Knowledge Base

With EthoVision XT

Behavioral phenotyping is a cornerstone of epilepsy research in zebrafish models. EthoVision® XT offers an advanced platform for the automated tracking and analysis of zebrafish behavior across multiple parameters. The accurate detection and tracking of the zebrafish larvae in all the different possible setups allow for a detailed analysis of locomotor activity, obtaining data of multiple parameters related to seizure-like behavior.

First, the cumulative percentage of variance of different variables was assessed, unveiling that only 2 principal components (PC) already explain about 80% of the total variance of the sample. Considering a wide range of behavioral parameters, a principal component analysis (PCA) based on two PCs was performed, selecting the key kinematic measurements related with an epileptic-like behavior that can also be considered a high activity behavior regarding the locomotor parameters.

Of all the considered parameters, only seven were considered key variables with higher projection and contribution (cos2 > 0.75). These key variables are associated with a modification of the larval position or orientation in the space and include:

• Maximum velocity (mm/s)
• Number of angle turns
• Maximum acceleration (mm/s²)
• Mobility in the arena (%)
• Cumulative duration (s) of three different mobility states: immobile, mobile, and highly mobile

Capturing High Phenotypic Variability

Epilepsy is characterized by a high phenotypic heterogeneity, ranging from normal phenotypes to simple febrile seizures and severe epileptic encephalopathies. This heterogeneity highlights the need to have a specialized system and to develop an analysis that allows a statistical distinction of non-epileptic and epileptic-like behavior. To this end, the behavior of multiple crispants of genes related with different childhood epilepsies were analyzed.

Considering the previously selected variables, a total sample of 563 larvae and 2209 events related with light-induced epileptic stimuli were analyzed, and the childhood epilepsy genes crispants activity was shown to be different to the behavior of the used controls (scramble). Within this plot, despite the high dispersion of the different measurements — related with the high heterogeneity of the epileptic manifestations — the scramble observations clustered in the center of the PC plot, showing non-epileptic phenotypes, while different observations remote to these positions, corresponding to crispants observations, were potential outliers that could be defining an epileptic-like fingerprint.

Behavioral Fingerprint

Analyzed larvae could be classified in different groups depending on their activity level, and it was speculated that the crispants of genes having a positive association with photosensitive epilepsy would have an increased representation in the more active group.

To explore this possibility, the Mahalanobis distance (MD) of all the observations was measured to evaluate how each flash response differs from the average of the entire population. MD measures the distance between a sample point and the distribution of all the sample points, and it is defined by two parameters:

• Distance Magnitude: The numerical MD value
• P-Value: A statistical measurement to validate the hypothesis (outliers considered with p-value ≤ 0.05)

Two regions of activity were distinguished and classified within the population:

• Low Activity Region: MD < 2.5 and p-value > 0.1, corresponding to a non-epileptic phenotype
• High Activity Region: MD > 2.5 and p-value < 0.1, corresponding to an epileptic-like phenotype

This observation suggests that larvae with increased photosensitivity are prone to display a highly aberrant behavior in response to exposure to flashes of light, constituting a population of outliers distinguished by their higher motor activity. These extreme behaviors are assumed to correspond to epileptic-like seizures characterized by aberrant and excessive movements.

Transferring these observations of the two differentiated groups back to the PC plot reveals that most of the observations corresponding to an epileptic-like phenotype overlap with the crispants of childhood epilepsy genes.

Therefore, the study demonstrated the efficacy of the EthoVision XT software and the precision of the Noldus equipment in defining a fingerprint for mild and severe childhood epilepsy genes that present a photosensitive epileptic-like behavior through an exhaustive analysis of key parameters detected, tracked, and analyzed, even though it is known to be a highly heterogeneous condition.

Source: EthoVision XT 18 - THC - Trial and Hardware Control, Noldus Information Technology