were observed.
72
HIPPOCAMPUS VOL. 1, NO. 1, JANUARY 1991
-r
A
C
I D
B
P-
I'
4 P
L
I
120 m"
50 msec
100 msec
2 0 mV
2.0 nA
LOCAL CIRCUIT INTERACTIONS IN KA-LESIONED HIPPOCAMPUS / Nakaiima et al.
in turn. was depolarized to produce a train of action potentials. Neither depolarization nor hyperpolarization was evoked in the “postsynaptic” cell. A similar result was obtained in paired pyramidal cell recordings from normal slices. Finally. in none of the cells was there any indication of electrotonic coupling. as would be indicated by electrotonic transfer of injected hyperpolarizing current. Interconnections between basket ceWpyramida1 cell pairs, and cell properties Although basket cells are relatively rare in hippocampus, they are easily recognizable by their electrophysiological features. and can be impaled by electrode penetrations at the pyramidaleioriens border (Schwartzkroin and Mathers. 1978). In slices from 5 KA-AB rats, we obtained recordings from 9 pyramidal cellibasket cell pairs (Table 1) (Fig. 4A). representing 9 pyramidal cells and 6 basket cells. In this population of pyramidal cells, all cells responded to afferent stimulation with bursts of action potentials and no IPSP: the response was similar whether the stimulating electrode was located in stratum radiatum o r in the alveus. Figure 4B shows stimulusinduced bursting in a pyramidal cell (upper trace) and interneuron (lower trace) pair. Notable in this pair is the longerlasting burst evoked in the interneuron as compared to the pyramidal cell. In neither cell was there any sign of an IPSP. Although 4 of the 6 interneurons responded to stimulation with such a burst of action potentials in the KA-AB tissue. 2 of the 6 basket cells failed to discharge action potentials in response to radiatum stimulation; only a small EPSP was apparent in these cells, consistent with the weak afferent efficacy reported in a previous study (Franck et al., 1988). In slices with normal physiology, we obtained recordings from 20 pyramidal cell-basket cell pairs, representing 20 pyramidal and 8 basket cells. In contrast to the results from KA-AB slices. all 8 of these basket cells responded to afferent stimulation with graded EPSPs. which could be maximized to trigger action potentials. All 20 of the pyramidal cells recorded in normal tissue responded to stimulation with an EPSP and/or single action potential followed by an IPSP: no burst discharges were seen. The limited pyramidal cell populations analyzed in these pyramidal celliinterneuron pairings were similar to the larger
.
73
population analyzed in the pyramidal cellipyramidal cell pairs with respect to membrane potential and input resistance. That is, pyramidal cells from KA-AB slices had a higher Ri,, than cells from control slices (Table I ) . Unlike the pyramidal cells, both the resting membrane potential of the basket cells recorded in KA-AB slices and their input resistance did not differ from cells in control slices (Ri, = 41.7 ? 7.2 MIL in KA-AB versus 41.7 & 10.1 MR in control, Table I ) . Input resistance of the basket cells was somewhat higher than that of pyramidal cells recorded in the same tissues. Synaptic interactions between pyramidal cells and basket cells were studied in 20 neuron pairs in control tissue and 9 neuron pairs in KA-AB tissue. Significantly more pyramidalbasket cell interactions were observed in electrophysiologically normal tissue than in KA-AB tissue (Fisher Exact Probability = ,031). To determine whether one o r both synapses in the recurrent inhibitory circuit were affected by CA3 lesions. the data were also analyzed according to the directionality of the interaction. In 8 of the 20 pairs examined in control slices. current-evoked action potential discharge in the pyramidal cell elicited EPSPs and/or action potentials in the paired basket cell. Figure 4C shows an example of such an interaction. Note that the relationship between pyramidal cell discharge and activation in the interneuron was episodic; the interneuron did not follow in a I-to-I manner, but rather fired discharges associated with only 2 of the 4 spikes in the pyramidal cell. Of the 9 pyramidal cell-interneuron pairs recorded in KA-AB tissue, only I pair showed excitatory drive from the pyramidal cell to the interneuron (Figure 4D). I n this case. too. the efficacy of the pyramidal cell drive was relatively weak; the interneuron was activated only when a pair of pyramidal cell spikes occurred at short intervals (at the very beginning of the train in Figure 4D). In 6 of 20 pairs in normal slices, activation of the interneurons with depolarizing current led to a slow hyperpolarizing potential in the paired pyramidal cell (Fig. 5A,B). Figure 5B shows an example of such an interaction. The hyperpolarization typically was smooth in waveform and small in amplitude, and outlasted the duration of the presynaptic spike train. Amplitude of the hyperpolarization could be influenced by altering the resting potential of the postsynaptic pyramidal cell. as would be expected for an inhibitory synaptic poten-
Fig. 4. (A) Diagrammatic representation of paired recording between a pyramidal cell ( P ) and basket cell interneuron (I) in which the P to I synapse was tested. (B) Simultaneous recordings from a pyramidal cell (top) and an interneuron (bottom). showing their responses to stratum radiatum stimulation in a KA-AB slice. In the pyramidal cell (RMP = -58 mV), the response consisted of 2 action potentials, but no sign of a hyperpolarizing IPSP. The same stimulus evoked a somewhat longer discharge in the interneuron; the EPSP component of the response is somewhat exaggerated due to injection of .2 nA hyperpolarizing current (from a RMP of - 64 mV). (C) Pyramidal cell-interneuron pair recorded in an electrophysiologically normal slice. The pyramidal cell was stimulated with depolarizing current to evoke 4 action potentials; the first and the last pyramidal cell action potential triggered action potentials in the interneuron, whereas the middle two pyramidal cell spikes had no effect on the interneuron. Interneuron stimulation in this pair had no effect on the pyramidal cell (not shown). Pyramidal cell RMP = -56 mV; interneuron RMP = -58 mV. (D) A similar pyramidal cell-interneuron pair recorded in a KA-AB slice. The pyramidal cell was depolarized to produce a train of action potentials. As in C, discharge in the pyramidal cell could evoke action potentials in the interneuron. However, the drive appeared quite weak: only when 2 pyramidal cell action potentials occurred with a short interspike interval (at the very beginning of the trace) was there a linked response in the interneuron. Interneuron stimulation in this pair produced no effect on the pyramidal cell (not shown). This example shows the only case in which there was clear synaptic coupling between a pyramidal cell and interneuron in KA-AB slices. Pyramidal cell RMP = -62 mV; interneuron RMP = -60 mV.
74 HIPPOCAMPUS VOL.
1, NO. 1, JANUARY 1991
A
C
I
I
PI 20 mV (1) 40 mV (p)
100 msec
2.0 nA
Fig. 5. (A) Diagrammatic representation of paired recording between interneuron and pyramidal cell in which the I to P
synapse was tested. (B) Pyramidal cell and interneuron recorded simultaneously in a control slice. Activation of the interneuron with injected depolarizing current led to a smooth, slow hyperpolarizing response in the pyramidal cell. l h e effect was barely noticeable when the pyramidal cell was at resting membrane potential ( - 60 mV), but was clear when the membrane was depolarized (-2 nA), as shown here. In this pair, pyramidal cell activation evoked no response in the interneuron (not shown). Interneuron RMP = -58 mV. (C) Pyramidal cell-interneuron pair recorded from a KA-AB slice. Neither spontaneous (*) nor evoked interneuron discharge had any effect on the pyramidal cell. Pyramidal cell RMP = -66 mV; interneuron RMP = -56 mV.
tial. No such hyperpolarizations were elicited in the 9 pyramidal cells recorded from KA-AB tissue when stimulating the presynaptic interneuron (Fig. 5C). A small number of pairs (3), all from normal tissue, showed reciprocal interactions. In these cases, pyramidal cell activation produced excitatory events in the idterneuron, and
interneuron activation produced inhibitory events in the pyramidal cell. While the total number of interacting neuron pairs was significantly greater in normal tissue than in KA-AB tissue, statistical comparisons of the directionality of the interactions were not significant (Fisher Exact Test pyramid to interneu-
LOCAL CIRCUIT INTERACTIONS IN KA-LESIONED HIPPOCAMPUS / Nakajima et a[. ron P = .13; interneuron to pyramid p = .08). These data, however, are suggestive that both synapses in the CAI recurrent pyramidal-to-basket cell circuit are relatively dysfunctional in the CA3 lesion-induced hyperexcitable hippocampus.
Multiple population responses evoked by distant stimulation In the course of assessing each slice for its epileptogenicity, stimulating electrodes were moved throughout the CAI stratum radiatum to evoke population responses. We found that the location of the stimulating electrode relative to the recording electrode was an important determinant of whether multiple population spikes were evoked in KA-AB slices. In 5 experiments, this relationship was explored systematically. A stimulation site close to the recording electrode (
Local Circuit Synaptic Interactions Between CAl Pyramidal Cells and Interneurons in the Kainate-Lesioned Hyperexcitable Hippocampus S. Nakajima,*$ J. E. Franck,* D. Bilkey,§ and P. A. Schwartzkroin,*T D e p a r t m e n t s o f "Neurological S u r g e r y , and +Physiology and Biophysics, University of W a s h i n g t o n , Seattle, WA 98195 U.S.A., $ D e p a r t m e n t o f N e u r o p s y c h i a t r y , University o f K y o t o Prefectural University of Medicine, K y o t o , J a p a n , and $ D e p a r t m e n t of Psychology, University of O t a g o , D u n e d i n ,
New Zealand
ABSTRACT Following kainate (KA)-induced lesions of subfield CA3-a lesion relevant to human temporal lobe epilepsy-remaining pyramidal cells in CAI display synchronous hyperexcitability associated with a loss of synaptic inhibition. Despite this loss, inhibitory interneurons in CAI remain viable, and the density and function of GABAergic receptors on the CAI pyramidal cells are maintained at approximately normal levels. To further evaluate inhibition in this system, the authors examined interactions between pyramidal cells and inhibitory interneurons in paired intracellular recordings. Recordings were carried out in rat hippocampal slices 2-4 weeks following bilateral intraventricular KA injections. The frequency of synaptic interactions between CAI basket cells and pyramidal cells was lower in hyperexcitable slices than in controls: both synapses in the recurrent inhibitory circuit appeared to be involved. No recurrent excitatory interactions were seen between pyramidal cell pairs in lesioned or normal slices. The weakened interconnections between pyramidal cells and interneurons are consistent with the decreased inhibition previously found in this model. Unexpectedly, strong stimulation, which may directly activate local inhibitory circuitry, was effective in reducing hyperexcitability in KA-lesioned slices. These data suggest that development of recurrent excitatory connections among CAI hippocampal pyramidal cells contribute little to tissue excitability. and support the hypothesis that a functional uncoupling between inhibitory interneurons and CAI pyramidal cells is responsible for the seizure-like activity typical of KA-lesioned hippocampus. The data are also consistent with the hypothesis that in the KA model, the structural circuitry needed for inhibition in CAI is maintained, and can be functionally activated by appropriate stimuli. Key words: hippocampus, kainic acid, epilepsy, paired intracellular recording, inhibitory interneuron
Kainic acid (KA) selectively destroys hippocampal CA3 pyramidal cells when injected into the ventricles in low doses (Nadler et al., 1978a; Lancaster and Wheal, 1982). This KAlesioned hippocampus bears structural similarities to the pathologic findings in human temporal lobe epilepsy (Margerison and Corsellis, 1966; Corsellis and Meldrum, 1976). Significantly, neurons that remain viable following this lesion-CAl pyramidal and dentate gyrus granule cells-develop long-term hyperexcitability and synchrony, and may ~
~
Correspondence and reprint requests to P. A . Schwartzkroin, Department of Neurological Surgery, RI-20, University of Washington, Seattle, WA 98195 U.S.A.
provide a model for the study of hippocampal epileptogenesis, which accompanies cell loss (Franck and Schwartzkroin, 1985; Ashwood et al., 1986a). Several electrophysiological studies have demonstrated that the primary feature of these hyperexcitable CAI pyramidal cells is the presence of both synaptically elicited and spontaneous synchronous bursting and a reduction o r loss of GABA-mediated inhibitory postsynaptic potentials (IPSPs) (Ashwood et al., 1983, 1986a; Franck and Schwartzkroin, 1985, Franck et al., 1988). Curiously, however, while functional inhibition is compromised, the inhibitory cells and their terminal distribution in area CAI survive this lesion, much as they do in human epileptic hippocampus (Babb et al.,
67
68
HIPPOCAMPUS VOL. 1, NO. 1, JANUARY 1991
1989): they are electrophysiologically viable, synthesize GABA, and nearby pyramidal cells maintain their responsivity to exogenous GABA application (Davies et al., 1985: Ashwood et al., 1986b: Franck et al., 1988). In addition to a loss of functional inhibition, the CAI region of these lesioned hippocampi may develop significant recurrent excitation. Detailed histologic studies show that the excitatory synaptic density in stratum radiatum and stratum oriens of area CAI is reduced by 80% 7 days after KA injection, mainly due to a loss of Schaffer collaterals and commissural fibers arising from CA3 pyramidal cells. The synaptic density returns to a near-normal level by 40-SO days post-lesion (Nadler et al., 1978b, 1980a,b). The source of these reinnervatingfibers is still unknown (Nadler et al., 1980b). However. in view of the abnormal hyperexcitability and synchrony, these newly formed synapses might originate from the axon collaterals of nearby CAI pyramidal cells (Nadler et al., 1980a; Franck et al., 1988). These observations suggest that the hyperexcitability and synchrony that develop in the CAI region of the chronic KA model of epileptogenesis may involve alterations in both the manner in which pyramidal cells interact with interneurons and the way in which they interact with each other. We studied simultaneously impaled pyramidal cell pairs and pyramidal cell-interneuron pairs in the CAI region of KA-lesioned hyperexcitable hippocampi to determine the following: ( I ) Does the loss of functional inhibition in this tissue result from a loss of one component of the bidirectional interaction between pyramidal cells and interneurons and (2) Is there an increase in recurrent excitatory collateralization in this hyperexcitable CA 1 region? The data presented here suggest that the loss of inhibition is correlated with decreased efficacy of both inhibitory synapses from interneurons onto pyramidal cells and excitatory synapses from pyramidal cells onto interneurons. Furthermore, the hyperexcitability and synchrony that develops in this tissue does not depend on aberrant recurrent excitation in CAI.
extra.
FW \I
Jio
mV
10 msec
Pyramids Cell
lnhbilory
Fig. 1 , Diagrammatic representation of the experimental paradigm. An extracellular recording electrode (extra) was positioned in the CAI pyramidal cell layer, and the stimulating electrode was placed into CAI stratum radiatum (stim). Inset shows a typical abnormal multiple population response to stratum radiatum stimulation (.2 mA, .3 Hz) in a KA-A13 slice. After determining whether the electrophysiologic response from such stimulation consisted of a normal (single population) response or an abnormal (multiple population) response, intracellular microelectrodes (intra) were introduced into the pyramidal cell region. close to the site at which the field potential response had been monitored. DG, dentate granule cells.
MATERIALS A N D METHODS Male Long Evans rats (50-60 days old) were anesthetized with a ketarnineixylazine mixture ( 3 : I , 0.15 ml/lOO g, i.m.) and KA was slowly administered into each ventricle (.6 pg in I p1 of salineihemisphere, pH 7.2-7.8) from a 10 pl Hamilton syringe positioned stereotaxically according to the atlas of Paxinos and Watson (1986); coordinates for injections were 0.6 mm posterior to bregma, 2.0 mm lateral to the midline, and 3.5 mm deep to the dura. Control animals underwent the same procedure, but received injections of buffered saline. The animals were sutured, treated with antibiotics, and allowed to recover from anesthesia for 2-4 weeks prior to electrophysiological studies. Rats were decapitated, and a section of brain containing the hippocampus was quickly removed. Four to eight transverse slices (450 p m thick) were cut (Vibroslicer, Frederick. Haer) from each dorsal hippocampus. Slices were placed in an interface-type recording chamber, warmed to 35°C. humidified with 95% 0,/50/0 C02,and perfused with modified Krebs-Ringer buffer (126 mM NaCI, 5 mM KCI, 2 mM CaCI,, 1.2 mM MgCI,, 1.2 mM NaH,P04, 26 mM NaHCO?, and 10 mM dextrose) (pH 7.4, flow rate I.O
mlimin). All physiologic recordings were performed by the same group of investigators. At the start of each electrophysiological recording session. extracellular field potential recordings were made with electrodes containing 2 M NaCl(1-5 M i l ) to determine the degree of abnormal excitability present in the CA1 region of each slice. With the recording electrode placed in the pyramidal cell layer, a bipolar stimulating electrode (insulated tungsten wire; .5 mm tip separation) was placed in stratum radiatum (or the alveus) to orthodrornically (or antidromically) activate CAI pyramidal cells (.05 msec duration; .01-.3 mA; .3-3.0 Hzj. In normal slices, this stimulation elicits a single population spike; the presence of 2 o r more population spikes is a reliable indicator of bursting activity at the single cell level (Wheal et al., 1984). Slices were categorized a s abnormal if stimulation elicited multiple ( 2 2 ) population spikes (Fig. 1). In each slice, paired intracellular recordings were made in the CAI region. Glass microelectrodes ( 4 M potassium acetate, 40-1 10 M i l ) were inserted into o r near stratum pyramidale. In those slices with abnormal burst discharge in the field, the intracellular electrodes were introduced in the re-
LOCAL CIRCUIT INTERACTIONS IN KA-LESIONED HIPPOCAMPUS / Nakajima et al. gion from which the field burst had been recorded (Fig. I). Two intracellular electrodes were placed within about 200 p n of each other, and the activity of impaled cells was manipulated to determine whether the cells were synaptically connected (see below). The impaled neuron was classified as either a pyramidal cell or interneuron on the basis of its physiological properties (Schwartzkroin, 1975; Schwartzkroin and Mathers, 1978). In attempts to penetrate interneurons, the microelectrode was generally introduced at the border of the stratum pyramidale and stratum oriens, where the “basket cell” interneuron type is most often located (Schwartzkroin and Mathers. 1978; Ashwood et al., 1984). Synaptic interactions between pairs of impaled neurons were determined by the following process. One neuron was maintained at resting potential, while the other was depolarized by injecting current pulses (100-200 msec duration, .21.0 nA amplitude, .3 Hz) that evoked spike trains. The potential “postsynaptic” cell was monitored for the occurrence of depolarizing or hyperpolarizing potentials that were timelocked to action potential discharge in the stimulated neuron. In order to uncover small synaptic potentials, the membrane potential of the “postsynaptic” cell was manipulated in depolarizing and hyperpolarizing directions (injection of DC current of up to 1 nA). This procedure was then repeated,
69
reversing the stimulated and “postsynaptic” cell to test connections in the opposite direction. In all pairs, hylperpolarizing current pulses (100 msec duration pulses, .5 nA) were also introduced in order to determine the input resistance of these neurons and to assess electrotonic communication between cells. In a small number of experiments, the effect of stimulus proximity on field potentials was evaluated. Abnormal, multiple population spikes were first obtained in response to stratum radiatum stimulation. The stimulating electrode was then moved closer to, and then more distant from, the irecording electrode. PopuIation spike responses to stimulation, for all locations, were monitored. As for the intracellular data, all field potential data were digitized and stored on VCR tape. At the end of each electrophysiological experiment, the monitored slices were fixed in 10% buffered formalin, and subsequently stained with a Nissl stain in order to determine the degree of cell damage in each slice.
RESULTS In the present study, intraventricular K A injections destroyed subfield CA3 in almost all experimental anirnals, leaving the CAI and dentate regions intact (Fig. 2). N o lesion was found in the hippocampus in 4% of the KA-injected animals
Fig. 2. Histology of a typical kainate lesion. Pyramidal cells in CA3 are gone. Nissl stain shows cell bodies of surviving CAI and dentate gyrus (DG) cell populations. Cresyl Violet x 38.
70
HIPPOCAMPUS VOL. 1, NO. 1, JANUARY 1991
(2 of 53); in both of these cases, only normal field potentials were evoked in hippocampal slices. Somewhat surprisingly, abnormal, multiple population responses were evoked in slices of only 45% of KA-injected animals (24 of 53) with verified lesions. Such responses were seen in at least 1 slice from each of these animals, but were generally not seen in every slice. In the remaining animals (27 of 53), none of the slices generated abnormal population spikes in response to radiatum or alveus stirnulation. The likelihood of abnormal electrophysiological responsiveness was not clearly associated with the degree of CA3 damage. Because a significant proportion of lesioned animals appeared normal electrophysiologically, we divided our slice population into 3 groups: ( I ) KA-induced structural lesion with abnormal electrophysiological excitability (KA-AB); (2) KA lesion with normal electrophysiological responsiveness
(Le., single population spike in response to stimulation; KANOR); and (3) non-KA-treated and electrophysiologically normal (NOR).
Single cell properties and local interactions between pyramidal cells In 17 slices from KA-AB rats, we obtained 39 pairs of intracellular recordings, which included 67 pyramidal cells. The cell properties of these neurons are compared with 34 pyramidal cells monitored in control tissue (Table 1). Since the properties of cells in unlesioned animals (NOR) was identical to that of cells in KA-lesioned, but electrophysiologically normal, slices (KA-NOR), the control group for statistical comparisons was composed of the KA-NOR and NOR groups. The mean resting membrane potential (RMP) of cells in slices from KA-AB animals was not different from that of cells in
Table 1. Comparison of Cell Properties and Local Circuit interactions in Kainate-lesioned and Normal Slices . _ _ _ ~ ~ ~ ~ _ _ _ _ _ ~ _ _ . _ _ ~ _ _ _ _ _ _ _ _ _ _ _
_
~
_
_
_
KA-NOR + NOR
-
~
-
~
~
~
_
KA-AB _
KA-NOR
____~.__
NOR
(Control)
013
0125
~
A
Interactions P+P Cell properties Pyramidal cells
0139
RMP (mV) R,, (MR)
0122
-59.5 t 5.4 37.2 t 10.8** ( n = 67)
-60.4 27.6
(n
* 4.9 s 7.5
=
30)
-58.0 2 5.2 20.4 2 4.4 (n = 4)
-60.1 26.7
* 5.0 * 7.6
(n = 34)
B
Interactions P+I I+P Total interacting pairs Cell properties Pyramidal cells
119 019 119*
217 317 517
6/13 3113 6113
8120 6/20 11120
RMP (mV) Rin (Mn)
-58.9 '' 4.4 36.6 5 7.8**
RMP Rin
-59.3 t 3.0 41.7 7.2 (n = 6)
-58.3 t 3.6 27.9 s 6.1 (n = 7) -55.0 t 3.5 41.3 s 8.7 (n = 4)
-58.4 Z 5.0 24.6 2 7.6 (n = 13) -59.2 2 2.8 42.1 5 12.8 (n = 4)
-58.4 ? 4.5 25.8 t 7.1 (n = 20) -57.1 5 3.7 41.7 5 10.1 (n = 8)
(n
lnterneurons
(mV) (Ma)
_
_
=
_
9)
_
~
.
~
_
_
_
~
* P = 0.03, Fisher Exact Probability relative to control. Three normal pairs had bidirectional interactions, and the totals count these pairs
only once. **P < 0.01, z-test relative to control
Table 1. (A) Pyramidal cell properties and pyramidal cell-pyramidal cell interactions in KA-treated, electrophysiologically abnormal slices (KA-AB) (column l ) , KA-treated, electrophysiologically normal slices (KA-NOR) (column 2), and normal slices (no KA treatment; NOR) (column 3). No differences in input resident and resting membrane potential were found between the KA-NOR and NOR slices, and thus the groups were combined to provide the control population (column 4). A comparison between KA-AB and control populations showed a significantly higher input resistance in pyramidal cells of the KA-AB tissue. In neither KA-AB nor control slices were any synaptic interactions between pyramidal cells (P -+ P) uncovered (number of pairs showing synaptic interactioninumber of pairs surveyed). (B) Interactions and cell properties of pyramidal cell-interneuron (basket cell) pairs in KA-AB and control populations. Input resistance for this population of pyramidal cells was higher than Ri, in controls. However, interneuron properties did not differ between electrophysiologically abnormal and normal slices. In control slices, 8 of 20 pairs showed an excitatory drive from the pyramidal cell onto the interneuron; in contrast, only 1 of 9 such pairs in the KA-AB slices showed such an interaction. Six of 20 pairs showed inhibition from the interneuron onto the pyramidal cell in control slices, whereas no pairs showed such interaction in the KA-AB slices. Resting membrane potential and input resistance measurements are given as means -C standard deviation. Statistical comparisons were made between the KA-AB and control populations.
_
_
_
~
LOCAL CIRCUIT INTERACTIONS IN KA-LESIONED HIPPOCAMPUS / Nakajima et al. slices from controls (59.5 ? 5.4 mV vs. 60.1 & 5.0 mV). However, as reported previously (Franck and Schwartzkroin, 1985), cells from KA-AB slices had significantly higher input resistance than did control cells (37.2 10.8 M a vs. 26.7 2 7.6 M i l , P < .01, t-test). Cellular responsiveness to stratum radiatum stimulation in control animals consisted of an initial excitatory postsynaptic potential (EPSP), often triggering a single action potential, followed by an inhibitory postsynaptic potential (IPSP). Alveus stimulation elicited an antidromic spike, followed by an IPSP. In contrast, in the majority of cells (69%, 46 of 67) recorded from KA-AB slices, stimulation at either site failed
*
B
I l l I
71
to elicit an IPSP. These cells responded to stimulation with multiple action potentials. It is of interest that in the remaining 21 cells recorded in KA-AB slices, stimulation at slow rates elicited only an EPSP or EPSP with single action potential. However, in all these cells, burst discharge was easily elicited when the stimulation rate was increased (from 0.3 to 3.0 Hz). Thus, even this latter cell population appeared to be marginally “epileptogenic”. In examining the synaptic connectivity between pairs of pyramidal cells, we could find neither excitatory nor inhibitory interactions in KA-AB slices. Figure 3 shows an example of the recordings from a pyramidal cell pair. Each cell,
/
L
L
P1
P2 I
20 mV
2.0 nA 100 m8ec Fig. 3 . (A) Diagrammatic representation of the tested circuitry between simultaneously impaled pairs of pyramidal cells (P). (B) Typical result of a pyramidal cell-pyramidal cell interaction test. Cell PI was first injected with depolarizing current (current monitor above recording trace), and the effect of the evoked spike train monitored in cell P2 (left). The current injection electrode was then switched, so that P2 was activated and PI monitored (right). The small spike-like potentials (*) in the recording traces represent artifactual capacitative coupling in the postsynaptic cell. P1 RMP = - 58 mV; P2 RMP = -62 mV. These cells were recorded from a KA-AB slice, but are typical, too, of pyramidal cell-pyramidal cell interactions in control slices; no interactions were observed.
72
HIPPOCAMPUS VOL. 1, NO. 1, JANUARY 1991
-r
A
C
I D
B
P-
I'
4 P
L
I
120 m"
50 msec
100 msec
2 0 mV
2.0 nA
LOCAL CIRCUIT INTERACTIONS IN KA-LESIONED HIPPOCAMPUS / Nakaiima et al.
in turn. was depolarized to produce a train of action potentials. Neither depolarization nor hyperpolarization was evoked in the “postsynaptic” cell. A similar result was obtained in paired pyramidal cell recordings from normal slices. Finally. in none of the cells was there any indication of electrotonic coupling. as would be indicated by electrotonic transfer of injected hyperpolarizing current. Interconnections between basket ceWpyramida1 cell pairs, and cell properties Although basket cells are relatively rare in hippocampus, they are easily recognizable by their electrophysiological features. and can be impaled by electrode penetrations at the pyramidaleioriens border (Schwartzkroin and Mathers. 1978). In slices from 5 KA-AB rats, we obtained recordings from 9 pyramidal cellibasket cell pairs (Table 1) (Fig. 4A). representing 9 pyramidal cells and 6 basket cells. In this population of pyramidal cells, all cells responded to afferent stimulation with bursts of action potentials and no IPSP: the response was similar whether the stimulating electrode was located in stratum radiatum o r in the alveus. Figure 4B shows stimulusinduced bursting in a pyramidal cell (upper trace) and interneuron (lower trace) pair. Notable in this pair is the longerlasting burst evoked in the interneuron as compared to the pyramidal cell. In neither cell was there any sign of an IPSP. Although 4 of the 6 interneurons responded to stimulation with such a burst of action potentials in the KA-AB tissue. 2 of the 6 basket cells failed to discharge action potentials in response to radiatum stimulation; only a small EPSP was apparent in these cells, consistent with the weak afferent efficacy reported in a previous study (Franck et al., 1988). In slices with normal physiology, we obtained recordings from 20 pyramidal cell-basket cell pairs, representing 20 pyramidal and 8 basket cells. In contrast to the results from KA-AB slices. all 8 of these basket cells responded to afferent stimulation with graded EPSPs. which could be maximized to trigger action potentials. All 20 of the pyramidal cells recorded in normal tissue responded to stimulation with an EPSP and/or single action potential followed by an IPSP: no burst discharges were seen. The limited pyramidal cell populations analyzed in these pyramidal celliinterneuron pairings were similar to the larger
.
73
population analyzed in the pyramidal cellipyramidal cell pairs with respect to membrane potential and input resistance. That is, pyramidal cells from KA-AB slices had a higher Ri,, than cells from control slices (Table I ) . Unlike the pyramidal cells, both the resting membrane potential of the basket cells recorded in KA-AB slices and their input resistance did not differ from cells in control slices (Ri, = 41.7 ? 7.2 MIL in KA-AB versus 41.7 & 10.1 MR in control, Table I ) . Input resistance of the basket cells was somewhat higher than that of pyramidal cells recorded in the same tissues. Synaptic interactions between pyramidal cells and basket cells were studied in 20 neuron pairs in control tissue and 9 neuron pairs in KA-AB tissue. Significantly more pyramidalbasket cell interactions were observed in electrophysiologically normal tissue than in KA-AB tissue (Fisher Exact Probability = ,031). To determine whether one o r both synapses in the recurrent inhibitory circuit were affected by CA3 lesions. the data were also analyzed according to the directionality of the interaction. In 8 of the 20 pairs examined in control slices. current-evoked action potential discharge in the pyramidal cell elicited EPSPs and/or action potentials in the paired basket cell. Figure 4C shows an example of such an interaction. Note that the relationship between pyramidal cell discharge and activation in the interneuron was episodic; the interneuron did not follow in a I-to-I manner, but rather fired discharges associated with only 2 of the 4 spikes in the pyramidal cell. Of the 9 pyramidal cell-interneuron pairs recorded in KA-AB tissue, only I pair showed excitatory drive from the pyramidal cell to the interneuron (Figure 4D). I n this case. too. the efficacy of the pyramidal cell drive was relatively weak; the interneuron was activated only when a pair of pyramidal cell spikes occurred at short intervals (at the very beginning of the train in Figure 4D). In 6 of 20 pairs in normal slices, activation of the interneurons with depolarizing current led to a slow hyperpolarizing potential in the paired pyramidal cell (Fig. 5A,B). Figure 5B shows an example of such an interaction. The hyperpolarization typically was smooth in waveform and small in amplitude, and outlasted the duration of the presynaptic spike train. Amplitude of the hyperpolarization could be influenced by altering the resting potential of the postsynaptic pyramidal cell. as would be expected for an inhibitory synaptic poten-
Fig. 4. (A) Diagrammatic representation of paired recording between a pyramidal cell ( P ) and basket cell interneuron (I) in which the P to I synapse was tested. (B) Simultaneous recordings from a pyramidal cell (top) and an interneuron (bottom). showing their responses to stratum radiatum stimulation in a KA-AB slice. In the pyramidal cell (RMP = -58 mV), the response consisted of 2 action potentials, but no sign of a hyperpolarizing IPSP. The same stimulus evoked a somewhat longer discharge in the interneuron; the EPSP component of the response is somewhat exaggerated due to injection of .2 nA hyperpolarizing current (from a RMP of - 64 mV). (C) Pyramidal cell-interneuron pair recorded in an electrophysiologically normal slice. The pyramidal cell was stimulated with depolarizing current to evoke 4 action potentials; the first and the last pyramidal cell action potential triggered action potentials in the interneuron, whereas the middle two pyramidal cell spikes had no effect on the interneuron. Interneuron stimulation in this pair had no effect on the pyramidal cell (not shown). Pyramidal cell RMP = -56 mV; interneuron RMP = -58 mV. (D) A similar pyramidal cell-interneuron pair recorded in a KA-AB slice. The pyramidal cell was depolarized to produce a train of action potentials. As in C, discharge in the pyramidal cell could evoke action potentials in the interneuron. However, the drive appeared quite weak: only when 2 pyramidal cell action potentials occurred with a short interspike interval (at the very beginning of the trace) was there a linked response in the interneuron. Interneuron stimulation in this pair produced no effect on the pyramidal cell (not shown). This example shows the only case in which there was clear synaptic coupling between a pyramidal cell and interneuron in KA-AB slices. Pyramidal cell RMP = -62 mV; interneuron RMP = -60 mV.
74 HIPPOCAMPUS VOL.
1, NO. 1, JANUARY 1991
A
C
I
I
PI 20 mV (1) 40 mV (p)
100 msec
2.0 nA
Fig. 5. (A) Diagrammatic representation of paired recording between interneuron and pyramidal cell in which the I to P
synapse was tested. (B) Pyramidal cell and interneuron recorded simultaneously in a control slice. Activation of the interneuron with injected depolarizing current led to a smooth, slow hyperpolarizing response in the pyramidal cell. l h e effect was barely noticeable when the pyramidal cell was at resting membrane potential ( - 60 mV), but was clear when the membrane was depolarized (-2 nA), as shown here. In this pair, pyramidal cell activation evoked no response in the interneuron (not shown). Interneuron RMP = -58 mV. (C) Pyramidal cell-interneuron pair recorded from a KA-AB slice. Neither spontaneous (*) nor evoked interneuron discharge had any effect on the pyramidal cell. Pyramidal cell RMP = -66 mV; interneuron RMP = -56 mV.
tial. No such hyperpolarizations were elicited in the 9 pyramidal cells recorded from KA-AB tissue when stimulating the presynaptic interneuron (Fig. 5C). A small number of pairs (3), all from normal tissue, showed reciprocal interactions. In these cases, pyramidal cell activation produced excitatory events in the idterneuron, and
interneuron activation produced inhibitory events in the pyramidal cell. While the total number of interacting neuron pairs was significantly greater in normal tissue than in KA-AB tissue, statistical comparisons of the directionality of the interactions were not significant (Fisher Exact Test pyramid to interneu-
LOCAL CIRCUIT INTERACTIONS IN KA-LESIONED HIPPOCAMPUS / Nakajima et a[. ron P = .13; interneuron to pyramid p = .08). These data, however, are suggestive that both synapses in the CAI recurrent pyramidal-to-basket cell circuit are relatively dysfunctional in the CA3 lesion-induced hyperexcitable hippocampus.
Multiple population responses evoked by distant stimulation In the course of assessing each slice for its epileptogenicity, stimulating electrodes were moved throughout the CAI stratum radiatum to evoke population responses. We found that the location of the stimulating electrode relative to the recording electrode was an important determinant of whether multiple population spikes were evoked in KA-AB slices. In 5 experiments, this relationship was explored systematically. A stimulation site close to the recording electrode (
