Oscillation and synchronization of neural activity is important in normal brain function but is also relevant to epileptogenesis. One of the most frequent forms of epilepsy originates in temporal lobe circuitry of which the entorhinal cortex (EC) is crucial. Because muscarinic receptor activation promotes oscillatory dynamics in EC neurons, we investigated in a brain slice preparation the effects of carbachol (CCh) on oscillatory population activity in the EC. We found that CCh produced epileptiform activity in EC, which according to field profile and current source density analysis was usually driven by layer V. In addition, localized CCh application and surgical isolation experiments demonstrated that EC layer II, but not layer III, can also independently generate synchronous population activity.
Intracellular recordings from EC principal cells during epileptiform activity demonstrated large-amplitude, synaptically driven depolarizing events and bursts of action potentials synchronized to the field spikes. In layer II neurons, the depolarizing events had a multiphasic reversal potential that suggested concurrent glutamatergic and GABAergic synaptic input. Interestingly, although the epileptiform activity required activation of AMPA but not NMDA receptors, small-amplitude field spikes persisted during block of fast excitatory neurotransmission. These field spikes were correlated to large-amplitude IPSPs in layer II neurons, and both activities were abolished by GABAA-receptor antagonism. Thus, in response to muscarinic activation, pools of EC interneurons discharge synchronously by a mechanism not necessarily involving principal cell activation.
Given the differential projection pattern of EC layers V and II toward the neocortex and hippocampus, respectively, their robust epileptogenic character may be of major importance in temporal lobe epilepsy.
The majority of intractable epilepsy cases involve foci located in the mesial aspect of the temporal lobes (Lothman et al., 1991). To understand the basic mechanisms of temporal lobe epilepsy and to suggest more efficacious pharmacological and/or surgical treatments (Engel, 1987), it is important to define a precise locus for the generation of ictal activity in the temporal lobe. In general, most research approaches concerning this disorder have focused on the contribution of the hippocampal formation (Lothman et al., 1991; Schwartzkroin, 1994). Recent experimental and clinical studies, however, have suggested that the entorhinal cortex (EC) may play a similar or more important role in temporal lobe epilepsy (Dasheiff and McNarmara, 1982; Walther et al., 1986; Wilson et al., 1988; Rutecki et al., 1989; Deutch et al., 1991;Goldring et al., 1992; Jones et al., 1992; Paré et al., 1992;Stringer and Lothman, 1992; Spencer and Spencer, 1994; Bragin et al., 1997).
The EC is known to be a “gateway” for the bi-directional passage of information in the neocortical-hippocampal-neocortical circuit (Van Hoesen, 1982; Witter et al., 1989; Lopes da Silva et al., 1990). Via a cascade of cortico-cortical projections, the superficial layers of the EC (II and III) receive an extensive input from polymodal sensory cortices (Jones and Powell, 1970; Van Hoesen and Pandya, 1975; Amaral et al., 1983; Deacon et al., 1983; Room and Groenewegen, 1986; Insausti et al., 1987; Reep et al., 1987) that is then conveyed to the hippocampal formation via the perforant path (Steward and Scoville, 1976). In turn, the hippocampal formation projects back on the deep layers of the EC (V–VI) (Swanson and Cowan, 1977), which provide output paths that reciprocate the input channels (Swanson and Kohler, 1986; Insausti et al., 1997). In addition, the deep layers of the EC also project massively on the EC superficial layers (Kohler, 1986b;Dolorfo and Amaral, 1997), thereby closing an EC–hippocampal loop. Thus, by virtue of its extensive projection systems, the EC network may act powerfully in the generalization of temporal lobe seizures.
The EC is also known to receive a profuse cholinergic input from the basal forebrain that terminates primarily in layers II and V (Lewis and Shute, 1967; Mellgren and Srebro, 1973; Milner et al., 1983; Alonso and Köhler, 1984; Lysakowski et al., 1989; Gaykema et al., 1990), precisely those layers that gate the main hippocampal input and output. It is well known that the cholinergic system promotes cortical activation and the expression of normal population oscillatory dynamics. In the EC, in vivo electrophysiological studies have shown that the cholinergic theta rhythm is generated primarily by cells in layer II (Mitchell and Ranck, 1980; Alonso and Garcı́a-Austt, 1987a,b; Dickson et al., 1995). In addition,in vitro studies have also shown that muscarinic receptor activation promotes the development of intrinsic oscillations in EC layer II neurons (Klink and Alonso, 1997a,b).
On the other hand, some evidence indicates that altered activity of the cholinergic system is relevant to epileptogenesis. Pharmacological activation of cholinergic receptors may promote and maintain epileptiform discharge (for review, see Turski et al., 1989), and cholinergic neurotransmission seems necessary for the normal progression of kindling (Cain, 1989). Upregulation in the levels and activity of both acetylcholine and its metabolic enzymes has also been reported in human epilepsy (Pope et al., 1947; Tower and McEarchern, 1949; Tower and Elliot, 1952; Coutinho-Netto et al., 1981; Kish et al., 1988). In the EC, expression of muscarinic receptor protein is increased in kindling-induced epileptic animals (Beldhuis et al., 1993), and the EC also shows a lower threshold for carbachol (CCh)-induced electrographic seizure activity than the hippocampus (Dickson, 1994).
Given the above information, and that CCh is known to induce rhythmic population activity in the hippocampus in vitro (Konopacki et al., 1987; MacVicar and Tse, 1989; Traub et al., 1992; Huerta and Lisman, 1993; Bianchi and Wong, 1994), we considered it important to test whether moderate muscarinic receptor activation would also trigger oscillatory population activity in the EC slice and to explore the relation of this activity to an epileptogenic state. Our study demonstrates that CCh induces, in the EC slice, large-amplitude rhythmic field activity correlated to the development of paroxysmal depolarizations in all EC principal neurons. We further demonstrate that muscarinic-induced epileptiform activity can be generated independently by both superficial (layer II) and deep laminae (layers V–VI). Moreover, CCh was found to synchronize not only principal cells but also inhibitory interneurons, and this even in the absence of fast excitatory neurotransmission (Michelson and Wong, 1994).
MATERIALS AND METHODS
General. Brain slices were prepared from male Long–Evans rats (100–250 gm) using standard procedures (Alonso and Klink, 1993). Animals were decapitated, and the brain was rapidly removed, blocked, and placed in a cold (4–6°C) oxygenated Ringer’s solution containing (in mm): 124 NaCl, 5 KCl, 1.2 NaH2PO4, 2.4 CaCl2, 1.3–2.6 MgSO4, 26 NaHCO3, and 10 glucose, pH 7.4, by saturation with 95% O2/5% CO2. Horizontal slices from the retrohippocampal region were cut at a thickness of 400 μm using a Vibratome (Pelco, Redding, CA) and typically did not include the hippocampal formation, because our focus was on oscillatory mechanisms within the EC. After at least a 1 hr recovery period during which they were submerged at room temperature, individual slices were transferred to an interface chamber maintained at 34 ± 1° and superfused at a rate of 1–2 ml/min. The borders and the cellular layers of the EC were discerned with a dissecting microscope and a transilluminatory light source.
Drugs were mixed fresh weekly and stored and refrigerated in stock solutions at 100- to 1000-fold concentrations. They were mixed to their final concentrations in graduated cylinders filled with Ringer’s solution that were attached to the superfusion tubing. Carbamylcholine (CCh), atropine sulfate, bicuculline methiodide, and picrotoxin (PTX) were purchased from Sigma (St. Louis, MO), and 6-cyano-7-nitroxaline-2,3-dione (CNQX), anddl(−)-2-amino-5-phosphonopentanoic acid (AP-5) were purchased from Tocris Cookson (Langford, UK). Salts used in the making of Ringer’s solution were all purchased from BDH (Toronto, Ontario, Canada).
Recording and stimulation. Recording electrodes were pulled from micropipette glass (World Precision Instruments, Sarasota, FL) on a Sutter Instruments puller (Novato, CA). Extracellular field electrodes were filled with Ringer’s solution (tip resistance 5–10 MΩ), and intracellular electrodes were filled with 2.5 mpotassium acetate (tip resistance 60–120 MΩ).
Field and intracellular signals were amplified using both channels of an Axoclamp 2A amplifier (Axon Instruments, Foster City, CA). Field signals were further amplified by a CyberAmp amplifier (Axon Instruments) to a final gain of 1000 and filtered at a bandpass of 0.1–1.0 kHz (some signals were recorded with the DC component and later filtered). Signals were visualized on-line using two digital storage oscilloscopes at slow and fast screen sweep speeds, and stored on VHS tape (Neuro-Corder, New York, NY) for off-line analysis using a 386-based computer.
Bipolar stimulation electrodes were constructed from two strips of fine tungsten wire glued along their length and insulated except at their tips. The two poles were positioned straddling the angular bundle obliquely (tip separation ∼300 μm) above the medial portion of the EC. Electrical stimulation was delivered using a square wave generator (MNI technical services) coupled to a constant current/stimulus isolation unit (World Precision Instruments), and stimuli consisted of 1 msec square waves of amplitude 2–5 V. Field potentials effects of stimulation were measured in layer II.
Field and intracellular activities were digitized from tape at sampling rates ranging from 1 to 50 kHz and plotted using a software acquisition package (Acqui, SICMU, Geneva, Switzerland). The amplitude of field spikes was calculated as the difference from peak to trough of the field waveform. Duration of field spikes was measured from the onset of negativity to the crossing point at the same potential level. In addition, field activity filtered at a bandpass of 1–40 Hz and sampled at 1 kHz was subjected to Fast Fourier Transform using a software package (Origin, Microcal Software, Northampton, MA) for analysis in the frequency domain. The predominant frequency of the signal was taken from the fundamental peak of the resulting spectra.
Field profiles. A stationary (reference) electrode was positioned in layer V while a roving electrode was moved along a plane orthogonal to the laminations of the EC in 50 μm stepwise increments. The reference and roving electrodes were separated in the medial axis by <50 μm, and care was taken to ensure that the trajectory of the roving electrode remained in the same column along the depth of the profile. Up to 16 field events were sampled at each level, and typically 10 events were averaged for each using the reference waveform for temporal alignment. The averaged field potential values were then stored in two-dimensional (space vs time) matrices and plotted as a three-dimensional contour plot (Origin).
Current source density (CSD) analysis (Mitzdorf, 1985) was conducted by direct computation of the second spatial derivative of the extracellular voltage potentials. Briefly, the potential values corresponding to all spatial depths at each time point were fitted with a seventh order polynomial. The second derivative of this function was then computed. Derivative functions for each time point were then stored in matrix form and plotted in the same fashion as for the extracellular potentials.
Horizontal field profiles were constructed in much the same manner as laminar profiles. A stationary (reference) electrode was positioned in either layer II or V, and the roving electrode was moved in 50–100 μm stepwise increments along the same medial-to-lateral axis within the confines of the same cellular layer. When averaging was conducted, traces were averaged as described above.
Laminar CCh application profiles. Laminar localized application of CCh was conducted during simultaneous field recordings made in both layers II and V by fast (10–50 msec) pressure pulses (Picospritzer, General Valve, Fairfield, NJ) to the back of a patch pipette containing 10 mm CCh in a 1% solution of Luxol Fast Blue in Ringer’s solution. The dye was added to visualize the area of the slice inundated to ensure that the area of application was highly localized and lamina specific. Only those applications that were confined to the lamina of interest were subsequently analyzed. In some experiments, 1 μm atropine sulfate was added to the superfusing Ringer’s solution during the applications to ensure that the recorded field effects were specific to the muscarinic actions of CCh.
Laminar isolation experiments. Cellular layers were separated physically by scalpel cuts that were made in a clear petri dish filled with oxygenated Ringer’s solution after the removal of the slice from the recording chamber. Slices were quickly replaced in the chamber after successful completion of the cuts and reperfused with CCh. In some cases, a broken pipette was used to separate cellular layers without removal from the recording chamber. In additional experiments, the isolation cuts were made before incubation with CCh.
Intracellular recordings. Intracellular impalements were made in layers II, III, and V. Only neurons fulfilling the following criteria were reported in the following study: resting membrane potential negative to −55mV, input resistance >30 MΩ, and overshooting spikes of amplitude >60 mV. Superficial EC neurons were identified electrophysiologically according to criteria outlined previously (Alonso and Klink, 1993; Dickson et al., 1997). Briefly, layer II stellate cells (SCs) showed profound time-dependent inward rectification in the hyperpolarizing direction and prominent subthreshold membrane potential oscillations and cluster spiking on DC depolarization. Layer II pyramidal-like cells showed less robust time-dependent inward rectification with hyperpolarizing pulses and no rhythmic subthreshold oscillations. In addition to their location in distinctly observable lamina, layer III pyramids showed high input resistance (<70 MΩ) and no, or only minor, time-dependent inward rectification with hyperpolarizing current pulses, and went easily into tonic firing with DC depolarization (Dickson et al., 1997).
General characteristics of CCh-induced population activity
Bath application of moderate concentrations of CCh (10–30 μm) induced rhythmic large-amplitude (>500 μV) field activity in 55% (67 of 122) of EC slices tested without previous control for field responses to angular bundle stimulation. The success rate of rhythmic population activity induction was much higher (20 of 21; ∼95%), however, when CCh was applied to EC slices selected for responding to angular bundle electrical stimulation, with an evoked field potential in layer II of at least 500 μV.
The induction and maintenance of rhythmic field activity by CCh was dependent on muscarinic receptor activation because it was blocked by atropine (300 nm–1 μm; n = 9). As illustrated in Figure 1, the CCh-induced field activity had an epileptiform morphology. It was typically initiated by a long-duration ictiform event (LDE: 5–40 sec in duration) (Fig. 1 A,B) and was usually followed by a slow periodic occurrence of short-duration ictiform events (SDEs: 0.1–4,0 sec in duration) (Fig. 1 A,C). Occasionally, however, LDEs would show a similar periodicity, although at slower frequencies (<0.05 Hz) (see Figs. 6, 9). Aside from their differences in duration and interevent frequency, both LDEs and SDEs were similar in their initial waveform pattern (Fig. 1 D).
As recorded in layer V, LDEs consisted of a train of negative-going population spikes superimposed on a large-amplitude (0.5–1.4 mV) negative DC potential (Fig. 1 A). As in the cases illustrated in Figure 1 B,E, LDEs were consistently multiphasic, having at least two and sometimes three phases of population spike activity with different spectral peak frequencies, which resembled the tonic and clonic phases characteristic of an electrographic seizure with a “transition” phase occasionally interposed between. The features of LDEs are summarized in Table1. Essentially, the tonic phase was characterized by a train of high frequency population spikes (14.4 Hz), whereas the transitional and clonic phases showed progressively slower frequencies of field spikes (7.5 and 4.0 Hz, respectively) (Fig.1 E). Similarly, the duration of individual population spikes tended to increase as the event progressed. The average amplitude of population spikes, however, was relatively constant over the three phases when summed over all cases but could show systematic variations in individual cases, such as the increase-decrease-increase sequence shown in Figure 1 E.
SDEs consisted of large-amplitude negative-going field potentials typically associated with a repetitive sequence of field spikes (Fig.1 A,C), often resembling those seen during the LDEs (Fig. 1 D). In fact, the average frequency of these field spikes (7.7 Hz) (Table 2) was virtually identical to the average population spike frequency summed over all phases of the LDEs (7.9 Hz) (Table 1). SDEs typically repeated with a clock-like rhythmicity (Fig. 1 A) at a frequency that ranged from 0.05 to 0.6 Hz for different cases. The quantified features of SDEs are summarized in Table 2.
Field profile and CSD analysis
In an attempt to clarify which EC layer acts as the generator of CCh-induced epileptiform activity, we examined the laminar distribution of field events to construct laminar field potential profiles and perform CSD analysis for the localization of the main sinks and sources of current (n = 6). A typical case is illustrated in Figure 2. As shown in Figure2 A, a stationary electrode (red) was positioned in layer V while a moving electrode (blue) was advanced in 50 μm steps from the pial surface (layer I) to the angular bundle in a direction normal to the cortical layers. Figure2 B illustrates the laminar field potential profile corresponding to the initial 60 msec of averaged SDEs recorded at distinct locations (indicated to the left). The corresponding reference recordings used to construct the profile have been superimposed at the top (red). It can be observed that a robust field negativity (corresponding to a population spike) appeared first in layers V–VI and then successively at later times in layer III and then in layer II. Using the latency values corresponding to the peak negativities in the respective lamina, the speed of propagation was calculated to be 110 ± 14 mm/sec. Although appearing first in layers V–VI, the field negativity in layer III showed both a greater amplitude and longer duration. The laminar field profile in Figure 2 B has been represented as a contour plot in the right panel of Figure 2 C. Field negativity has been coded in the hot (red) end of the spectrum and was initiated in the deep layers and propagated superficially to layer III and finally to layer II.
The left panel of Figure 2 C represents the contour plot for the CSD analysis of the field potential profile in B andC. Current sinks have been coded in the hot (red) end of the spectrum. Note that a distinct short-lasting (2–3 msec) current sink appeared first in layers V–VI. A second but much larger current sink developed next in layer III, where it extended over the full extent of the layer progressing from its deepest border with layer IV (lamina dessicans) to the transition with layer II, where it stopped. A third distinct and final current sink then developed in layer II. The sequential development of current sinks first in layers V–VI, then in layer III, and finally in layer II suggests that these epileptiform discharges are typically generated in the deep layers of EC and then propagated actively toward layers III and II after the known ascending pathways (Kohler, 1986a). As is the case in the CSD profile of Figure 2, it was a consistent observation that in layer V the initial current sink was always followed by two other sinks at a fixed interval of ∼5 msec, thus indicating that the population activity in layer V occurred as a high frequency (∼200 Hz) population oscillation. In contrast, in layer III the initial short-lasting current sink was followed at ∼5 msec by a second, maintained sink of very long duration (∼20 msec). On the other hand, in layer II, only the initial short-lasting current sink was typically observed.
We also wanted to examine whether the activity generated in layers V–VI occurred synchronously over the full horizontal extent of the layer or whether there was some specific focal zone from which the activity was propagated. With this in mind we constructed field potential profiles along the mediolateral axis of layer V (horizontal profiles, n = 6). A typical experiment is illustrated in Figure 3. The diagram in Figure3 A summarizes the experimental protocol. The reference electrode was positioned in the most medial portion of layer V, and a moving electrode was advanced at fixed steps (100 μm) along layer V. Representative recordings of SDEs recorded at increasing distances from the reference electrode are plotted in Figure 3 B. The top traces in this panel are a superimposition of all averaged reference recordings during the track. Note that SDEs recorded at increasing lateral distances from the reference recording occur at increasing latencies, thus indicating that in this case the epileptiform activity propagated in the mediolateral direction along layer V at a speed of ∼100 mm/sec (Fig. 3 D, open circles). Occasionally (such as the case illustrated in Fig. 3), SDEs could appear at the same recording site with two different waveforms (Fig.3 C). Although the events corresponding to one of the waveforms showed positive latencies at increasing lateral distances from the reference (B, C, D; open circles), the other showed negative latencies (C, D; filled circles) that indicated independent focal events showing opposing propagation directions in the mediolateral axis. Although independent, the speed of propagation of epileptiform activity in either the mediolateral (open circles) or lateromedial direction (filled circles) was almost identical (D). These results suggest the existence of multiple focal generators of epileptiform activity within layer V that can be propagated in either the lateral or medial direction in the horizontal axis of the slice.
Surgical isolations and pressure pulse application
Although the above data indicate that layers V–VI play a leading role in the generation of epileptiform activity in the EC, our laboratory has shown that CCh has a robust facilitatory action on the intrinsic oscillatory and bursting properties of EC layer II neurons (Klink and Alonso, 1997a,b). Also, projection neurons from layer II display a profuse net of recurrent collaterals that are likely to excite neighboring cells (R. Klink and A. Alonso, unpublished observations). Thus, the electrophysiology and anatomy of EC layer II suggests that this layer could act as an independent pacemaker for oscillatory dynamics. Indeed, in those cases (n = 10) in which we monitored the induction of epileptiform activity using CCh by making dual simultaneous recordings in both layer II and layer V, low-amplitude short-duration field events were observed occasionally (n = 2) in layer II before any field activity in layer V.
To examine the issue of whether EC layer II could independently generate synchronous population activity, we initially performed focal CCh applications in layers II, III, and V while we simultaneously recorded field activity in layers II and V (Fig.4 A) (n= 17). As illustrated in Figure 4 B, in 30% of the cases (n = 17), pressure pulse applications of CCh in layer II evoked synchronous population activity localized exclusively to layer II. On the other hand, when CCh was focally applied to layer V, a prolonged series of epileptiform events was consistently recorded (100%; n = 17) by both layer V and layer II electrodes (Fig. 4 D). This activity mimicked the onset of epileptiform discharges seen with bath perfusion of CCh in that the first event triggered by CCh was an LDE and was followed by short ictiform discharges. Focal CCh applications to layer III never resulted in the elicitation of epileptiform events, whereas applications to deep layer VI/angular bundle typically resulted in no activity or the onset of a small number of SDEs at long delays (Fig. 4 E). In all cases tested (n = 3), it was found that atropine blocked the activity elicited by focal CCh applications at both layer II and layer V sites (not shown). Thus, these results suggest that both EC layers II and V, but not layer III, contain the appropriate neuronal and circuit machinery to independently generate synchronous population activity.
To further clarify whether EC layer II could independently generate rhythmic population activity, we undertook surgical isolation experiments. In the intact EC slice, we first induced epileptiform activity with CCh to be ensured of the viability of the slice, and then under continuous CCh superfusion we cut the slice to separate the internal and external principal laminae. As shown in Figure5, preserved epileptiform activity resembling that seen previous to the cut could be observed in most cases, in both the piece of slice containing the deep layers V–VI (Fig. 5 A) (90%; n = 19) or the superficial layers I–III (Fig. 5 B) (32%; n = 19) laminae. In additional cases, with surgical cuts that selectively isolated layer II, epileptiform activity was maintained over this layer (Fig. 5 C) (3 of 12 cases). In no case, however, was activity maintained in a piece of slice containing only layer III (n = 10). In addition, field potential profiles conducted after surgical isolation of the superficial principal lamina (layers I–III; n = 3) indicated that the remaining field activity was generated by layer II only and volume conducted to the remaining layer III.
To further substantiate the above results, surgical cuts separating the deep and superficial laminae were also performed before perfusion with CCh was undertaken (n = 21). In most of these cases (n = 19), the piece of slice containing the deep laminae showed patterns of epileptiform activity similar to those of intact slices when perfused with CCh. Also, in several cases (n = 6) CCh application to the piece of slice containing layer II induced epileptiform activity (not shown). The above surgical isolation experiments indicate that both layers V–VI and layer II can act as independent pacemakers of synchronous population activity.
We also investigated the intracellular correlates of the CCh-induced epileptiform activity, particularly in neurons from layer II (n = 47), because they have been the most extensively characterized electrophysiologically, but also in neurons from layers III (n = 12) and V (n = 9). During both LDEs and SDEs, principal cells in all layers fired bursts of action potentials synchronized with the local negative-going field spikes (Figs. 6,7). As in the specific layer II cell illustrated in Figure 6, in neurons from all layers, membrane voltage manipulation always revealed that the field-associated bursts of action potentials were driven by depolarizing events of synaptic nature. It was also apparent that in neurons from all layers the epileptiform discharges were typically associated with a plateau depolarization (Fig. 7; note dotted line in A–D). This plateau depolarization was particularly prominent and considerably outlasted the field discharges in neurons from layers III and V in comparison to neurons from layer II (Fig. 7).
In layer II projections neurons, we explored the reversal potential of the synaptic events underlying the epileptiform discharges by means of DC current injection (n = 6). As in the case illustrated in Figure 8, the depolarizing synaptic events were always fully reversed at potentials positive to approximately −40mV; however, we could always detect at least two components that reversed at different potentials. There was an early phase that reversed at approximately −50mV (Fig. 8 A,arrow) and a later phase that reversed at membrane potentials ∼5–10 mV more positive (Fig. 8 A,asterisk) than the early phase. Among other factors, this multiphasic reversal potential may have resulted from the rather synchronous input onto projection cells of both glutamatergic- and GABAergic-mediated synaptic potentials. Consistent with this interpretation, three fast-spiking cells (presumably interneurons) that were briefly (1–3 min) recorded during CCh-induced population activity also fired in bursts synchronized to the negative-going population spikes of the epileptiform events (not shown).
Given the excitatory nature of the intracellular correlate of field epileptiform events, we sought to characterize their dependence on glutamatergic neurotransmission. Antagonism of NMDA receptor-mediated neurotransmission with AP-5 (30 μm) did not abolish the epileptiform discharges in any of the cases tested (n = 6) (Fig. 9); however, it did produce some consistent effects. First, block of NMDA neurotransmission decreased the duration of individual population spikes from 179 ± 40 to 105 ± 36 msec and increased their frequency from 7.0 ± 1.7 to 14.2 ± 4.1 Hz. Second, NMDA receptor block also produced an increase in the duration of the epileptiform events from an average of 4.3 ± 2.3 to 6.1 ± 1.5 sec (Fig. 9).
The excitatory component of the field activity was dependent, however, on AMPA receptor-mediated glutamatergic transmission. It could be blocked by CNQX alone (n = 5) or in combination with AP-5 (n = 10) (Fig.10). To our surprise, however, small-amplitude field activity always persisted in layer II after blockade of fast excitatory amino acid (EAA) neurotransmission (Fig.10 A,B). This activity was also present in surgically isolated layer II as well (n = 3), and typically consisted of positive-going field events (Figs. 10,11). The intracellular correlate of this activity in layer II projection cells were large-amplitude IPSPs (8 ± 2 mV at 61 ± 2 mV; n = 8) [for simplicity hereafter referred to as giant IPSPs (gIPSPs)] of a rather fast rate of rise (43 ± 21 msec) and slow decay (total duration, 903 ± 116 msec) (Fig. 10 A–C). After peaking, these gIPSPs continuously decreased toward the resting level (Fig.10 B,C); however, we could typically distinguish, particularly at hyperpolarized levels, a decrease in the rate of decay at ∼100 msec from the gIPSP onset (Fig. 10 C). Membrane hyperpolarization revealed that gIPSP peak reversed at approximately −75 mV and the late phase at ∼10 mV more negative (Fig.10 C) (n = 3).
Interestingly, we observed that when the extracellular and intracellular electrodes were separated by distances greater than ∼200 μm, there was a lack of synchrony between the field events and the intracellular gIPSPs. In addition, as in the case illustrated in Figure 10, even within this critical range there tended to be a lack of absolute consistency between the intracellular and extracellular recordings in terms of waveform and synchrony. To illustrate this point, all individual field events and gIPSPs from Figure10 A have been superimposed in Figure10 B at an expanded time scale. Note that gIPSPs could occur in the absence of a field event (Fig. 10 A,B, asterisk), or conversely field events could occur in the absence of gIPSP (Fig. 10 A, filled circle). In addition, the shape of the field waveform could occasionally vary, and although most gIPSPs were associated with a field event, they could occur at close but different time relations to the field event (Fig.10 B, right). Finally, we examined the effect of GABAA receptor antagonism with PTX (100 μm) or bicuculline (10 μm) on the field events and gIPSPs. As shown in Figure 10 D, in all cases (n = 5) GABAA receptor antagonism always abolished all field events and IPSPs that persisted during glutamatergic transmission block.
The above data suggest that gIPSPs may result from the synchronous firing of interneurons, the positive-going field events in layer II arising from the associated inhibitory currents in principal neurons. The lack of perfect synchrony between the gIPSPs and field events suggest, however, that there might be multiple pools of interneurons impinging on distinct (and perhaps overlapping) pools of principal cells. To further examine this possibility, we performed mediolateral and depth profiles of the layer II positive-going field events (Fig.11 A–D) (n = 5). Typically, a reference electrode was centered in layer II, and a second moving electrode recorded activity at increasing fixed 50 μm steps in the mediolateral or laminar axis (Fig. 11 A). Figure 11,B and D, demonstrates that as the distance from the reference electrode increased, the amplitude of the reference field event decreased and became practically undetectable at a distance of ∼200–300 μm. Despite large-amplitude changes at these distances, no measurable latency changes were apparent in the waveform recorded at the moveable electrode. At the same time, as the distance from the reference electrode increased, other field events not detected by the reference electrode appeared in the moving electrode (Fig.11 B, traces 4 and 5). Profiles conducted in both the medial and lateral directions with respect to the stationary electrode (Fig. 11 D) demonstrated similar findings. Laminar profile analysis also indicated that the layer II positive events reversed in the upper layer III to become quickly undetectable as the test electrode was advanced further toward layer IV (Fig. 11 B).
This report focused on the ability of muscarinic receptor activation with CCh to trigger rhythmic population activity in the EC slice. CCh, in moderate concentrations (10–30 μm), indeed triggered population activity in EC, which had an epileptiform character. Similar to that of the neocortex, epileptiform activity in EC was typically driven by layer V; however, an isolated layer II was also able to generate epileptiform events. Because EC layers II and V gate the cortical input and output, respectively, of the hippocampal formation (Sorensen and Shipley, 1979; Van Hoesen, 1982; Swanson and Kohler, 1986; Witter et al., 1986; Witter, 1989), our data suggest that they may act as powerful distributors of epileptiform activity throughout the temporal lobe. In addition, we also found that CCh synchronized firing of inhibitory interneurons by a mechanism not necessarily dependent on principal cell activation (Michelson and Wong, 1994).
The epileptiform activity induced by CCh in EC showed similarities to that observed in the combined entorhinal–hippocampal slice using several proconvulsive manipulations such as low Mg2+(Walther et al., 1986), high K+ (Bear and Lothman, 1993), GABAA receptor block (Jones and Heinemann, 1988), 4-aminopyridine (4-AP) (Avoli et al., 1996), and pilocarpine (Nagao et al., 1996). In all cases, LDEs appeared to be generated by the EC and were maintained exclusively in that region after its surgical isolation.
Recent clinical data have also highlighted the importance of EC in epilepsy. Sclerotic lesions, long thought to be limited to the hippocampus in temporal lobe epilepsy (Babb, 1991; Meenchke and Veith, 1991; Swanson, 1995), have recently been demonstrated to include cortex further lateral in the temporal lobe, including the EC (Gloor, 1991;Levesque et al., 1991; Du et al., 1993). Furthermore, EC stimulation generates field potentials in the hippocampus that resemble spontaneous interictal discharge (Rutecki et al., 1989), and limbic seizures recorded with depth or subdural strip electrodes demonstrate focal seizure generation in EC (Spencer and Spencer, 1994). Moreover, recent surgical evidence has pointed to the role of entorhinal removal in controlling intractable temporal lobe seizures (Goldring et al., 1992,1993; Fried, 1993; Feindel et al., 1996).
Our laminar field potential profile and CSD analysis clearly indicated that the CCh-induced epileptiform discharges were initiated in layer V from where they propagated actively toward layers III and II at a speed of ∼0.1 m/sec. In addition, horizontal field potential profiles also demonstrated that layer V could contain several foci from where epileptiform events propagated horizontally in either the medial or lateral direction, also at a speed of ∼0.1 m/sec (Miles et al., 1988;Chagnac-Amitai and Connors, 1989a; Traub et al., 1993). It thus appears that once the activity is initiated by a small pool of layer V neurons, it spreads rapidly throughout the EC. That layer V neurons are primarily responsible for the generation of CCh-induced epileptiform activity is consistent with previous studies in the neocortical slice (Chagnac-Amitai and Connors, 1989b; Silva et al., 1991; Prince, 1993). A role of EC layer V neurons in the generation of epileptiform activity in the Mg2+-free model has also been suggested recently (Jones and Heinemann, 1988).
A leading role of EC layer V in the generation of epileptiform events was definitively established by focal CCh applications and surgical isolation experiments. Focal CCh applications in layer V always resulted in robust synchronous activity that was actively propagated toward superficial layers. In addition, a surgically isolated layer V continued to robustly generate epileptiform activity. The spread of epileptiform activity from layer V to layers III and II is consistent with the known unidirectional organization of the EC layer connections in which each EC layer innervates every layer situated superficially to it (Kohler, 1986b).
Interestingly, our experiments also pointed out that EC layer II, but not layer III, also has the capacity to independently generate hypersynchronous rhythmic population activity. Indeed, focal CCh applications to layer II frequently resulted in locally generated long trains of rhythmic population activity that did not actively spread to deeper layers, and a surgically isolated layer II was often found to sustain short or long ictiform discharges. Thus, in the EC, both layers V and II seem to have the adequate neuronal and circuit machinery for the generation of hypersynchronous population activity. In the neocortex, an isolated layer II/III has recently been shown to generate population events at a slow frequency in response to kainic acid application (Flint and Connors, 1996).
EC layer II contains two morphologically and electrophysiologically distinct principal cell types: the SCs and pyramidal-like cells (Alonso and Klink, 1993). The SCs generate rhythmic subthreshold membrane potential oscillations, the manifestation of which is facilitated by muscarinic receptor activation (Klink and Alonso, 1997b). The pyramidal-like cells behave more similarly to regularly spiking neocortical neurons, and muscarinic receptor activation facilitates in them the generation of intrinsic bursting activity (Klink and Alonso, 1997b). Both neuronal types display a profuse net of recurrent axonal collaterals that distribute extensively over layer II (Kohler, 1986b) (R. Klink and A. Alonso, unpublished observations). EC layer II thus appears to contain the appropriate neuronal and circuit mechanisms for the local promotion of oscillatory dynamics, and it is thus not surprising that CCh triggers synchronized population oscillations locally in layer II.
The inability of EC layer III to independently generate synchronized population oscillations in response to CCh was particularly evidenced by the inability of focal layer III CCh applications to induce population oscillations, in spite of the fact that focal CCh applications do cause direct depolarizing responses in these neurons (R. Klink and A. Alonso, unpublished observations). This finding is somewhat consistent with the absence of intrinsic oscillatory neurons in layer III (Dickson et al., 1997) and the more restricted horizontal projections of layer III as compared with those of layer II (Kohler, 1986b). Our CSD and intracellular recording analysis indicate, however, that during the epileptiform events layer III neurons are subjected to very strong excitatory synaptic bombardment (presumably from layer V neurons) (Scharfman, 1996; Gloveli et al., 1997). It might be that a massive NMDA receptor activation in layer III cells, caused by hyperactivation of layer V neurons, is at the basis of the robust layer III neuronal degeneration observed in temporal lobe epilepsy (Du et al., 1993, 1995). It is tempting to speculate that during chronic epilepsy the loss of the layer III input onto layer II (Kohler, 1986b) may lead to sprouting of the recurrent collaterals of layer II cells, thereby enhancing feedback excitation and thus the epileptiform tendency of layer II.
Our intracellular analysis demonstrated that principal neurons in all EC layers discharged bursts of action potentials synchronized with the local negative-going field spikes. In all cases, membrane hyperpolarization revealed that these bursts were driven by large-amplitude synaptic events. In layer II neurons, these synaptic events were always fully reversed at approximately −40 mV, and we could always detect an early component that reversed at slightly more hyperpolarized levels. This suggests that the epileptogenic synaptic events are probably the compound action of both glutamatergic-mediated EPSPs and GABAergic-mediated IPSPs. Similar data have been reported during perfusion of hippocampal slices with 4-AP (Rutecki et al., 1987), and simultaneous firing of interneurons and principal cells has been described in the high potassium model of epilepsy in the hippocampus in vitro (McBain, 1994), as well as during paroxysmal events in vivo (Steriade et al., 1994; Bragin et al., 1995, 1997; Steriade and Contreras, 1995). Moreover, in epileptic human mesial temporal lobe tissue, the observation of synchronized synaptic potentials with similar reversal properties as observed in the present study has also led to the suggestion of concomitant activation of inhibitory and excitatory synaptic input (Schwartzkroin and Knowles, 1984).
The maintenance of the ictiform activity induced by CCh appeared selectively dependent on fast EAA neurotransmission, because it was only modulated by NMDA receptor antagonism but blocked by AMPA-receptor antagonism. This pharmacological profile is similar to that reported for CCh-induced rhythmic population oscillations in the hippocampus (MacVicar and Tse, 1989). In the EC, however, small-amplitude, positive-going field events persisted in layer II during combined block of both AMPA and NMDA receptors. These events occurred periodically and were correlated with gIPSPs in layer II principal cells and abolished by GABAA receptor antagonism. Recurrent gIPSPs in principal cells during EAA neurotransmission block have also been observed in the neocortical and hippocampal slice in the presence of 4-AP (Aram et al., 1991; Perrault and Avoli, 1992; Michelson and Wong, 1994). Also, recently, 4-AP application to EC slices has been shown to induce long-lasting depolarizations in principal neurons that persisted during EAA neurotransmission block and were markedly depressed by GABAA receptor antagonism (Lopantsev and Avoli, 1996). Furthermore, synchronized synaptic potentials observed in epileptic human mesial temporal lobe tissue were also found to be dependent on fast GABAergic neurotransmission (Schwartzkroin and Haglund, 1986).
The mechanism of 4-AP-induced synchronization of inhibitory neurons in the hippocampus has been studied in detail by Michelson and Wong (1994). These authors reported that in the presence of 4-AP and during blockade of EAA neurotransmission, gIPSPs occur rhythmically in hippocampal principal cells. These gIPSPs were invariably triphasic in appearance, and PTX blocked the fast initial event (GABAAmediated) and left in isolation a slower monophasic event (GABAB mediated). These authors also concluded that hippocampal GABAergic interneurons may become synchronized via (1) recurrent interneuron collaterals and the depolarizing action of synaptically activated GABAA receptors and (2) electrotonic coupling. Similar mechanisms may also operate in EC (Buhl and Jones, 1993; Wouterlood et al., 1995). Even though the gIPSPs in EC layer II neurons were not triphasic in appearance, they did demonstrate a fast rise time (∼43 msec) compatible with GABAA receptor activation (Michelson and Wong, 1991) and a slow decay compatible with GABAB receptor activation. In addition, the early phase reversed at approximately −75 mV, whereas the slow late phase reversed ∼10 mV more negative.
The fact that the gIPSPs in layer II neurons were correlated with the locally generated positive field events suggests that these events result from inhibitory synaptic currents in projection cells. Because there was not always a perfect synchrony between the field events and the gIPSPs (Fig. 10), and because field events recorded at sites separated by distances greater than ∼200 μm tended to be uncorrelated (Fig. 11), this suggests that these events arise from the synchronous activation of discrete, localized pools of interneurons that innervate discrete, localized pools of principal cells (Freund and Buzsaki, 1996). Thus, inhibitory networks may play a fundamental role in the organization of population dynamics (Llinas et al., 1991; Cobb et al., 1995; Jefferys et al., 1996; Wang and Buzsaki, 1996) and functional domains within the EC.
This work was supported by a grant from the Canadian Medical Research Council. Dr. Dickson was supported by a fellowship from the Natural Sciences and Engineering Council of Canada. Dr. Alonso is a Montreal Neurological Institute Killam Scholar
Correspondence should be addressed to Dr. A. Alonso, Department of Neurology and Neurosurgery, Montreal Neurological Institute and McGill University, Montreal, Quebec, Canada H3A 2B4.