Abstracts

Rationale: Channelopathy-associated epilepsies are clinically diverse, ranging from mild self-limiting seizures that resolve in adulthood to severe developmental and epileptic encephalopathies (DEEs) that are often resistant to current pharmacological treatment. In particular, mutations in a number of different ion channel genes can result in severe disease for which disease modifying treatments are needed.

Methods: Here, we generated human neuronal models across three different ion channelopathies, KCNQ2, KCNT1, and SCN8A. We produced millions of differentiated cortical excitatory neurons derived from human induced pluripotent stem cell (iPSC) lines and characterized their functional features using our propietary system BRITE^TM (Bioengineered NeuRonal Insight-driven Therapeutic Engine) which includes high-throughput optical electrophysiology imaging and machine learning (ML)-based analytics for phenotyping. Neuronal excitability is assayed with genetically-encoded light-sensitive proteins, CheRiff and QuasAr, for simultaneous optical stimulation and recording of neuronal action potentials. Measurements are acquired in 96- or 384-well plates on a fully-automated microscope allowing for voltage recordings from 300-500 neurons in a single movie and >500,000 neurons per day. Q-State’s unsupervised ML pipeline extracts >700 functional features from each movie, representing spike shape and timing properties in response to an array of stimulation protocols. The application of ML to these robust data sets then allows for identification of disease-associated phenotypes across thousands of iPS cell-derived neurons.

Results: For KCNQ2 and SCN8A, we used collections of 15 isogenic cell lines genetically-edited with CRISPR/Cas9 to express either gain-of-function (KCNQ2-R201C, SCN8A-R850Q, SCN8A-R1872L) or loss-of-function (KCNQ2-T274M) disease-causing variants. Functional phenotypic analyses of >10,000 neurons differentiated from the KCNQ2 cell lines revealed that 185 electrophysiological parameters showed opposing effects in the gain-of-function as compared to the loss-of-function. For the SCN8A cell lines, phenotypic analyses revealed significant differences in spiking frequency and rheobase for the R1872L variant. For KCNT1, we explored two gain-of-function variants (KCNT1-G288S and KCNT1-A934T) which differ in severity in patients. Our system identified similar, yet differentially-graded electrophysiological phenotypes, with a stronger phenotype in the neurons expressing the more severe variant (KCNT1-A934T) with significant differences in such parameters as spike width, rheobase, and afterhyperpolarization. This KCNT1-associated phenotype was also confirmed in neurons differentiated from a set of patient cell lines. In each case, ongoing assessments are evaluating both approved and novel compounds (small molecules and antisense oligonucleotides) for phenotypic rescue.

Conclusions: The identification of the channelopathy-associated phenotypes presented here can enable the evaluation and development of therapeutics. The phenotypic analyses described here can further be applied to other channelopathy-associated diseases.

Funding: R43NS117263