HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) All Versions of this Article: 92/6/3582    most recent 01028.2003v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (14) Citing Articles via Google Scholar Google Scholar Articles by Bausch, S. B. Articles by McNamara, J. O. Search for Related Content PubMed PubMed Citation Articles by Bausch, S. B. Articles by McNamara, J. O.

REPORT

Contributions of Mossy Fiber and CA1 Pyramidal Cell Sprouting to Dentate Granule Cell Hyperexcitability in Kainic Acid–Treated Hippocampal Slice Cultures

¹Departments of Medicine (Neurology), ²Neurobiology, and ³Pharmacology and Molecular Cancer Biology, Duke University Medical Center, Durham, North Carolina 27710

Submitted 24 October 2003; accepted in final form 19 July 2004

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Axonal sprouting like that of the mossy fibers is commonly associated with temporal lobe epilepsy, but its significance remains uncertain. To investigate the functional consequences of sprouting of mossy fibers and alternative pathways, kainic acid (KA) was used to induce robust mossy fiber sprouting in hippocampal slice cultures. Physiological comparisons documented many similarities in granule cell responses between KA- and vehicle-treated cultures, including: seizures, epileptiform bursts, and spontaneous excitatoty postsynaptic currents (sEPSCs) >600pA. GABAergic control and contribution of glutamatergic synaptic transmission were similar. Analyses of neurobiotin-filled CA1 pyramidal cells revealed robust axonal sprouting in both vehicle- and KA-treated cultures, which was significantly greater in KA-treated cultures. Hilar stimulation evoked an antidromic population spike followed by variable numbers of postsynaptic potentials (PSPs) and population spikes in both vehicle- and KA-treated cultures. Despite robust mossy fiber sprouting, knife cuts separating CA1 from dentate gyrus virtually abolished EPSPs evoked by hilar stimulation in KA-treated but not vehicle-treated cultures, suggesting a pivotal role of functional afferents from CA1 to dentate gyrus in KA-treated cultures. Together, these findings demonstrate striking hyperexcitability of dentate granule cells in long-term hippocampal slice cultures after treatment with either vehicle or KA. The contribution to hilar-evoked hyperexcitability of granule cells by the unexpected axonal projection from CA1 to dentate in KA-treated cultures reinforces the idea that axonal sprouting may contribute to pathologic hyperexcitability of granule cells.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
The hippocampus is a common locus for seizures and anomalous synapse formation associated with temporal lobe epilepsy. The most well documented and well studied synaptic rearrangement in the hippocampus is mossy fiber sprouting. Mossy fiber sprouting is often found in tissue taken from patients and numerous animal models with temporal lobe epilepsy (Cronin and Dudek 1988; de Lanerolle et al. 1989; Houser et al. 1990; Sutula et al. 1989; Mello et al. 1992; Sloviter 1992; Sutula et al. 1988; Tauck and Nadler 1985; Wuarin and Dudek 1996) and is a pronounced expansion of a small normal projection of granule cell mossy fiber axons into the supragranular layers of the dentate gyrus (de Lanerolle et al. 1989; Houser et al. 1990; Laurberg and Zimmer 1981; Mello et al. 1992; Molnar and Nadler 1999; Okazaki et al. 1995; Represa et al. 1993; Ribak and Peterson 1991; Seress 1992; Sutula et al. 1988, 1989; Tauck and Nadler 1985) and within the hilus (Sutula et al. 1998; Wenzel et al. 2000). Anatomical and physiological studies have shown that sprouted mossy fibers form synapses onto GABAergic neurons and to a larger degree onto granule cells (Buckmaster et al. 2002; Franck et al. 1995; Kotti et al. 1997; Molnar and Nadler 1999; Okazaki et al. 1995; Represa et al. 1993; Ribak and Peterson 1991; Sloviter 1992; Sutula et al. 1988, 1989; Wenzel et al. 2000; Wuarin and Dudek 1996; Zhang and Houser 1999), suggesting that granule cells form synapses between themselves and thus create an aberrant recurrent excitatory network. Such a recurrent network could increase granule cell excitability (Cronin et al. 1992; Okazaki et al. 1999; Patrylo and Dudek 1998; Tauck and Nadler 1985; Wuarin and Dudek 1996) and compromise the granule cells' normal ability to serve as a barrier to invasion of abnormal activity into the hippocampus (Behr et al. 1998; Collins et al. 1983). Sprouted mossy fiber synapses onto GABAergic neurons further suggest that recurrent inhibition may dampen hyperexcitability associated with a recurrent excitatory granule cell network (Cronin et al. 1992; Franck et al. 1995; Sloviter 1992). These ideas are consistent with the more general hypotheses that recurrent excitatory networks induce hyperexcitability that can contribute to seizure genesis and that seizures can be kept under control by GABAergic inhibition.

In line with these general hypotheses, mossy fiber sprouting has been associated with granule cell hyperexcitability in standard slices prepared acutely (acute slices) from epileptic human and rat hippocampus. Hilar stimulation elicits multiple population spikes in a small percentage of these slices in physiological buffer. A larger percentage of slices displays burst discharges after application of a GABA_A receptor antagonist or elevation of extracellular potassium (Cronin et al. 1992; Franck et al. 1995; Hardison et al. 2000; Masukawa et al. 1992; Okazaki and Nadler 2001; Patrylo and Dudek 1998; Tauck and Nadler 1985; Williams et al. 2002; Wuarin and Dudek 1996). By contrast, hilar stimulation elicits a single antidromic population spike in most slices from control human and rat hippocampus in both physiological buffer and after suppression of GABA_A receptor mediated inhibition. However, whether mossy fiber sprouting is the only aberrant recurrent excitatory network contributing to granule cell hyperexcitability remains uncertain.

The relative contribution of mossy fiber sprouting to limbic seizures and epileptogenesis also remains a topic of debate. In support of a role for mossy fiber sprouting in limbic epilepsy, Dudek and colleagues (Patrylo and Dudek 1998; Wuarin and Dudek 1996) documented seizures in acute slices from kainic acid (KA)–treated epileptic rats with robust mossy fiber sprouting using a combination of a GABA_A receptor antagonist and elevated extracellular K⁺. Furthermore, Gorter et al. (2001) and Zhang et al. (2002) found a positive correlation between mossy fiber sprouting and the frequency of spontaneous seizures in animal models in vivo. However, evidence against a causal role for mossy fiber sprouting in limbic epilepsy is the observation that modest mossy fiber sprouting can be associated with an absence of limbic seizures (Cassell and Brown 1984; Nissinen et al. 2001; Timofeeva and Peterson 1999), that the degree of mossy fiber sprouting is not a reliable indicator of the rate of epileptogenesis (Elmer et al. 1997; Nissinen et al. 2001; Zhang et al. 2002) or the severity, frequency, or number of seizures (Mohapel et al. 2000; Nissinen et al. 2001; Proper et al. 2000; Sutula et al. 1989), and that spontaneous limbic seizures can occur in the absence of mossy fiber sprouting (Franck et al. 1995; Gorter et al. 2001; Houser et al. 1990; Longo and Mello 1997, 1998; Masukawa et al. 1995; Mohapel et al. 2000; Proper et al. 2000; Sutula et al. 1989; Williams et al. 2002; Zhang et al. 2002). Moreover, many laboratories have studied extracellular granule cell layer field potentials in acute slices from rat or resected human epileptic hippocampus (Cronin et al. 1992; Franck et al. 1995; Isokawa et al. 1991, 1997; Masukawa et al. 1991; Molnar and Nadler 1999; Okazaki et al. 1999; Tauck and Nadler 1985; Urban et al. 1990; Williamson et al. 1995) and none reported seizures involving granule cells in either physiological buffer or after application of GABA_A receptor antagonists, despite documented mossy fiber sprouting in many of the studies. Thus the relative contribution of mossy fiber sprouting to granule cell hyperexcitability and limbic epilepsy remains controversial.

This controversy led us to further investigate the role of mossy fiber sprouting in granule cell hyperexcitability and seizures and to examine the possibility that alternative hippocampal pathways can contribute to these events. In this study, we performed electrophysiological analyses of granule cell excitability and morphological analyses of the synaptic organization in an in vitro organotypic hippocampal slice culture model of KA-induced temporal lobe epilepsy and mossy fiber sprouting. Application of KA to hippocampal slice cultures induces seizures, Ammon's horn sclerosis, and mossy fiber sprouting (Routbort et al. 1999), which are similar to observations from both human epilepsy and in vivo models. In contrast to acute slices isolated from epileptic brain, sprouting takes place within the confines of the culture so synaptic connections are not severed immediately before study.

Portions of this manuscript were presented previously in abstract form (Bausch and McNamara 1997, 1999; Bausch et al. 1998). Previously published results from vehicle-treated cultures (Bausch and McNamara 2000) are provided for ease of comparison.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Organotypic slice cultures

Slice cultures were prepared using the method of Stoppini et al. (1991) as described previously (Bausch and McNamara 2000; Routbort et al. 1999). Briefly, postnatal day 11 Sprague–Dawley rat pups (Zivic-Miller, Zenople, PA) were anesthetized with pentobarbital and decapitated. The brains were removed and hippocampi were cut into 400-µm transverse sections using a McIlwain tissue chopper and placed into Gey's balanced salt solution (GBSS, Gibco BRL) supplemented with 6.5 mg/ml glucose. The middle 4–6 slices of each hippocampus (with the entorhinal cortex removed) were placed onto tissue culture membrane inserts (Millipore) in a tissue culture dish containing medium consisting of 50% minimum essential medium, 25% Hank's buffered salt solution, 25% heat-inactivated horse serum, 0.5% GlutaMax II (all from Gibco BRL), 10 mM HEPES, and 6.5 mg/ml glucose (pH 7.2). Medium was changed 2 to 4 times per week. Cultures were maintained at 37°C under room air + 5% CO₂. After 10 days in vitro (10 DIV), cultures were treated for 48 h with 6 or 7 µM kainic acid (Tocris Cookson) diluted in medium. Vehicle-treated cultures were treated similarly, but kainic acid was omitted. All treatment of animals was according to National Institutes of Health and institutional guidelines.

Electrophysiological recording

A portion of the tissue culture insert membrane containing a single slice culture was placed in a submerged recording chamber mounted to a Zeiss Axioskop microscope. Slice cultures were superfused at room temperature with a recording buffer composed of (in mM): NaCl 120, KCl 3.5, MgSO₄ 1.3, CaCl₂ 2.5, NaH₂PO₄ 1.24, NaHCO₃ 25.6, and glucose 10, equilibrated with 95% O₂ and 5% CO₂. Bicuculline methiodide (BMI, 10 µM; Sigma), D(–)-2-amino-5-phosphonopentanoic acid (D-APV, 50 µM; Tocris Cookson) and 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX, 10 µM; Tocris Cookson) or 6-nitro-7-sulfamoylbenzo[f]quinoxaline-2,3-dione (NBQX, 10 µM; Tocris Cookson) were diluted immediately before use in recording buffer and applied by bath superfusion. Recording pipettes (whole cell, 2–5 M; extracellular, 1–3 M) were filled with 3 M NaCl for extracellular recordings; with (in mM): Cs-gluconate 100, CsCl 30, HEPES 10, EGTA 10, MgCl₂ 3, adenosine 5'-triphosphate (Na₂ATP, Sigma) 2, lidocaine N-ethyl bromide (QX-314; Research Biochemicals International) 5 (pH 7.2 with CsOH) for whole cell voltage-clamp recordings; or with (in mM): K-gluconate 100, KCl 30, HEPES 10, EGTA 10, MgCl₂ 3, Na₂ATP 2 (pH 7.2 with KOH) for whole cell current-clamp recordings. Evoked responses were elicited by stimulation (0.3-ms square pulse, 0.03 Hz, 20–700 µA) using a Grass stimulator and a concentric bipolar electrode (MCE-100, Rhodes Medical Supply) placed in the hilar mossy fiber pathway immediately adjacent to the CA3c pyramidal cell layer (illustrated in Fig. 6A). Data were collected using an Axopatch 1D amplifier (2 kHz analog filter) and pCLAMP or Axotape software.

View larger version (23K): [in this window] [in a new window]   FIG. 6. Connections between dentate granule cells and neurons from other hippocampal subregions contribute to the short-latency hilar-evoked EPSP. In KA-treated slice cultures, hilar-evoked extracellular field potential responses were recorded in the dentate granule cell layer in the presence of BMI (10 µM) as described in the METHODS. Responses were measured at a stimulus intensity sufficient to evoke a maximal response (200–700 µA). A: acute knife cuts were performed 45 min to 1 h before recording as described in the text and were placed between the dentate gyrus and CA1 (CA1 cut) or CA3a/b (CA3 cut). A combination of the CA1 and CA3 cuts (DG cut) was used to completely isolate the dentate gyrus/CA3c region from the rest of the hippocampal formation. B: short-latency EPSP duration was measured from the point immediately after the antidromic population spike to a point at 50% of the peak amplitude (excluding any spikes) during the EPSP decay. This point was chosen because, in many recordings, longer-latency PSPs were superimposed on the short-latency EPSP decay before it reached 80–100% return to baseline. Short-latency EPSP amplitude was measured from baseline to peak positivity. In recordings showing a negative shift in potential immediately after the antidromic population spike (see Fig. 7, BMI), short-latency EPSP amplitude and duration were assigned 0 mV and 0 ms, respectively. CA1 and DG cut, but not the CA3 cut, significantly decreased the duration (C) and amplitude (D) of the short-latency EPSP in KA-treated cultures. Numbers and error bars represent the means ± SE in the number of slice cultures in parentheses. *, Significantly different from no cut and CA3 cut (ANOVA with LSD post hoc comparison; P < 0.05).

Dentate granule cell layer field potential recordings were deemed acceptable if hilar stimulation yielded an action potential spike, which immediately followed the stimulus artifact with a response threshold 100 µA (e.g., Fig. 3A). This spike could be abolished with tetrodotoxin (TTX; 1 µM; Calbiochem; data not shown) and was called an antidromic population spike because of the very short latency and lack of an underlying field excitatory postsynaptic potential (EPSP). Neither the amplitude of the antidromic population spike nor the shape of the waveform was used as a criterion for acceptable recordings. Evoked responses were measured at a stimulus intensity sufficient to elicit a maximal response (200–700 µA). PSPs were defined as waveforms that were blocked by antagonists of fast synaptic transmission (BMI + APV + CNQX or NBQX). Knife cuts between the dentate gyrus and different regions of the hippocampal formation were performed with a scalpel blade under a Reichert Stereo Zoom dissecting microscope as described in the RESULTS and depicted in Fig. 6A. Data from cut and uncut cultures were obtained from different slice cultures because knife cuts disrupted the integrity of the recordings (data not shown) and data obtained after removal and repositioning of electrodes is difficult to interpret because field potential waveforms can be altered drastically by slight changes in recording electrode location.

View larger version (30K): [in this window] [in a new window]   FIG. 3. Granule cells in KA- and vehicle-treated hippocampal slice cultures were hyperexcitable and hyperexcitability was under GABAA receptor–mediated control. Hilar-evoked extracellular field potential responses in KA- and vehicle-treated slice cultures were recorded in the dentate granule cell layer as described in the METHODS. A: in physiological recording buffer, hilar stimulation (stimulus artifact; *) elicited an antidromic population spike (open arrow) followed by either a short-latency postsynaptic potential (PSP) (A1, arrowhead), a second smaller population spike (A2, curved arrow), multiple spikes (A3, curved arrow), or an isolated antidromic population spike (A4, open arrow). B: representative traces show that application of BMI (10 µM) increased the short-latency evoked excitatory PSP (EPSP, arrowhead) and caused the appearance of longer-latency EPSPs (arrows). Amplitude of the short-latency evoked EPSP generally was above threshold for action potential firing, as indicated by the superimposed population spike. (BMI-induced increase in amplitude and decrease in duration of the short-latency EPSP is not representative of the mean.) Antidromic population spike (open arrow) is immediately adjacent to the stimulation artifact (*). Inset in KA-BMI is an enlargement of the longer-latency EPSPs and illustrates an example of the asynchronous small spikes that were superimposed on longer-latency EPSPs in 75% of KA-treated cultures. C: representative traces show that application of BMI also caused the appearance of spontaneous bursts of EPSPs and population spikes in KA- and vehicle-treated cultures. Traces in B and C were recorded in the same slice cultures. Horizontal scale bar in A applies to all traces in A. Scale bars in KA-BMI apply to all KA traces in B and in C apply to all traces in C.

TOLUIDINE BLUE. After recordings, slice cultures were stained with toluidine blue to document neuronal loss. Previously we found that loss of CA3a/b pyramidal cells was reliably associated with mossy fiber sprouting (Routbort et al. 1999). Although detection of mossy fibers by Timm stain would have provided a more direct approach to assess mossy fiber sprouting, staining quality was poor after 0.5–1 h recordings in both vehicle- and KA-treated cultures. Vehicle- and KA-treated cultures were always stained with toluidine blue concurrently under identical experimental conditions. Briefly, as described previously (Routbort et al. 1999), cultures were fixed with 3% glutaraldehyde in 0.1 M phosphate buffer (PB [pH 7.4]), permeabilized with 0.5% Triton X-100, stained with 0.5% toluidine blue, treated with 70% ethanol followed by 95% ethanol containing 0.2% glacial acetic acid, mounted onto subbed glass slides, dehydrated, cleared, and coverslipped. Cell layers were scored subjectively as: 3, many prominently stained neurons; 2, sparser number of stained neurons; 1, very sparse number of scattered stained neurons; 0, no stained neurons in cell layer using an Axiovert 135 microscope at 40 x magnification. Neurons were differentiated from glia based on their larger nuclear size. CA3c was defined as the CA3 pyramidal cell layer located between the blades of the dentate granule cell layer. CA3a/b was defined as the CA3 pyramidal cell layer excluding the CA3c region.

TIMM STAIN. Cultures were stained with a Timm stain for heavy metals according to a protocol modified from Danscher and Zimmer (1978), as described previously (Routbort et al. 1999). Vehicle- and KA-treated cultures were always stained concurrently under identical experimental conditions. Briefly, cultures were fixed with 1% sodium sulfide in 0.1 M PB containing 1 mM EDTA followed by 0.3% glutaraldehyde in PB. Cultures were then postfixed in 70% ethanol and developed in a solution of 0.085% silver nitrate, 1.7% hydroquinone, and 30% gum arabic (all wt/vol) in 0.2 M citrate buffer. Finally, cultures were mounted onto subbed glass slides, dehydrated, cleared, and coverslipped.

NEUROBIOTIN. Individual neurons were visualized and filled with neurobiotin using whole cell recording techniques, as described previously (Bausch and McNamara 2000). Neurobiotin (0.4 or 0.5%; Vector) was added to the pipette solution immediately before use. After diffusion of the neurobiotin-containing solution into the neuron for 20–45 min, cultures were fixed overnight with 4% paraformaldehyde in 0.1 M PB, removed from the insert membrane, sunk in 30% sucrose in 0.1 M PBS (PB containing 0.15 M NaCl and 2.7 mM KCl [pH 7.4]) and stored frozen at –70°C. Briefly, thawed cultures were treated with PBS containing 10% methanol and 0.6% H₂O₂, blocked with PBS containing 2% bovine serum albumin (BSA) and 0.75% Triton X-100 and incubated in ABC elite (Vector) diluted in PBS containing 2% BSA and 0.1% Triton X-100 according to kit instructions overnight at 4°C. Cultures were then treated with 0.05% 3,3'-diaminobenzidine (DAB, Sigma), 0.028% CoCl_2, 0.02% nickel ammonium sulfate, and 0.00075% H₂O₂ in PBS until staining was evident. Cultures were then mounted onto subbed glass slides, dehydrated, cleared in xylenes, and coverslipped. Camera lucida reconstructions were drawn using a Zeiss Axioskop microscope at 250 x magnification. Regions were defined as described above.

Statistical analysis

All data analyses were performed with investigators blinded to experimental groupings. Numbers and error bars represent the means ± SE in the stated number of slice cultures except where otherwise stated. All statistical analysis was performed with Sigma Stat software. Data fitting a nonparametric distribution were tested for significance using the Kruskal–Wallis ANOVA by ranks test with Dunn's post hoc comparison when comparing multiple groups or Mann–Whitney rank sum test when comparing 2 experimental groups. Data fitting a normal parametric distribution were tested for significance using an ANOVA with least significance difference (LSD) or Holm–Sidak post hoc comparisons when comparing multiple groups or a t-test when comparing 2 experimental groups. The chi-square test was used to test for a significant difference in spontaneous EPSC amplitude frequencies. Spearman rank order correlation was used to test for correlations between toluidine blue staining and electrophysiological measures.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
In vitro model of temporal lobe epilepsy and mossy fiber sprouting

As described previously (Routbort et al. 1999), organotypic hippocampal slice cultures isolated at P11, treated at 10 DIV with KA for 48 h and then maintained long term (40–60 days) in vitro exhibited complete loss of CA3a/b pyramidal cells, loss of CA3c pyramidal cells in a subset of cultures, but minimal loss of dentate granule cells (Fig. 1B). KA-treated cultures also displayed robust sprouting of mossy fibers into the inner dentate molecular layer as shown by Timm stain (Fig. 1D) and quantitative analysis of single granule cells filled with neurobiotin (Routbort et al. 1999). Vehicle-treated cultures showed intact principal cell layers (Fig. 1A) and a relative paucity of mossy fiber sprouting (Fig. 1C). To begin to investigate the functional significance of mossy fiber sprouting and the potential role of alternative hippocampal pathways in granule cell hyperexcitability and seizures, electrophysiological and histological comparisons between vehicle- and KA-treated slice cultures were conducted. All experiments on KA- and vehicle-treated cultures were performed concurrently, including our previously published results from vehicle-treated cultures (Bausch and McNamara 2000).

View larger version (85K): [in this window] [in a new window]   FIG. 1. Kainic acid (KA) treatment in vitro caused cell loss and mossy fiber sprouting. Vehicle-treated (A, C) and KA-treated (B, D) hippocampal slice cultures were stained with either toluidine blue (A, B) or Timm stain (C, D) about 50 days after treatment [60 days in vitro (DIV)] as described in the METHODS. Toluidine blue staining revealed (A) a normal distribution of pyramidal cells and dentate granule cells in vehicle-treated cultures, but (B) a loss of CA3a/b pyramidal cells and a slight decrease in the number of granule cells in KA-treated cultures. CA3c pyramidal cells were spared in this representative culture and in about 50% of KA-treated cultures. Timm staining revealed (C) a relative paucity of zinc-positive mossy fibers in the inner molecular of vehicle-treated cultures. D: mossy fiber sprouting into the inner molecular layer (arrows) was increased after KA treatment. DG, dentate gyrus; g, dentate granule cell layer; h, hilus; m, molecular layer. Scale bar in A is for A and B, 500 µm; that in C is for C and D, 100 µm.

SPONTANEOUS SEIZURES. Our previous work in vehicle-treated long-term hippocampal slice cultures (40–60 DIV) documented spontaneous electrographic seizures involving granule cells in physiological buffer and after GABA_A receptor blockade with the antagonist BMI (Bausch and McNamara 2000). Electrographic seizures showed similarities to electroencephalographic (EEG) waveforms that occur during tonic–clonic motor seizures in animals (Traynelis and Dingledine 1988). Therefore in the first series of experiments, field potential recordings from the dentate granule cell layer were conducted in an effort to detect possible alterations in spontaneous seizures in KA-treated cultures. Electrographic seizures were defined as a burst of rhythmic activity 3 s in duration which evolved over time and exhibited an abrupt onset and an abrupt termination (Bausch and McNamara 2000). An increase in spontaneous seizure incidence, frequency, or duration in KA-treated compared with vehicle-treated cultures would suggest that mossy fiber sprouting contributes to exacerbation of seizures.

Physiological Buffer.    Recordings conducted for 46 ± 2 min in physiological buffer disclosed spontaneous electrographic seizures in 29% of KA-treated slice cultures (5 of 17 cultures, Fig. 2A). The mean seizure duration was 201 ± 200 s. These results are not statistically different from data reported previously for vehicle-treated long-term hippocampal slice cultures, which showed a single electrographic seizure in 22% of vehicle-treated cultures (2 of 9 cultures) in physiological buffer and a mean seizure duration of 32 s (Bausch and McNamara 2000). However, statistical analysis on seizure duration was confounded by the incidence of only two seizures in vehicle-treated cultures.

View larger version (27K): [in this window] [in a new window]   FIG. 2. Seizures were observed in the granule cell layer of hippocampal slice cultures. Extracellular field potentials were recorded in the dentate granule cell layer of a KA-treated hippocampal slice culture (60 DIV) in both (A) physiological recording buffer and (B) after application of BMI, (10 µM) as described in the METHODS. A1: trace from a representative recording illustrates the transient high-frequency spiking observed in the granule cell layer of 29% of KA-treated slice cultures in physiological buffer. A2: an expanded time scale of the high-frequency spiking shown in A1 reveals rhythmic spiking with an abrupt onset superimposed on a flat baseline. B1: trace from a representative recording shows the high-frequency spiking observed in the granule cell layer in 65% of KA-treated slice cultures during wash-in of BMI (note difference in scale from A). These seizures displayed both "tonic" and "clonic" phases (Traynelis and Dingledine 1988). B2: an expanded time scale of B1 reveals the rhythmic spiking with an abrupt onset superimposed on a positive field potential shift, which represents the "tonic" phase of the seizure. B3: another expanded time scale of B1 illustrates the epileptiform bursting that represents the "clonic" phase of the seizure. Vertical scale bar in A1 applies to all traces in A; in B1 applies to all traces in B. Horizontal scale bar in B3 applies to B2 and B3.

HILAR-EVOKED FIELD POTENTIAL RESPONSES. Next, the responses of granule cells to hilar stimulation were recorded in the granule cell layer in both physiological buffer and BMI to investigate whether the excitability of the granule cells themselves was altered in KA-treated cultures. An increase in the incidence of excitatory postsynaptic potentials (EPSPs) or multiple population spikes in KA-treated compared with vehicle-treated cultures would suggest that mossy fiber sprouting enhances granule cell excitability.

PHYSIOLOGICAL BUFFER. In physiological buffer, hilar stimulation in KA-treated cultures (n = 25) elicited an antidromic population spike followed by: a short-latency EPSP (44% of cultures), a second population spike (35%), multiple population spikes (6%), or an isolated antidromic population spike (15%) (Fig. 3A). These results are not statistically different from data reported previously for vehicle-treated slice cultures (short-latency EPSP, 55%; second population spike, 21%; multiple population spikes, 24%; n = 28; Bausch and McNamara 2000). However, the short-latency EPSP in KA-treated cultures was decreased by 75% in amplitude and 20% in duration when compared with vehicle-treated cultures (n = 28; Bausch and McNamara 2000) (ANOVA with LSD post hoc comparison; P < 0.05), which most likely reflects partial loss of afferent inputs and slight loss of dentate granule cells after KA treatment (Routbort et al. 1999).

BMI. The presence of BMI during hilar stimulation in KA-treated cultures (n = 21) increased the percentage of cultures exhibiting multiple population spikes to 98%, increased the mean short-latency evoked EPSP duration by 6-fold, and elicited longer latency hilar-evoked EPSPs (evoked epileptiform bursting, Fig. 3B, KA-BMI) in 96% of slice cultures. Again, these results are not statistically different from previous data for vehicle-treated long-term hippocampal slice cultures (multiple population spikes in 86% of cultures; 8-fold increase in short-latency evoked EPSP duration; evoked epileptiform bursting, in 92% of cultures; n = 21; Bausch and McNamara 2000). Although the mean short-latency EPSP amplitude and duration were both decreased by 39% in KA-treated cultures, this difference was not significant.

These findings show that granule cell excitability was not increased in KA-treated compared with vehicle-treated cultures. Furthermore, together with seizure data, these findings show that GABAergic inhibition was intact in KA-treated hippocampal slice cultures and that granule cell excitability was under GABAergic control.

EPSPs in KA-treated cultures required glutamatergic transmission

Our previous study in vehicle-treated hippocampal slice cultures demonstrated that hilar-evoked and spontaneous EPSPs required glutamatergic transmission mediated by both NMDA and AMPA/KA type glutamate receptors (Bausch and McNamara 2000). In KA-treated cultures, hilar-evoked and spontaneous EPSPs were abolished by a combination of D-APV and NBQX or CNQX and individually the effects of the NMDA receptor antagonist, D-APV (n = 3 slice cultures), and the AMPA/KA receptor antagonist, NBQX (n = 10 slice cultures) on hilar-evoked and spontaneous EPSP amplitude, duration, shape, number, and frequency were not statistically different from data previously reported for vehicle-treated hippocampal slice cultures (data not shown). These findings demonstrate a requirement for both NMDA and AMPA/KA receptor activation and suggest that the relative contributions of these receptors to hilar-evoked and spontaneous EPSPs were not significantly altered in KA-treated compared with vehicle-treated hippocampal slice cultures.

Thus many of our initial findings from KA-treated cultures were indistinguishable from those reported previously for vehicle cultures (Bausch and McNamara 2000). These findings suggest that increased mossy fiber sprouting does not exacerbate electrographic seizures or granule cell excitability in hippocampal slice cultures or that the physiological effects of a recurrent mossy fiber pathway were masked by more powerful pathways. However, several notable differences between vehicle- and KA-treated cultures were observed.

Granule cell firing properties were altered in KA-treated hippocampal slice cultures

In the presence of BMI, hilar stimulation evoked a short-latency EPSP and longer latency EPSPs in both KA- and vehicle-treated cultures. However, blinded analyses revealed small asynchronous spikes (Fig. 3B, KA-BMI, inset) in 75% of cultures and large synchronous population spikes superimposed on longer-latency EPSPs in 25% of KA-treated cultures. This finding is opposite to observations from vehicle-treated slice cultures, where large synchronous population spikes were superimposed on longer-latency EPSPs in 75% of cultures (Fig. 3B, vehicle-BMI) and small asynchronous spikes were seen in 25% of cultures (P 0.001, Mann–Whitney rank sum test). Similar results were found for spontaneous epileptiform bursts of EPSPs (Fig. 3C and data not shown). These data suggest 4 possibilities. First, that after KA treatment granule cells no longer fire synchronously to produce a population spike, or second, that the slight loss of granule cells decreased population spike amplitude. These explanations are unlikely because, in KA-treated cultures, asynchronous small spikes were seen on longer-latency EPSPs even when large population spikes were superimposed on the short-latency EPSP. The remaining possibilities are that after KA treatment, granule cell membrane properties were altered or granule cells received more asynchronous excitatory synaptic inputs.

To examine granule cell membrane properties, whole cell current-clamp recordings were conducted within 2 min of establishing whole cell configuration. Granule cell resting membrane potential (RMP), input resistance (R_in), and action potential (AP) threshold were not significantly altered in KA-treated (RMP = –64 ± 5 mV, n = 13; R_in = 153 ± 17 M, n = 8; AP threshold = –49 ± 3.5 mV, n = 7) compared with vehicle-treated cultures (from Bausch and McNamara 2000: RMP = –66 ± 8 mV, n = 10; R_in = 135 ± 11 M, n = 11; AP threshold = –44 ± 1.2 mV, n = 9). However, the number of action potentials elicited by a 200pA input was increased by 150% in KA-treated (AP = 7.5, n = 8) compared with vehicle-treated slice cultures (AP = 3, n = 10; P < 0.01, Mann–Whitney rank sum test). Such a change should render granule cells more likely to fire action potentials in KA-treated cultures and thus alone cannot account for the decreased incidence of large-population spikes on longer-latency EPSPs.

Amplitude distribution of spontaneous EPSCS in KA-treated cultures was shifted

To investigate the possibility that granule cells receive more asynchronous excitatory synaptic inputs after KA treatment, excitatory postsynaptic currents (EPSCs) were measured in granule cells using whole cell voltage-clamp recordings. EPSCs were isolated from inhibitory postsynaptic currents (IPSCs) by recording in the presence of the GABA_A receptor antagonist BMI.

VEHICLE-TREATED CULTURES. EPSCs were recorded first in vehicle-treated cultures to provide a baseline for comparison with KA-treated cultures. Granule cells in vehicle-treated cultures received spontaneous EPSCs (sEPSCs) with a frequency of 0.7 ± 0.2 Hz that fell into 2 amplitude distributions (Fig. 4A, vehicle). Of all sEPSCs, 82% were <200 pA (small); 18% were >600 pA (large) (Fig. 4A, left). Small-amplitude sEPSCs displayed a fast rise and decay with a single peak (Table 1 and data not shown). Large-amplitude sEPSCs (Fig. 4B1) showed a steep rise and prolonged decay (Table 1), with 0–15 large peaks (median, 3) superimposed on the decay. Large-amplitude sEPSCs were similar in shape to both hilar-evoked EPSCs (compare Fig. 4B1 with Fig. 4B3, vehicle) and field EPSPs (compare Fig. 4B1 with Fig. 3, B and C, vehicle).

View larger version (30K): [in this window] [in a new window]   FIG. 4. Distribution of spontaneous excitatory postsynaptic current (sEPSC) amplitudes was shifted in granule cells from KA-treated cultures. Spontaneous EPSCs and hilar-evoked EPSCs were recorded in the presence of BMI (10 µM) in dentate granule cells of vehicle- and KA-treated slice cultures using whole cell voltage clamp as described in the METHODS. Spontaneous EPSC amplitudes were measured from baseline to peak negativity and divided into 25 pA bins. A: amplitude histograms of sEPSCs showed that granule cells in both vehicle- and KA-treated cultures received small (200 pA) and very large (>600 pA) sEPSCs. However, KA-treated cultures received significantly fewer small sEPSCs. (Note difference of y-axis scale in vehicle compared with KA.) Midamplitude sEPSCs (201–600 pA) were observed only in granule cells from KA-treated cultures. B: representative traces show that very large spontaneous (B1 and B2) and evoked EPSCs (B3) exhibited a steep rise and a slow decay with multiple large peaks superimposed on the decay. C: representative traces show the shape of the midamplitude spontaneous (C1 and C2) and evoked EPSCs (C3) in KA-treated cultures. Stimulation artifacts were removed in B3. *, significantly different from vehicle (chi-square test, P < 0.001). Number of cells is indicated in parentheses. Scale bars in B1 apply to B1 and B2 and those in C1 apply to C1 and C2.

Despite these similarities, alterations in sEPSCs <600 pA were observed in KA-treated cultures. The proportion of small-amplitude (<200 pA) sEPSCs was significantly decreased; area, half-width, and decay of small-amplitude sEPSCs were significantly increased and sEPSCs in the 200–600 pA (midamplitude) range appeared in KA-treated cultures (Table 1, Fig. 4A; note difference of y-axis scale in vehicle compared with KA). Midamplitude sEPSCs were not observed in vehicle-treated cultures but were similar in shape to hilar-evoked EPSCs in the same KA-treated cultures (compare Fig. 4C1 with Fig. 4C3). The rise time for midamplitude sEPSCs was significantly greater than for small-amplitude EPSCs; decay times fell between those of small- and large-amplitude sEPSCs (Table 1). Taken together, these data suggest increased numbers of active synapses, increased numbers of summated EPSCs, and fewer single events or altered receptor properties at existing synapses. However, although midamplitude sEPSCs showed a single peak (Fig. 4C2) in 50% of KA-treated cultures, the other 50% exhibited midamplitude sEPSCs with multiple small peaks on the rise and decay (Fig. 4C1), suggestive of multiple asynchronous excitatory synaptic inputs or asynchronous glutamate release from different sites on the same synaptic bouton (Jonas et al. 1993). This asynchronous activity could account for the greater percentage of field potential recordings from KA-treated cultures showing asynchronous small spikes superimposed on longer-latency EPSPs.

Synaptic contributions from multiple subfields contribute to granule cell responses

To investigate possible sources of synaptic contributions to granule cell responses, anatomical labeling of individual neurons and electrophysiological recordings combined with knife cuts were performed.

LABELING. Our previous work in vehicle-treated cultures documented reciprocal axonal projections between individual CA1 pyramidal cells and dentate granule cells (Bausch and McNamara 2000). Changes in the prevalence of reciprocal connections could alter granule cell responses, so single CA1 pyramidal cells and dentate granule cells were filled with neurobiotin. In KA-treated cultures, 25% (2/8) of granule cells sent axon collaterals to str. lacunosum-moleculare of CA1 and 33% (5/15) of CA1 pyramidal cells sent axon collaterals to the dentate gyrus. These findings are similar to previously reported results from vehicle-treated cultures, in which 20% of granule cells sent axon collaterals to str. lacunosum-moleculare of CA1 and 22% of CA1 pyramidal cells sent axon collaterals to the dentate gyrus (Bausch and McNamara 2000). These data show that reciprocal projections between CA1 and the dentate gyrus are present but not increased in KA-treated slice cultures. However, further analyses of CA1 pyramidal cells did document a significant 2-fold increase in total axonal length of CA1 pyramidal cells in KA-treated compared with vehicle-treated cultures (Fig. 5, A and B), a change most prominent in str. pyramidale (Fig. 5, A and C). Axonal sprouting of CA1 pyramidal cells in acute slices from epileptic rats has been associated with increased excitability of CA1 pyramidal neurons (Ashwood et al. 1986; Esclapez et al. 1999; Franck and Schwartzkroin 1985; Lancaster and Wheal 1984; Meier and Dudek 1996; Meier et al. 1992; Perez et al. 1996; Smith and Dudek 2001, 2002; Wu and Leung 2003), raising the possibility that hyperexcitability in CA1 could contribute to asynchronous granule cell responses in KA-treated cultures.

View larger version (38K): [in this window] [in a new window]   FIG. 5. CA1 pyramidal cells in KA-treated cultures show robust axonal sprouting. Single CA1 pyramidal cells in vehicle- and KA-treated slice cultures were filled with neurobiotin using whole cell techniques as described in the METHODS. A: camera lucida drawings show representative CA1 pyramidal cells from (top) vehicle- and (middle) KA-treated hippocampal slice cultures. Bottom: illustrates the most sprouted CA1 pyramidal cell from a KA-treated culture. Axons are shown as thin black lines. Somata and dendrites are illustrated with thick blue lines. B and C: total axonal length (B) and axonal length within CA1 regions (C) were measured from camera lucida drawings of neurobiotin-filled CA1 pyramidal cells (n = 7 vehicle; n = 13 KA). gc, granule cell layer; so, str oriens; sp, str pyramidale; sr, str radiatum. *, Different from vehicle (ANOVA with LSD post hoc comparison, P < 0.05); **, different from vehicle (t-test, P < 0.01). Scale bar in A is for all drawings, 200 µm.

In the presence of BMI, hilar stimulation evoked a short-latency EPSP and longer-latency EPSPs (Fig. 6B) in 96% of uncut KA-treated cultures (Table 2). The CA3 cut did not significantly alter the hilar-evoked short-latency EPSP (Fig. 6, C and D) or the percentage (90%) of cultures showing evoked longer-latency EPSPs (Table 2), consistent with the elimination of CA3a/b pyramidal cells by KA treatment. In contrast, the CA1 cut and the combination CA1/CA3 cut (DG cut) virtually abolished hilar-evoked short-latency EPSP duration (Fig. 6C) and amplitude (Fig. 6D) and decreased the percentage of cultures exhibiting hilar-evoked longer-latency EPSPs to 22 and 9%, respectively (Table 2). These data suggest that functional synaptic connections from CA1 contribute prominently to hilar-evoked granule cell EPSPs in KA-treated hippocampal slice cultures. Furthermore, these findings stand in sharp contrast to those reported previously for vehicle-treated cultures, in which CA1 and combination CA1/CA3 cuts (DG cut) decreased short-latency EPSP duration and amplitude by only about 75 and 40%, respectively (Bausch and McNamara 2000), and decreased the percentage of cultures exhibiting hilar-evoked longer-latency EPSPs by 8 and 20%, respectively (Table 2). These differences indicate that the principal source of excitatory afferents mediating the hilar-evoked EPSPs and action potentials of the dentate granule cells arise from CA1 in KA-treated cultures and from neurons intrinsic to the dentate gyrus/CA3c region in vehicle-treated cultures. It seems likely that the afferents in dentate gyrus/CA3c region responsible for responses in vehicle-treated cultures were destroyed by KA.

View larger version (13K): [in this window] [in a new window]   FIG. 7. Negative field potential shift observed after the CA1 and DG cut in KA-treated cultures was mediated predominantly by NMDA receptors. Hilar-evoked extracellular field potential responses were recorded in the dentate granule cell layer of CA1 and DG cut KA-treated slice cultures as described in the METHODS. Knife cuts were performed 45 min to 1 h before recording, as illustrated in Fig. 6A. A distinct negative shift in potential was unmasked after GABAA receptor blockade with BMI (10 µM). Further addition of the NMDA receptor antagonist, D-APV (50 µM), to recording buffer containing BMI almost completely abolished this negative shift. Stimulation artifacts were removed from all traces. Scale bars apply to all traces.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

Similarities between KA-treated hippocampal slice cultures and in vivo models and human temporal lobe epilepsy

One question arising from this study is whether morphological alterations and granule cell hyperexcitability observed in long-term KA-treated hippocampal slice cultures share similarities with changes associated with limbic epilepsy in animal models or humans.

MORPHOLOGY. We showed previously that KA application to hippocampal slice cultures induced seizures, cell death, and mossy fiber sprouting, similar to that seen after KA administration to rodents in vivo (Routbort et al. 1999). We now extend this finding to include robust sprouting of CA1 pyramidal cell axons. The magnitude of KA-induced CA1 pyramidal cell axonal sprouting in slice cultures was consistent with the approximate doubling of total axonal length in single CA1 pyramidal cells filled with neurobiotin in acute slices from epileptic KA or pilocarpine treated rats when compared with controls (Esclapez et al. 1999; Perez et al. 1996; Smith and Dudek 2001). However, total axonal length for CA1 pyramidal cells in vehicle-treated cultures was 2.7 and 9 times greater than that reported by Esclapez et al. (1999) and Smith and Dudek (2001), respectively, for slices from control rats, suggesting that slice culture preparation alone also elicits axonal sprouting of CA1 pyramidal cells.

PHYSIOLOGY. One notable difference between KA-treated hippocampal slice cultures and acute slices isolated from animal models or humans with temporal lobe epilepsy is the incidence of seizures involving granule cells. In our study, spontaneous seizures were observed in 29% of cultures in physiological buffer and in 65% of cultures in the presence of the GABA_A receptor antagonist BMI. These findings stand in contrast to those of other investigators who have not reported seizures involving granule cells in slices isolated acutely from epileptic rat or resected human epileptic hippocampus (Cronin et al. 1992; Franck et al. 1995; Isokawa et al. 1991, 1997; Masukawa et al. 1991; Molnar and Nadler 1999; Okazaki et al. 1999; Tauck and Nadler 1985; Urban et al. 1990; Williamson et al. 1995) even in the presence of GABA_A receptor antagonists. However, our findings are similar to those of Dudek and colleagues (Patrylo and Dudek 1998; Wuarin and Dudek 1996) who detected seizures in about 70% of slices acutely isolated from KA-treated epileptic rats when slices were incubated in a GABA_A receptor antagonist with elevated extracellular K⁺ (6–9 mM). Unlike seizures recorded by Dudek and collegues in acute slices, however, seizures in slice cultures were not associated with a dramatic DC potential shift.

Aside from seizures, granule cell hyperexcitability in KA-treated hippocampal slice cultures shares many similarities with granule cell hyperexcitability in acute slices isolated from epileptic rat and resected human epileptic hippocampus, including burst discharges, alterations in sEPSCs, and negative field potential shifts. Like findings reported here, in physiological buffer hilar stimulation elicits multiple population spikes in a small percentage of hippocampal slices isolated from human limbic epilepsy patients and rats made "epileptic " with kainate or pilocarpine. A larger percentage of slices displays burst discharges after application of a GABA_A receptor antagonist or elevation of extracellular potassium (Cronin et al. 1992; Franck et al. 1995; Hardison et al. 2000; Masukawa et al. 1992; Okazaki and Nadler 2001; Patrylo and Dudek 1998; Tauck and Nadler 1985; Williams et al. 2002; Wuarin and Dudek 1996). Similar to our results, 200 pA–1.7 nA sEPSCs and increased sEPSC amplitude distribution compared with controls have been recorded in granule cells of acute slices from KA- and pilocarpine-treated rats (Okazaki et al. 1999; Simmons et al. 1997; Williams et al. 2002; Wuarin and Dudek 2001) and both unimodal and complex EPSCs with multiple peaks have been evoked in granule cells in a subset of slices from pilocarpine-treated rats. Also consistent with our data, negative field potential shifts have been evoked in the granule cell layer in acute slices isolated from resected human epileptic hippocampus (Isokawa and Fried 1996; Masukawa et al. 1992) and KA-treated rats (Patrylo and Dudek 1998, 1999; Williams et al. 2002; Wuarin and Dudek 1996), which could be abolished with a combination of NMDA and AMPA/KA receptor antagonists (Isokawa and Fried 1996; Patrylo and Dudek 1998). Thus many of the physiological features of granule cells in KA-treated hippocampal slice cultures also have been associated with limbic epilepsy in animal models or humans.

Differences between KA-treated hippocampal slice cultures and in vivo models and human temporal lobe epilepsy

Also of major interest are the differences between long-term KA-treated hippocampal slice cultures and acute slices isolated from animal models or humans with limbic epilepsy.

MORPHOLOGY. One of the most striking differences in hippocampal slice cultures compared with acute slices is the extent of synaptic connectivity. We reported previously on the profound expansions of normal pathways in hippocampal slice cultures (Bausch and McNamara 2000). Similarly, Gahwiler et al. (1997) and Pavlidis and Madison (1999) have estimated that normal synaptic connections in hippocampal slice cultures are increased 10-fold relative to acute slices. However, in addition to making normal synapses, neurons in slice culture also may form aberrant synapses, which may include the reciprocal CA1 pyramidal cell to dentate granule cell synapse suggested by our results in both vehicle- and KA-treated slice cultures. This extensive synaptic reorganization in hippocampal slice cultures compared with acute slices provides a complex backdrop for the study of the physiological consequences of mossy fiber sprouting caused by KA treatment.

PHYSIOLOGY. As expected, we documented here that granule cells in KA-treated long-term hippocampal slice cultures were hyperexcitable compared with acute slices from normal rats. Hyperexcitability was manifested as spontaneous seizures, interictal bursting, and large-amplitude sEPSCs, which do not occur in granule cells in acute slices from normal rats (Cronin et al. 1992; Okazaki et al. 1999; Patrylo and Dudek 1998; Simmons et al. 1997; Wuarin and Dudek 1996). Surprisingly, despite increased mossy fiber sprouting and increased numbers of action potentials evoked by depolarizing currents in KA-treated compared with vehicle-treated cultures, granule cell hyperexcitability in KA- and vehicle-treated cultures was similar. This lack of a difference between vehicle- and KA-treated cultures in the incidence of seizures, epileptiform bursts, multiple population spikes, very large amplitude sEPSCs, and the requirement of glutamatergic transmission raises the possibility that common synaptic sources are responsible for each. The absence of CA3a/b pyramidal cells in KA-treated cultures precludes CA3a/b and the paucity of mossy fiber sprouting in vehicle-treated cultures suggests a nonmossy fiber pathway. Based on our anatomical and physiological data using knife cuts in KA- and vehicle-treated slice cultures (Bausch and McNamara 2000; this study), neurons intrinsic to the dentate gyrus/CA3c or in CA1 are likely sources of synaptic inputs. Correlations between granule cell responses after knife cuts and neuronal survival in KA-treated cultures suggest that synaptic networks incorporating both CA3c pyramidal neurons and granule cells, not mossy fiber sprouting itself, are critical to hilar-evoked longer-latency synaptic responses. Thus circuits including CA1 and CA3c pyramidal cells appear to be more important than mossy fiber sprouting for hyperexcitability in KA-treated hippocampal slice cultures. This stands in contrast to acute slices isolated from animal models or humans with limbic epilepsy, where mossy fiber sprouting has been proposed to predominate. However, regardless of the relative importance of a single pathway, findings from both acute hippocampal slices and hippocampal slice cultures strongly support the overall concept that axonal sprouting and synaptic reorganization can contribute to abnormal synchronous activity.

Potential synaptic contributions to granule cell hyperexcitability in the epileptic hippocampus

The presence of seizures, epileptiform activity, sEPSC amplitude distribution shifts, and negative field potential shifts in tissue isolated from humans or animal models with limbic epilepsy and mossy fiber sprouting has led to the hypothesis that a recurrent granule cell network formed by sprouted mossy fibers contributes to these events. In support of this idea, numerous studies have provided evidence for a functional recurrent granule cell network after mossy fiber sprouting. Most notably, a greater number of EPSP/Cs in single granule cells is observed after stimulation of adjacent granule cells with glutamate microdrops or laser-evoked release of caged glutamate in acute slices isolated from rats treated with KA, pilocarpine, or kindling paradigms (Lynch and Sutula 2000; Molnar and Nadler 1999; Williams et al. 2002; Wuarin and Dudek 1996, 2001). However, although the importance of synaptic glutamatergic transmission in seizures, epileptiform activity, sEPSCs, and negative field potential shifts is well documented, the extent to which synaptic connections of sprouted mossy fibers per se contributes to this hyperexcitability is unclear. Our data in hippocampal slice cultures suggest that synaptic connections from CA1, nongranule cell neurons intrinsic to the dentate gyrus and potentially from CA3, can all contribute to granule cell hyperexcitability. The extent to which alterations in CA1 pyramidal cells may contribute to granule cell hyperexcitability in humans is unclear, given that a projection from CA1 to the dentate gyrus has not been documented in humans and the CA1 region is most susceptible to damage in the human epileptic hippocampus. However, recent evidence suggests that circuits involving hilar mossy cells and CA3 pyramidal cells do survive in the sclerotic hippocampus of humans and animal models (Eid et al. 2002; Scharfman et al. 2001) and electrophysiological recordings by Scharfman (1993, 1994a,b) in acute hippocampal slices isolated from normal rats have provided strong evidence for a polysynaptic projection from CA3 pyramidal cells to dentate granule cells by a hilar "mossy" cell intermediate. Thus alterations in a CA3 pyramidal cell—mossy cell—granule cell circuit also may contribute to granule cell hyperexcitability in the human epileptic hippocampus.

In summary, although a majority of research on granule cell hyperexcitability in the epileptic hippocampus has revolved around the recurrent excitatory granule cell network formed by sprouted mossy fiber axons, the results from the present study support the idea that synaptic reorganization in alternative networks can also influence hippocampal function and contribute to dentate granule cell hyperexcitability. Indeed, numerous synaptic rearrangements have been associated with limbic seizures, including sprouting of granule cells (Houser et al. 1990; Mello et al. 1992; Nadler et al. 1980; Sutula et al. 1988, 1989; Tauck and Nadler 1985), CA1 pyramidal cells (Bausch and McNamara 1998, 1999