Intrinsic Connectivity of the Rat Subiculum: II. Properties of Synchronous Spontaneous Activity and a Demonstration of Multiple Generator Regions

Abstract

Brain structures that can generate epileptiform activity possess excitatory interconnections among principal cells and a subset of these neurons that can be spontaneously active (“pacemaker” cells). We describe electrophysiological evidence for excitatory interactions among rat subicular neurons. Subiculum was isolated from presubiculum, CA1, and entorhinal cortex in ventral horizontal slices. Nominally zero magnesium perfusate, picrotoxin (100 μM), or NMDA (20 μM) was used to induce spontaneous firing in subicular neurons. Synchronous population activity and the spread of population events from one end of subiculum to the other in isolated subicular subslices indicate that subicular pyramidal neurons are coupled together by excitatory synapses. Both electrophysiological classes of subicular pyramidal cells (bursting and regular spiking) exhibited synchronous activity, indicating that both cell classes are targets of local excitatory inputs. Burst firing neurons were active in the absence of synchronous activity in field recordings, indicating that these cells may serve as pacemaker neurons for the generation of epileptiform activity in subiculum. Epileptiform events could originate at either proximal or distal segments of the subiculum from ventral horizontal slices. In some slices, events originated in both proximal and distal locations and propagated to the other location. Finally, propagation was supported over axonal paths through the cell layer and in the apical dendritic zone. We conclude that subicular burst firing and regular spiking neurons are coupled by means of glutamatergic synapses. These connections may serve to distribute activity driven by topographically organized inputs and to synchronize subicular cell activity.

It is widely held that brain regions possessing some critical density of recurrent excitatory connectivity and that pacemaker neurons are capable of generating epileptiform activity (Heinemann, 1987). One of the best studied regions is area CA3 (for reviews, see Traub and Miles, 1991; Traub et al., 1999). Although pyramidal cells in CA1 also possess some recurrent excitatory connections, the density of these connections is apparently not sufficient to support the generation of spontaneous activity in isolated pieces of CA1 tissue. Alternatively, the differences in the intrinsic burst firing of individual pyramidal cells in CA3 vs. CA1 may be more of a factor in CA1’s inability to generate spontaneous activity (Wong and Stewart, 1992; cf. Tancredi et al., 1988). The intrinsic connectivity may be sufficient, but pacemaker neurons may be deficient.

Subiculum has been shown to generate spontaneous epileptiform activity in the presence of a convulsant when isolated from surrounding brain regions (Behr and Heinemann, 1996; Stewart, 1999), but less is known about the local connectivity of subicular cells than CA1 or CA3 cells. Subiculum contains at least two electrophysiological classes of principal neurons (reviewed in the first study; e.g., Stewart and Wong, 1993; Taube, 1993). In the first study of this pair, we presented morphologic data from electrophysiologically classified subicular neurons that included the first description of axon collateralization by each cell type. Numerous varicosities and axonal extensions along collaterals in, and apical to the cell layer, suggested that each electrophysiological cell class makes multiple synaptic contacts within the subiculum itself (see p. 490–505, this issue; Harris and Stewart 1999; also Shepherd and Harris, 1998).

To support the conclusion that some contacts made by principal neurons in subiculum were excitatory contacts onto other principal neurons in subiculum, we examined the behavior of each principal cell type during the generation of epileptiform events in the isolated subiculum. A preliminary subset of these results has been published (Stewart, 1999). Specifically, we sought to determine which cell types discharged in relation to synchronous “population” events in field potential recordings. In addition, we sought to define the properties of propagation and the paths within subiculum that support propagation of activity as physiological demonstrations of an intrinsic circuitry in the subiculum. Finally, by defining the patterns of activity that can be generated by subicular neurons when they are isolated from other brain regions, we can contrast this activity with the activity generated by larger circuits of neurons that include subiculum (e.g., Funahashi et al., 1999; Harris and Stewart, 2001).

MATERIALS AND METHODS

Our methods have been published in detail elsewhere (Funahashi and Stewart, 1997; Funahashi, et al., 1999; Stewart, 1999; Harris et al., 2001(this issue)). The experiments were performed on male Sprague-Dawley albino rats (150–250 g). The methods for slice preparation are described in the companion paper.

Isolation of brain regions

Subsections of slices were made before and after placement into the recording chamber. When subsections were made before placement, a slice was placed on filter paper moistened with perfusate. Cuts were made by using a microsurgical knife under a dissecting microscope. When subsections were made after placement in the recording chamber, a microknife was mounted in a micromanipulator and pulled or successively stabbed through the slice. Similar disconnection procedures have been used by us (Funahashi and Stewart, 1997; Funahashi et al., 1999) and others (e.g., Behr and Heinemann, 1996). Placements of the cuts were judged by eye and confirmed after some experiments with thionin stain. The compact cell layer of CA1 and layer II presubiculum are readily distinguished from the other cortical gray matter and the white matter of the alveus and angular bundle. Knife cuts separating CA1 from subiculum, presubiculum from subiculum, and entorhinal cortex from subiculum can all be made with confidence in the living tissue. One advantage of knife cuts over methods that transiently inactivate brain regions is that the disconnections can be examined histologically after experiments were completed. Isolated subiculum remained in place in relation to the surrounding structures. The ability to clearly see the CA1 cell layer made it possible to accurately estimate the top of the subicular cell layer. The fissure and the alveus were also readily observed in the living slice. This meant that there were good indicators of the top, middle, and bottom of the subiculum for making and reproducing partial transections (shown in Fig. 5). Thirty minutes to 1 hour was allowed after making subsections for equilibration before recordings began. Response stability was monitored for the time slices were in the recording chamber.

Fig. 5.

A–D: Illustration of the subicular subslice and the types of transections used to examine the spread of activity between distal and proximal parts. Slices are horizontal. Subiculum (Sub) was isolated from the surrounding brain regions by microknife cuts between CA1 and subiculum, presubiculum and subiculum, and through the alveus. Additional cuts were made in some experiments in subiculum itself to separate proximal (near CA1) from distal (near presubiculum) portions. Scale bar = 1 mm.

Extracellular stimulation and recording techniques

Extracellular recording electrodes were stainless steel (Roboz, Rockville, MD) with tip impedances (at 1 kHz) of 0.8–1.1 MΩ. Signals, referred to the bath, were amplified (model 1800, A-M Systems, Inc., Everett, WA), filtered (0.1 Hz to 10 kHz, −6 dB/octave), and digitized (Digidata 1200, Axon Instruments, Foster City, CA).

Extracellular stimulating electrodes were parallel bipolar electrodes (150-μm-diameter stainless steel, 0.5-mm tip exposure, 0.2-mm electrode separation; FHC, Brunswick, ME). Stimulating pulses were put through constant-current isolation units (Isolator-10, Axon Instruments, Foster City, CA) at 0.05 to 0.5 Hz. All electrodes were placed under direct visual guidance with the dissecting microscope. Intracellular recording techniques are described in the companion paper.

Pharmacologic manipulations

A nominally zero magnesium perfusate was used as a convulsant in most slices (Traub et al., 1994). Magnesium chloride was replaced with calcium chloride. Other slices were exposed to picrotoxin (100 μM), a γ-aminobutyric acid_A receptor antagonist, or N-methyl-D-aspartate (NMDA; 20 μM), an NMDA receptor agonist to induce spontaneous activity. 3-[2-carboxypiperazin-4-yl]-propyl-1-phosphonic acid (CPP; 10 μM), an NMDA receptor antagonist, was used to confirm the role of NMDA receptors in the induction of epileptiform activity. Picrotoxin was obtained from Sigma (St. Louis, MO). NMDA and CPP were obtained from Tocris (Ballwin, MO). All compounds were dissolved in artificial cerebrospinal fluid.

RESULTS

Experiments were performed on 62 ventral horizontal slices from 28 rat brains. The location and plane of section are the same for all experiments described in this study. Slices had one to three field potential recordings taken from entorhinal cortex (layer V or layer III), subiculum (near CA1, “proximal,” midway between CA1 and presubiculum, and/or near presubiculum, “distal”), and/or CA3. Every slice had at least one field electrode in subiculum. Recordings from entorhinal cortex or CA3 were used to confirm isolation of subiculum from other regions of the slice. Spontaneous activity in subiculum was synchronous with either entorhinal activity or CA3 activity in intact horizontal slices. No slices in this study showed subicular events synchronous with both regions, although this has been reported even at the single cell level (Stewart and Wong, 1993). Microknife cuts used to isolate subiculum desynchronized entorhinal cortex or CA3 from subiculum (Fig. 1). Subiculum generated its own epileptiform activity. Behr and Heinemann (1996) reported spontaneous epileptiform activity in isolated subiculum after exposure to zero magnesium media.

Fig. 1.

Spontaneous epileptiform events in subiculum and entorhinal cortex are synchronous in intact slices and independent after microknife cuts to isolate subiculum. Top: Field potential recordings from proximal subiculum (near CA1) and layer V of medial entorhinal cortex in a ventral horizontal slice exposed to picrotoxin (100 μM) before (left) and after (right) microsurgical isolation of the subiculum with cuts between presubiculum and subiculum, CA1, and subiculum, and through approximately the center of the alveus between entorhinal cortex and subiculum. Events before isolation consisted of a primary burst and a long series of afterdischarges (seven in the figure). After isolation of subiculum, events in entorhinal cortex continued to include multiple afterdischarges, but subicular events were reduced to a single population burst. Note also that the intervals between events increased in subiculum but decreased in entorhinal cortex. Bottom: Isolated subicular subslice exposed to N-methyl-D-aspartate (NMDA, 20 μM; left) and a mixture of NMDA and 3-[2-carboxypiperazin-4-yl]-propyl-1-phosphonic acid (CPP, 10 μM; right). CPP eliminated all activity induced by NMDA. Calibrations: 250 μV, 1 sec.

The epileptiform activity in the isolated subiculum was different from its activity when connections with surrounding regions remained intact (Fig. 1, top). A different pattern activity was most apparent when picrotoxin was used as the convulsant agent. Afterdischarges were seen in isolated subiculum from only one slice exposed to picrotoxin. All other recordings from picrotoxin-exposed isolated subiculum subslices showed single interictal-type events (n = 10 slices exposed to picrotoxin). By contrast, exposure of slices to nominally zero magnesium media (n = 49 slices) or NMDA (20 μM; n = 4 slices) induced more frequent discharges, and these typically included multiple afterdischarges (compare top and bottom panels of Fig. 1). CPP (10 μM) eliminated all activity in NMDA-treated slices (n = 4) but left smaller primary burst discharges in picrotoxin-treated slices (n = 2).

Activity of identified bursting and regular spiking cells during epileptiform activity

Intracellular recordings were taken from subicular neurons in isolated subicular subslices to identify the electrophysiological cell classes that participate in epileptiform burst discharges. Both the burst firing class and the regular spiking class of pyramidal cells discharged multiple action potentials during spontaneous epileptiform events. Figure 2 shows the activity of burst firing neurons in isolated subiculum. In Figure 2A₁–4 and B₁, different responses by different burst firing neurons are shown. In Figure 2A, it can be seen that the threshold response to depolarizing current injection can be a burst of action potentials, a doublet of action potentials or single action potentials. Although many burst firing neurons discharge a burst of action potentials at threshold and need to be maintained at depolarized levels to convert their threshold responses to doublets or single spikes (e.g., Stewart and Wong, 1993), some cells can spontaneously switch firing modes (Fig. 2A₁,A₂,A₃). We emphasize that our classification considers any cell that can give a burst response to injected current to be an intrinsically bursting neuron. Cells are classified as regular spiking cells if they cannot be made to burst in response to current injection. In other words, intrinsically bursting cells can burst or spike singly. Regular spiking cells can only fire single spikes to injected current. The classification is discussed in more detail in the first study of this pair of articles.

Fig. 2.

Firing of an intrinsically bursting neuron associated with population events in field recordings from isolated subiculum. A: Responses from four different intrinsically bursting neurons to illustrate variations in firing patterns. A₁,A₂,A₃: Responses range from single spikes to full bursts. A₄: Cell with a high threshold shows sag to steady state level in response to hyperpolarizing current pulse. B,C: Same intrinsically bursting cell firing in relation to synchronous field responses induced by zero magnesium media in an isolated piece of subiculum. B₁: The cell’s response to current injection (0.2 nA, 100 msec). B₂: The location of the cell deep in the cell layer with its principal division of the apical dendrite at the top of the cell layer (B₃). B₄: Simultaneous intracellular and field potential recordings show the firing relation of this cell to a spontaneous epileptiform event. In C₁–C₃, attempts were made to alter the cell’s firing by applying depolarizing (C₁) or hyperpolarizing (C₃) current. Note that differences in the appearance of action potentials recorded at different sweep speeds vary because of differences in the sampling interval used for digitizing waveforms. Action potentials in very slow sweep speed can have their peak amplitudes underestimated. Calibrations: 50 mV for intracellular recordings and 0.5 mV for field in B₄, 0.2 mV for field in C; 100 msec in B₁, 300 msec in B₄, 100 msec in C. Scale bars = 300 μm in B₂, μm 150 B₃. Horizontal line in C₁–C₃ is −70mV.

The firing of burst firing cells in relation to epileptiform events recorded in the field is also shown in Figure 2B₄,C₁–C₃. During the primary burst (the first large negative-going event in the field potential recordings) and each afterdischarge (the smaller negative-going events after the first event), the cell fired a burst or doublet of action potentials. Depending on the level of polarization of the cell membrane, more or fewer single spikes were discharged during the primary burst and changes could also be seen in the number of spikes within the bursts of action potentials exhibited by the cell. It was not possible to hyperpolarize bursting neurons to reduce the number of spikes to less than a doublet of action potentials. A spike doublet was the minimal action potential response as hyperpolarizing current was applied to the cell. Even in burst firing neurons that exhibited single spiking or burst firing at threshold, action potential doublets with <10-msec interspike intervals were the minimal response during epileptiform discharges. The firing by regular spiking neurons was different during hyperpolarizing current injection.

Regular spiking neurons fired multiple action potentials during the primary burst and afterdischarges recorded in the field (Fig. 3). The number of action potentials depended on the level of polarization, much like that in burst firing neurons. In regular spiking neurons, however, it was possible to reduce the number of spikes fired during field population spikes and increase the interspike intervals to >10 msec (e.g., Fig. 3B). Similar high frequency firing by regular spiking cells in response to orthodromic activation has been reported (Stewart, 1997), but this activity was not considered to be intrinsic burst firing because action potentials could be eliminated from the orthodromic burst by hyperpolarizing the recorded neuron.

Fig. 3.

A: Regular spiking cell shows firing in relation to population events in field recordings from isolated subiculum. Note that individual spikes from the cell’s response can be dropped by hyperpolarization of the membrane. Arrowheads identify excitatory postsynaptic potentials that were seen in this cell in between synchronous events. No firing was seen in between synchronous events at resting membrane potential. B: Faster sweep during hyperpolarizing current injection. Inset: responses to current injection. Calibrations: (A) 50 mV, 0.5 mV, 4 sec; (B) 50 mV, 400 μsec; (inset) 20 mV, 100 msec.

In one slice, we recorded three burst firing neurons in the absence of any field evidence of epileptiform activity. Field activity was sampled throughout the subiculum, as we did routinely in all experiments. Special attention was paid to field activity in the vicinity of the intracellularly recorded cell. No field events were recorded at any location. Spontaneous activity by these intrinsically bursting cells at rest (Fig. 4, top trace; beginning of second trace; bottom trace) was characterized by bursts of three to six action potentials or larger events consisting of multiple clusters of action potentials. Tonic depolarization of the cells eliminated the brief bursts of action potentials, inducing instead many more single action potentials (Fig. 4, middle two traces). The switch from burst firing mode to single spiking mode has been described previously (Mattia et al., 1993; Stewart and Wong, 1993; Taube, 1993). Despite the switch from brief bursts at about 1/second to single spikes at approximately 6–20/second, the large compound events still appeared.

Fig. 4.

Burst firing cell firing in the absence of any field activity, suggesting that burst firing cells may function as pacemaker neurons for the generation of epileptiform activity. Simple and compound bursts are seen in each sweep (12.8 seconds per sweep). A simultaneous field potential recording showed no activity. Membrane depolarization with current injection converted the cell’s firing from bursts to single spiking at a higher rate. Occasional bursts were still seen, suggesting that some bursts may have been synaptically driven (see Stewart, 1997). At less positive potentials, at rest and at potentials tested that were hyperpolarized from rest, burst were seen. Reference line is approximately −30 mV. Calibration: 40 mV, 1 sec.

Use of microknife cuts and partial transections to examine propagation of activity in subiculum

Figure 5A shows a horizontal slice with the basic cuts used to isolate the subiculum from presubiculum, CA1, and entorhinal cortex. No cuts were placed at or near the fissure to isolate subiculum from dentate. In the three slices tested, no activity was seen in dentate, indicating a lack of electrophysiological coupling of the two regions across the fissure. The set of three cuts was used to isolate subiculum. Additional cuts were made or extended as shown in Figure 5B,C,D to isolate proximal and distal portions of the subiculum. These different types of partial transection were used to examine paths for propagation within subiculum.

As described previously, most of the horizontally directed axon collaterals were located in the cell layer. One method of progressive disconnection used partial transections that began at the hippocampal fissure (see e.g., Fig. 5B) and were extended toward, and eventually through the cell layer. These cuts were made in seven slices, progressing in stages from the fissure to approximately the top of the cell layer in all seven, with extension through the cell layer, and finally to the ventricle in five of seven slices. Figure 6 shows an example of this kind of experiment. As shown in the figure, it was common for distal subiculum (black traces and black electrode marker) to lead proximal subiculum (gray traces and gray electrode marker) suggesting an origin for epileptiform events in the distal subiculum with propagation into the proximal subiculum (Fig. 6A shows activity in isolated subiculum before any attempt was made to disconnect the proximal region from the distal region). Although this was a common pattern, it was not the rule (as shown in later figures). Partial transections compromising spread through the apical dendritic zone never desynchronized proximal and distal recording sites nor eliminated activity at one or both locations (Fig. 6B–D). Not until transections included the basal dendritic zone did the recordings show that proximal and distal locations were no longer coupled (Fig. 6E). In the example shown in Figure 6, only distal subiculum remained active (black traces).

Fig. 6.

A–E: Example of slices disconnected with cuts beginning at the fissure and extending down to the apical side of the cell layer and ultimately through to the alveus. Only the distal side remains spontaneously active. Simultaneous field recordings from proximal (gray) and distal (black) recording locations in isolated subiculum in zero magnesium media. Three different sweep speeds are used to show synchrony of multiple events (e.g., A₁), presence of afterdischarges in each event (e.g., A₂) and temporal relation of events on the two recording electrodes (e.g., A₃). Transection involving only the apical dendritic region does not substantially alter the synchronization of the two locations. Progressive involvement of the cell layer ends up eliminating the activity on the proximal electrode, suggesting a single distal generator region in this slice. Voltage calibation: 0.2 mV. Time calibration: (A₁–E₁) 5 sec; (A₂–E₂) 0.1 sec; (A₃–D₃) 0.05 sec. Scale bar in E₃ = 0.5 mm.

Partial transections that began at the alveus (Fig. 5C) and were extended apically through the cell layer to the fissure were similarly ineffective at desynchronizing proximal and distal subicular recording sites until transections were complete (Fig. 7). Twelve slices had partial transections that began at the deep white matter and extended up through approximately the level of the top of the cell layer in one or two stages. Five of the 12 had the transections extended to the fissure. Figure 7A shows activity in the isolated subiculum before any additional cuts were made. A small partial transection beginning at the alveus is shown in Figure 7B, with progressive extension through the cell layer (Fig. 7C) to complete transection (Fig. 7D). We noted in several slices that transections that involved the basal dendritic zone and the cell layer (e.g., like that shown in Fig. 7C) led to a delayed coupling of the proximal and distal recording sites. Events were still apparently propagated from one end to the other, but conduction delays increased from <10 msec to >50 msec. In one slice, the proximal site led the distal site when the onsets of the primary discharge were compared. However, afterdischarges occurred first at the distal site. In this slice, the transection beginning in the basal dendritic zone eliminated late afterdischarges from the proximal recording site without desynchronizing the primary burst or the early afterdischarges.

Fig. 7.

A–D:Example of slices disconnected with cuts beginning at the alveus and extending apical-ward to ultimately reach the fissure. This experiment is similar to that shown in Figure 6, but transection started near the alveus and progressed to reach the fissure. Cuts progressing through the cell layer do not eliminate synchrony, suggesting that pathways in the apical dendritic zone can mediate synchronization. Full transection leaves both the proximal and distal parts spontaneously active, indicating at least two independent generator regions in this slice. See Fig. 6 for calibrations.

That spontaneous epileptiform events could originate at proximal or distal sites is well illustrated by slices in which events were found on both sides of a full transection (Fig. 7D) and in slices where events were found to start at different sites (Fig. 8). In Figure 7, full transection desynchronizes activity recorded on the proximal and distal sides of the transection but did not eliminate activity from either side as full transections were found to do in other slices (e.g., Fig. 6). In several slices, the onset and appearance of field activity was clear enough and distinct enough to permit discrimination of events originating proximally, propagating to the distal site, from events originating distally, propagating to the proximal site (Fig. 8B). The rate of propagation is shown in Figure 8A for one slice that had events start near the distal end and in another slice that had events start at both the distal and proximal ends (Fig. 8B). Propagation rate was not found to differ for events moving one direction or the other.

Fig. 8.

Examination of the spread of activity in a subicular subslice in which two generator regions were present. Graphs plot the latency difference between events recorded on field electrodes with different interelectrode distances. A: A slice with a single generator region. The delay between events on the two electrodes is a simple function of the distance between electrodes. B: A slice with two generator regions. At each location it was possible to differentiate two different general shapes for population spikes. Plots of latency vs. electrode separation reveal proximal events sometimes leading distal events and other time lagging distal events, indicating that events could sometimes started at a proximal generator site and other times started at a distal generator site. We found that most slices had distal generator sites.

DISCUSSION

We describe electrophysiological evidence for excitatory interactions among rat subicular neurons. This evidence is (1) subicular principal cells of both electrophysiological classes fire synchronously during epileptiform events generated in the isolated subiculum; (2) synchronous events propagate away from their place of origin in subiculum to other parts of subiculum; and (3) synchronous discharges depend on glutamate receptors. Synchronous population activity and the spread of population events from one end of subiculum to the other in isolated subicular subslices indicate that neurons are coupled together. Both electrophysiological classes of subicular pyramidal cells (bursting and regular spiking) exhibited synchronous activity. In the previous study, we showed that both electrophysiological classes of cells had axon collaterals with numerous axonal extensions and varicosities (light microscopic evidence for synaptic contacts) in and apical to the cell layer. Taken together, the morphologic and electrophysiological data indicate that both subicular cell classes get and give excitatory connections within the subiculum itself.

An additional finding was that burst firing neurons were found to be active in the absence of synchronous activity in field recordings. This finding indicates that burst firing neurons may serve as pacemaker neurons for the generation of epileptiform activity in subiculum. In the previous study, we described the collateralization of axons within the subiculum that would permit a pacemaker neuron to activate neighboring neurons. In the previous study, we showed that neurons located deep in the cell layer had relatively narrowly distributed ascending axon collaterals. We suggested that this conferred a columnar organization on the subiculum. A pacemaker neuron located deep in the cell layer may, thereby, be able to activate other cells above it to entrain the cells of a part of the subiculum.

Epileptiform events could originate at either proximal or distal parts of the subiculum from ventral horizontal slices. In some slices, events originated in both proximal and distal locations and propagated to the other location. Finally, propagation was supported over axonal paths through the cell layer and in the apical dendritic zone. By comparison with the axonal arbors of deep neurons, subicular cells located in the superficial part of the cell layer had axon collaterals that reached longer distances within the subicular cell layer. These extensive collaterals may serve to organize activity within a region of subiculum and to relay this activity from one part of subiculum to another.

Epileptiform activity in the subiculum

Behr and Heinemann (1996) were the first to demonstrate spontaneous epileptiform activity in the isolated subiculum. Knife cuts were used to isolate subiculum from CA1 and presubiculum. Spontaneous epileptiform discharges were found after treatment with zero magnesium perfusate. The findings reported herein confirm the results of this earlier report and extend the results in several ways. First, we demonstrate the involvement of both burst firing and regular spiking neurons in these epileptiform discharges. Second, we characterize the spread of epileptiform activity within the subiculum, identifying pathways in the cell layer and in the apical dendritic zone. Third, we demonstrate that multiple locations within subiculum can serve as the seizure focus. Fourth, we show that subicular burst firing neurons can serve as “pacemaker” neurons for the generation of epileptiform activity. Finally, we demonstrate the effectiveness of picrotoxin and NMDA as convulsant agents in subiculum.

The morphologic data (Harris et al., companion paper) suggested that paths for propagation existed within the cell layer and in the apical dendritic zone. After finding that epileptiform events could originate near the CA1 border and/or near the presubiculum border and spread to the opposite side, we used partial transections to examine the pathways for the spread of activity in subiculum. Partial transections that extended from the fissure down to the apical edge of the cell layer did not block propagation but did reduce the number of afterdischarges in recordings on the proximal and distal sides of the transection. Cuts extending from the alveus to the apical side of the cell layer also failed to block propagation, but were associated with both a loss of afterdischarges and a significant increase in conduction delay. These results suggest that the pathway mediated by axons running in the cell layer is more effective for relaying events along the CA1–presubiculum axis. The changes in the number of afterdischarges are consistent with reductions in the amount of excitatory input and in the duration of excitation. An alvear pathway does not seem to be significant because our cuts to isolate subiculum from entorhinal cortex ran in the alveus and most likely damaged many of these fibers. More significantly, from the morphologic data, we did not see collaterals of axons that had entered the alveus come back into the subiculum. The connections in subiculum occurred by means of collaterals that ran in and above the cell layer.

Sustained depolarization of burst firing cell dendrites caused multiple burst discharges by these cells (Funahashi et al., 1999). This, together with their pacemaker capabilities, suggests that afterdischarges in the isolated subiculum may result from sustained synaptic depolarization of the apical dendrites of burst firing cells. The partial transections could reduce the amount of excitatory input to cells by eliminating activity on paths in the apical dendritic zone or the cell layer. Reductions in the duration of excitation could result from disruption of circuits within subiculum, which can repeatedly activate some cells.

Greene and Totterdell (1997) found relatively more burst firing neurons closer to the borders of subiculum with CA1 and presubiculum, although we found a similar proportion all along the CA1–presubiculum axis. A concentration of intrinsic bursting neurons at the borders would suggest that these regions might be more likely to serve as seizure foci. We did find that spontaneous epileptiform events tended to originate nearer to one or both borders than toward the midpoint of the CA1–presubiculum axis. Interestingly, some slices had independent seizure foci at both sides and events propagated back and forth. What makes this especially interesting is the possibility that activity may be relayed back and forth between different parts of the subiculum.

Several burst firing neurons were identified as potential pacemaker neurons by being spontaneously active in the absence of a synchronous population discharge. Regular spiking neurons were only found to fire in relation to the synchronous discharges, although one cell did show small excitatory postsynaptic potentials in between synchronous events. It was interesting that the discharges of burst firing neurons at rest and unrelated to the synchronous population discharges were bursts of action potentials. These could be converted to regular spikes by depolarizing the cell membrane.

We have taken these results as electrophysiological evidence for excitatory interconnections between subicular pyramidal cells, demonstrating that both the burst firing and regular spiking cell classes receive excitatory input. In the previous study characterizing the morphologic properties of subicular cells, we showed light microscopic evidence suggestive of synaptic contacts within subiculum by both classes of subicular principal neurons. We conclude that subicular principal cells are sufficiently interconnected to permit synchronous activity and the propagation of action potentials within subiculum. The presence of pacemaker neurons in addition to interconnections among principal cells account for the subiculum’s ability to generate spontaneous epileptiform activity.

Interactions of subiculum with other hippocampal formation regions

Subiculum is known to relay epileptiform activity that originates in area CA3 in brain slices. It has been shown that CA3 leads CA1, and CA1 leads the subiculum (e.g., Stewart and Wong, 1993). The connection between CA3 and CA1 is the third synapse of the trisynaptic pathway (Andersen et al., 1971). The connection between CA1 and subiculum is arguably the most important next synapse in the hippocampal formation circuitry. As we discussed earlier, subiculum is also the target of inputs from entorhinal cortex and presubiculum. Entorhinal cortex can generate its own epileptiform activity (reviewed in Scharfman, 2000), and single subicular neurons can follow activity originating in CA3 or entorhinal cortex (Stewart and Wong, 1993). Presubiculum can generate epileptiform activity on its own (Funahashi and Stewart, 1997), and subicular neurons are excited by presubicular afferents (Stewart, 1997; Funahashi et al., 1999). At least three cortical afferent systems converge in subiculum: CA3 to CA1 to subiculum, entorhinal cortex, and presubiculum.

Subiculum can relay its activity out to all three afferent systems. Subicular inputs to entorhinal cortex are the best known system to complete an entorhinal–hippocampal circuit that includes the trisynaptic pathway (reviewed in Witter et al., 1989, 2000). Excitatory subicular outputs to the presubiculum have been demonstrated (Funahashi and Stewart, 1997; Stewart, 1997; Funahashi et al., 1999) and even a functional back projection to CA1 has been shown (Berger et al., 1980; Finch et al., 1983; Köhler, 1985; Harris and Stewart, 2001).

The variety of circuits, small and large, that include subiculum highlight its position in the hippocampal formation as a relay, integrator, and distributor of activity from and to many cortical and subcortical structures. One possible function for the interconnections among subicular cells and its reciprocal connectivity with CA1, presubiculum, and entorhinal cortex is suggested by the typical firing patterns of cells seen in freely behaving animals.

Recordings from hippocampal and parahippocampal neurons of freely behaving animals have identified several behavioral correlates of the firing of individual cells. One of the best known correlates is the relation of cell firing to an animal’s location in its environment. Cells firing in a particular place in an environment are called place cells (Muller, 1996; Muller et al., 1999). The firing of place cells is characteristically a stream of action potentials for the time an animal is in its place field. The animal can be moving through the field or stationary in the field. Spike trains will depend on the time spent in the field but typically are on the order of hundreds of milliseconds in duration. Recordings taken from cells along the trisynaptic pathway have not demonstrated that CA1 cells, for example, follow CA3 cells in their firing. In addition, adjacent cells rarely have similar firing properties. This is surprising in view of the degree to which the relays (e.g., CA3 to CA1 or CA1 to subiculum) are organized. We believe that the circuits within and between regions of the hippocampal formation serve to sustain firing levels in groups of neurons, forming a collectively active system whose activity represents something such as a place or an environment. A rich system of interconnections and reentrant circuits can eliminate issues of conduction delays, making it a good or efficient system to build cognitive representations. Unfortunately, this same system of interconnections and re-entrant circuits makes these regions especially vulnerable to the kind of self-activation that manifests itself as epilepsy. With so many possible foci for seizure activity, only a much clearer understanding of the interactions of cells in and between regions will permit therapeutic strategies that can eliminate the hypersynchrony of epilepsy without eliminating the synchrony needed for normal functioning.