An Optogenetic Kindling Model of Neocortical Epilepsy

Epileptogenesis is the gradual process by which the healthy brain develops epilepsy. However, the neuronal circuit changes that underlie epileptogenesis are not well understood. Unfortunately, current chemically or electrically induced epilepsy models suffer from lack of cell specificity, so it is seldom known which cells were activated during epileptogenesis. We therefore sought to develop an optogenetic variant of the classical kindling model of epilepsy in which activatable cells are both genetically defined and fluorescently tagged. We briefly optogenetically activated pyramidal cells (PCs) in awake behaving mice every two days and conducted a series of experiments to validate the effectiveness of the model. Although initially inert, brief optogenetic stimuli eventually elicited seizures that increased in number and severity with additional stimulation sessions. Seizures were associated with long-lasting plasticity, but not with tissue damage or astrocyte reactivity. Once optokindled, mice retained an elevated seizure susceptibility for several weeks in the absence of additional stimulation, indicating a form of long-term sensitization. We conclude that optokindling shares many features with classical kindling, with the added benefit that the role of specific neuronal populations in epileptogenesis can be studied. Links between long-term plasticity and epilepsy can thus be elucidated.

Following this procedure, coronal slices were mounted using coverslips with a 40 µl bolus of ProLong Gold Antifade mount (P10144, Life Technologies, Burlington, ON, Canada).
Sections were imaged using a Fluoview FV1000 confocal laser scanning microscope and Fluoview software (Olympus Canada, Richmond Hill, ON, Canada). Analysis of antibodylabelled slices was performed manually using ImageJ 54 and Igor Pro (Wavemetrics Inc., Lake Oswego, OR, USA). For GFAP, ~6 sections with two measurements each, covering all six cortical layers in M1, were analysed per animal. M1 was compared to a non-labelled cortical region. For NeuN, four or five sections were analysed with 7-8 measurements per animal. M1 cell counts were carried out across the six cortical layers.

Automated electrographic seizure detection
To independently detect seizures offline, a simple automatic seizure detection software algorithm was developed. Although this approach had the disadvantage of missing the occasional seizure that was detected by inspection, it was preferred because it was unbiased and -as opposed to manual inspection -invariably gave the same results when rerun on the same EEG sweeps. We manually inspected all automatically detected electrographic seizures.
EEG Fourier power traces were first converted to z-score sweeps. For z-scoring purposes, the background power levels were determined from the median of at least 64 one-second-long EEG segments recorded in the absence of laser stimulation. The median was used to automatically exclude the occasional movement artifact, because such artifacts resulted in massive responses that typically saturated the amplifier for about a second at a time.
To detect electrographic seizures, a combined threshold and duration criterion was applied. If power exceeded z-score 4 for longer than 4 seconds, then this event was deemed a seizure. EYFP and ChR2 control animals were used to determine these threshold values, whereby which no seizures were detected in the control mice. We estimated a false negative rate of ~9% by direct inspection of EEGs. Z-scores above 100 were always rejected as movement artifacts. The z-score threshold crossing was taken as the start of an electrographic seizure. Seizure duration was automatically determined from seizure start to the first z-score downstroke threshold crossing. No attempt was made to merge events separated by a brief time of relative inactivity, so automatically detected electrographic seizures were likely underestimated both in terms of number and duration.

Measurement of light scattering in cortical tissue
To measure light scattering properties in cortical tissue, we dissected whole brains from P30-45 mice and placed them on top of plain microscope slides (12-550-A3, Fisher Scientific, Nepean, ON, Canada). Next, we mounted the same 1.25-mm ferrule and the same 445-nm laser used for stimulation on the stereotax we performed our surgeries on.
After, we mounted the sensor from a Thorlabs power meter (PM100D meter, S121C sensor, Thorlabs, Newton, NJ, USA) directly below the slide. We descended the ferrule into the brain in 100 µm increments while measuring the power that was able to reach the sensor.

Page 6 of 18
We took 3 measurements at each depth and repeated the experiment with two different brains.

Statistics
The results are reported as the mean ± SEM. Significance levels are denoted using asterisks (* p < 0.05, ** p < 0.01, *** p < 0.001). Unless otherwise stated, we used Student's t test for equal means for all pairwise comparisons. If an equality of variances F test gave p < 0.05, we employed the unequal variances t test. Individual data sets were tested using a one-sample t test. For multiple comparisons, pairwise comparisons were carried out if one-way ANOVA suggested this at the p < 0.05 significance level. Equal or unequal variances ANOVA was used depending on Bartlett's test for equal variances.
Wilcoxon-Mann-Whitney's non-parametric test was always used in parallel to the t test, with similar outcome. Multiple comparisons were corrected post hoc using Bonferroni-Dunn's method. Statistical tests were performed in Igor Pro 7 or 8 unless otherwise stated.
For circular statistics (Supplementary Fig. S4), we used the circstat toolbox in MATLAB in conjunction with the Watson U 2 test 57 .