Abstract

Sharp wave-ripples and interictal events are physiological and pathological forms of transient high activity in the hippocampus with similar features. Sharp wave-ripples have been shown to be essential in memory consolidation, whereas epileptiform (interictal) events are thought to be damaging. It is essential to grasp the difference between physiological sharp wave-ripples and pathological interictal events to understand the failure of control mechanisms in the latter case. We investigated the dynamics of activity generated intrinsically in the Cornu Ammonis region 3 of the mouse hippocampus in vitro, using four different types of intervention to induce epileptiform activity. As a result, sharp wave-ripples spontaneously occurring in Cornu Ammonis region 3 disappeared, and following an asynchronous transitory phase, activity reorganized into a new form of pathological synchrony. During epileptiform events, all neurons increased their firing rate compared to sharp wave-ripples. Different cell types showed complementary firing: parvalbumin-positive basket cells and some axo-axonic cells stopped firing as a result of a depolarization block at the climax of the events in high potassium, 4-aminopyridine and zero magnesium models, but not in the gabazine model. In contrast, pyramidal cells began firing maximally at this stage. To understand the underlying mechanism we measured changes of intrinsic neuronal and transmission parameters in the high potassium model. We found that the cellular excitability increased and excitatory transmission was enhanced, whereas inhibitory transmission was compromised. We observed a strong short-term depression in parvalbumin-positive basket cell to pyramidal cell transmission. Thus, the collapse of pyramidal cell perisomatic inhibition appears to be a crucial factor in the emergence of epileptiform events.

Introduction

Brain states are characterized by behaviour-associated coordinated alternation of distinct EEG patterns in different cortical regions (Sirota et al., 2003; Buzsaki, 2006; Isomura et al., 2006). For instance, cortical slow oscillations are the result of the alternation of low activity down states and high activity up states (Steriade et al., 1993a, b, 2001). A similar alternation of activity can be observed in the hippocampus during cycles of theta-embedded gamma oscillations (Soltesz and Deschenes, 1993; Bragin et al., 1995) and during physiological sharp wave-ripples (Buzsaki, 1986; Ylinen et al., 1995). This suggests that the generation of recurring transient high activity (and therefore synchronous) events is an inherent and general property of healthy cortical networks. Physiological sharp wave-ripples (that are different from pathological transient events observed in epileptic patients often referred to as sharp-waves by clinicians) can be considered to be their most simple manifestation in the hippocampus, and were shown to be important in memory consolidation (Girardeau et al., 2009; Jadhav et al., 2012). In the epileptic hippocampus, different pathological forms of transient high activity events, including interictal, pre-ictal or ictal events (referred to as epileptic events) can be observed and are considered damaging (Engel, 1996; Aldenkamp et al., 2005; Holmes and Lenck-Santini, 2006; Zhou et al., 2007).

Hippocampal slices can produce spontaneously-emerging in vivo-like sharp wave-ripples (Kubota et al., 2003; Ellender et al., 2010), whereas epileptic events can be induced upon pharmacological intervention [e.g. increasing excitability with high K+ (Moody et al., 1974; Traynelis and Dingledine, 1988), applying 4-aminopyridine (Rutecki et al., 1987; Louvel et al., 1994), decreasing or eliminating inhibition (Schwartzkroin and Prince, 1977; Traub and Wong, 1983; Hablitz, 1984; Miles et al., 1984, 1988) or omitting Mg2+ (Mody et al., 1987; Jones and Heinemann, 1988; Dreier and Heinemann, 1991)]. As previously observed, the firing patterns of different hippocampal neurons were found to be modified during epileptic events. Most neurons increased their firing frequency, but some cells became silent, likely as a result of a depolarization block during the pathological events (Kawaguchi, 2001; Bikson et al., 2003; Ziburkus et al., 2006; Cammarota et al., 2013).

In the present study, our aim was to clarify some basic differences between physiological and pathological transient high activity events. As in vitro hippocampal slices can generate several different forms of transient high activity events, including both sharp wave-ripples and interictal events, we induced transitions from sharp wave-ripples to interictal events by different epileptiform activity-inducing treatments to answer the following questions: (i) What is the phenomenological difference between physiological sharp wave-ripples and pathological interictal events? (ii) How do the same identified neurons behave during sharp wave-ripples and interictal events? and (iii) What are the underlying mechanisms resulting in the transition from the physiological to the pathological network state?

We found numerous differences in basic cellular and network parameters when comparing sharp wave-ripples and interictal events [primarily the collapse of parvalbumin-positive basket cell (PVBC)-mediated inhibition in the epileptiform event-producing state]. These changes lead to a reorganization of synchrony and neuronal firing patterns, and result in physiological sharp wave-ripples being replaced by interictal events.

Materials and methods

Animals were kept and used according to the regulations of the European Community’s Council Directive of 24 November 1986 (86/609/EEC). Experimental procedures were reviewed and approved by the Animal Welfare Committee of the Institute of Experimental Medicine, Hungarian Academy of Sciences, Budapest.

CD1 and Bl6 mice of both sexes (postnatal day 19–40) were used in the experiments. To measure selectively from cells containing the Ca2+ binding protein parvalbumin, transgenic mice expressing enhanced green fluorescent protein (eGFP) controlled by the parvalbumin promoter were also used in this study (Meyer et al., 2002). CCK (cholecystokinin)-expressing interneurons were sampled in slices prepared from CCK dsRed transgenic mice (Supplementary Table 1 and Supplementary Fig. 1; for experimental details of the characterization of this mouse strain see the Supplementary material), where the expression of red fluorescent protein was under the control of the CCK promoter. In all cases, the mice were decapitated under deep isoflurane anaesthesia. The brain was removed into ice cold cutting solution, which had been bubbled with 95% O2/5% CO2 (carbogen gas) for at least 30 min before use. For contents of solutions see Table 1. Horizontal hippocampal slices of 200- or 450-µm thickness were cut using a vibratome (Leica VT1000S or VT1200S), and slices were placed into an interface-type holding chamber for recovery. This chamber contained standard artificial CSF (Table 1) saturated with carbogen at 35°C that gradually cooled to room temperature. After incubation for at least 90 min, slices were transferred individually into a submerged-style recording chamber equipped with a dual superfusion system (Hajos et al., 2009) where slices were placed on a metal mesh and two separate fluid inlets allowed artificial CSF to flow both above and below the slices at a rate of 3–3.5 ml/min for each flow channel, at 32–34°C.

Table 1

Composition of extra-and intracellular solutions

Extracellular solutions (in mM)
 
          
 Sucrose NaCl KCl NaHCO3 CaCl2 MgCl2 NaH2PO4 Glucose    
Artificial CSF 126 3.5 26 1.6 1.2 1.25 10    
Cutting 205 2.5 26 0.5 1.25 10    

 
Intracellular solutions (in mM)
 
pH 7.39, osmolarity of 285 mOsm/l
 
     
 K-gluconate CsCl MgCl2 HEPES NaCl Mg-ATP ATP GTP Creatine phosphate QX-314 Biocytin 
 
Intra 1 110  40 0.3 0.20% 
Intra 2 80 (Cs-gluconate) 60 10  0.20% 
Intra 3 138 10 0.4 10 0.2 0.20% 
Extracellular solutions (in mM)
 
          
 Sucrose NaCl KCl NaHCO3 CaCl2 MgCl2 NaH2PO4 Glucose    
Artificial CSF 126 3.5 26 1.6 1.2 1.25 10    
Cutting 205 2.5 26 0.5 1.25 10    

 
Intracellular solutions (in mM)
 
pH 7.39, osmolarity of 285 mOsm/l
 
     
 K-gluconate CsCl MgCl2 HEPES NaCl Mg-ATP ATP GTP Creatine phosphate QX-314 Biocytin 
 
Intra 1 110  40 0.3 0.20% 
Intra 2 80 (Cs-gluconate) 60 10  0.20% 
Intra 3 138 10 0.4 10 0.2 0.20% 

Standard patch electrodes were used in all recording configurations (i.e. in whole-cell patch-clamp, loose-patch and field potential recordings). Pipette resistances were 3–6 MΩ when filled with the intrapipette solution (Table 1) or with artificial CSF.

Field recordings and neuronal firing

Data were recorded with a Multiclamp 700B amplifier (Molecular Devices). Local field potentials were monitored in stratum pyramidale of the CA3 area using artificial CSF-filled patch pipettes. For the recording of cell firing, individual neurons in CA3 were concomitantly recorded in loose-patch mode for ∼20–35 min. Neurons were identified visually using differential interference contrast microscopy (Olympus BX61W). Then, the pipette was withdrawn and whole-cell patch-clamp recordings were performed on the same cells with another pipette filled with intrapipette solution 1 (Table 1). Access resistance was in the range of 5–20 MΩ. Only recordings where the access resistance did not change >25% during the recording were included in the study. Postsynaptic potentials and action potentials were recorded in current clamp mode, by de- and hyperpolarizing cells to different membrane potentials (from −70 mV to −30 mV, 5 mV each step). The depolarization was carried out by applying a maintained current injection for 1.5–2 min for each step. The resting membrane potential was recorded immediately after break-in. Both field and unit recordings were low-pass filtered at 2 kHz using the built-in Bessel filter of the amplifier. Data were digitized at 6 kHz with a PCI-6042E board (National Instruments) using EVAN 1.3 software (courtesy of Prof. Istvan Mody, UCLA, CA), and were analysed offline with custom-made programs written in MATLAB 7.0.4 and Delphi 6.0 by AIG.

Multichannel local field potential recordings

The local field potential (concomitantly at different sites) was recorded with a laminar multi-electrode array placed on the surface of the hippocampal slice, parallel to the orientation of pyramidal cell dendrites spanning all hippocampal layers (24 channels, 50 µm inter-contact distance, Neuronelektród Kft.). We used a custom-made referential amplifier system (band-pass 0.1 Hz to 7 kHz) (Ulbert et al., 2001, 2004a). Signals were digitized with a 16 bit resolution analogue-to-digital converter (National Instruments) and recorded at 20 kHz sampling rate on each channel using a custom-made virtual instrument in LabView (National Instruments). Current source density calculations were made using the three-point formula smoothed by Hamming window (Ulbert et al., 2001). Results are depicted by heat maps using custom-made software.

Stimulation-evoked postsynaptic currents

To record stimulation-evoked currents, 200-µm thick slices were used to minimize spontaneous network activity. Evoked inhibitory- and excitatory postsynaptic currents were recorded in pyramidal cells at a holding potential of −70 mV. A stimulating electrode made of theta glass was placed into stratum radiatum to activate Schaffer collaterals or inhibitory fibres, or into the border of strata pyramidale and lucidum to evoke inhibition with a perisomatic origin. To record inhibitory and excitatory postsynaptic currents, intrapipette solutions 2 and 3 were used, respectively (Table 1). When recording inhibitory postsynaptic currents, the artificial CSF contained 10 µM NBQX (2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo[f]quinoxaline-2,3-dione) and 50 µM AP5 (2R-amino-5-phosphonovaleric) acid to block fast excitatory transmission; when excitatory postsynaptic currents were recorded, the borders of CA3a-b and CA3b-c were cut to decrease the network size and minimize network activity. Data were digitized at 6 kHz with a PCI-6042E board (National Instruments) using Stimulog software (courtesy of Prof. Zoltan Nusser, IEM, Budapest), and were analysed off-line using the Evan software.

Inhibitory synaptic transmission

For paired recordings, we used 200-µm thick slices to examine changes in perisomatic inhibition (in artificial CSF with normal versus high K+). For the presynaptic cell, intrapipette solution 3 was used; the postsynaptic cell was recorded with intrapipette solution 2. The artificial CSF contained 10 µM NBQX and 50 µM AP5 acid to block fast excitatory transmission and exclude epileptic events that would interfere with the measurement of transmission. Presynaptic interneurons were held in current clamp mode around a membrane potential of −50 mV, and stimulated by a train of 30 action potentials at 150 Hz followed by four action potentials at 300 Hz (similar to the firing pattern recorded in loose-patch mode), 2.5–3.5 nA. Pyramidal cells were clamped at a holding potential of −50 mV (to mimic the depolarized state in elevated K+). Series resistance was frequently monitored; cells for which the series resistance changed >25% during recording were discarded from further analysis.

Statistics

Throughout the manuscript we applied non-parametric tests as data usually did not show a normal distribution. Statistical tests used were the following: Wilcoxon paired test, Mann-Whitney U-test, Kruskal-Wallis ANOVA, Friedman ANOVA and Kolmogorov-Smirnov test.

Further information on data analysis is available in the Supplementary material.

Results

Sharp wave-ripple-generating states can be switched into epileptiform event-generating states using four different epileptogenic treatments

Physiological sharp wave-ripples are spontaneously generated in 450-µm thick mouse (postnatal days 20–40) hippocampal slices in artificial CSF (Hajos et al., 2009) with features matching sharp wave-ripples recorded in vivo (see Supplementary material and Supplementary Fig. 2). With four different epileptogenic treatments: high K+ (8.5 mM, n = 86), 4-aminopyridine (30 µM, n = 8), 0 Mg2+ (n = 19) or gabazine (2 µM, n = 23), we could evoke transitions from the sharp wave-ripple-generating state to epileptiform activity-generating states (Fig. 1). As shown in Fig. 1A, the elevation of extracellular K+ gradually eliminated sharp wave-ripples and evoked a state characterized by a featureless EEG (transitory phase), followed by a state with recurring epileptiform events (Fig. 1) defined as large amplitude interictal-like events accompanied by high multi-unit activity. Both sharp wave-ripples and interictal events could be recorded in hippocampal CA3 minislices in high K+ after cutting off the dentate gyrus and area CA1, indicating that the CA3 region on its own can generate these two types of network activities (n = 4, not shown).

Figure 1

Transition from sharp-wave ripples to epileptiform events and their differences. (A) Interictal events were induced by elevating extracellular K+ concentration. A highly active, desynchronized state separates the physiological, transiently highly active sharp wave-ripple state from the pathological, transiently highly active interictal events. Note that an increase of multi-unit frequency precedes interictal events. Upper traces: local field potential (LFP) of the transitory phase. Lower traces: plot of multi-unit frequency demonstrating network activity during the transition. (B) Enlarged image of a sharp wave-ripple and accompanying multi-unit (MU) activity at two time scales, as well as the transitory phase and interictal events and the underlying multi-unit activities appearing in four different epilepsy models. (C) Physiological and pathological transient high activity events are plotted to compare their correlation. A negative correlation is present in the rate of amplitude of the two types of events (n = 26). (D) The rate of sharp wave-ripples (SWRs) and interictal events (IIEs) at three different phases of the transition from several experiments. During control conditions sharp wave-ripple rate is high [and epileptiform event (EE) rate is zero], but after the pharmacological intervention it disappears completely. After the transitory phase (when the rate of both sharp wave-ripples and interictal events is zero) the rate of interictal events starts to increase, whereas sharp wave-ripple rate remains at zero. Note that for slices able to generate both sharp wave-ripples and interictal events, the interictal events cannot be seen until sharp wave-ripples disappear completely (n = 25). Sharp wave-ripples (E) and early interictal events (F) may seem similar (see insets); however, compared to sharp wave-ripples, interictal events (even early ones) have a larger amplitude, a longer duration (top trace) and are accompanied by several-fold higher multi-unit activity (bottom trace). Analysis of current source densities (centre) showed only minor shifts in the organization of sinks (red) and sources (blue), especially in the second half of an event.

Figure 1

Transition from sharp-wave ripples to epileptiform events and their differences. (A) Interictal events were induced by elevating extracellular K+ concentration. A highly active, desynchronized state separates the physiological, transiently highly active sharp wave-ripple state from the pathological, transiently highly active interictal events. Note that an increase of multi-unit frequency precedes interictal events. Upper traces: local field potential (LFP) of the transitory phase. Lower traces: plot of multi-unit frequency demonstrating network activity during the transition. (B) Enlarged image of a sharp wave-ripple and accompanying multi-unit (MU) activity at two time scales, as well as the transitory phase and interictal events and the underlying multi-unit activities appearing in four different epilepsy models. (C) Physiological and pathological transient high activity events are plotted to compare their correlation. A negative correlation is present in the rate of amplitude of the two types of events (n = 26). (D) The rate of sharp wave-ripples (SWRs) and interictal events (IIEs) at three different phases of the transition from several experiments. During control conditions sharp wave-ripple rate is high [and epileptiform event (EE) rate is zero], but after the pharmacological intervention it disappears completely. After the transitory phase (when the rate of both sharp wave-ripples and interictal events is zero) the rate of interictal events starts to increase, whereas sharp wave-ripple rate remains at zero. Note that for slices able to generate both sharp wave-ripples and interictal events, the interictal events cannot be seen until sharp wave-ripples disappear completely (n = 25). Sharp wave-ripples (E) and early interictal events (F) may seem similar (see insets); however, compared to sharp wave-ripples, interictal events (even early ones) have a larger amplitude, a longer duration (top trace) and are accompanied by several-fold higher multi-unit activity (bottom trace). Analysis of current source densities (centre) showed only minor shifts in the organization of sinks (red) and sources (blue), especially in the second half of an event.

Interictal-like events were also observed in the other three models (Fig. 1B). In all four models, a transitory state that separated sharp wave-ripples and interictal events appeared, with similar properties among models (Supplementary Fig. 3). The duration of transition varied greatly among experiments within and between models. The shortest time necessary (from adding the pharmacological agent until the first epileptiform event) was 58 s (high K+ model), whereas the longest was 2218 s in the 0 Mg2+ model; the median time and IQR (for the four models together) was 538 s (313–560), n = 136. We observed a complex reorganization of multi-unit activity during the transitory state in all models. After the transitory phase the activity evolved into interictal events, and in certain models into more complex epileptic forms (for a more detailed description of the three models see the Supplementary material and Supplementary Fig. 3).

After demonstrating that we can successfully induce sharp wave-ripple-to-interictal event transitions in four different ways, we focused our experiments to uncover the details of the transition to identify the accompanying changes in parameters and processes, with the aim of revealing the underlying mechanisms.

Sharp wave-ripples and interictal events are different transient high activity events in the high K+ model

Analysing the occurrence of sharp wave-ripples and interictal events in a large set of slices, we found that most slices producing large amplitude sharp wave-ripples produced either small amplitude interictal events or no epileptic events at all. Conversely, slices with small, infrequently-emerging sharp wave-ripples or no sharp wave-ripples were more likely to generate epileptic events, present as large amplitude interictal events, indicating that the capability of a slice to generate either sharp wave-ripples or interictal events is likely to be inversely related. To support this finding, the amplitude of sharp wave-ripples and interictal events was quantified (Supplementary material). Regression analysis showed a significant negative correlation between the amplitude/presence of these events (Fig. 1C, P = 0.013, R = 0.48, n = 26).

In cases when slices did produce sharp wave-ripples under control conditions and interictal events in high K+, the two event types never appeared interleaved. Therefore, we quantified how sharp wave-ripples are replaced by interictal events (n = 25) (Fig. 1D). The two types of oscillations excluded each other and were always separated by the transitory phase, strengthening the notion that they represent different network phenomena (median duration of this transitory phase was 315 s with IQR 170–459 s).

Because interictal events (especially early ones, Fig. 1F) could easily be mistaken for sharp wave-ripples (Fig. 1E) we quantified the dissimilarities: First, there is a significant difference in the amplitude of the two event types; 135 µV (IQR 124.9–141.1) for sharp wave-ripples and 344 µV (IQR 299–402) for early interictal events (Mann-Whitney U-test, P = 0.039, differences were studied thoroughly in 10 slices). Second, sharp wave-ripples and early interictal events can be separated based on their duration: sharp wave-ripples lasted 46 ms (IQR 42.2–57.4), whereas early interictal events lasted 104 ms (IQR 89.0–115) (Mann-Whitney U-test, P = 0.026, n = 26). In addition, a significant difference was found in the underlying multi-unit activity: it was 170 Hz (IQR 150–190) for sharp wave-ripples and 275 Hz (IQR 160–282) for interictal events (compared within experiments, Mann-Whitney U-test, P < 0.001).

Early interictal events evolved into late events that are more persistent, and will therefore be examined in more detail in the present study (henceforth late interictal events are referred to as interictal events). We also compared their amplitude, duration and other features to those of sharp wave-ripples (for sharp wave-ripple values see above). A significant difference was found in the amplitude and duration of the two event types; interictal event amplitude was 640 µV (IQR 512–692) (Mann-Whitney U-test, P = 0.014,), duration was 129 ms (IQR 104–157) (Mann-Whitney U-test, P = 0.008, n = 26). The third difference we found was that the period separating events from each other was 637 ms (IQR 338–813) for sharp wave-ripples and 1112 ms (IQR 862–1794) for interictal events (Mann-Whitney U-test, P = 0.02). Finally, a significant difference was found in the underlying multi-unit activity, which was 170 Hz (IQR 150–190) for sharp wave-ripples compared to 305 Hz (IQR 233–466) for interictal events, respectively (compared within experiments, Mann-Whitney U-test, P < 0.001).

Differences have been described in the high-frequency component of sharp wave-ripples versus interictal events (Bragin et al., 2002; Foffani et al., 2007; Engel et al., 2009; Levesque et al., 2011), but we could not find a systematic difference in the frequency of this component of the two types of events (using wavelet transformation), although oscillations during the peak of interictal events tended to be of slightly higher frequency and less regular than the ripples of sharp wave-ripples (not shown). Current source density analysis of the events did not demonstrate a significantly different picture either, although for the interictal events the initial source was spreading into stratum oriens and there were altered long-lasting sinks (red) and sources (blue) (Fig. 1E and F) in the later phase, similar to what was found in epileptic human tissue (Ulbert et al., 2004b; Wittner et al., 2009).

After describing basic differences between the two events, we analysed the transitory phase to understand how network synchrony becomes disorganized and rearranges later into a new form of transient high activity events.

Rearrangement of synchrony during the transitory phase separates sharp wave-ripples from interictal events

Multi-unit activity during sharp wave-ripples and interictal events was organized into robust, transient synchronous bursts. However, during the transitory phase leading from sharp wave-ripples to interictal events the multi-unit activity gradually became asynchronous, and only after a certain time did it rearrange into a new form of synchrony (interictal events, Fig. 2A and B), presumably when the level of population firing activity and its synchrony reached a threshold level (de la Prida et al., 2006) and recovery dynamics after the previous event started to dominate (Staley et al., 2001). The time-binned autocorrelogram of multi-unit firing (Fig. 2D) displays how the clustered firing of sharp wave-ripples had dissolved during the transitory phase and regrouped into another synchrony during interictal events. To quantify and visualize the loss of synchrony leading to the transitory phase and gain of synchrony preceding interictal events, the instantaneous frequency of multi-unit activity was normalized to its low-pass filtered average. This measure clearly shows how often the firing exceeds baseline activity during the synchronous bursts. Thus the disruption and rearrangement of activity became visible (Fig. 2C). The local minima and maxima of the multi-unit firing frequency showed large differences during sharp wave-ripples and interictal events (Fig. 2C), but approached each other during the transition period, suggesting a steady, elevated, but less structured activity, instead of high synchronies interspersed with silent periods. We also calculated the ‘burstiness’ of multi-unit activity (Supplementary material), which showed a similar U-shaped curve (Fig. 2E). As a simple measure of fluctuation we plotted the standard deviation (SD) of the local field potential or the multi-unit instantaneous frequency. We found that during sharp wave-ripples the SD values are relatively high and stable, during the transitory phase they drop and then eventually build up again to reach values higher than during sharp wave-ripples (Fig. 2F and G). It is important to note that synchronization of multi-unit activity starts to increase long before (in the experiment shown, 2–3 min before) the field potential fluctuation associated with interictal events appears (period indicated with a box in Fig. 2F), indicating once more that an increase in multi-unit activity leads the reorganization of network activity, and gross changes in the local field potential only appear later.

Figure 2

Reorganization of synchrony during sharp-wave ripple to epileptiform event transition induced by high K+. During the transition the synchrony of multi-unit activity drops and then builds up again until the network reaches the level of synchrony where interictal events start. (A) Local field potential (LFP) during the transitory phase. Areas emphasized with grey bars [sharp wave-ripple (SWR), T1, T2, T3 and interictal event (IIE)] indicate different phases of the transition and are magnified in B. The raster plot of multi-units and the multi-unit frequency are shown below the local field potential. (C) Upper trace shows how multi-unit frequency (grey) and its low-pass-filtered baseline (black) increases during the transition. The lower graph shows the baseline-normalized instantaneous frequency fluctuation. (D) Time-binned autocorrelogram of multi-unit activity showing that the synchrony of firing during sharp wave-ripples falls apart during the transitory period and reorganizes into a different synchrony during interictal events. (E) A U-shaped curve of ‘burstiness’ of multi-unit frequency shows that during the transition phase, synchrony decreases in the system. (F) A decrease and gradual recovery can be seen both in the standard deviation (SD) of local field potential values and in the SD of instantaneous multi-unit frequency. Note that the SD of multi-unit frequency starts to increase several minutes earlier than the SD of the local field potential signal (framed area), and a high level of synchrony evolves before anything is seen in the local field potential. (G) Changes in the SD of multi-unit instantaneous frequency from sharp wave-ripples to the transition phase and to interictal events in six recordings. Scale bars: B upper = 100 µV, lower = 60 Hz; time scale = 500 ms.

Figure 2

Reorganization of synchrony during sharp-wave ripple to epileptiform event transition induced by high K+. During the transition the synchrony of multi-unit activity drops and then builds up again until the network reaches the level of synchrony where interictal events start. (A) Local field potential (LFP) during the transitory phase. Areas emphasized with grey bars [sharp wave-ripple (SWR), T1, T2, T3 and interictal event (IIE)] indicate different phases of the transition and are magnified in B. The raster plot of multi-units and the multi-unit frequency are shown below the local field potential. (C) Upper trace shows how multi-unit frequency (grey) and its low-pass-filtered baseline (black) increases during the transition. The lower graph shows the baseline-normalized instantaneous frequency fluctuation. (D) Time-binned autocorrelogram of multi-unit activity showing that the synchrony of firing during sharp wave-ripples falls apart during the transitory period and reorganizes into a different synchrony during interictal events. (E) A U-shaped curve of ‘burstiness’ of multi-unit frequency shows that during the transition phase, synchrony decreases in the system. (F) A decrease and gradual recovery can be seen both in the standard deviation (SD) of local field potential values and in the SD of instantaneous multi-unit frequency. Note that the SD of multi-unit frequency starts to increase several minutes earlier than the SD of the local field potential signal (framed area), and a high level of synchrony evolves before anything is seen in the local field potential. (G) Changes in the SD of multi-unit instantaneous frequency from sharp wave-ripples to the transition phase and to interictal events in six recordings. Scale bars: B upper = 100 µV, lower = 60 Hz; time scale = 500 ms.

So far we have described the phenomenological and behavioural differences of the hippocampal CA3 area during sharp wave-ripples, the transition phase and interictal events, and have defined certain features differentiating them from one another. However, to understand the mechanisms responsible for transition we need to clarify the effects of high K+ application on cellular and network features and parameters.

Classification of the recorded CA3 neurons

To uncover the spiking behaviour of distinct neuron types in CA3 during interictal events, we recorded local field potentials simultaneously with action potentials in loose-patch mode in neurons under visual guidance, and subsequently postsynaptic potentials (and action potentials) in whole-cell mode, followed by anatomical identification of neurons. Based on the dendritic and axonal arborization, recorded neurons were grouped into five anatomical types: pyramidal cells, PVBCs and axo-axonic cells, CCK-expressing basket cells and a mixed group of dendritic layer innervating cells (Freund and Buzsaki, 1996; Klausberger and Somogyi, 2008). The firing properties of these groups in relation to interictal events were compared (Fig. 3). For detailed morphological descriptions see the Supplementary material and Supplementary Fig. 4.

Figure 3

PVBCs stop firing during epileptiform events, whereas other cells increase their firing rate. (A) Firing of anatomically identified CA3 neurons during sharp wave-ripples (SWRs) recorded in loose patch mode. Somata and dendrites of cells are shown in black, axons in red. Firing of the cells can be seen in the lower rows. (B and C) Firing of neurons during early and late interictal events (IIEs), respectively. All examined neurons increased their firing rate and changed their firing pattern during early interictal events. During late interictal events they increased their firing rate further, and some cells (PVBCs) stopped firing at the peak. Below the local field potential the relative power in the 150–400 Hz band is plotted to show the duration of high-frequency oscillation during the interictal events. High frequency oscillation (HFO) coincided with pyramidal cell firing and the silent phase of PVBCs. For quantification, firing of neurons was separated into three phases of 100 ms: before (b), during (d) and after (a) the peak of the interictal event. PVBCs stopped firing at the peak (grey area, approximately ‘during’ phase) of interictal events. PV+ = parvalbumin-positive.

Figure 3

PVBCs stop firing during epileptiform events, whereas other cells increase their firing rate. (A) Firing of anatomically identified CA3 neurons during sharp wave-ripples (SWRs) recorded in loose patch mode. Somata and dendrites of cells are shown in black, axons in red. Firing of the cells can be seen in the lower rows. (B and C) Firing of neurons during early and late interictal events (IIEs), respectively. All examined neurons increased their firing rate and changed their firing pattern during early interictal events. During late interictal events they increased their firing rate further, and some cells (PVBCs) stopped firing at the peak. Below the local field potential the relative power in the 150–400 Hz band is plotted to show the duration of high-frequency oscillation during the interictal events. High frequency oscillation (HFO) coincided with pyramidal cell firing and the silent phase of PVBCs. For quantification, firing of neurons was separated into three phases of 100 ms: before (b), during (d) and after (a) the peak of the interictal event. PVBCs stopped firing at the peak (grey area, approximately ‘during’ phase) of interictal events. PV+ = parvalbumin-positive.

Activity of identified cell types is different during physiological and pathological transient high activity events

We examined the firing behaviour of identified CA3 hippocampal interneurons and pyramidal cells during the transition in the high K+ model. First we examined early interictal events. All neuron types increased their firing rate, and some showed decreased spike amplitude (Fig. 3B). However, this altered firing pattern was changed further as early interictal events evolved into late interictal events.

A noticeable difference among early and late interictal events was that the high-frequency oscillation (Boksa et al., 1998) in the local field potential during the peak of the events was significantly longer during late interictal events (81.5 ms, IQR 54.38–92.25) than during the early interictal events (27.25 ms, IQR 22.5–40.63) (Wilcoxon paired test, P < 0.01). This difference can be seen in the plot of relative power in the 150–400 Hz band in the traces below the local field potential in Fig. 3B and C.

The firing rate of all studied neurons changed during interictal events compared to sharp wave-ripples (Figs 3A–C and 4A and Table 2). As the firing pattern of different neurons varied systematically during phases of a single interictal event, we defined three phases where firing properties were analysed separately: 100 ms before the peak of the event, 100 ms during the event (after the multi-unit peak) and 100 ms immediately after the event (Figs 3C, 4A and B).

Figure 4

Firing rate varies among different phases of epileptiform events induced by high K+. (A) Spike distribution histograms of individual neurons during sharp wave-ripples (SWR; grey: individual traces, dashed red: average) and interictal events (IIE; black: individual traces, red: average) show that the firing pattern becomes altered and the firing rate increases. (B) Statistical comparison of the average number of spikes fired by different neurons during sharp wave-ripples (s), in the ‘before’ (b), ‘during’ (d) and ‘after’ (a) phases. Note that the spike number significantly increases from sharp wave-ripples to interictal events (marked with asterisks). Firing also differs greatly among different phases of interictal events. (C) Changes in normalized firing probability and amplitude for different neuron types (normalized to before phase). Upper graphs compare changes in normalized values between the before and the during phases, whereas lower traces compare changes among the before and after phases. Grey area indicates a decrease (<100%), asterisks indicate significance at P < 0.05). PV+BC = parvalbumin-positive basket cell; CCK+BC = cholecystokinin-positive basket cell; PC = pyramidal cell; AAC = axo-axonic cell; DC = dendritic layer innervating cell.

Figure 4

Firing rate varies among different phases of epileptiform events induced by high K+. (A) Spike distribution histograms of individual neurons during sharp wave-ripples (SWR; grey: individual traces, dashed red: average) and interictal events (IIE; black: individual traces, red: average) show that the firing pattern becomes altered and the firing rate increases. (B) Statistical comparison of the average number of spikes fired by different neurons during sharp wave-ripples (s), in the ‘before’ (b), ‘during’ (d) and ‘after’ (a) phases. Note that the spike number significantly increases from sharp wave-ripples to interictal events (marked with asterisks). Firing also differs greatly among different phases of interictal events. (C) Changes in normalized firing probability and amplitude for different neuron types (normalized to before phase). Upper graphs compare changes in normalized values between the before and the during phases, whereas lower traces compare changes among the before and after phases. Grey area indicates a decrease (<100%), asterisks indicate significance at P < 0.05). PV+BC = parvalbumin-positive basket cell; CCK+BC = cholecystokinin-positive basket cell; PC = pyramidal cell; AAC = axo-axonic cell; DC = dendritic layer innervating cell.

Table 2

Spiking characteristics of different hippocampal CA3 neurons during sharp wave-ripples and during different phases of epileptiform events, and values of significance when comparing them

 PC PV+ BC AAC CCK+ BC DC 
Spike/sharp wave-ripple 2.9 (0.9–5.1) 2.1 (0.9–2.5) 0.6 (0.2–0.8) 0.8 (0.4–1.8) 
Spike/interictal event 11.8 (4.7–11.6) 6.7 (7.7–11.5) 12.8 (7.1–18.6) 15.1 (11.5–14.7) 11.7 (8.9–15.9) 
Wilcoxon test P-values <0.001 0.038 0.008 0.03 <0.001 
Median number of action potentials across events in different phases 
Action potentials before phase 2.5 (2.4–4.9) 4.5 (3.5–8.1) 7 (3.9–8) 4.89 (4.1–5.8) 3.69 (3.3–5.3) 
Action potentials during phase 8.5 (4–9.3) 0.76 (0.5–1) 4.18 (2.6–4.7) 6.88 (1.6–5.7) 5.13 (0.6–1.6) 
Action potentials after phase 0.17 (0.1–4.2) 1.1 (0.4–1.5) 3.74 (3.3–5.4) 2.5 (1.7–3.2) 3.1 (1.5–4.2) 
Change in firing rate (spike number in before phase: 100%) 
Before to during 211 (148.5–325%) 12.5 (7.6–23.9%) 100.4 (58.2–136.2%) 129 (107–154%) 129 (33.1–200.1%) 
Before to after 6.7 (3.5–8.9%) 36.81 (14.1–37%) 76.6 (73.7–121.8%) 47.9 (33.3–65.1%) 69.1 (55.5–93%) 
During to after (during was 100%) 4.3 (2.1–7.7%) 223.4 (58.1–421%) 89.5 (34.2–126.6%) 41 (38.9–46.9%) 62.9 (32.6–99.7%) 
Friedman ANOVA <0.001 0.007 0.8 0.015 0.011 
Post hoc corrected Wilcoxon test P-values 
Before to during 0.026 0.024 0.68 0.29 0.4 
Before to after 0.016 0.02 0.5 0.13 0.03 
During to after 0.014 0.024 0.92 0.043 0.42 
Change in spike amplitude (amplitude in before phase: 100%) 
Before to during 54.4% (66.7–87.6) 65.3% (49.5–75) 54.89% (54.9–68.4) 98.5% (96.4–103.1) 73.14% (63.9–86.7) 
Before to after 78.4% (61.2–86.3) 71.2% (61.2–76.5) 66.94% (66.3–76.4%) 111.6% (107;113.3%) 89.87% (77.4–98%) 
During to after (during was 100%) 109.8% (76.5–112.6) 109.1% (102–128.9) 125.9 %(108.9–138.6) 106% (102.5–117.4) 116.1% (103–128.5) 
Wilcoxon test P-values 
Before to during 0.009 0.005 0.043 0.004 
Before to after 0.018 0.005 0.028 0.144 0.012 
During to after 0.57 0.05 0.075 0.043 0.004 
 PC PV+ BC AAC CCK+ BC DC 
Spike/sharp wave-ripple 2.9 (0.9–5.1) 2.1 (0.9–2.5) 0.6 (0.2–0.8) 0.8 (0.4–1.8) 
Spike/interictal event 11.8 (4.7–11.6) 6.7 (7.7–11.5) 12.8 (7.1–18.6) 15.1 (11.5–14.7) 11.7 (8.9–15.9) 
Wilcoxon test P-values <0.001 0.038 0.008 0.03 <0.001 
Median number of action potentials across events in different phases 
Action potentials before phase 2.5 (2.4–4.9) 4.5 (3.5–8.1) 7 (3.9–8) 4.89 (4.1–5.8) 3.69 (3.3–5.3) 
Action potentials during phase 8.5 (4–9.3) 0.76 (0.5–1) 4.18 (2.6–4.7) 6.88 (1.6–5.7) 5.13 (0.6–1.6) 
Action potentials after phase 0.17 (0.1–4.2) 1.1 (0.4–1.5) 3.74 (3.3–5.4) 2.5 (1.7–3.2) 3.1 (1.5–4.2) 
Change in firing rate (spike number in before phase: 100%) 
Before to during 211 (148.5–325%) 12.5 (7.6–23.9%) 100.4 (58.2–136.2%) 129 (107–154%) 129 (33.1–200.1%) 
Before to after 6.7 (3.5–8.9%) 36.81 (14.1–37%) 76.6 (73.7–121.8%) 47.9 (33.3–65.1%) 69.1 (55.5–93%) 
During to after (during was 100%) 4.3 (2.1–7.7%) 223.4 (58.1–421%) 89.5 (34.2–126.6%) 41 (38.9–46.9%) 62.9 (32.6–99.7%) 
Friedman ANOVA <0.001 0.007 0.8 0.015 0.011 
Post hoc corrected Wilcoxon test P-values 
Before to during 0.026 0.024 0.68 0.29 0.4 
Before to after 0.016 0.02 0.5 0.13 0.03 
During to after 0.014 0.024 0.92 0.043 0.42 
Change in spike amplitude (amplitude in before phase: 100%) 
Before to during 54.4% (66.7–87.6) 65.3% (49.5–75) 54.89% (54.9–68.4) 98.5% (96.4–103.1) 73.14% (63.9–86.7) 
Before to after 78.4% (61.2–86.3) 71.2% (61.2–76.5) 66.94% (66.3–76.4%) 111.6% (107;113.3%) 89.87% (77.4–98%) 
During to after (during was 100%) 109.8% (76.5–112.6) 109.1% (102–128.9) 125.9 %(108.9–138.6) 106% (102.5–117.4) 116.1% (103–128.5) 
Wilcoxon test P-values 
Before to during 0.009 0.005 0.043 0.004 
Before to after 0.018 0.005 0.028 0.144 0.012 
During to after 0.57 0.05 0.075 0.043 0.004 

PV+BC = parvalbumin-positive basket cell; CCK+BC = cholecystokinin-positive basket cell; PC = pyramidal cell; AAC = axo-axonic cell; DC = dendritic layer innervating cell.

Numbers in bold are significant P-values (<0.05); values in brackets are IQR.

Almost all neurons showed a greatly increased firing rate during interictal events compared with sharp wave-ripples, with the exception of PVBCs, where the maximal firing rate only slightly exceeded the firing rate during sharp wave-ripples (Fig. 4A and Table 2). In Table 2, spike numbers are described during the entire event, (duration was ∼300–400 ms). As sharp wave-ripples last ∼100 ms, it may be more appropriate to compare spike numbers fired during sharp wave-ripples to spike numbers fired during either the ‘before’, ‘during’ or ‘after’ phases. In this way we could compare spike numbers fired over similar epochs.

Pyramidal cells (n = 12) fired with low spiking probability during sharp wave-ripples [usually no spikes were detected, but in a larger, recently published data set we encountered pyramidal cells firing in association with sharp wave-ripples (Hajos et al., 2013)]. The firing probability significantly increased during interictal events; moreover, pyramidal cells fired bursts of action potentials between interictal events. The firing rates of pyramidal cells varied significantly between different phases of the interictal events, with a significant rise during the event, followed by a significant drop immediately after (Friedman ANOVA and post hoc Wilcoxon paired test; Figs 3C and 4A and Table 2).

PVBC (n = 10) and axo-axonic cells (n = 6) fired numerous action potentials during sharp wave-ripples, and fired with a somewhat higher frequency 100–150 ms before the large negative peak of the interictal events (‘before’). However, when the field event reached its negative peak (‘during’), most PVBC cells and axo-axonic cells dropped their firing rate, and spike amplitude decreased gradually (Fig. 4B and C and Table 2). In all PVBCs and one axo-axonic cell, this decrease continued until action potentials were no longer detectable (Table 2). After the interictal event, when the local field potential amplitude was close to baseline, the firing of the cells progressively recovered, and the spike number increased. Significant changes among phases were found for PVBCs when before, during and after phases were compared (Friedman ANOVA and post hoc Wilcoxon paired test, Fig. 4A–C and Table 2), but not for axo-axonic cells.

Most CCK-positive basket cells (n = 5) and dendritic layer innervating cells (n = 15), unlike the previous cells, fired with a moderate probability during sharp wave-ripples, and increased their firing rate further during interictal events (Fig. 3A–C and Table 2). Close to the initial negative peak of interictal events, CCK-positive basket cells and dendritic layer innervating cells started firing, continued to do so during the entire event, and decreased their firing rate after the event (Friedman ANOVA and Wilcoxon paired test) (Fig. 4B and C and Table 2).

Next, we compared the normalized changes (given as %) in firing rate (between the interictal event phases) among the five neuron groups, where 100% was the number of spikes produced in the ‘before’ phase. When examining changes between the before and during phases, PVBC values were significantly smaller than those of pyramidal cells, CCK-positive basket cells and dendritic layer innervating cells; axo-axonic cell values were significantly smaller than pyramidal cell values. However, no other groups showed significant differences (differences among groups were tested with Kruskal-Wallis ANOVA followed by post hoc Mann-Whitney U-test with Bonferroni correction). When firing rate changes were compared between before and after phases, pyramidal cell and PVBC values were significantly smaller than axo-axonic cell and dendritic layer innervating cell values (Kruskal-Wallis ANOVA, Mann Whitney U-test; Fig. 4C, Tables 2 and 3).

Table 3

Statistics on spiking characteristics of CA3 neurons

Kruskal-Wallis ANOVA: P = 0.011  PC PV+ BC AAC CCK+ BC 
 
Firing rate change: before to during PC     
 PV + BC <0.001    
 AAC 0.046 0.062   
 CCK + BC 0.137 0.014 0.27  
 DC 0.084 0.002 0.35 0.89 

 
Kruskal-Wallis ANOVA: P = 0.006  PC PV + BC AAC CCK + BC 

 
Firing rate change: before to after PC     
 PV + BC 0.27    
 AAC 0.023 0.023   
 CCK + BC 0.056 0.143 0.39  
 DC 0.036 0.018 0.48 0.35 

 
Kruskal-Wallis ANOVA: P < 0.001  PC PV + BC AAC CCK + BC 

 
Spike amplitude change: before to during PC     
 PV + BC 0.24    
 AAC 0.098 0.85   
 CCK + BC 0.08 0.012 0.004  
 DC 0.16 0.043 0.13 0.028 

 
Kruskal-Wallis ANOVA: P < 0.001  PC PV + BC AAC CCK + BC 

 
Spike amplitude change: before to after PC     
 PV + BC 0.015    
 AAC 0.22 0.7   
 CCK + BC 0.06 0.012 0.028  
 DC 0.5 0.008 0.028 0.038 
Kruskal-Wallis ANOVA: P = 0.011  PC PV+ BC AAC CCK+ BC 
 
Firing rate change: before to during PC     
 PV + BC <0.001    
 AAC 0.046 0.062   
 CCK + BC 0.137 0.014 0.27  
 DC 0.084 0.002 0.35 0.89 

 
Kruskal-Wallis ANOVA: P = 0.006  PC PV + BC AAC CCK + BC 

 
Firing rate change: before to after PC     
 PV + BC 0.27    
 AAC 0.023 0.023   
 CCK + BC 0.056 0.143 0.39  
 DC 0.036 0.018 0.48 0.35 

 
Kruskal-Wallis ANOVA: P < 0.001  PC PV + BC AAC CCK + BC 

 
Spike amplitude change: before to during PC     
 PV + BC 0.24    
 AAC 0.098 0.85   
 CCK + BC 0.08 0.012 0.004  
 DC 0.16 0.043 0.13 0.028 

 
Kruskal-Wallis ANOVA: P < 0.001  PC PV + BC AAC CCK + BC 

 
Spike amplitude change: before to after PC     
 PV + BC 0.015    
 AAC 0.22 0.7   
 CCK + BC 0.06 0.012 0.028  
 DC 0.5 0.008 0.028 0.038 

PV+BC = parvalbumin-positive basket cell; CCK+BC = cholecystokinin-positive basket cell; PC = pyramidal cell; AAC = axo-axonic cell; DC = dendritic layer innervating cell.

Numbers in bold are significant P-values (<0.05).

Levels of significance when comparing relative changes of firing rate and amplitude.

Among cell groups. As post hoc test, Mann-Whitney U-test is used with Bonferroni correction.

In the next step, we studied the extracellular spike amplitude evolution. Although only PVBCs and one axo-axonic cell decreased their firing amplitude to zero, all recorded neuron types showed somewhat decreased spike amplitude in the during phase. When normalized amplitudes were compared among the three activity phases for each neuron group (amplitude in the before phase was 100%), a significant decrease was found for pyramidal cells, PVBCs, axo-axonic cells and dendritic layer innervating cells, whereas CCK-positive basket cells showed significant differences only when comparing the during phase to the after phase (Wilcoxon paired test). These data suggested that all neurons received a massive depolarization, causing a decrease in their spike amplitudes; however, the severity of the decrease differed greatly among cell types (Fig. 4C and Table 2).

Finally, changes in normalized spike amplitudes were compared among neuron groups, where 100% was the amplitude of spikes produced in the before phase. The decreased spike amplitude values of PVBCs and axo-axonic cells were significantly smaller than that of CCK-positive basket cells and dendritic layer innervating cells. In addition, significant differences were found between CCK-positive basket cells and dendritic layer innervating cells (normalized spike amplitude was significantly smaller for dendritic layer innervating cells, Kruskal-Wallis ANOVA, Mann Whitney U-test; Fig. 4C, Tables 2 and 3).

Membrane potential changes of hippocampal CA3 neurons during interictal events

The decrease in extracellular spike amplitudes and the cessation of firing suggested that cells might receive a strong depolarization and some interneurons would enter into depolarization block during interictal events. To strengthen this hypothesis we recorded the activity of neurons in whole-cell current clamp mode simultaneously with local field potential recordings. The value of the resting membrane potential and depolarization during interictal events was estimated in three different ways. This redundancy was necessary because our recordings were carried out in whole-cell mode, and even the most carefully chosen intrapipette solution can alter the intracellular ion milieu, and thus the firing of a cell. First, we compared the firing pattern recorded in loose-patch mode to the action potential pattern in whole-cell mode at different membrane potentials (Fig 5A–C). As shown in Fig. 5D, the intracellularly-recorded firing matched the loose-patch-recorded firing best when cells were held at potentials around −30 to −40 mV between interictal events using a constant injected current. When the membrane potential was recorded in I = 0 mode immediately after break-in, the membrane potential between interictal events was −35.0 mV (IQR −40.3 to −30.7) for pyramidal cells (n = 8); −39.7 mV (IQR −40.2 to −39.6) for PVBCs (n = 5); −45.1 mV (IQR −45.3 to −43.1) for axo-axonic cells (n = 3), −41.8 mV (IQR −42.5 to −40.9) for CCK-positive basket cells (n = 3) and −29.7 mV (IQR −30.4 to −29.1) for dendritic layer innervating cells (n = 3). We found no significant difference among cell groups (P = 0.46, Kruskal-Wallis ANOVA), or between membrane potentials estimated with the two methods (P = 0.318, paired sample Wilcoxon test; to enhance readability, results are summarized in Fig. 5 and Table 4). Finally, we calculated the approximate depolarization caused by the increase of extracellular K+ according to the Nernst equation, which resulted in a depolarization of 23 mV (calculated with 8.5 mM K+ in the extracellular solution compared to 3.5 mM). As the extracellular K+ concentration likely increases transiently during interictal events (as a result of elevated firing) (Gnatkovsky et al., 2008), this result may underestimate the actual depolarization, which may reach ∼30 mV according to previous estimates (Frohlich et al., 2008; Cressman et al., 2009), relative to a resting potential of −64 ± 1 mV under control conditions (Spruston and Johnston, 1992). With all calculating methods, the depolarization of the membrane potential in the high K+ solution was about +25–35 mV compared to the estimated control membrane potential in artificial CSF.

Figure 5

PVBC firing is blocked as a result of strong depolarization. (AC) Three examples show that in the epileptogenic artificial CSF the membrane potential of neurons is approximately −40 mV before interictal events (IIEs, baseline), as their loose-patch-recorded spiking matches the firing pattern recorded at −40 mV in current clamp during interictal events. Upper pair of traces show loose-patch-recorded firing (black) during an epileptiform event (grey: local field potential). Similar sequences of spikes were recorded from the neuron at −40 mV (middle traces), but not at −70 mV (lower traces). The similarity to the −40 mV potential was true for all examined cell types. A pyramidal cell (A), a PVBC (B) and a CCK+ basket cell (C) are shown. (D) The membrane potential where neurons fired with a similar pattern to that seen in loose patch mode is indicated with the lower striped grey box and whiskers (baseline), whereas the membrane potential recorded immediately after break-in (in I = 0 mode) is shown with white box and whiskers. Higher grey box and whiskers plots indicate the membrane potential recorded during the interictal event peak. This value was markedly higher compared to the former two time points, and was more variable among cells. (E) Change in spiking rate as a function of depolarization is shown for CA3 neurons. Some neurons increase their firing rate upon depolarization [pyramidal cell (PC), CCK+ basket cell (BC), dendritic layer innervating cell) whereas others decrease it (PVBC, axo-axonic cell). To compare neurons with inherently different spiking rates, membrane potential-dependent changes were normalized to the peak firing rate of individual neurons. Grey line indicates 100% as maximal firing. The figure shows that PVBCs and axo-axonic cells fire already maximally close to normal resting potential (note that PVBCs may completely stop firing, whereas, axo-axonic cells only drop their firing rate until a certain point). Other neurons, most importantly pyramidal cells, can increase their firing rate with depolarization. PV+BC = parvalbumin-positive basket cell; CCK+BC = cholecystokinin-positive basket cell; AAC = axo-axonic cell; DC = dendritic layer innervating cell; CC pot = membrane potential in current clamp mode; LP = loose patch.

Figure 5

PVBC firing is blocked as a result of strong depolarization. (AC) Three examples show that in the epileptogenic artificial CSF the membrane potential of neurons is approximately −40 mV before interictal events (IIEs, baseline), as their loose-patch-recorded spiking matches the firing pattern recorded at −40 mV in current clamp during interictal events. Upper pair of traces show loose-patch-recorded firing (black) during an epileptiform event (grey: local field potential). Similar sequences of spikes were recorded from the neuron at −40 mV (middle traces), but not at −70 mV (lower traces). The similarity to the −40 mV potential was true for all examined cell types. A pyramidal cell (A), a PVBC (B) and a CCK+ basket cell (C) are shown. (D) The membrane potential where neurons fired with a similar pattern to that seen in loose patch mode is indicated with the lower striped grey box and whiskers (baseline), whereas the membrane potential recorded immediately after break-in (in I = 0 mode) is shown with white box and whiskers. Higher grey box and whiskers plots indicate the membrane potential recorded during the interictal event peak. This value was markedly higher compared to the former two time points, and was more variable among cells. (E) Change in spiking rate as a function of depolarization is shown for CA3 neurons. Some neurons increase their firing rate upon depolarization [pyramidal cell (PC), CCK+ basket cell (BC), dendritic layer innervating cell) whereas others decrease it (PVBC, axo-axonic cell). To compare neurons with inherently different spiking rates, membrane potential-dependent changes were normalized to the peak firing rate of individual neurons. Grey line indicates 100% as maximal firing. The figure shows that PVBCs and axo-axonic cells fire already maximally close to normal resting potential (note that PVBCs may completely stop firing, whereas, axo-axonic cells only drop their firing rate until a certain point). Other neurons, most importantly pyramidal cells, can increase their firing rate with depolarization. PV+BC = parvalbumin-positive basket cell; CCK+BC = cholecystokinin-positive basket cell; AAC = axo-axonic cell; DC = dendritic layer innervating cell; CC pot = membrane potential in current clamp mode; LP = loose patch.

Table 4

Membrane potential characteristics of different CA3 neurons during epileptiform events

Membrane potential closest to loose patch pattern PC PV + BC AAC CCK + BC DC 
Membrane potential I = 0 −30.8 mV (−32–−30.3) −39.7 mV (−40.2–−39.6) −45.1 mV (−45.3–−43) −41.8 mV (−42.5–−40.9) −29.7 mV (−30.4;−29.1) 
Membrane potential at interictal event peak −13.1 mV (−15.4–−11.9) −17.6 mV (−19.9–0) −19.9 mV (−20.2–−15.2) −29.9 mV (−33.8–−29.1) −28 mV (−29.1–−27.8) 
Membrane potential where firing probability is the highest −35.1 mV (−36.4–−33.8) −40.4 mV (−45.1–−29.8) −40.4 mV (−45.2–−40.2) −44.8 mV (−45.1–−42.3) −30.4 mV (−32.8–−30.1) 

 
IPSG (nS) % PV + BC AAC CCK + BC   

 
High K+ (control: 100%) 45.3% (33.4–49.1) 37.3% (11.2–56.5) 73.44% (54.3–93)   
Washout 57.5% (38.4–88.9) 61.7% (54.5–139) 93.7% (72.2–167.5)   

 
Amplitude (pA) % PV + BC AAC CCK + BC   

 
High K+ (control: 100%) 42.2% (26.3–51.7) 65.3% (51.9–70.4) 76.9% (73.7–80.8)   
Washout 98.3% (93–137.8) 97.5% (88.1–111.2) 98.9% (91.8–106)   
Membrane potential closest to loose patch pattern PC PV + BC AAC CCK + BC DC 
Membrane potential I = 0 −30.8 mV (−32–−30.3) −39.7 mV (−40.2–−39.6) −45.1 mV (−45.3–−43) −41.8 mV (−42.5–−40.9) −29.7 mV (−30.4;−29.1) 
Membrane potential at interictal event peak −13.1 mV (−15.4–−11.9) −17.6 mV (−19.9–0) −19.9 mV (−20.2–−15.2) −29.9 mV (−33.8–−29.1) −28 mV (−29.1–−27.8) 
Membrane potential where firing probability is the highest −35.1 mV (−36.4–−33.8) −40.4 mV (−45.1–−29.8) −40.4 mV (−45.2–−40.2) −44.8 mV (−45.1–−42.3) −30.4 mV (−32.8–−30.1) 

 
IPSG (nS) % PV + BC AAC CCK + BC   

 
High K+ (control: 100%) 45.3% (33.4–49.1) 37.3% (11.2–56.5) 73.44% (54.3–93)   
Washout 57.5% (38.4–88.9) 61.7% (54.5–139) 93.7% (72.2–167.5)   

 
Amplitude (pA) % PV + BC AAC CCK + BC   

 
High K+ (control: 100%) 42.2% (26.3–51.7) 65.3% (51.9–70.4) 76.9% (73.7–80.8)   
Washout 98.3% (93–137.8) 97.5% (88.1–111.2) 98.9% (91.8–106)   

Transmission parameters of different perisomatic−pyramidal cell pairs among different conditions.

Values in brackets are IQR.

PC = pyramidal cell; AAC = axo-axonic cell; DC = dendritic layer innervating cell; IPSG = inhibitory postsynaptic conductance.

Even though the baseline membrane potential showed no significant difference between cell groups in high K+ when interictal events occurred, the membrane potential further depolarized, and the magnitude of this deflection varied among neuron groups. As the level of this depolarization may determine whether a cell enters into depolarization block or not, we compared the maximum of the low-pass-filtered (30 Hz) membrane potential that different cell types reached during interictal event peaks (in several cases this was the value of spike threshold, but in other cases membrane potential increased further during spiking, see Fig. 5B PVBC). Pyramidal cells and PVBCs experienced the largest depolarization [for pyramidal cells −15.5 mV (IQR −21.5 to −11.9), for PVBCs −17.6 mV (IQR −19.9 to 0)]. However, only the PVBC depolarization differed significantly from the depolarization of other cell types, namely from CCK-positive basket cells (P = 0.037, Mann-Whitney U-test), and dendritic layer innervating cells (P = 0.036, Mann-Whitney U-test) (Fig. 5E and Table 4), indicating that the strong depolarization can be a factor responsible for the depolarization block of PVBCs. This strong transient depolarization cannot be the result of the temporal change in K+ concentration (because of elevated firing during the peak of interictal events) as different cell types showed different depolarization levels.

To uncover firing characteristics of distinct cell types at different depolarization levels (threshold and depolarization block threshold), neurons were recorded in current clamp mode at different membrane potentials from −70 to −30 mV (5 mV steps, Fig. 5E) and the number of spikes (Supplementary Table 2) were compared between different neuron types and different potentials. We found that when pyramidal cells, CCK-positive basket cells and dendritic layer innervating cells were held around −70 mV, they fired with a low firing probability compared to what we recorded in loose-patch mode (around −40 to −45 mV). For all three cell types, their firing probability was lowest at −70 mV. However, as we gradually depolarized these cells, the spiking frequency increased during interictal events (Fig. 5E). Conversely, in the case of PVBCs and axo-axonic cells, the firing frequency was quite high at −70 mV and gradually decreased as the cells were depolarized (lowest at −30 mV for PVBCs and −40 mV for axo-axonic cells). When the depolarization of PVBCs reached approximately −45 mV, they decreased their firing, and entered into depolarization block during interictal events, similar to loose-patch recordings.

Firing pattern of pyramidal cells and parvalbumin-positive basket cells during interictal events in other models

As depolarization block seemed to be an important event in the generation of interictal events, we examined how the two key cell types, pyramidal cells and PVBCs, fire during interictal events in the other three models (Fig. 6). We found that both cell types, similar to their behaviour in high K+, strongly increased their firing frequency during interictal events. In the 4-aminopyridine and 0 Mg2+ models, PVBCs first increased their firing frequency and then entered into depolarization block around the peak of the interictal events. While PVBCs were inactive, high frequency oscillation appeared in the local field potential and pyramidal cells fired. In the gabazine model, although the firing did not stop at this stage, the amplitude of the spikes dropped temporarily, suggesting strong intracellular depolarization. Pyramidal cell firing again coincided here with the high frequency oscillation period of the local field potential.

Figure 6

Firing pattern of pyramidal cells and parvalbumin-positive basket cells in other models. The firing of pyramidal cells (A) and PVBCs (B) was recorded in loose patch mode (upper rows) simultaneously with local field potential recordings (middle rows) in three further models of interictal events (IIEs). The relative power in the 150–400 band (base-normalized to the period before the interictal events) was also calculated and plotted (lower trace) to indicate the period when high frequency oscillation (Boksa et al., 1998) was present in the local field potential (LFP). Pyramidal cells mostly fired at (and after) the peak of interictal events. PVBCs cease to fire (similarly to the high K+ model) at this stage in the 4-aminopyridine (4-AP) and zero Mg2+ models (suggesting they received strong depolarization), but only decrease their spike amplitude in the gabazine model with a continued firing. The high-frequency oscillation in the local field potential coincided with strong pyramidal cell firing in all models. (C) The firing frequencies in the different models during different phases for the two cell types. HFO = high frequency oscillation; PC = pyramidal cell; PV+BC = parvalbumin-positive basket cell.

Figure 6

Firing pattern of pyramidal cells and parvalbumin-positive basket cells in other models. The firing of pyramidal cells (A) and PVBCs (B) was recorded in loose patch mode (upper rows) simultaneously with local field potential recordings (middle rows) in three further models of interictal events (IIEs). The relative power in the 150–400 band (base-normalized to the period before the interictal events) was also calculated and plotted (lower trace) to indicate the period when high frequency oscillation (Boksa et al., 1998) was present in the local field potential (LFP). Pyramidal cells mostly fired at (and after) the peak of interictal events. PVBCs cease to fire (similarly to the high K+ model) at this stage in the 4-aminopyridine (4-AP) and zero Mg2+ models (suggesting they received strong depolarization), but only decrease their spike amplitude in the gabazine model with a continued firing. The high-frequency oscillation in the local field potential coincided with strong pyramidal cell firing in all models. (C) The firing frequencies in the different models during different phases for the two cell types. HFO = high frequency oscillation; PC = pyramidal cell; PV+BC = parvalbumin-positive basket cell.

As in the previous set of experiments we proved that progressive drop in extracellular action potential amplitude is the result of depolarization block of firing as a result of strong intracellular depolarization, we did not make systematic experiments for the other three models. Nevertheless, we measured the membrane potential in some cells and found that, in agreement with the findings of the cell-attached recordings, cells are strongly depolarized in the 4-aminopyridine [resting membrane potential (RMP) −35 ± 7.1 mV, n = 2 for PVBC cells, RMP: −39 ± 6.5 mV, n = 3 for pyramidal cells) and 0 Mg2+ models (RMP: −40 ± 4 mV, n = 3 for PVBC cells, RMP: −54 ± 5.9 mV, n = 3 for pyramidal cells) but not in the gabazine model (RMP: −55 ± 8.7 mV, n = 3 for PVBC, RMP: −53 ± 2.7 mV, n = 2 for pyramidal cells).

Firing recorded in loose-patch mode (in all models) and intracellular potentials (examined in detail only in the high K+ model) correlated closely with phases of the epileptic field potential. Notably, we observed interaction among firing of different cell types. Therefore, we investigated in detail the correlation between the field signal and the relative timing of the firing of different cell types.

Stages of an epileptic event correlate with intracellular potentials

Analysing local field potential features recorded simultaneously with intracellular potentials of pyramidal cells and interneurons with no injected current (I = 0) in the high K+ model, we found four characteristic phases of interictal event evolution (Fig. 7):

  • During the first phase, a mild negative deflection of the local field potential was associated with a small depolarization of pyramidal cells and a significant depolarization and firing frequency increase in the PVBCs, accompanied by an increase in multi-unit activity (Prida and Sanchez-Andres, 1999; de la Prida et al., 2006).

  • In the second phase, a steep negative shift visible on the local field potential was associated with a more pronounced pyramidal cell depolarization, and a steep depolarization of PVBCs accompanied by accelerated firing and drop in action potential amplitude. The multi-unit activity increased further.

  • The onset of the third phase was defined by the blockade of PVBC firing and by a simultaneous strong depolarization of pyramidal cells associated with multiple action potentials (Trevelyan et al., 2006). The high frequency oscillation (see also Fig. 3B and C) that appears in this phase is most probably the population spike of the active pyramidal cells. By the end of the phase, after an initial increase, the multi-unit activity and the pyramidal cell firing started to drop.

  • During the fourth phase, the local field potential slowly returned to baseline through a negative period, the firing of PVBCs gradually recovered, pyramidal cells became repolarized to a baseline membrane potential, and stopped at a certain level of repolarization. At the same time the multi-unit activity returned to baseline (Fig. 7).

Figure 7

Build-up of depolarization during epileptiform events and its effect on neuronal firing: pyramidal cells start to fire massively when inhibition from PVBCs collapses. During the late interictal events we studied, four phases can be distinguished: (1) Primary depolarization, PVBCs depolarize and start firing, pyramidal cells start to depolarize, multi-unit activity starts to increase. There is a slow negative deflection in the local field potential. (2) Secondary depolarization, the frequency of PVBC firing further increases while the amplitude drops as the cell depolarizes even further, pyramidal cells depolarize further but do not fire yet, multi-unit activity increases heavily, steep negative drop in local field potential appears (LFP). (3) Interictal event builds up, the power in the 150–400 Hz band increases, PVBCs cease to fire (depolarization block) and pyramidal cells start firing as a result of the additional depolarization due to loss of inhibition, multi-unit activity stagnates and starts to drop. A high frequency, large amplitude component (most probably pyramidal cell extracellular spikes, units) appears in the local field potential, accompanied by a positive envelope. (4) The local field potential normalizes, PVBC firing gradually recovers as cells exit the depolarization block, pyramidal cells stop firing and multi-unit activity drops. The figure illustrates the behaviour of the two cell types to compare them; the local field potential was recorded simultaneously with the PVBC, whereas the pyramidal cell was recorded in another experiment. However, in our experiments, pyramidal cells and PVBCs fired during the given phases as illustrated. HFO = high frequency oscillation; PV+ = parvalbumin-positive.

Figure 7

Build-up of depolarization during epileptiform events and its effect on neuronal firing: pyramidal cells start to fire massively when inhibition from PVBCs collapses. During the late interictal events we studied, four phases can be distinguished: (1) Primary depolarization, PVBCs depolarize and start firing, pyramidal cells start to depolarize, multi-unit activity starts to increase. There is a slow negative deflection in the local field potential. (2) Secondary depolarization, the frequency of PVBC firing further increases while the amplitude drops as the cell depolarizes even further, pyramidal cells depolarize further but do not fire yet, multi-unit activity increases heavily, steep negative drop in local field potential appears (LFP). (3) Interictal event builds up, the power in the 150–400 Hz band increases, PVBCs cease to fire (depolarization block) and pyramidal cells start firing as a result of the additional depolarization due to loss of inhibition, multi-unit activity stagnates and starts to drop. A high frequency, large amplitude component (most probably pyramidal cell extracellular spikes, units) appears in the local field potential, accompanied by a positive envelope. (4) The local field potential normalizes, PVBC firing gradually recovers as cells exit the depolarization block, pyramidal cells stop firing and multi-unit activity drops. The figure illustrates the behaviour of the two cell types to compare them; the local field potential was recorded simultaneously with the PVBC, whereas the pyramidal cell was recorded in another experiment. However, in our experiments, pyramidal cells and PVBCs fired during the given phases as illustrated. HFO = high frequency oscillation; PV+ = parvalbumin-positive.

These phases could be distinguished in the field and cell-attached recordings of the early and late high K+-induced interictal events (compare Fig. 3B and C), as well as in the other three models. It was the relative length and strength of the phases that were different in the early versus late interictal events.

As we have shown above, the firing pattern of neurons becomes altered during the states that generate epileptiform activity. This can be the result of either changes in cellular parameters important in signal integration, or alterations in the parameters of excitatory and inhibitory transmission. In the next steps we set out to reveal the possible basis of the observed alterations.

High K+ application alters cellular and network parameters

First we measured basic parameters of pyramidal cells (n = 9) in high K+ and compared them to control conditions. As shown in Fig. 8A we found that the membrane potential of cells depolarized from −59.2 mV (IQR −62.1 to −56.4) to −35.5 mV (IQR −41.2 to −27.1), their input resistance decreased from 68.2 MΩ (IQR 42.1–114.8) to 29.1 MΩ (IQR 28.8–44.2), and the threshold of the current injection required to induce at least one action potential (during a step protocol with 800 ms long de-and hyperpolarizing steps) decreased from 177 pA (IQR 153–213) to 37 pA (IQR −156 to 124).

Figure 8

High K+ increases excitability, and decreases inhibitory transmission relative to excitatory transmission. (A) When the extracellular K+ concentration is increased, pyramidal cells become more depolarized (top), input resistance of cells drops to 49 ± 4% (middle), and excitability increases (bottom), as cells fire the first action potential at a smaller membrane current in high K+ compared with control conditions (n = 7). (B–E) shows the increase in the ratio of excitatory to inhibitory transmission in high K+. The charge of inhibitory postsynaptic currents (IPSCs) evoked by stimulating perisomatic inhibitory axons (stratum lucidum/pyramidale) dropped after high K+ application (B). Similarly, in the case of dendritic inhibitory postsynaptic currents evoked by stimulating stratum radiatum the charge decreased in high K+ (C). However, when excitatory postsynaptic currents (EPSCs) were evoked in high K+ the charge increased (D). (E) Summary of the effects on synaptic transmission. (F and G) Paired recordings show that inhibitory charge decreases in high K+ in the case of PVBCs (n = 7) and axo-axonic cells (n = 7), but remains fairly intact in the case of CCK-positive basket cells (n = 5). H) Compared with control conditions (black), short-term depression becomes more pronounced in high K+ (grey) for the first 10 peaks (afterwards the synapse efficacy drops greatly) for PVBCs, whereas it does not change in the case of axo-axonic cells. In the case of CCK-positive basket cells, in high K+ the depression was less pronounced. Moreover plasticity could transiently switch from depression to facilitation. Asterisks illustrate significance at P < 0.05. AP = action potential; PV+ = parvalbumin-positive; uIPSC = unitary IPSC.

Figure 8

High K+ increases excitability, and decreases inhibitory transmission relative to excitatory transmission. (A) When the extracellular K+ concentration is increased, pyramidal cells become more depolarized (top), input resistance of cells drops to 49 ± 4% (middle), and excitability increases (bottom), as cells fire the first action potential at a smaller membrane current in high K+ compared with control conditions (n = 7). (B–E) shows the increase in the ratio of excitatory to inhibitory transmission in high K+. The charge of inhibitory postsynaptic currents (IPSCs) evoked by stimulating perisomatic inhibitory axons (stratum lucidum/pyramidale) dropped after high K+ application (B). Similarly, in the case of dendritic inhibitory postsynaptic currents evoked by stimulating stratum radiatum the charge decreased in high K+ (C). However, when excitatory postsynaptic currents (EPSCs) were evoked in high K+ the charge increased (D). (E) Summary of the effects on synaptic transmission. (F and G) Paired recordings show that inhibitory charge decreases in high K+ in the case of PVBCs (n = 7) and axo-axonic cells (n = 7), but remains fairly intact in the case of CCK-positive basket cells (n = 5). H) Compared with control conditions (black), short-term depression becomes more pronounced in high K+ (grey) for the first 10 peaks (afterwards the synapse efficacy drops greatly) for PVBCs, whereas it does not change in the case of axo-axonic cells. In the case of CCK-positive basket cells, in high K+ the depression was less pronounced. Moreover plasticity could transiently switch from depression to facilitation. Asterisks illustrate significance at P < 0.05. AP = action potential; PV+ = parvalbumin-positive; uIPSC = unitary IPSC.

Next we measured the effect of high K+ on excitatory and inhibitory transmission. In pyramidal cells we recorded inhibitory postsynaptic currents evoked with local electric stimulation at the border of strata pyramidale and lucidum (to measure changes in perisomatic inhibition), or in stratum radiatum (to estimate the alteration in dendritic inhibition). In both cases we found a significant decrease in inhibitory postsynaptic current amplitude to 45% of control (IQR 29.4–54.9) for perisomatic inhibition (P = 0.004 paired Wilcoxon test, Fig. 8B) and to 58.5% of control (36.2; 78.9) for dendritic inhibition (P = 0.002 paired Wilcoxon test, Fig. 8C). Then excitatory postsynaptic potentials evoked in stratum radiatum were recorded in pyramidal cells. When K+ was elevated, we found a significant increase in excitatory postsynaptic potential amplitude to 132.6% of control (IQR 92.3–216.1) (P = 0.002, paired Wilcoxon test, Fig. 8D and E).

The question arose if altered action potential shape (amplitude and/or width, charge transfer) can be responsible for the changes in excitatory and inhibitory transmission. We examined how these parameters changed during high K+ wash in (for more details see Supplementary material). As in both inhibitory cells and in pyramidal cells we saw similar changes (Supplementary Fig. 6), this cannot be a mechanism responsible for the simultaneous increase of excitatory transmission and decrease of inhibitory transmission.

These data indicate that the efficiency of synaptic inhibition is decreased, whereas excitatory synaptic transmission is increased in high K+. We observed the strongest depression in perisomatic inhibition, which is considered to be the most important in the control of pyramidal cell firing (Cobb et al., 1995; Miles et al., 1996). Therefore, we carried out paired recordings of monosynaptically coupled perisomatic inhibitory-pyramidal cell pairs (PVBC-pyramidal cell, axo-axonic cell-pyramidal cell and CCK-positive basket cell-pyramidal cell) to uncover the exact changes affecting the transmitter release of different perisomatic interneurons.

The strength and short-term depression of parvalbumin-containing basket cells inhibitory action is modulated by high K+ application

Presynaptic cells were targeted in slices prepared from transgenic mice expressing fluorescent markers in parvalbumin- or CCK-containing neurons. Postsynaptic currents were evoked by a train of action potentials triggered in the presynaptic cell, similar to the firing of these cells recorded during interictal events (30 action potentials with 150 Hz followed by four action potentials with 300 Hz). We compared the peak amplitudes and inhibitory charges recorded in pyramidal cells in control conditions and in elevated K+ (probably because of plastic processes in high K+, washout did not typically result in a complete reversal of the effects of treatment; see Fig. 8D). Inhibitory postsynaptic currents in pyramidal cells evoked by PVBCs (n = 7) and axo-axonic cells (n = 7) significantly decreased in amplitude in high K+. However, inhibitory postsynaptic currents in pyramidal cells evoked by CCK-positive basket cells (n = 5) remained fairly intact (for inhibitory postsynaptic current peak amplitudes: P = 0.03 for PVBC-pyramidal cell pairs, P = 0.03 for axo-axonic cell-pyramidal cell pairs and P = 0.44 for CCK-positive basket cell-pyramidal cell pairs; for inhibitory charge: P = 0.03 for PVBC-pyramidal cell pairs, P = 0.03 for axo-axonic cell-pyramidal cell pairs and P = 0.625 for CCK-positive basket cell-pyramidal cell pairs, paired Wilcoxon tests; to enhance readability, results are summarized in Fig. 8F and Table 4). We also found that both during control conditions and in high K+ the inhibitory postsynaptic currents often disappeared before the train ended during the train from PVBCs and axo-axonic cells, indicating that neurotransmission cannot be sustained at the high frequency of firing throughout interictal events. In contrast, a sustained transmission and asynchronous release was found (Hefft and Jonas, 2005; Szabo et al., 2010) in case of CCK-positive basket cells, even in the presence of high K+.

Our final question was whether short-term depression becomes altered in high K+. Therefore, the amplitude of the first 10 inhibitory postsynaptic currents was compared to the amplitude of the first inhibitory postsynaptic current (Pn/P1), illustrated in Fig. 8H. In the case of PVBCs (n = 7), short-term depression became more pronounced in high K+ (grey) for the first 10 peaks compared to control conditions (black, P < 0.01, Kolmogorov-Smirnov), whereas no such change appeared in the case of axo-axonic cells (n = 7, P = 0.42, Kolmogorov-Smirnov). In contrast, when inhibitory postsynaptic currents originating from CCK-positive basket cells (n = 5) were studied in high K+, the depression was less pronounced. Moreover, plasticity could transiently switch from depression to facilitation (P < 0.01, Kolmogorov-Smirnov, Fig. 8G).

Thus, perisomatic inhibition provided by parvalbumin-positive interneurons becomes largely ineffective during an epileptic event, whereas the inhibitory transmission of CCK-positive basket cells remains fairly intact or occasionally, even slightly increased.

Discussion

In the present study we wished to describe the difference between physiological sharp wave-ripples and interictal events. Unfortunately, the term sharp wave-ripple used by neurobiologists for physiological events is misleading for neurologists/clinicians who use the term ‘sharp wave’ to identify an EEG element occurring in association with epilepsy. This term was borrowed by the biologist from the clinical EEG nomenclature, but it identifies a healthy pattern essential in learning and memory formation.

We have shown that: (i) in vitro sharp wave-ripples and interictal events are indeed different network phenomena with distinct properties. Upon pharmacological interference sharp wave-ripples disappear and, following a transitory phase, the network activity reorganises into a new form of activity in all examined models; (ii) During interictal events all CA3 neurons fire with an increased firing rate compared with sharp wave-ripples. However, the firing of PVBCs and some axo-axonic cells stops (except for the gabazine model) as a result of depolarization block before the climax of the event; (iii) During interictal events the firing of PVBCs and pyramidal cells are complementary; i.e. pyramidal cells start firing when PVBCs get into depolarization block, whereas dendritic inhibitory cells fire strongly during all phases of the interictal events; (iv) In high K+ the balance of excitation to inhibition is shifted: inhibitory transmission is compromised, excitation is enhanced, and the integrative properties of pyramidal cells also change, resulting in higher excitability; (v) Inhibition collapses for several synergistic reasons: first, as PVBCs and axo-axonic cells enter into depolarization block, they stop firing action potentials; second, even when action potentials are generated, gamma-aminobutyric acid (GABA) release is greatly decreased (this is true for all recorded inhibitory cells, but mainly for PVBCs and axo-axonic cells); and, finally, the short-term depression of inhibitory postsynaptic currents originating from PVBCs is increased in high K+.

Physiological sharp wave-ripples and interictal events are distinct types of transient high activity events

Though the field potential signal of sharp wave-ripples and interictal events recorded in stratum pyramidale may look similar in shape, they differ in several features (amplitude, duration, accompanying multi-unit activity, firing pattern of neurons). As opposed to sharp wave-ripples, during interictal events all pyramidal cells are repetitively active. We proved that interictal events never evolve from sharp wave-ripples, but are separated by a featureless transitory phase, where coordinated firing characteristic of sharp wave-ripples is disrupted, the baseline activity (multi-unit activity) increases and disorganized firing evolves.

Highly active low synchrony states similar to the one observed during the transition phase were found in other in vitro epilepsy models induced by decreasing GABAergic inhibition (Cohen et al., 2006; de la Prida et al., 2006), adding 4-aminopyridine (Perreault and Avoli, 1991; Barbarosie and Avoli, 1997), omitting Mg2+ (Whittington et al., 1995; Huberfeld et al., 2011) or decreasing Ca2+ (Bikson et al., 2003), strengthening the hypothesis that sharp wave-ripples and epileptic events represent different types of network activity. However, this activity did not persist as a stable network state. When the population activity reached a critical level, a new type of synchrony, interictal events (as well as at later stages epileptic events of different complexity), appeared in our slices, similar to results published earlier (Khosravani et al., 2005). The results of the theoretical paper by Brunel and Wang (2003) might explain why the two types of transient synchrony are mutually exclusive, as well as the presence of the unstructured gap between them. They explored the effect of changing excitatory and inhibitory transmission parameters, showing that network dynamics could be pushed from one type of oscillation to a mechanistically different one with an unorganized/asynchronous state in between. They reasoned that there are parameter ranges where the network cannot generate a synchronous state, as the proper timing of recurrent feed-back mechanisms is not ensured.

Inhibitory control especially from parvalbumin-positive basket cells fails during interictal events

At the beginning of an interictal event, the spontaneous firing of highly excitable pyramidal cells may reach a level of run-away excitation (Lux and Heinemann, 1978; Frohlich et al., 2008) and the build-up of excitation in the initiating pyramidal cell population is manifested as the first step of depolarization both in inhibitory neurons and pyramidal cells (Fig. 8). Although in our set of pyramidal cells we could hardly see any spiking in association with sharp wave-ripples, in a recently published, larger data set we did see a subset of pyramidal cells firing before and during sharp wave-ripples (Hajos et al., 2013). Furthermore, as shown in Supplementary Fig. 5 and Figs 5 and 7, phase 1 and 2, intracellular recordings reveal that sharp wave-ripples and interictal events are preceded by a build-up of depolarizing potentials, indicating increasing excitatory neuron activity.

In parallel with the increase in excitation, inhibitory neuron populations start to be activated. There is a level of excitation, however, when PVBCs and some axo-axonic cells enter into depolarization block (in most models), and the build-up of excitation enters an uncontrolled state where all pyramidal cells (relieved from perisomatic inhibition) fire repetitively. In the 0 Mg2+ model it has previously been shown that pyramidal cells do activate, and interictal events can propagate with a larger speed when inhibition is terminated (Trevelyan et al., 2006, 2007).

Similar inactivation (depolarization block) of fast-spiking cells has previously been described in cortical slices using the 0 Mg2+ or the 4-aminopyridine model (Kawaguchi, 2001; Cammarota, 2013). The validity of our finding is further supported by recent clinical findings in epileptic patients showing that inhibitory cells enter into depolarization block at the beginning of seizures (Ahmed et al., 2012). In addition, recent animal studies have shown that decreasing the activity of pyramidal cells could delay electrographic and behavioural initiation of status epilepticus (Sukhotinsky et al., 2013) or decrease paroxysmal activity in cell culture (Tonnesen et al., 2009). Besides inhibiting pyramidal cells, activating parvalbumin-containing cells can also reduce seizure frequency of epilepsy (Krook-Magnuson et al., 2013), indicating that restoring the activity of PVBCs could effectively decrease pyramidal cell firing and control network activity. It is a future task, however, to build a clinical approach aiming to normalize the firing of strategically crucial neuron types.

Why is inhibitory control sufficient during sharp wave-ripples, whereas it fails during interictal events? It seems that inhibition is compromised in the high K+ state at three stages: (i) there is a general decrease of inhibitory transmission strength even for single action potentials, especially for perisomatic inhibition (Fig. 8B and C); (ii) the transmission of parvalbumin-positive cells, characterized by multiple high-frequency spiking during an epileptiform event suffers a strong (for PVBCs an almost complete) short-term depression (Fig. 8G); and (iii) the most effective inhibitory neurons, PVBCs, enter into depolarization block before the peak of interictal events (Figs 3–5). Although dendritic inhibitory neurons and CCK-positive basket cells keep firing with increasing frequency and their transmission is potentiated somewhat, it seems that they cannot control the runaway firing of pyramidal cells during interictal events (they actually might even promote pyramidal cell firing by reducing their entry into depolarization block, see below).

Features of parvalbumin-positive basket cells that make them vulnerable to excess excitation

The next question is why PVBCs and some axo-axonic cells get into depolarization block whereas other interneurons and pyramidal cells escape. Depolarization block of neurons was observed in pioneering in vivo studies of neocortical and hippocampal seizures using intracellular recordings from single unidentified cells (Kandel and Spencer, 1961; Matsumoto and Marsan, 1964). The involvement of Na+ channels and persistent sodium currents has been shown to be critical in the evolution of depolarization block (Bikson et al., 2003). The transition from sustained spiking to depolarization block can be described using dynamical systems theory (Izhikevich, 2007; Dovzhenok and Kuznetsov, 2012). Although no detailed mathematical analysis of the relationship between cellular parameters and the input current required for depolarization block has been performed, the conditions leading to depolarization block likely depend on the properties (such as the voltage-dependence and kinetics of channel activation and inactivation) and densities of spike-generating currents, but may also be influenced by slower (e.g. adaptation) currents. The properties of both Na+ and K+ channels are known to be different in PVBCs and pyramidal cells (Martina and Jonas, 1997; Martina et al., 1998). Another important factor which determines whether depolarization block occurs is the net input current (synaptic current) received by the neuron, which may differ substantially among the different cell types. In fact, we showed here that during interictal events, PVBCs and axo-axonic cells reach more depolarized membrane potentials compared with other cell types. This can be explained by the fact that parvalbumin-positive cells receive significantly more excitatory (∼15 000) inputs, balanced with a weak inhibition (6%) compared with CCK-positive (∼5000 excitatory input, 35% inhibition) and dendritic inhibitory neurons (∼2600 excitatory input, 29% inhibition; Gulyas et al., 1999; Matyas et al., 2004). On the other hand, pyramidal cells have a dendritic input organization that is rather similar to that of parvalbumin-positive cells. They receive a large amount of excitatory input (∼30 000 synapses) that is balanced only by a weak inhibition (5.3%; Megias et al., 2001); still, they do not get into depolarization block during interictal events. One of the reasons for this is probably the observed differences in spike-generating currents as discussed above; another important difference may be the presence of slower adaptation currents (such as M-type and slow after hyperpolarization K+ currents) in pyramidal cells, but not in PVBCs. Most importantly, the perisomatic input organization of the two cell types is different. PVBCs do receive perisomatic and somatic excitatory inputs, whereas pyramidal cells are devoid of them, and therefore might not experience the same depolarization block as PVBCs do (Gulyas et al., 1999; Megias et al., 2001). It has been shown that in neurons not receiving perisomatic excitation, the soma and the axon are electrotonically distant (Rancz and Hausser, 2010), and therefore might not experience the same depolarization as neurons with perisomatic excitatory input. Thus, a unique combination of cellular and connectivity parameters in PVBCs may act synergistically to explain the presence of depolarization block during interictal events. As a result, the enhanced excitation effectively activates pyramidal cells without causing a depolarization block, and without PVBC-mediated control over spiking, the network activity further increases.

Spike time histograms showed that except for PVBCs, all neuron types could considerably increase their firing rate during interictal events compared to sharp wave-ripples. An important implication of our results is that PVBCs, by receiving a large amount of excitatory input and expressing the proper combination of ion channels, are tuned to be able to fire maximally during physiological transient high activity events, i.e. sharp wave-ripples. On the other hand, a detrimental consequence of this fine tuning is that when they receive a pathologically high level of excitation, they enter into depolarization block, i.e. the ‘fuse blows’.

Absolute and relative changes in cellular and network properties underlie the switch from healthy to pathological synchrony in the high K+ model

Combining the results of Brunel and Wang (2003) with our findings (long-term changes of cellular and network parameters and transient changes in inhibitory transmission, firing patterns of different cell types) as well as with earlier results demonstrating that pyramidal cells and interneurons become activated at different times during interictal events (Trevelyan et al., 2006; Ziburkus et al., 2006; Spampanato and Mody, 2007), we suggest the following sequence of events during the evolution of interictal events:

  • Initiation stage: deflections in the local field potential, increasing spiking of neurons (multi-units and cell-attached spikes) as well as two-phased depolarization of intracellular potentials suggest that excitatory activity starts to build up gradually at the beginning of an interictal events (phase 1 and 2 in Fig. 7), similar to the case of sharp wave-ripples (Hajos et al., 2013). However, as the strength and the balance of excitatory and inhibitory transmission, as well as cellular excitability, is shifted by high K+ application (compared with sharp wave-ripples), although inhibition is still present in this phase, it cannot restrain the excitation from building-up further in the recurrent system of CA3.

  • Pyramidal cell firing/high frequency oscillation stage: as the activity in the system increases beyond the physiological level, perisomatic inhibition totally fails (phase 3) through different synergistic mechanisms (depolarization block of PVBCs, short-term depression of inhibitory postsynaptic currents) and most pyramidal cells start to fire at high frequency. These high frequency, synchronized action potentials manifest as high frequency oscillations in the local field potential (Dzhala and Staley, 2004; Foffani et al., 2007). It is important to note that as perisomatic inhibitory transmission has collapsed at this point, synchronous inhibitory postsynaptic potentials do not contribute to local field potential generation in stratum pyramidale, unlike in the case of the generation of the ripples of sharp wave-ripples (Csicsvari et al., 1999; Klausberger et al., 2003; Le Van Quyen et al., 2008; Hajos et al., 2013).

  • Termination stage: examining phase 4 in Fig. 7 might give a clue as to why interictal events terminate. Here we can see that pyramidal cell firing accommodates and starts to slow down, whereas PVBCs are still in depolarization block (the same is visible in the drop of multi-unit frequency for all models in Fig. 1B). So it is the refractoriness of the pyramidal cells that is the first step in the termination. As pyramidal cells fire less, inhibition regains control and terminates interictal events. Paradoxically, the depolarization block of PVBCs (evidently a refractory mechanism) can help terminate interictal events, as while the cells are not firing their inhibitory transmission might recover from the strong depression, and when the cells start firing again their inhibition is effective. The decreasing firing frequency of PVBCs in phase 4 (Fig. 7) indicates that their excitatory drive decreases (note that before they stop firing as a result of depolarization block their firing frequency keeps increasing, so they do not slow because of accommodation, but because of decreasing drive). Most probably several refractory mechanisms are engaged by the end of phase 3 due to the repetitive high frequency firing of the pyramidal cells. Collapse of glutamatergic transmission or refractoriness of the firing might be elements that result in decreased pyramidal cell firing and the recovery of inhibition.

Isomorphic mechanisms might lead to interictal events in different models and play a role in the generation of more complex epileptic events

We measured in detail how cellular and network parameters change in the high K+ model. We found that the excitability of neurons, as well as the absolute and relative values of excitatory and inhibitory transmission become altered. Other epilepsy-inducing treatments used also evoke changes in these critical parameters. 4-Aminopyridine application causes a blockade of voltage-gated K+ channels (Glover, 1982; Rudy, 1988; Choquet and Korn, 1992), and thus changes input resistance and membrane potential. Zero Mg2+ activates N-methyl-d-aspartate receptors and induces an increase in excitatory transmission, leading to direct depolarization (Flatman et al., 1983; Coan and Collingridge, 1985; Mody et al., 1987) and long term potentiation of synaptic transmission (Kauer et al., 1988). Gabazine directly increases excitability by blocking synaptic and non-synaptic GABAA receptors, thus eliminating inhibition and depolarizing cells (Heaulme et al., 1987; Aradi and Maccaferri, 2004; Wlodarczyk et al., 2013), as well as by increasing input resistance (Cope et al., 2005). Recent work by Aivar and Prida (personal communication) also demonstrates changes in the strength and ratio of excitatory and inhibitory transmission in an in vitro epileptic model induced by low Ca2+ concentration.

The findings of Brunel and Wang (2003), especially in the light of their later study (Geisler et al., 2005), leaves enough room for possible parameter changes to push the system into different dynamics. The work of Marder (2011) also emphasizes that in the case of neurons and networks, several different sets of parameter combinations can result in similar or identical behaviours. Recent results of genetic studies revealed that epilepsies with similar symptomatology can be caused by highly distinct mutations in different genes (Poduri and Lowenstein, 2011; Allen et al., 2013).

These ideas drove us to suggest a common framework for all the studied models: As a first step, shifts of different nature in excitability, and a change in the ratio of excitation versus inhibition, are induced in the different epilepsy induction models, resulting in a pathological, uncontrolled increase of activity during the initiation phase of interictal events.

As a second step, inhibition fails during the pyramidal cell firing/high frequency oscillation stage in all but the gabazine model. After the activity builds up to a sufficiently high level, excitatory drive makes PVBCs fire at a non-physiologically high frequency, resulting in strong short-term depression of inhibitory transmission that releases pyramidal cells from inhibition, which then start firing at high frequency, generating the high frequency oscillation in the local field potential. Inhibition is further damaged because PVBCs enter into an additional depolarization block that evolves because elevated K+, 4-aminopyridine and 0 Mg2+ causes direct strong depolarization of the membrane that is further boosted by the strong excitation during the initiation stage.

PVBCs do not enter depolarization block in the gabazine model, probably because although gabazine causes some depolarization by inhibiting synaptic and non-synaptic GABAA currents (Cope et al., 2005), this depolarization may not be enough to push the membrane into depolarization block. Yet, the pyramidal cell firing/high frequency oscillation stage is present in this model too. We believe that the initiation stage starts due to compromised inhibition (so the lack of PVBC depolarization block is irrelevant), and at some stage the uncontrolled build-up of excitation causes masses of pyramidal cells to reach the threshold for high frequency firing, and the interictal event enters the pyramidal cell firing/high frequency oscillation stage.

In these models, the last step, the termination stage of interictal events, is possibly driven by the same refractory mechanisms in the excitatory system as in the high K+ model, as we see a similar drop in pyramidal cell firing by the end of the second stage.

To explain all stages of the interictal events, we invoked two layers of parameter changes on different time scales. The prolonged parameter modulation, the difference between the healthy and the epileptic state, manifests on the several-minute time-scale of epileptiform activity induction. In epileptic patients this is probably the original reason why epilepsy starts: later events are consequences. The alteration of parameters means that interictal events are initiated instead of sharp wave-ripples. As a second layer, short term changes (collapse of inhibition, refractoriness of excitation on the 200–1000 ms scale of an interictal event) allow uncontrolled firing and later termination of a single cycle. If we want to explain the evolution of epileptiform event forms (early and late interictal, preictal and ictal), we have to invoke a third layer of changes that is a superimposed slow drift in the parameters (on the 1–5-min scale), most probably evoked by pathologically high activity (changes in K+ levels, metabolic exhaustion, potentiation of synaptic weights, etc.). This drift results in changes in the initiation rate and recurrent structure of interictal event-like bursts, as well as in the relative length of their different stages (e.g. high frequency oscillation stage is longer during late than during early interictal events). We propose that preictal and ictal events are combinations of repetitive, degenerated interictal events. There must be a secondary refractory mechanism associated with the third layer as well, as repetitive ictal events are almost always followed by silent periods lasting for minutes.

Acknowledgements

Prof. Péter Halász (National Institute of Neurosciences, Budapest, Hungary) and Dr. Dániel Fabó (National Institute of Neurosciences, Budapest, Hungary) kindly helped us to classify in vitro epileptiform activities. We are grateful to those who were kind to read the manuscript: Marco de Curtis, Norbert Hájos and Richard Miles. We also thank Dr. Hannah Monyer for generously providing the PV eGFP mice. The authors thank The authors also thank the Nikon Microscopy Center at IEM, Nikon Austria GmbH and Auro-Science Consulting Ltd for kindly providing microscopy support. Katalin Lengyel, Erzsebet Gregori and Győző Goda for excellent technical support.

Funding

This work was supported by:, Hungarian Scientific Research Fund (OTKA K83251 and OTKA 81357), European Research Council Advanced grant for T.F.F. (ERC-2011-ADG-294313) (SERRACO), National Office for Research and Technology NKTH-ANR, Neurogen and Multisca, European Union Seventh Framework Program (NeuroSeker) and TÁMOP-4.2.1.B-11/2/KMR-2011-0002

Supplementary material

Supplementary material is available at Brain online.

Abbreviations

    Abbreviations
  • CA

    cornu ammonis

  • PVBC

    parvalbumin-positive basket cell

References

Ahmed
OJ
Truccolo
W
Donoghue
JA
Eskandar
EN
Cosgrove
GR
Potter
NS
, et al.  . 
Human inhibitory single neurons switch off before dramatic increases in seizure intensity
 , 
2012
San Diego
AES Meeting
Aldenkamp
AP
Beitler
J
Arends
J
van der Linden
I
Diepman
L
Acute effects of subclinical epileptiform EEG discharges on cognitive activation
Funct Neurol
 , 
2005
, vol. 
20
 (pg. 
23
-
8
)
Allen
AS
Berkovic
SF
Cossette
P
Delanty
N
Dlugos
D
Eichler epileptiform event
, et al.  . 
De novo mutations in epileptic encephalopathies
Nature
 , 
2013
, vol. 
501
 (pg. 
217
-
21
)
Aradi
I
Maccaferri
G
Cell type-specific synaptic dynamics of synchronized bursting in the juvenile CA3 rat hippocampus
J Neurosci
 , 
2004
, vol. 
24
 (pg. 
9681
-
92
)
Barbarosie
M
Avoli
M
CA3-driven hippocampal-entorhinal loop controls rather than sustains in vitro limbic seizures
J Neurosci
 , 
1997
, vol. 
17
 (pg. 
9308
-
14
)
Bikson
M
Hahn
PJ
Fox
JE
Jefferys
JG
Depolarization block of neurons during maintenance of electrographic seizures
J Neurophysiol
 , 
2003
, vol. 
90
 (pg. 
2402
-
8
)
Boksa
P
Wilson
D
Rochford
J
Responses to stress and novelty in adult rats born vaginally, by cesarean section or by cesarean section with acute anoxia
Biol Neonate
 , 
1998
, vol. 
74
 (pg. 
48
-
59
)
Bragin
A
Jando
G
Nadasdy
Z
Hetke
J
Wise
K
Buzsaki
G
Gamma (40-100 Hz) oscillation in the hippocampus of the behaving rat
J Neurosci
 , 
1995
, vol. 
15
 
1 Pt 1
(pg. 
47
-
60
)
Bragin
A
Mody
I
Wilson
CL
Engel
J
Jr
Local generation of fast ripples in epileptic brain
J Neurosci
 , 
2002
, vol. 
22
 (pg. 
2012
-
21
)
Brunel
N
Wang
XJ
What determines the frequency of fast network oscillations with irregular neural discharges? I
Synaptic dynamics and excitation-inhibition balance. J Neurophysiol
 , 
2003
, vol. 
90
 (pg. 
415
-
30
)
Buzsaki
G
Hippocampal sharp waves: their origin and significance
Brain Res
 , 
1986
, vol. 
398
 (pg. 
242
-
52
)
Buzsaki
G
Rhythms of the Brain
 , 
2006
New York
Oxford University Press
Cammarota
M
Losi
G
Chiavegato
A
Zonta
M
Carmignoto
G
Fast spiking interneuron control of seizure propagation in a cortical slice model of focal epilepsy
J Physiol
 , 
2013
, vol. 
591
 
Pt 4
(pg. 
807
-
22
)
Choquet
D
Korn
H
Mechanism of 4-aminopyridine action on voltage-gated potassium channels in lymphocytes
J Gen Physiol
 , 
1992
, vol. 
99
 (pg. 
217
-
40
)
Coan
EJ
Collingridge
GL
Magnesium ions block an N-methyl-D-aspartate receptor-mediated component of synaptic transmission in rat hippocampus
Neurosci Lett
 , 
1985
, vol. 
53
 (pg. 
21
-
6
)
Cobb
SR
Buhl
EH
Halasy
K
Paulsen
O
Somogyi
P
Synchronization of neuronal activity in hippocampus by individual GABAergic interneurons
Nature
 , 
1995
, vol. 
378
 (pg. 
75
-
8
)
Cohen
I
Huberfeld
G
Miles
R
Emergence of disinhibition-induced synchrony in the CA3 region of the guinea pig hippocampus in vitro
J Physiol
 , 
2006
, vol. 
570
 
Pt 3
(pg. 
583
-
94
)
Cope
DW
Hughes
SW
Crunelli
V
GABAA receptor-mediated tonic inhibition in thalamic neurons
J Neurosci
 , 
2005
, vol. 
25
 (pg. 
11553
-
63
)
Cressman
JR
Jr.
Ullah
G
Ziburkus
J
Schiff
SJ
Barreto
E
The influence of sodium and potassium dynamics on excitability, seizures, and the stability of persistent states: I. Single neuron dynamics
J Comput Neurosci
 , 
2009
, vol. 
26
 (pg. 
159
-
70
)
Csicsvari
J
Hirase
H
Czurko
A
Mamiya
A
Buzsaki
G
Oscillatory coupling of hippocampal pyramidal cells and interneurons in the behaving Rat
J Neurosci
 , 
1999
, vol. 
19
 (pg. 
274
-
87
)
de la Prida
LM
Huberfeld
G
Cohen
I
Miles
R
Threshold behavior in the initiation of hippocampal population bursts
Neuron
 , 
2006
, vol. 
49
 (pg. 
131
-
42
)
Dovzhenok
A
Kuznetsov
AS
Exploring neuronal bistability at the depolarization block
PloS One
 , 
2012
, vol. 
7
 pg. 
e42811
 
Dreier
JP
Heinemann
U
Regional and time dependent variations of low Mg2+ induced epileptiform activity in rat temporal cortex slices
Exp Brain Res
 , 
1991
, vol. 
87
 (pg. 
581
-
96
)
Dzhala
VI
Staley
KJ
Mechanisms of fast ripples in the hippocampus
J Neurosci
 , 
2004
, vol. 
24
 (pg. 
8896
-
906
)
Ellender
TJ
Nissen
W
Colgin
LL
Mann
EO
Paulsen
O
Priming of hippocampal population bursts by individual perisomatic-targeting interneurons
J Neurosci
 , 
2010
, vol. 
30
 (pg. 
5979
-
91
)
Engel
J
Jr
Introduction to temporal lobe epilepsy
Epilepsy Res
 , 
1996
, vol. 
26
 (pg. 
141
-
50
)
Engel
J
Jr.
Bragin
A
Staba
R
Mody
I
High-frequency oscillations: what is normal and what is not?
Epilepsia
 , 
2009
, vol. 
50
 (pg. 
598
-
604
)
Flatman
JA
Schwindt
PC
Crill
WE
Stafstrom
CE
Multiple actions of N-methyl-D-aspartate on cat neocortical neurons in vitro
Brain research
 , 
1983
, vol. 
266
 (pg. 
169
-
73
)
Foffani
G
Uzcategui
YG
Gal
B
Menendez de la Prida
L
Reduced spike-timing reliability correlates with the emergence of fast ripples in the rat epileptic hippocampus
Neuron
 , 
2007
, vol. 
55
 (pg. 
930
-
41
)
Freund
TF
Buzsaki
G
Interneurons of the hippocampus
Hippocampus
 , 
1996
, vol. 
6
 (pg. 
347
-
470
)
Frohlich
F
Bazhenov
M
Iragui-Madoz
V
Sejnowski
TJ
Potassium dynamics in the epileptic cortex: new insights on an old topic
Neuroscientist
 , 
2008
, vol. 
14
 (pg. 
422
-
33
)
Geisler
C
Brunel
N
Wang
XJ
Contributions of intrinsic membrane dynamics to fast network oscillations with irregular neuronal discharges
J Neurophysiol
 , 
2005
, vol. 
94
 (pg. 
4344
-
61
)
Girardeau
G
Benchenane
K
Wiener
SI
Buzsaki
G
Zugaro
MB
Selective suppression of hippocampal ripples impairs spatial memory
Nat Neurosci
 , 
2009
, vol. 
12
 (pg. 
1222
-
3
)
Glover
WE
The aminopyridines
Gen Pharmacol
 , 
1982
, vol. 
13
 (pg. 
259
-
85
)
Gnatkovsky
V
Librizzi
L
Trombin
F
de Curtis
M
Fast activity at seizure onset is mediated by inhibitory circuits in the entorhinal cortex in vitro
Ann Neurol
 , 
2008
, vol. 
64
 (pg. 
674
-
86
)
Gulyas
AI
Megias
M
Emri
Z
Freund
TF
Total number and ratio of excitatory and inhibitory synapses converging onto single interneurons of different types in the CA1 area of the rat hippocampus
J Neurosci
 , 
1999
, vol. 
19
 (pg. 
10082
-
97
)
Hablitz
JJ
Picrotoxin-induced epileptiform activity in hippocampus: role of endogenous versus synaptic factors
J Neurophysiol
 , 
1984
, vol. 
51
 (pg. 
1011
-
27
)
Hajos
N
Ellender
TJ
Zemankovics
R
Mann
EO
Exley
R
Cragg
SJ
, et al.  . 
Maintaining network activity in submerged hippocampal slices: importance of oxygen supply
Eur J Neurosci
 , 
2009
, vol. 
29
 (pg. 
319
-
27
)
Hajos
N
Karlocai
MR
Nemeth
B
Ulbert
I
Monyer
H
Szabo
G
, et al.  . 
Input-output features of anatomically identified CA3 neurons during hippocampal sharp wave/ripple oscillation in vitro
J Neurosci
 , 
2013
, vol. 
33
 (pg. 
11677
-
91
)
Heaulme
M
Chambon
JP
Leyris
R
Wermuth
CG
Biziere
K
Characterization of the binding of [3H]SR 95531, a GABAA antagonist, to rat brain membranes
J Neurochem
 , 
1987
, vol. 
48
 (pg. 
1677
-
86
)
Hefft
S
Jonas
P
Asynchronous GABA release generates long-lasting inhibition at a hippocampal interneuron-principal neuron synapse
Nat Neurosci
 , 
2005
, vol. 
8
 (pg. 
1319
-
28
)
Holmes
GL
Lenck-Santini
PP
Role of interictal epileptiform abnormalities in cognitive impairment
Epilepsy Behav
 , 
2006
, vol. 
8
 (pg. 
504
-
15
)
Huberfeld
G
Menendez de la Prida
L
Pallud
J
Cohen
I
Le Van Quyen
M
Adam
C
, et al.  . 
Glutamatergic pre-ictal discharges emerge at the transition to seizure in human epilepsy
Nat Neurosci
 , 
2011
, vol. 
14
 (pg. 
627
-
34
)
Isomura
Y
Sirota
A
Ozen
S
Montgomery
S
Mizuseki
K
Henze
DA
, et al.  . 
Integration and segregation of activity in entorhinal-hippocampal subregions by neocortical slow oscillations
Neuron
 , 
2006
, vol. 
52
 (pg. 
871
-
82
)
Izhikevich
E
Dynamical systems in neuroscience
 , 
2007
Cambridge, MA
The MIT Press
Jadhav
SP
Kemere
C
German
PW
Frank
LM
Awake hippocampal sharp-wave ripples support spatial memory
Science
 , 
2012
, vol. 
336
 (pg. 
1454
-
8
)
Jones
RS
Heinemann
U
Synaptic and intrinsic responses of medical entorhinal cortical cells in normal and magnesium-free medium in vitro
J Neurophysiol
 , 
1988
, vol. 
59
 (pg. 
1476
-
96
)
Kandel
ER
Spencer
WA
The pyramidal cell during hippocampal seizure
Epilepsia
 , 
1961
, vol. 
2
 (pg. 
63
-
9
)
Kauer
JA
Malenka
RC
Nicoll
RA
NMDA application potentiates synaptic transmission in the hippocampus
Nature
 , 
1988
, vol. 
334
 (pg. 
250
-
2
)
Kawaguchi
Y
Distinct firing patterns of neuronal subtypes in cortical synchronized activities
J Neurosci
 , 
2001
, vol. 
21
 (pg. 
7261
-
72
)
Khosravani
H
Pinnegar
CR
Mitchell
JR
Bardakjian
BL
Federico
P
Carlen
PL
Increased high-frequency oscillations precede in vitro low-Mg seizures
Epilepsia
 , 
2005
, vol. 
46
 (pg. 
1188
-
97
)
Klausberger
T
Magill
PJ
Marton
LF
Roberts
JD
Cobden
PM
Buzsaki
G
, et al.  . 
Brain-state- and cell-type-specific firing of hippocampal interneurons in vivo
Nature
 , 
2003
, vol. 
421
 (pg. 
844
-
8
)
Klausberger
T
Somogyi
P
Neuronal diversity and temporal dynamics: the unity of hippocampal circuit operations
Science
 , 
2008
, vol. 
321
 (pg. 
53
-
7
)
Krook-Magnuson
E
Armstrong
C
Oijala
M
Soltesz
I
On-demand optogenetic control of spontaneous seizures in temporal lobe epilepsy
Nat Commun
 , 
2013
, vol. 
4
 pg. 
1376
 
Kubota
D
Colgin
LL
Casale
M
Brucher
FA
Lynch
G
Endogenous waves in hippocampal slices
J Neurophysiol
 , 
2003
, vol. 
89
 (pg. 
81
-
9
)
Le Van Quyen
M
Bragin
A
Staba
R
Crepon
B
Wilson
CL
Engel
J
Jr
Cell type-specific firing during ripple oscillations in the hippocampal formation of humans
J Neurosci
 , 
2008
, vol. 
28
 (pg. 
6104
-
10
)
Levesque
M
Bortel
A
Gotman
J
Avoli
M
High-frequency (80-500 Hz) oscillations and epileptogenesis in temporal lobe epilepsy
Neurobiol Dis
 , 
2011
, vol. 
42
 (pg. 
231
-
41
)
Louvel
J
Avoli
M
Kurcewicz
I
Pumain
R
Extracellular free potassium during synchronous activity induced by 4-aminopyridine in the juvenile rat hippocampus
Neurosci Lett
 , 
1994
, vol. 
167
 (pg. 
97
-
100
)
Lux
HD
Heinemann
U
Ionic changes during experimentally induced seizure activity
Electroencephalogr Clin Neurophysiol Suppl
 , 
1978
, vol. 
34
 (pg. 
289
-
97
)
Marder
E
Variability, compensation, and modulation in neurons and circuits
Proc Natl Acad Sci USA
 , 
2011
, vol. 
108
 
Suppl 3
(pg. 
15542
-
8
)
Martina
M
Jonas
P
Functional differences in Na+ channel gating between fast-spiking interneurones and principal neurones of rat hippocampus
J Physiol
 , 
1997
, vol. 
505
 
Pt 3
(pg. 
593
-
603
)
Martina
M
Schultz
JH
Ehmke
H
Monyer
H
Jonas
P
Functional and molecular differences between voltage-gated K+ channels of fast-spiking interneurons and pyramidal neurons of rat hippocampus
J Neurosci
 , 
1998
, vol. 
18
 (pg. 
8111
-
25
)
Matsumoto
H
Marsan
CA
Cortical Cellular Phenomena in Experimental Epilepsy: Ictal Manifestations
Exp Neurol
 , 
1964
, vol. 
9
 (pg. 
305
-
26
)
Matyas
F
Freund
TF
Gulyas
AI
Convergence of excitatory and inhibitory inputs onto CCK-containing basket cells in the CA1 area of the rat hippocampus
Eur J Neurosci
 , 
2004
, vol. 
19
 (pg. 
1243
-
56
)
Megias
M
Emri
Z
Freund
TF
Gulyas
AI
Total number and distribution of inhibitory and excitatory synapses on hippocampal CA1 pyramidal cells
Neuroscience
 , 
2001
, vol. 
102
 (pg. 
527
-
40
)
Meyer
AH
Katona
I
Blatow
M
Rozov
A
Monyer
H
In vivo labeling of parvalbumin-positive interneurons and analysis of electrical coupling in identified neurons
J Neurosci
 , 
2002
, vol. 
22
 (pg. 
7055
-
64
)
Miles
R
Toth
K
Gulyas
AI
Hajos
N
Freund
TF
Differences between somatic and dendritic inhibition in the hippocampus
Neuron
 , 
1996
, vol. 
16
 (pg. 
815
-
23
)
Miles
R
Traub
RD
Wong
RK
Spread of synchronous firing in longitudinal slices from the CA3 region of the hippocampus
J Neurophysiol
 , 
1988
, vol. 
60
 (pg. 
1481
-
96
)
Miles
R
Wong
RK
Traub
RD
Synchronized afterdischarges in the hippocampus: contribution of local synaptic interactions
Neuroscience
 , 
1984
, vol. 
12
 (pg. 
1179
-
89
)
Mody
I
Lambert
JD
Heinemann
U
Low extracellular magnesium induces epileptiform activity and spreading depression in rat hippocampal slices
J Neurophysiol
 , 
1987
, vol. 
57
 (pg. 
869
-
88
)
Moody
WJ
Futamachi
KJ
Prince
DA
Extracellular potassium activity during epileptogenesis
Exp Neurol
 , 
1974
, vol. 
42
 (pg. 
248
-
63
)
Perreault
P
Avoli
M
Physiology and pharmacology of epileptiform activity induced by 4-aminopyridine in rat hippocampal slices
J Neurophysiol
 , 
1991
, vol. 
65
 (pg. 
771
-
85
)
Poduri
A
Lowenstein
D
Epilepsy genetics–past, present, and future
Curr Opin Genet Dev
 , 
2011
, vol. 
21
 (pg. 
325
-
32
)
Prida
LM
Sanchez-Andres
JV
Nonlinear frequency-dependent synchronization in the developing hippocampus
J Neurophysiol
 , 
1999
, vol. 
82
 (pg. 
202
-
8
)
Rancz
EA
Hausser
M
Dendritic spikes mediate negative synaptic gain control in cerebellar Purkinje cells
Proc Natl Acad Sci USA
 , 
2010
, vol. 
107
 (pg. 
22284
-
9
)
Rudy
B
Diversity and ubiquity of K channels
Neuroscience
 , 
1988
, vol. 
25
 (pg. 
729
-
49
)
Rutecki
PA
Lebeda
FJ
Johnston
D
4-Aminopyridine produces epileptiform activity in hippocampus and enhances synaptic excitation and inhibition
J Neurophysiol
 , 
1987
, vol. 
57
 (pg. 
1911
-
24
)
Schwartzkroin
PA
Prince
DA
Penicillin-induced epileptiform activity in the hippocampal in vitro prepatation
Ann Neurol
 , 
1977
, vol. 
1
 (pg. 
463
-
9
)
Sirota
A
Csicsvari
J
Buhl
D
Buzsaki
G
Communication between neocortex and hippocampus during sleep in rodents
Proc Natl Acad Sci USA
 , 
2003
, vol. 
100
 (pg. 
2065
-
9
)
Soltesz
I
Deschenes
M
Low- and high-frequency membrane potential oscillations during theta activity in CA1 and CA3 pyramidal neurons of the rat hippocampus under ketamine-xylazine anesthesia
J Neurophysiol
 , 
1993
, vol. 
70
 (pg. 
97
-
116
)
Spampanato
J
Mody
I
Spike timing of lacunosom-moleculare targeting interneurons and CA3 pyramidal cells during high-frequency network oscillations in vitro
J Neurophysiol
 , 
2007
, vol. 
98
 (pg. 
96
-
104
)
Spruston
N
Johnston
D
Perforated patch-clamp analysis of the passive membrane properties of three classes of hippocampal neurons
J Neurophysiol
 , 
1992
, vol. 
67
 (pg. 
508
-
29
)
Staley
KJ
Bains
JS
Yee
A
Hellier
J
Longacher
JM
Statistical model relating CA3 burst probability to recovery from burst-induced depression at recurrent collateral synapses
J Neurophysiol
 , 
2001
, vol. 
86
 (pg. 
2736
-
47
)
Steriade
M
To burst, or rather, not to burst
Nat Neurosci
 , 
2001
, vol. 
4
 pg. 
671
 
Steriade
M
Nunez
A
Amzica
F
Intracellular analysis of relations between the slow (<1 Hz) neocortical oscillation and other sleep rhythms of the electroencephalogram
J Neurosci
 , 
1993a
, vol. 
13
 (pg. 
3266
-
83
)
Steriade
M
Nunez
A
Amzica
F
A novel slow (<1 Hz) oscillation of neocortical neurons in vivo: depolarizing and hyperpolarizing components
J Neurosci
 , 
1993b
, vol. 
13
 (pg. 
3252
-
65
)
Sukhotinsky
I
Chan
AM
Ahmed
OJ
Rao
VR
Gradinaru
V
Ramakrishnan
C
, et al.  . 
Optogenetic delay of status epilepticus onset in an in vivo rodent epilepsy model
PloS One
 , 
2013
, vol. 
8
 pg. 
e62013
 
Szabo
GG
Holderith
N
Gulyas
AI
Freund
TF
Hajos
N
Distinct synaptic properties of perisomatic inhibitory cell types and their different modulation by cholinergic receptor activation in the CA3 region of the mouse hippocampus
Eur J Neurosci
 , 
2010
, vol. 
31
 (pg. 
2234
-
46
)
Tonnesen
J
Sorensen
AT
Deisseroth
K
Lundberg
C
Kokaia
M
Optogenetic control of epileptiform activity
Proc Natl Acad Sci USA
 , 
2009
, vol. 
106
 (pg. 
12162
-
7
)
Traub
RD
Wong
RK
Synchronized burst discharge in disinhibited hippocampal slice. II. Model of cellular mechanism
J Neurophysiol
 , 
1983
, vol. 
49
 (pg. 
459
-
71
)
Traynelis
SF
Dingledine
R
Potassium-induced spontaneous electrographic seizures in the rat hippocampal slice
J Neurophysiol
 , 
1988
, vol. 
59
 (pg. 
259
-
76
)
Trevelyan
AJ
Sussillo
D
Watson
BO
Yuste
R
Modular propagation of epileptiform activity: evidence for an inhibitory veto in neocortex
J Neurosci
 , 
2006
, vol. 
26
 (pg. 
12447
-
55
)
Trevelyan
AJ
Sussillo
D
Yuste
R
Feedforward inhibition contributes to the control of epileptiform propagation speed
J Neurosci
 , 
2007
, vol. 
27
 (pg. 
3383
-
7
)
Ulbert
I
Halgren
E
Heit
G
Karmos
G
Multiple microelectrode-recording system for human intracortical applications
J Neurosci Methods
 , 
2001
, vol. 
106
 (pg. 
69
-
79
)
Ulbert
I
Heit
G
Madsen
J
Karmos
G
Halgren
E
Laminar analysis of human neocortical interictal spike generation and propagation: current source density and multi-unit analysis in vivo
Epilepsia
 , 
2004a
, vol. 
45
 
Suppl 4
(pg. 
48
-
56
)
Ulbert
I
Magloczky
Z
Eross
L
Czirjak
S
Vajda
J
Bognar
L
, et al.  . 
In vivo laminar electrophysiology co-registered with histology in the hippocampus of patients with temporal lobe epilepsy
Exp Neurol
 , 
2004b
, vol. 
187
 (pg. 
310
-
8
)
Whittington
MA
Traub
RD
Jefferys
JG
Erosion of inhibition contributes to the progression of low magnesium bursts in rat hippocampal slices
J Physiol
 , 
1995
, vol. 
486
 
Pt 3
(pg. 
723
-
34
)
Wittner
L
Huberfeld
G
Clemenceau
S
Eross
L
Dezamis
E
Entz
L
, et al.  . 
The epileptic human hippocampal cornu ammonis 2 region generates spontaneous interictal-like activity in vitro
Brain
 , 
2009
, vol. 
132
 
Pt 11
(pg. 
3032
-
46
)
Wlodarczyk
AI
Sylantyev
S
Herd
MB
Kersante
F
Lambert
JJ
Rusakov
DA
, et al.  . 
GABA-independent GABAA receptor openings maintain tonic currents
J Neurosci
 , 
2013
, vol. 
33
 (pg. 
3905
-
14
)
Ylinen
A
Bragin
A
Nadasdy
Z
Jando
G
Szabo
I
Sik
A
, et al.  . 
Sharp wave-associated high-frequency oscillation (200 Hz) in the intact hippocampus: network and intracellular mechanisms
J Neurosci
 , 
1995
, vol. 
15
 
1 Pt 1
(pg. 
30
-
46
)
Zhou
JL
Lenck-Santini
PP
Zhao
Q
Holmes
GL
Effect of interictal spikes on single-cell firing patterns in the hippocampus
Epilepsia
 , 
2007
, vol. 
48
 (pg. 
720
-
31
)
Ziburkus
J
Cressman
JR
Barreto
E
Schiff
SJ
Interneuron and pyramidal cell interplay during in vitro seizure-like events
J Neurophysiol
 , 
2006
, vol. 
95
 (pg. 
3948
-
54
)