Qin, W.; Beevis, J.; Wilde, M.; Stednitz, S.; Arnold, J.; Ma, M.; Favre-Bulle, I. A.; Peterson, A.; Hoffman, E. J.; Scott, E.

Despite its prevalence and clinical impacts, epilepsy remains incompletely understood in terms of the population dynamics that mediate seizure susceptibility, initiation, and propagation across brain-wide networks. In this study, we have performed calcium imaging in zebrafish, brain-wide and at cellular resolution, at baseline and as seizures are induced using the GABAA receptor antagonist pentylenetetrazol (PTZ). We have then modeled the network architecture in wild-type and scn1lab-/- larvae, which are seizure-prone and serve as a model for Dravet syndrome. scn1lab-/- larvae show increased pair-wise correlations between neurons when exposed to PTZ, and graph analyses of these correlations revealed genotype-specific network alterations during seizures, identifying regions and metrics linked to seizure onset. Using generative network modeling, we then explored the wiring rules that govern activity in these networks, identifying specific network properties linked to seizure susceptibility that were only detectable using large-scale, cellular-resolution data. Even at baseline in the absence of seizures, these rules differed by genotype in a way that enabled the identification of scn1lab-/- larvae and predicted individuals' seizure risk independently of their observable phenotype. These findings uncover the cellular-resolution network properties of a zebrafish model of Dravet syndrome and establish a predictive framework for seizure susceptibility grounded in multi-scale functional connectivity.