MODELING EPILEPSY AND SEIZURE-LIKE ACTIVITY USING HUMAN IN VITRO NEURONAL NETWORKS

Epilepsy affects over 65 million people, of which more than 20 million remain resistant to current antiseizure medications (ASM). Despite the validated animal models of acute and chronic seizures used in preclinical drug discovery, many clinical trials fail partially due to low human relevance. While human cell–based in vitro epilepsy models cannot capture the full complexity of in vivo systems, they provide a complementary translational approach to existing animal models. Advanced models integrating patient-derived induced pluripotent stem cells (hiPSCs), microelectrode arrays, and microfluidic engineering allow real-time monitoring of spontaneous and induced seizure-like activity (SLA) within defined neuronal networks and circuits.

Our recent studies using hiPSC-derived cortical networks revealed variant-specific and clinically relevant network phenotypes, with heterogeneous excitatory–inhibitory cultures showing heightened sensitivity to Dravet Syndrome-associated dysfunction. Furthermore, induction of SLA using kainic acid (KA) enabled controlled, dose-dependent modulation of network activity without evidence of cytotoxicity. Neuronal networks exhibited concentration-dependent recovery following KA exposure. At higher concentrations, KA treatment was associated with increased secretion of epilepsy-associated tRNA-derived small RNA fragments, suggesting a link between SLA and stress-related molecular responses.

Seizure- and brain-on-chip systems extend these capabilities by enabling spatially organized, axonally connected networks that model seizure initiation and propagation. To take a leap forward, we are leveraging our previously validated brain-on-chip, Modular Platform for Epilepsy Modelling In Vitro (MEMO) to reveal patient-specific functional phenotypes between compartmentalized neuronal networks at circuitry level. This platform represents a critical step towards precision medicine, improved drug discovery, and more predictive, human-relevant epilepsy modelling.