High-frequency network activity, global increase in neuronal activity, and synchrony expansion precede epileptic seizures in vitro

Abstract

How seizures start is a major question in epilepsy research. Preictal EEG changes occur in both human patients and animal models, but their underlying mechanisms and relationship with seizure initiation remain unknown. Here we demonstrate the existence, in the hippocampal CA1 region, of a preictal state characterized by the progressive and global increase in neuronal activity associated with a widespread buildup of low-amplitude high-frequency activity (HFA) (>100 Hz) and reduction in system complexity. HFA is generated by the firing of neurons, mainly pyramidal cells, at much lower frequencies. Individual cycles of HFA are generated by the near-synchronous (within approximately 5 ms) firing of small numbers of pyramidal cells. The presence of HFA in the low-calcium model implicates nonsynaptic synchronization; the presence of very similar HFA in the high-potassium model shows that it does not depend on an absence of synaptic transmission. Immediately before seizure onset, CA1 is in a state of high sensitivity in which weak depolarizing or synchronizing perturbations can trigger seizures. Transition to seizure is characterized by a rapid expansion and fusion of the neuronal populations responsible for HFA, associated with a progressive slowing of HFA, leading to a single, massive, hypersynchronous cluster generating the high-amplitude low-frequency activity of the seizure.

Figure 1.

Low-calcium seizures and HFA. A, Repeated seizure activity. B, Detail of low-calcium seizure. Seizures are superimposed on large negative DC shift. C, Periods between seizures are characterized by the presence of low-amplitude HFA. D, Short burst of HFA with superimposed multiunit activity. E, Bandpass filtered (80–250 Hz) data showing only HFA. F, Raw data were high-pass filtered (>600 Hz) to reveal multiunit activity (MUA).

Figure 2.

Temporal profile of HFA. A, Example of period between seizures (DC removed, high-pass >10 Hz filter). B, Isolated interictal period. C, Bandpass (80–250 Hz) filtered interictal period demonstrates increases in amplitude and especially in incidence of bursts of HFA preceding the seizure. D, Corresponding wavelet spectrogram shows increase (pale gray to white) in power, especially in frequency band (80–250 Hz) mainly ∼185 Hz. E, Corresponding summated power of 80–250 Hz band and first moment of power spectra. F, Time course of summated power of 80–250 Hz band in 21 slices (mean ± SEM; gray area is SEM). The increase power starts >30 s before the seizure. After the end of the seizure, HFA occurs and quickly decreases in power. G, Because the period between seizures in each slice is not equal, interictal periods were each divided into 100 epochs of equal length, and the mean values of summated power for individual epochs was obtained. H, I, First moment (H) and median (I) of power spectra show that buildup in HFA is accompanied by progressive drop in frequency, with sudden drop immediately preceding the seizure. J, Temporal profile of incidence of >5 and >7 SD HFA cycles. Preceding the seizure, there is increase in number of >5 and >7 SD oscillations, but the steeper increase in >7 SD suggests also an increase in amplitude of the HFA preceding the seizure (period between seizures divided into 10 segments). K, Ratio between numbers of >7 and >5 SD cycles reveal an increase in cycles with higher amplitude.

Figure 3.

Preictal and ictal changes in Gaussian process entropy rate. A, Two seizures and intervening interictal period. B, Signal bandpass filtered at 80–250 Hz shows buildup in HFA. C, Phase space diagram constructed from the activity during the middle part of the interictal period has a visually irregular (complex) pattern. D, Activity closer to the seizure onset is still characterized by an irregular, but expanding, pattern in phase space. E, Seizure onset is characterized by a progressive expansion and more regular pattern of trajectories. F, Regular pattern of trajectories during the seizure, which resembles a three- or more dimensional limit cycle. G, GPER, with GPER progressively decreasing before the seizure and a substantial drop during the seizure. H, Bar graph shows changes in interictal GPER across 10 slices. I, Changes in GPER during seizures across 10 slices.

Figure 4.

Synchrony profile of HFA and seizures on a local scale. A, Recording with closely spaced electrodes (25 μm). Period between seizures is characterized by the presence of HFA in all channels. Preceding the seizure are two short prebursts (asterisks). B, GSI slightly increases during preictal period. Transient increases can be observed during the prebursts. Seizure onset is characterized by a sudden increase in GSI, rising up to 0.9. C, Detail of recording between seizures shows HFA present in all channels (i). Phase-synchronization matrix (ii) shows two statistically significant clusters (p < 0.01). Synchronization indices for the individual clusters are 0.18 and 0.09. Synchronization index of largest cluster corresponds to GSI. Second spatial derivative of voltage shows the presence of several local sinks (iii). D, Seizure onset is characterized by expansion of one of the clusters and coalescence with adjacent ones. GSI is 0.38. Second derivative of voltage shows expansion and spread of local sinks. E, Seizure activity is characterized by high-amplitude low-frequency activity during which whole recorded area generates activity synchronously and creates one large hypersynchronous cluster. GSI reaches 0.83. Note the continuous current sinks and sources propagating across the area of recording. F, G, Values of GSI between seizures (F) and during seizures (G). There is a modest, but significant, progressive increase in synchrony during preictal period. H, Preictal changes characterized by increase in amplitude are associated also with increased spatial extent (averaged HFA from interictal period divided into 5 segments).

Figure 5.

Synchrony profile of HFA and seizures on a global scale. A, Recording with large separation between electrodes (200 μm) shows widespread buildup in HFA. B, Between seizures, GSI values are very low. C, Detail of activity during preictal period shows the widespread presence of HFA, which lacks synchrony between electrodes. D, Seizure onset is characterized by an expansion of synchrony, which is rapid but slower then on the local scale. E, During the seizure, the entire CA1 can generate epileptic activity in near complete synchrony (GSI of 0.95). F, G, GSI between (F) and during (G) seizures on the large (global) scale.

Figure 6.

Cellular mechanisms of HFA and dynamics of cellular firing. A, Averaged pyramidal cell action potential (gray area represents SEM; n = 84 cells). B, Averaged interneuronal action potential (gray area represents SEM; n = 27 cells). C, D, Normalized and averaged autocorrelograms of firing of pyramidal cells (C) and of interneurons (D). E, F, Averaged normalized cross-correlograms between HFA cycle and firing of pyramidal cells (E) and interneurons (F). G, Phase histograms of firing probability, during HFA cycles, show increase in probability of pyramidal cell firing during the HFA cycle, with increased probability of firing mainly during the trough of HFA cycle. H, Phase histogram for interneurons show only mild increase of firing during the HFA cycle. I, Coincident firing. Coincident firing was observed 43 times between the illustrated pair of neurons. Histogram shows the number (y-axis) of coincident firings (x-axis) observed in 200 surrogates. Results suggest that coincident firing between these two neurons is not attributable to random coincidence (p < 0.001). J–L, Interictal period between two seizures is characterized by a progressive buildup in HFA. M, Activity of pyramidal cells (PYR) and interneurons (INT) derived from three closely spaced tetrodes reveals a progressive increase in neuronal firing and recruitment of cells. N, Transition to seizures is characterized by progressive increase in multiunit activity (n = 24; 8 slices).

Figure 7.

Sensitivity to external perturbations. A, Seizure triggered by local application of 10 mm glutamate. B, During preictal period immediately preceding the next expected seizure, weak external depolarizing electric fields (−4 V/m, 0.75 s) can trigger seizure (C). D, Repeated medium strength (−8 V/m) pulses failed to trigger seizures until later in the interictal period. E, Strong depolarizing field (−40 V/m) reliably triggered seizures early in the interictal period. F, Repeated pulses (as in D) demonstrate the relationship between field strength and shortening of the interictal period. G, Sensitivity of CA1 area to applied fields, of all strengths, as a function of stage during the interictal period (divided into 5 equal segments based on the interictal period measured in the absence of fields). The closer to the predicted seizure onset, the higher the percentage of applied fields of all strengths capable of inducing seizures.

Figure 8.

Preseizure changes in the high-potassium model. A, Repeated seizure activity. B, Between seizures, low-amplitude HFA is present with superimposed multiunit activity (MUA). C, Episode of HFA bandpass filtered (80–250 Hz) data. D, High-pass filtering (>600 Hz) shows multiunit activity (MUA). E, Isolated interictal period shows progressive buildup of low-amplitude HFA. F, Bandpass (80–250 Hz) filtered interictal period demonstrates increases in amplitude and incidence of HFA preceding the seizure. G, Corresponding wavelet spectrogram shows progressive increase (pale gray to white) in power, especially in frequency band (80–250 Hz) centered around ∼195 Hz (arrows). H, Time course of summated power of 80–250 Hz band in 16 slices (mean ± SEM, gray area is SEM). I, Frequency profile shift toward low frequencies preceding the seizures, expressed by the temporal profile of the first moment of the spectrum.