Spike-induced cytoarchitectonic changes in epileptic human cortex are reduced via MAP2K inhibition

Interictal spikes are electroencephalographic discharges that occur at or near brain regions that produce epileptic seizures. While their role in generating seizures is not well understood, spikes have profound effects on cognition and behavior, depending on where and when they occur. We previously demonstrated that spiking areas of the human neocortex show sustained MAPK activation in superficial cortical layers I-III and are associated with microlesions in deeper cortical areas characterized by reduced neuronal nuclear protein (NeuN) staining and increased microglial infiltration. Based on these findings, we chose to investigate additional neuronal populations within microlesions, specifically inhibitory interneurons. Additionally, we hypothesized that spiking would be sufficient to induce similar cytoarchitectonic changes within the rat cortex and that inhibition of MAPK signaling, using a MAP2K inhibitor, would not only inhibit spike formation but also reduce these cytoarchitectonic changes and improve behavioral outcomes. To test these hypotheses, we analyzed tissue samples from 16 patients with intractable epilepsy who required cortical resections. We also utilized a tetanus toxin-induced animal model of interictal spiking, designed to produce spikes without seizures in male Sprague-Dawley rats. Rats were fitted with epidural electrodes, to permit EEG recording for the duration of the study, and automated algorithms were implemented to quantify spikes. After 6 months, animals were sacrificed to assess the effects of chronic spiking on cortical cytoarchitecture. Here, we show that microlesions may promote excitability due to a significant reduction of inhibitory neurons that could be responsible for promoting interictal spikes in superficial layers. Similarly, we found that the induction of epileptic spikes in the rat model produced analogous changes, including reduced NeuN, calbindin, and parvalbumin-positive neurons and increased microglia, suggesting that spikes are sufficient for inducing these cytoarchitectonic changes in humans. Finally, we implicated MAPK signaling as a driving force producing these pathological changes. Using CI-1040 to inhibit MAP2K, both acutely and after spikes developed, resulted in fewer interictal spikes, reduced microglial activation, and less inhibitory neuron loss. Treated animals had significantly fewer high-amplitude, short-duration spikes, which correlated with improved spatial memory performance on the Barnes maze. Together, our results provide evidence for a cytoarchitectonic pathogenesis underlying the epileptic cortex, which can be ameliorated through both early and delayed MAP2K inhibition. These findings highlight the potential role of CI-1040 as a pharmacological treatment that could prevent the development of epileptic activity and reduce cognitive impairment in both patients with epilepsy and those with non-epileptic spike-associated neurobehavioral disorders.

Methods
40 two-month-old male Sprague-Dawley rats were utilized for this study. Briefly, rats were anesthetized and 7 electrode holes were drilled through the skull (3 left and 3 right electrode holes located at the following locations relative to bregma: AP +4 mm, ML 3.5 mm; AP -1 mm, ML 3.5 mm; AP -6 mm, ML 3.5 mm; 1 reference electrode hole above the nasal sinus: AP +10.5 mm, ML 0.5 mm left). Before electrode screw placement, rats received an injection of either 1 uL of sterile PBS alone (sham rats) or 80 ng of tetanus toxin (TeNT rats) in 1 uL of sterile PBS into the left somatosensory cortex (corresponding to the site of the left middle electrode: AP -1 mm, ML 3.5 mm left, DV -1.5 mm). Epidural electrode screws were then placed, and rats were fitted with head cap apparatuses.

Beginning on postoperative day 5, electroencephalography (EEG) recordings were conducted every 5 days through postoperative day 35. From post-operative days 35-84, EEG recordings were conducted once weekly. After postoperative day 84, EEG recordings occurred once per month until the 6-month study endpoint.

Rats in both the TeNT and sham groups were further subdivided into 3 drug treatment groups: early drug treatment with an MEK inhibitor (post-operative days 0-7), delayed drug treatment (post-operative days 15-21), or no drug treatment.

EEG data was acquired at a sampling rate of 1000 Hz using Stellate HARMONIE software (Version 6.0, Stellate Systems Inc.) and converted into European Data Format (EDF) files for subsequent visualization (EDFbrowser, Version 1.76; Free Software Foundation) and data analysis. To analyze EEG data for the presence of interictal spikes, we implemented the spike-detection algorithm included with this dataset. Briefly, EDF files were imported into MATLAB using the edfread function. EDF files were filtered with a 1-35 Hz fourth-order Butterworth bandpass filter. Each electrode was analyzed for the presence of spikes; spikes were defined as high-amplitude events with negative polarity, shorter than 200 ms in duration, with a maximum voltage greater than one standard deviation above the median background signal. Both spikes (< 70 ms) and sharp waves (70-200 ms) were included in the overall spike count. Spikes were then divided into two half waves (Fig. 2), for which amplitude, duration, and slope were calculated. The estimation of these metrics depends on the accurate identification of the start and end points (edges) of each spike. These can be identified visually as the locations of the first trough points (local minima) that occur just before (start) and after (end) the point of maximum voltage. After identifying the spike location (point of maximum voltage), edge detection was performed in three essential steps. First, the raw signals were filtered using a fourth-order Butterworth highpass filter with a 7 Hz cutoff frequency to eliminate the confounding effect of low-frequency background or prominent slow waves. Next, the filtered signals were smoothed using a 12-point moving average window. Finally, the raw, filtered, and smoothed signals were inverted, and potential edge points were identified using the findpeaks function. "True" edges were designated as those that were detected consistently across the raw, filtered, and smoothed signals. Moreover, the maximum amplitude and total duration for each spike was determined. For each recording day, spike counts were normalized using the total recording length (mean spikes/h).