Segregation of Seizures and Spreading Depolarization across Cortical Layers

Summary

Objective:

Cortical spreading depolarization (SD) and seizures are often co-occurring electrophysiological phenomena. However, the cross-layer dynamics of SD during seizures and the effect of SD on epileptic activity across cortical layers remain largely unknown.

Methods:

We explored the spatial-temporal dynamics of SD and epileptic activity across layers of the rat barrel cortex using DC silicone probe recordings during flurothyl-induced seizures.

Results:

SD occurred in half of the flurothyl-evoked seizures. SD always started from the superficial layers and spread downwards either through all cortical layers or stopping at the L4/L5 border. In cases without SD, seizures were characterized by synchronized population firing across all cortical layers through the entire seizure. However, when SD occurred, epileptic activity was transiently silenced in layers involved with SD but persisted in deeper layers. During partial SD, epileptiform activity persisted in deep layers through the entire seizure with positive signals at the cortical surface reflecting passive sources of population spikes generated in deeper cortical layers. During full SD, the initial phase of SD propagation through the superficial layers was similar to partial SD, with suppression of activity at the superficial layers and segregation of seizures to deep layers. Further propagation of SD to deep layers resulted in a wave of transient suppression of epileptic activity through the entire cortical column. Thus, vertical propagation of SD through the cortical column creates dynamic network states during which epileptiform activity is restricted to layers without SD.

Significance:

Our results point to the importance of vertical SD spread in the SD-related depression of epileptiform activity across cortical layers.

Introduction

Cortical spreading depolarization (SD) is a propagating wave of collective neuronal depolarization in the cortical grey matter that is associated with a wave of spreading depression of electrical activity. SD occurs in migraine, traumatic and ischemic brain injury and epilepsy^1–5. SD is generated by massive ionic fluxes across neuronal and glial cell membranes mediated by synaptic and voltage-gated ion channels and an accumulation of extracellular potassium concentration. Neurons almost completely depolarize during SD and lose the ability to fire action potentials due to inactivation of voltage-gated sodium channels resulting in depression of all types of electrical activity in the region invaded by SD.

Various factors promoting SD (high-potassium, low magnesium, enhanced synaptic activity, glutamate receptor agonists) also promote seizures and moreover, seizures and SD frequently co-occur in various experimental models and patients⁶. Spreading depression was first described by Leão as a slowly propagating wave of complete cortical electrogram flattening following epileptic discharge evoked by tetanic electrical stimulation during alternate-current (AC) EEG recordings⁷. Large negative shifts were later described during SD⁸ using full-band direct current (DC)-coupled recordings, which are now considered as the gold standard for SD detection⁹. Termination of tetanic stimulation or high-potassium induced seizures by SD has been shown in the rat hippocampus^10–15. SD also curtails epileptic discharges induced by application of penicillin to the rat cortex¹⁶, and mechanically-induced SD transiently suppresses spontaneously occurring spike-wave discharges in WAG/Rij rats¹⁷. DC recordings have also revealed co-occurrence of SD and seizures in patients, with SD typically terminating epileptic discharges or producing transient suppression of ongoing epileptic activity as shown in acute brain injury caused by trauma, and spontaneous subarachnoid and/or intracerebral haemorrhage¹⁸. Yet, SD is also associated with seizures occurring during the rebound from SD resulting in a series of spreading seizures^7;18–20.

While previous studies mainly addressed SD propagation in two-dimensional horizontal cortical space, SD also shows remarkable three-dimensional dynamics with a variety of vertical propagation patterns across cortical layers. In the hippocampus, SD displays robust layer-specific features with the maximal negative LFP shifts and associated sinks observed in the layers containing apical and basal dendrites of CA1 pyramidal cells^11;15;21. In high-potassium and trauma-induced neocortical SD models, SD is typically initiated in the superficial cortical layers and spreads through the entire cortical depth at a rate of 4–6 mm/min²² that is comparable to the horizontal velocity of SD propagation^1;3;23;24. Yet, under certain conditions, SD may be restricted to superficial or deep cortical layers, with a hallmark SD separation border located between the granular and infragranular layers^22;25. How seizure-associated SD is initiated and spreads across layers of the cortex remains unknown. Also, little is known about how SD impacts epileptic activity through different cortical layers. Here, we addressed these questions using silicone probe DC recordings of the local field potential (LFP) and multiple unit activity (MUA) across layers of a cortical column in the rat somatosensory barrel cortex in a flurothyl seizure model. We found that flurothyl-induced epileptic activity is often associated with SD that initiates in the superficial layers and slowly spreads downwards, either through all cortical layers or stopping at the L4/L5 border. Downstream SD propagation strongly impacted epileptic activity within a column with full suppression of activity in the layers recruited by SD, and segregation of seizures to the layers located below the SD-involved cortex. Thus, vertical propagation of SD through the cortical column during seizure creates dynamic network states during which epileptic activity is restricted to layers non-recruited by SD, whereas in the layers invaded by SD epileptic activity is suppressed. The results of the present study point to the importance of vertical SD spread in the SD-related depression of epileptic activity across cortical layers.

Materials and Methods

This work was carried out in accordance with EU Directive 2010/63/EU for animal experiments, and all animal-use protocols were approved by the French National Institute of Health and Medical Research (INSERM, protocol N007.08.01) and Kazan Federal University on the use of laboratory animals (ethical approval by the Institutional Animal Care and Use Committee of Kazan State Medical University N9–2013). Wistar rats of both sexes from postnatal days [P] 25–60 were used. Intracortical recordings of the flurothyl-induced seizures were performed on head restrained urethane-anaesthetized (1.2–1.5 g/kg, i.p.) rats as described previously²⁸ with some modifications. In brief, during the surgical procedure rats were anesthetized with isoflurane (1.5%) using a Fluotec4 (Surgivet/Anesco) anaesthesia apparatus. The skin and periosteum were removed from the skull, which was then covered with a layer of dental acrylic, except an area ~ 2 mm in diameter above the primary somatosensory barrel cortex at the coordinates selected according to the animal age^33;34. The nuchal muscles were cut from the skull to reduce movement artefacts. Following 1–2 hrs recovery the rat was positioned in the stereotaxic apparatus, and a metal ring was fixed to the skull by dental cement and via ball-joint to a magnetic stand. During recordings, rats were placed on a heated platform (37°C) using an automatic temperature controller (TC-344B; Warner Instruments, Hamden, CT). Flurothyl (Sigma) was applied through a mask (0.1 ml for 90 s)²⁸ once (n=12 rats) or twice with a 1.5 hr interval (n=13 rats). Concomitant piezo-recordings of the hindlimb movements were used to monitor motor manifestations.

Recordings of the local field potential (LFP) and multiple unit activity (MUA) were performed using linear multichannel arrays (silicone probes, 413 μm² iridium electrode surface, 100 μm electrode separation distance, Neuronexus Technologies, USA). The signals were amplified and filtered at 0 Hz-9 kHz using a DigitalLynx (Neuralynx, USA) amplifier set in DC mode (input range ±131 mV), digitized at 32 kHz and saved on a PC for post-hoc analysis using custom-written functions in MATLAB (MathWorks, USA). DC offsets were compensated at the beginning of recordings using a compensation chain between the combined reference/ground electrode and the ground input of the amplifier as described previously²². The whiskers were trimmed to a length of 0.8–1.5 mm and were stimulated using a piezoelectric bending actuator (Noliac, Denmark) using 200 ms square pulses with 5–10 s intervals. A needle (22G) was glued to the end of a piezo actuator and the tip of the whisker was inserted into the blunt tip of the needle. The principal whisker (PW) was identified by the shortest latency MUA responses in layer 4 evoked by single whisker deflection.

Experimental data were processed using MATLAB environment (MathWorks, USA). Raw signals, local field potentials and extracellular spikes were analyzed by custom-developed, MATLAB-based programs including ExpressAnalysis and Eview (AZ, https://github.com/AndreyZakharovExp). For LFP analysis, the wide-band signal was downsampled to 1000 Hz using the mean function resample. Positive polarity is shown as up in all figures. Filtering with cutting frequencies >1Hz was performed by mean Chebyshev type II digital filters (cheby2 function). For cutting frequencies below 1 Hz the digital RC-filter was used. For SD detection, DC traces were lowpass filtered at < 3Hz and SD fronts were detected from the first derivative of the lowpass filtered LFP trace at threshold of > 3 μV/ms, and events exceeding 5 mV in amplitude and 6 seconds in duration were considered as SD. Negative waves occurring synchronously on several channels and eliminated by common referencing were excluded from analysis. For action potential detection, the raw wide-band signal was filtered (bandpass 300–5000 Hz) and negative deflections exceeding 5 standard deviations calculated over the most silent 1 s length segment of the filtered trace were considered as spikes (MUA). For population spikes (PSs) analysis, LFP was highpass filtered at 2 Hz and common reference subtraction and current-source density (CSD) analysis across the cortical depth was used to eliminate volume conduction and localize sinks and sources. Common reference was calculated as the LFP value at each time point averaged across all recording sites of the silicon probe. CSD was computed according to a differential scheme for the second spatial derivative of LFP along the probe axis. PSs were detected at a threshold of 100 standard deviations of LFP in L5–L6. LFP-kinematogram crosscorrelation was computed after signal bandpass filtering in the range 2–1000 Hz in 100 ms long fragments near PS onset.

Statistical analysis was performed using the Matlab Statistics toolbox. The two-sided signed rank test for the matched samples was performed to assess the significance of differences between groups of data with the level of significance kept at p < 0.05.

Results

Extracellular local field potential (LFP) and multiple unit activity (MUA) were recorded during flurothyl-evoked activity across layers of a cortical barrel column in the primary somatosensory barrel cortex of head-restrained, urethane-anesthetized 1–2 month old rats n=25 rats). Electrode location was verified by current-source density (CSD) profile and short-latency MUA in L4 and L5/6 border during principal vibrissa – evoked response^26;27. Extracellular LFP signals were recorded in DC mode which is essential for a detection of slow potential shifts during SD as described previously²².

Inhalation of flurothyl evoked ictal-like electrographic discharges in the barrel cortex, which started 29±2 s after the onset of flurothyl inhalation and lasted for 130±8 s (n=38; Fig. 1A,B). SD was not observed during seizures in about half of cases (55%; “no SD” case on Fig. 1A; n=21/38). In the other cases, SD occurred once (n=17) or occasionally twice (n=3) per seizure. While SD always involved superficial electrodes, it sometimes remained restricted to the superficial channels (“Partial SD” on Fig. 1A; n=7/17 of the first SDs) or occurred through entire cortical depth (“Full SD” on Fig. 1A; n=10/17 of the first SDs) (Fig. 1C). Occurrence of SD varied in time relatively to the flurothyl exposure without any difference in partial SD (53±14 s; n=7) and full SD (46±6 s; n=10; P>0.05) delays from the onset of flurothyl inhalation (pooled data for the first SDs) (Fig. 1B,D). Latency of SD occurrence in relation to seizures also varied in time, with delay of SD onset from seizure onset averaging 28±13 s (n=7; partial SD) and 21±6 s (n=10; full SD), and with a delay of seizure end from SD offset of 73±26 s (n=7; partial SD) and 53±17 s (n=10; full SD) (Fig. 1B,D).

Figure 1. Occurrence and properties of SD during flurothyl-evoked seizures.

(A) DC-coupled recordings of the flurothyl-evoked epileptic discharges at different depths of a cortical barrel column with a 16 channel silicone probe (100 μm vertical separation distance between the electrodes) in three different animals. Left (“no SD”): seizures not associated with SD. Middle (“Partial SD”): seizures associated with partial SD restricted to the superficial layers. Right (“Full SD”): seizures associated with SD propagating top-down through all cortical layers. Below, 1Hz highpass filtered LFP from the top channel at the cortical surface and hindlimb movement mechanogram. The horizontal blue bar above traces shows the period of flurothyl inhalation. (B) Temporal organization of seizures (grey bars), partial SDs (yellow bars) and full SDs (red bars) in 38 flurothyl applications. Time=0 corresponds to the onset of flurothyl inhalation. (C) Circular histogram of SD occurrence during flurothyl-evoked seizures. (D) Group data on seizure and SD onsets and offsets (in the superficial layers) relative to the time of flurothyl inhalation. (E) Depth profiles of partial SD (top) and full SD (bottom) parameters. Blue circles show the mean and their diameter shows SE of cortical depth. Shaded areas show SE. (F) Partial (left) and full (right) SD onsets as a function of cortical depth in the barrel cortex. Red circles indicate SD earliest onset point, blue circles indicate SD latest onset point. Green squares show mean±SE of the SD onset start and end time and depth. (G) Corresponding histogram of SD onset start (left) and end (right) depths. Note that SDs always start at the superficial layers and either wane in L4 or at the L4/5 border (partial SD) or fully propagate through all layers (full SD).

SD was first observed in the superficial layers at a cortical depth of 132±32 μm (partial SD) and 195±41 μm (full SD) (Fig. 1F, G, red circles and lines). SD further spread to the cortical depth either terminating at 710±50 μm (L4/5 border) within 7.8±1.2 s in cases of partial SD (n=9, (first and second SDs counted)) or at 1578±37 μm (grey/white matter border) within 21.2±2.1 s in cases of full SD (n=12) (Fig. 1F, G, blue circles and lines). The average velocity of SD propagation was 5.4±0.7 mm/min and 4.4±0.6 mm/min in cases of partial and full SD, respectively. SD amplitude attained 23.5±3.5 mV and 25.6±3.2 mV in the superficial layers in cases of partial and full SD, respectively, and 31.1±3.9 mV in the deep layers in cases of full SD. SD duration was higher in superficial layers (partial SD: 37±6 s; full SD: 41±4 s) compared to deep layers (full SD: 29±1 s; Fig 1E).

We further explored the effects of SD on vertical organization of epileptic activity across cortical layers of the barrel column. During seizures without SD, flurothyl-induced epileptic activity was characterized by synchronization of activity across all layers through the entire course of seizure (Fig.2). Population spikes (PSs) in the barrel cortex were highly correlated with hindlimb jerks throughout seizures, with the jerks delayed from PSs by 29±2 ms suggesting that epileptic spikes are synchronized through the sensorimotor cortex and drive generalized motor events. Current-source density analysis of PSs revealed that sinks in the deep layers preceded sinks in the superficial layers during initial (Fig. 2H), tonic (Fig. 2I) and clonic (Fig. 2J) phases and persisted in superficial and deep layers through the entire course of the seizure (Fig. 2C). Elevated multiple unit activity persisted in all cortical layers through the seizure (Fig. 2D, E, G), and MUA in the deep layers preceded MUA in superficial layers (Fig. 2H, I, J) in agreement with the results of CSD analysis. This set of seizures without SD was associated with only 1.7±1.3 mV and 4.3±1.1 mV negative LFP shifts in the superficial and deep layers, respectively (Fig. 2F)²⁰. Thus, flurothyl-induced seizures without SD were characterized by PS activity initiated in the deep layers, synchronizing activity through all cortical layers through the entire time course of the seizure.

Figure 2. SD-lacking flurothyl-evoked epileptic discharges in a cortical barrel column.

(A) DC-coupled recordings of the flurothyl-evoked epileptic discharge not associated with SD recorded at different depths of a cortical barrel column with a 16 channel silicone probe (100 μm vertical separation distance between the electrodes). Bottom, piezo-recordings of the hindlimb movements. Cntr sd, signal deviation at baseline. (B) Correlation between the population spikes (popspikes) and hindlimb jerks. Right panel, peak crosscorrelation of hindlimb jerks (bandpass 2–1000 Hz) versus popsikes in deep layers of the barrel cortex. Each point corresponds to an individual popspike. Peak values of the crosscorrelation coefficient are color-coded. Vertical position shows time lag between the peak of movement from the popspike. Time scale is shown on the left panel, T=0 corresponds to the peak of popspike. Left panel, histogram of time lags between jerks and popspikes through the whole seizure. Crosscorrelations were calculated for the absolute LFP and kinematographic values within a time window of ±100 ms from the popspike. (C) Popspike sinks and sources in superficial (~200 μm, blue lines) and deep (~1200 μm, red lines) layers calculated as an average CSD value within 15 ms after the popspike onset. Note that popspikes are essentially associated with sinks both in superficial and deep layers through the entire time course of seizure. (D) MUA density at different cortical depths calculated within sliding 100 ms long time windows. (E) MUA density during individual popspikes in superficial (~200 μm, blue lines) and deep (~1200 μm, red lines) layers. Note that popspikes are associated with MUA bursts both in superficial and deep layers through the entire time course of seizure. (F-G) Average LFP (F) and MUA density (G) in the superficial (~200 μm, blue lines) and deep (~1200 μm, red lines) layers in “no SD” group (n=21 seizures). Horizontal bars above indicate the period of flurothyl application (cyan) and seizure (grey). Shaded areas and error bars show SE. Note lack of large negative LFP shifts characteristic of SD and persistence of MUA through the whole seizure. (H-J) Examples of popspikes during the initial phase of seizure (H), tonic phase (I) and clonic phase (J). Top: example traces of LFP with common reference subtraction (black lines) and MUA (vertical red bars above traces) at various cortical depths and hindlimb kinematogram (bottom trace), bottom left: average popspikes’ LFP (black traces) overlaid on a color-coded CSD map, bottom right: color-coded depth profile of MUA density triggered by popsikes. Note that popspikes are associated with sinks and MUA bursts in superficial and deep layers through all phases of the flurothyl-evoked tonic-clonic discharge in the barrel cortex. A-D and H-J: data from one animal; F-G: group data on 21 seizures without SD.

A different profile of epileptic activity was observed in cases of seizure associated with SD (Fig. 3&4). In cases with partial SD, during which a large negative LFP shift was only observed in the superficial but not deep layers (Fig. 3A, F), epileptic PSs at the seizure onset were associated with sinks and MUA discharges through all layers as in cases without SD (Fig. 3C–H). Occurrence of partial SD was associated with a complete suppression of epileptic activity at superficial layers and segregation of epileptic events to the deep layers non-invaded by SD (Fig. C–G, I). Onset of activity suppression at superficial layers followed vertical top-down SD propagation whereas recovery of epileptic activity followed a rebound of SD (Fig. 3D,E). Interestingly, fast LFP PS activity was not suppressed at superficial layers despite SD and total MUA suppression (Fig. 1A and 3I). These positive PS-related LFP events at superficial layers were passive in their nature, however, reflecting passive propagation of PSs in deep layers (Fig. 3I, C). On rebound of SD, synchronization of epileptic activity through the entire column returned to normal until the end of the seizure (Fig. 3J). Association of epileptic activity with motor jerks was maintained through the seizure (Fig. 3B).

Figure 3. Partial SD transiently suppresses epileptic activity in superficial but not deep cortical layers.

Layout is the same as in Figure 2. Note the differences: (A) large negative SD shift occurs during seizure in the superifical but not deep layers (panel A in an individual animal and panel F in a “partial SD” group, n=7). (B) Coupling between popspikes and myoclonic hindlimb jerks is maintained during seizure. (C) Popspikes display a switch from sinks to sources during SD in superficial but not deep layers. (D-E) Overall MUA (D) and MUA per popspike (E) is completely and transiently depressed in superficial layers during partial SD but persists in the deep layers. (F-G) group data on negative LFP shift associated with SD in superficial layers (F, blue line) and suppression of MUA during SD restricted to superficial layers (G). (H) At the seizure onset prior to SD, popspikes display sinks and MUA bursts in superficial and deep layers. (I) Popspikes during SD in superficial layers display positive LFP signals and sources and no unit firing at the cortical surface. Popspike sinks and MUA bursts are only observed in deep layers. (J) Synchronized activity in superficial and deep layers recovers after cessation of SD. F-G: group data on 7 seizures with “Partial SD”.

Figure 4. Dynamic epileptic states during full SD propagation.

Layout is the same as in Figures 2&3. Note the differences: (A) Large negative SD shift starts in the superifical layers and propagates down the entire depth of the cortical column. (B) Motor manifestations are transiently suppressed when SD recruits deep layers. (C) During propagation of SD through the superficial layers, popspikes display a switch from sinks to sources in superficial layers. When SD arrives to the deep layers, popspikes in deep layers also show a transient switch from sinks to sources. (D-E) Overall MUA (D) and MUA per popspike (E) are completely depressed in the superficial layers but persist in the deep layers during recruitment of superficial layers by SD as in the case of “Partial SD”. Further invasion of SD into the deep layers is associated with full suppression of MUA through the entire column. (F-G) group data on negative LFP shift associated with SD first in the superficial layers (blue) followed by SD in the deep layers (red) (F) and suppression of MUA during SD first in the superficial layers and then also in the deep layers (G). (H) At the seizure onset prior to SD, popspikes display sinks and MUA bursts in superficial and deep layers. (I) Popspikes during SD in superficial layers display positive LFP signals and sources and no unit firing at the cortical surface. Popspikes sinks and MUA bursts are only observed in the deep layers as in the case of “partial SD”. (J) Complete propagation of SD to the deep layers is associated with a quasi-complete suppression of popspikes and MUA bursts in all layers. (K) Synchronized activity in superficial and deep layers recovers after cessation of SD. F-G: group data on 10 seizures with “Full SD”.

Full SD propagation caused more profound changes in ictal activity than in cases with partial SD (Fig. 4). The initial part of the seizures was similar to partial SD cases, with initially full propagation of PSs before SD onset (Fig. 4H), followed by suppression of activity at the superficial layers and its compartmentalization to deep layers when SD invaded superficial layers (Fig. 4I). Deeper SD propagation was associated with a descending wave of suppression of epileptic activity also in deep cortical layers (Fig. 4J, which shows a transition from deep PSs to complete electrocerebral inactivity during full SD). As in cases with partial SD, epileptic activity during a transitory state when SD involved only superficial but not deep layers, was characterized by negative LFP events and MUA bursts during PSs generated in deep layers and positive LFP deflections, passive sources and MUA silence in superficial layers (Fig. 4I, C–G). Recovery from complete translayer depression to vertically generalized PSs (Fig. 4K) largely followed a rebound from the DC-LFP shifts during SD (Fig. 4A,D and F,G). In the case illustrated on Fig. 4A, hindlimb jerks were transiently suppressed during SD in the recorded barrel column (Fig. 4B) suggesting spread of SD to the motor cortex, but this was not consistently observed in all animals.

Finally, we explored whether and how filter settings affect SD associated with flurothyl–evoked epileptic discharges. As shown on Figure 5A, highpass filtering of LFP traces profoundly inhibited and modified the time course of large negative LFP shifts during SD recorded in DC–mode. The effects of highpass filtering on various parameters of SD in superficial layers included a decrease in SD peak amplitude (half-depression observed at 0.048±0.006 Hz), average SD plateau amplitude measured during 3 s periods of maximal negativity (half-depression observed at 0.032±0.002 Hz), SD front slope (half-depression observed at 0.099±0.001 Hz) and SD half-width (half-depression observed at 0.014±0.01 Hz). These observations are in agreement with previous studies indicating that full band DC recordings are essential for the detection and adequate assessment of SD⁹.

Figure 5. Effects of highpass filtering on SD.

(A) Example traces of flurothyl-evoked SD in the superficial cortical layers at different highpass filter settings. Note that SD is severely perturbed by highpass filtering in a cutoff frequency dependent manner, and that SD is barely seen during (conventional for AC-EEG) >1Hz highpass filtering. (B) Dependence of various SD parameters on highpass filter frequency. Group data from 21 SDs.

Discussion

The main findings of the present study are that SD occurring during flurothyl-evoked seizures is initiated in the superficial layers, that it propagates either partially or fully through the cortical depth, and that vertical SD propagation creates dynamic network states during which epileptic activity is fully suppressed in the cortical layers involved with SD but persists in the deeper non-involved layers.

Partial SD associated with SD restricted to the superficial layers and depression of epileptic activity in the superficial but not deep cortical layers is the main observation of the present study. Indeed, the basic concept of SD implies that any type of activity is fully depressed in the cortical region invaded by SD. Here, we show that the partial SD that was observed in nearly half of SDs occurring during flurothyl – evoked seizures did not halt epileptic activity in the cortical column. Indeed, recordings by electrodes at the cortical surface (that closely match ECoG, or EEG signals) detected SD transients without suppression of epileptic-like PS LFP activity (e.g. see Fig. 1A and 3I). However, depth-recordings revealed that these epileptiform events at the cortical surface reflect passive sources of PSs generated in deep cortical layers to which SD failed to propagate, and that neuronal firing is completely suppressed during SD in superficial layers. A similar transient state of epileptic activity supported by the deep layers with the surface layers being silenced was also observed during the initial phase of full SDs, during which SD had invaded the superficial but not yet deep layers. Together, these findings are in agreement with the concept of the instrumental role of SD in spreading depression but also point to importance of the vertical aspects of SD propagation and associated phenomena including spreading depression, which appears to be only partial in cases of partial vertical SD propagation.

SD during flurothyl-evoked seizures displayed several characteristic layer-specific features including initiation in the superficial layers, vertical spread from superficial to deep layers at a velocity of 4–6 mm/min, and longer SD duration at the cortical surface. These layer-specific features of SD occurring during flurothyl-evoked seizures are similar to those induced by high-potassium and traumatic brain injury (pinprick and dura incision)²². Our findings are in agreement with easier induction and faster propagation of SD^7;29 and higher frequency of recurrent SDs²³ in the superficial cortical layers, and with preferential propagation of SD through the superficial layers in slices of the neocortex in vitro^23;30–32. In addition, we found that about half of SDs generated in superficial layers arrested at the boarder between superficial and deep layers, whereas fully propagating SDs often slowed down at a depth of ~1000 μm (example shown on Fig. 4A). This is in keeping with a kind of “barrier“ for vertical SD propagation and segregation of SDs in the superficial and deep cortical layers previously reported for recurrent SDs evoked by continuous remote high-potassium application^22;25. In future studies it would be of interest to determine the mechanisms underlying the higher propensity of the superficial layers to SD and the nature of the ”barrier’ for vertical SD propagation.

Occurrence of SD in nearly half of the flurothyl-evoked seizures supports the idea that seizures and SD are often co-occurring electrophysiological phenomena. In the present study, SD typically interrupted epileptic activity in a layer-specific manner, or totally terminated epileptic discharge contributing to postictal depression as described previously both in the animal models and human patients^10–16;18. In some cases, SD occurred shortly after seizure onset with most epileptic activity occurring on rebound from SD that more resembles spreading convulsions also reported in the animal models and patients^7;18–20. Wherever SD occurred, it was always associated with a complete depression of epileptic activity in the cortical layers involved in SD indicating that this phenomenon is an important event in patterning seizures. In keeping with strong filtering of slow LFP shifts associated with SD during conventional AC-EEG recordings (Fig. 5), DC recordings appear essential for the comprehensive interpretation of the epileptic activity patterns.

In conclusion, we have shown that SD, which accompanies flurothyl-induced seizures in nearly half of cases, strongly affects epileptic activity in a layer-specific manner. The relatively slow and often-incomplete vertical top-down SD propagation creates dynamic states in the cortical column with epileptic activity completely suppressed in the cortical layers involved with SD but persisting in the layers below the SD. The present results point to the importance of SD assessment during electrographic recordings in epileptic patients. In future studies, it would be of interest to determine the occurrence of SD in relation to various types of seizures in patients and in different animal models, the mechanisms involved in triggering SD by seizure and the impact of SD on epileptogenesis and on the efficiency of epilepsy treatment.

Key Point Box.

• Flurothyl-evoked seizures are often associated with Spreading Depolarization (SD) in the rat barrel cortex

• SD always initiates in the superficial cortical layers and spreads downwards either through all cortical layers or stopping at the L4/L5 border

• SD creates dynamic states with epileptiform activity suppressed in the SD-involved layers but persisting below the SD-involved cortex

• Thus, vertically propagating SD strongly impacts epileptiform activity across cortical layers