Abstract
LaCoursiere CM, Ullmann JFP, Koh HY, Turner L, Baker CM, Robens B, Shao W, Rotenberg A, McGraw CM, Poduri AH. iScience. 2024 Jun 5;27(7):110172. doi: 10.1016/j.isci.2024.110172. PMID: 39021799; PMCID: PMC11253282. Hundreds of novel candidate human epilepsy-associated genes have been identified thanks to advancements in next-generation sequencing and large genome-wide association studies, but establishing genetic etiology requires functional validation. We generated a list of >2200 candidate epilepsy-associated genes, of which 48 were developed into stable loss-of-function zebrafish models. Of those 48, evidence of seizure-like behavior was present in 5 (arfgef1, kcnd2, kcnv1, ubr5, and wnt8b). Further characterization provided evidence for epileptiform activity via electrophysiology in kcnd2 and wnt8b mutants. Additionally, arfgef1 and wnt8b mutants showed a decrease in the number of inhibitory interneurons in the optic tectum of larval animals. Further, RNA sequencing revealed convergent transcriptional abnormalities between mutant lines, consistent with their developmental defects and hyperexcitable phenotypes. These zebrafish models provide strongest experimental evidence supporting the role of ARFGEF1, KCND2, and WNT8B in human epilepsy and further demonstrate the utility of this model system for evaluating candidate human epilepsy genes.
Commentary
Next-generation sequencing has provided unprecedented opportunities for the diagnosis of genetic forms of epilepsy. 1 Indeed, the number of genes reported to be associated with epilepsy has increased precipitously to >2000, though many still require validation. 2 These advances therefore led to new challenges: the need for high-throughput models to verify pathogenicity, define biological mechanisms, and test precision therapies. 1
Multiple model systems have been used to study epilepsy-associated genes, each with strengths and weaknesses. 1 Rodent models, particularly mice, have been a frequent choice for research on genetic epilepsies, given the capacity to study the impact of genetic variants on brain networks, and intercellular dynamics such as those between neurons and glia. However, mouse models can take years to develop and characterize. In some cases, mice bearing genetic variants that cause epilepsy in humans don’t exhibit the cardinal feature of epilepsy, seizures. Human-derived neurons and organoids present another important approach, providing the unprecedented opportunity for direct access to human tissues and cell types—but these are also less amenable to high throughput screening of hundreds of new genes. Further, organoids do not yet recapitulate the full and complex brain networks that generate seizures, though advances have been made with the generation of assembloids. 3
Here, LaCoursiere et al. 2 skillfully address gaps posed by the model systems above, using another animal model, zebrafish (Danio rerio) for high throughput screening of the pathogenicity of epilepsy genes. Advantages of this model articulated by the authors include high genetic homology with humans, clutch sizes of up to several hundred fish, rapid development, short life span, efficiency of husbandry, and the occurrence of behavioral seizures. 2 The utility of zebrafish to model epilepsy was previously demonstrated, including in seminal studies relevant to Dravet syndrome4,5 and other catastrophic childhood epilepsies. 6 From ∼2200 genes reported to be epilepsy-associated, compiled from multiple databases, the authors identified genes meeting the following criteria: (1) close homology between zebrafish and humans; (2) low tolerance to loss-of-function predicted with in silico modeling; and (3) genes that had not been previously studied in animal models. This yielded 81 genes; stable zebrafish lines could be generated using CRISPR/Cas9-mediated gene disruption (loss of function insertion or deletion in a protein coding region of the gene) in 48 of these. It was assumed that genetic variants that would not transmit from F0 to F1 were embryonic lethal. The 48 lines generated were then subjected to pathogenicity screening (functional validation), by measuring swimming behavior at baseline, during photic stimulation, and during exposure to subthreshold pentylenetetrazole, 7 using previously established criteria for swimming behavior that would be considered seizure. 7 Zebrafish with loss of function in 5 gene paralogs—arfgef1, kcnd2, kcnv1, ubr5, and wnt8b—exhibited a higher frequency of seizure-like swim patterns, though there were differences in whether spontaneous or provoked seizures occurred. A subset (zebrafish with loss of kcnd2 and wnt8b) exhibited evidence of spontaneous hyperexcitability (increased spikes and bursting activity) during tectum local field potential recordings. Two others, arfgef1 and wnt8b mutants, exhibited decreased numbers of interneurons within the larval optic tectum. Single nucleus RNA sequencing revealed convergence in the 2 mutants that had exhibited hyperexcitability (kcnd2 and wnt8b), with upregulation of pathways related to axon sprouting and synaptogenesis.
The authors note caveats to the employed protocol, mainly related to the low sensitivity (ability to detect true positives, resulting in few false negatives) of the behavioral filter. Screening tests ideally have high sensitivity and specificity (ability to reject true negatives, resulting in few false positives). A substantial number of genes were not included in the analysis because viable zebrafish lines could not be generated, likely due to embryonic lethality. Presumably, some pathogenic genes did not pass this first hurdle. The second level of screening, which employed video monitoring of swimming behavior, relied on the assumption that increased velocity and distance of swimming would be reliable predictors of seizure-like activity. This was a reasonable assumption based on prior work demonstrating validity in zebrafish models of epilepsy. 8 However, some epilepsy-associated genes may also cause motor impairment, leading to decreased swimming occurring alongside seizures, such as chd2 and stxbp1a, as noted by the authors. It is also not clear how many genes that are thought to be highly associated with epilepsy and/or very likely pathogenic were not detected by this behavioral screening. The existence of gene paralogs in zebrafish may have a compensatory effect, obscuring pathogenicity. 9 As the authors point out, the list of 5 gene hits should not be considered exhaustive among the 2200 original genes. Considering all of the above, 1 reasonable next step might be to generate a list of “gold standard” epilepsy genes—those known to cause epilepsy with high penetrance in humans and quantify the specificity and sensitivity of this screening protocol in a zebrafish model of the same genetic epilepsies. Of note, zebrafish lines have already been created for numerous genes with severe childhood epilepsy. 6 Screening measures could then be adjusted to optimize sensitivity and specificity. One consideration might be to insert pathogenic variants identified in humans, rather than complete knockout of genes.
Among the 48 genes that screened positive based on behavioral criteria, further test results related to intrinsic excitability were variable (arfgef1 mutants exhibited reduced numbers of interneurons, e.g., but no evidence of spontaneous hyperexcitability). In the ideal screening process, additional confirmatory tests might be tailored to the putative functions of the gene in question, and need not be high throughput. For example, the activity and kinetics of an ion channel expressed in a heterologous cell line; or the impact of a gene implicated in neuronal structure on brain development, assessed in organoids or mice. The author's use of RNAseq from zebrafish brain is particularly useful to rapidly gain insight into mechanisms related to genes with likely pathogenicity.
Even with an optimized screening protocol, it is possible that intrinsic differences between mammals and zebrafish might lead to false positives or negatives. For example, rodent models exhibit the layered cortical brain structure shared with other mammals including humans, while zebrafish brains lack this complexity. 10 Recordings and cellular observations come from the optic tectum, which may not feature the same network structures and dynamics that give rise to seizures in human epilepsy. Such intrinsic limitations could be quantified by comparison with “gold standard genes” associated with epilepsy.
While we have provided ideas to further build on this study by LaCoursiere et al., it should not be understated that this work represents a promising and rigorous starting point for a critically needed, novel modality in the toolkit for screening epilepsy genes. The approach led by LaCoursiere et al. could be powerfully used in combination with other model systems described above, to streamline screening and characterization of epilepsy genes. This work establishes an important new research dimension using zebrafish to study epilepsy, adding to to the known utility of zebrafish for drug screening in forms of epilepsy where the pathogenic gene is already established.4–6
