Testing of putative antiseizure medications in a preclinical Dravet syndrome zebrafish model

Abstract

Dravet syndrome is a severe genetic epilepsy primarily caused by de novo mutations in a voltage-activated sodium channel gene (SCN1A). Patients face life-threatening seizures that are largely resistant to available anti-seizure medications. Preclinical Dravet syndrome animal models are a valuable tool to identify candidate anti-seizure medications for these patients. Among these, scn1lab mutant zebrafish, exhibiting spontaneous seizure-like activity, are particularly amenable to large-scale drug screening. Thus far, we have screened more than 3000 drug candidates in scn1lab zebrafish mutants, identifying valproate, stiripentol, and fenfluramine e.g. Food and Drug Administration-approved drugs, with clinical application in the Dravet syndrome population. Successful phenotypic screening in scn1lab mutant zebrafish is rigorous and consists of two stages: (i) a locomotion-based assay measuring high-velocity convulsive swim behaviour and (ii) an electrophysiology-based assay, using in vivo local field potential recordings, to quantify electrographic seizure-like events. Historically, nearly 90% of drug candidates fail during translation from preclinical models to the clinic. With such a high failure rate, it becomes necessary to address issues of replication and false positive identification. Leveraging our scn1lab zebrafish assays is one approach to address these problems. Here, we curated a list of nine anti-seizure drug candidates recently identified by other groups using preclinical Dravet syndrome models: 1-Ethyl-2-benzimidazolinone, AA43279, chlorzoxazone, donepezil, lisuride, mifepristone, pargyline, soticlestat and vorinostat. First-stage locomotion-based assays in scn1lab mutant zebrafish identified only 1-Ethyl-2-benzimidazolinone, chlorzoxazone and lisuride. However, second-stage local field potential recording assays did not show significant suppression of spontaneous electrographic seizure activity for any of the nine anti-seizure drug candidates. Surprisingly, soticlestat induced frank electrographic seizure-like discharges in wild-type control zebrafish. Taken together, our results failed to replicate clear anti-seizure efficacy for these drug candidates highlighting a necessity for strict scientific standards in preclinical identification of anti-seizure medications.

Keywords: anti-seizure medications; dravet; drug screening; epilepsy; zebrafish larvae.

Graphical abstract

Figure 1

Behavioral phenotypic screening of candidate ASMs in WT and scn1lab larvae. (A) Representative baseline swim velocities (mm/s) over time for 5 dpf WT and scn1lab mutant larvae. Threshold for behavioral seizure events indicated by dotted line (28 mm/s). Scn1lab larvae (N = 60 across 3 replicates) have significantly (P = 0.014) higher maximum velocities on average compared to WTs (N = 70 across 3 replicates), all larvae are represented as individual data points. (B) Timeline for behavioral tracking acquisition (top). Heatmap of the percent change in average swim velocity from baseline after treatment with candidate ASMs at three different concentrations. Significant changes from vehicle control are indicated by stars with an N = 30 or 40 per condition across minimum three trials for WT and scn1lab larvae. (C) Behavioral tracking plots for WT larvae showing baseline activity followed by ASM treatment for drugs that caused a significant change in swimming velocity compared to DMSO controls. The plot below highlights the normalized velocity in percent, after ASM treatment for each larvae recorded. AA43279 significantly increased swim velocity in WT larvae at 100 µM (N = 40 across three replicates, P < 0.0001). (D) Behavioral tracking plots for scn1lab larvae showing baseline activity followed by ASM treatment for drugs that caused a significant change in swimming velocity compared to DMSO controls. Donepezil increased swim velocities for scn1lab larvae (N = 40 across 4 replicates, P < 0.01). 1-EBIO, chlorzoxazone and lisuride significantly reduced swim velocities (N = 30–40 across 3–4 replicates, P < 0.05–0.0001). The plot below highlights the normalized velocity in percent, after ASM treatment for each larvae recorded. An unpaired t-test or ANOVA was performed for statistical analysis, P < 0.05 = *, P < 0.005 = **, P < 0.0005 = ***, P < 0.0001 = ****.

Figure 2

Toxicology assay for larvae following behavioral assessment. (A) Percent survival of WT and scn1lab larvae (N = 30 or 40 per condition with minimum independent 3 replicates). Percentage of larvae that did not survive are overlayed. Vorinostat at 100 µM significantly reduced scn1lab survival compared to DMSO-treated controls. (B) Radar plot quantifying touch responses of WT larvae (N = 30 or 40 per condition with minimum independent three replicates). Dotted lines segment each candidate ASM and toxicity is plotted as a percentage of larvae. Toxic ASMs are labelled. (C) Radar plot showing candidate ASM toxicity for scn1lab larvae (N = 30 or 40 per condition with minimum independent three replicates). (D) Radar plot of WT and scn1lab toxicity overlayed. 1-EBIO and Chlorzoxazone were toxic for both populations. Donepezil uniquely impacted WT larvae. One-way ANOVA was performed for statistical analysis, P < 0.01 = *, P < 0.0001 = **.

Figure 3

Electrophysiological recordings of larvae treated with candidate ASMs. (A) Donut plots of untreated WT (left, N = 10) and scn1lab (right, N = 22) larvae showing the distribution of type 0 (normal), type 1 (interictal-like) and type 2 (ictal-like) electrical activity along with representative LFP traces. (B) Heatmap of average LFP scores for both WT and scn1lab larvae after treatment with candidate ASMs at three different concentrations showing a significant increase in activity for WT larvae treated with AA43279 (N = 7, P = 0.01) and pargyline (N = 7, P = 0.05). (C) Violin plot showing scored electrophysiological recordings from individual WT larvae after treatment with candidate ASMs at each concentration. AA43279, Chlorzoxazone and Pargyline significantly induced abnormal activity in WT larvae compared to control DMSO treatment (N = 2–4 per condition across 2–3 replicates, P < 0.0001). 1-EBIO-induced type 2 and vorinostat-induced type 1 events in 1 out of 4 larvae. (D) Violin plot showing scored electrophysiological recordings from individual scn1lab larvae. No drugs reliably prevented seizure activity compared to control DMSO-treated larvae, with only some drugs having a mild modulatory effect on seizure activity (N = 2–6 per condition across 2–4 replicates). Statistical tests include Kruskal-Wallis and One-way ANOVA, P < 0.05 = *, P < 0.0001 = **, all other data were not significantly different. Scale for traces are 1 s by 0.1 mV.

Figure 4

In-depth assessment of soticlestat treatment of seizure activity for WT and scn1lab larvae. (A) Swimming behaviour for individual WT (N = 60 across three replicates) and scn1lab (N = 40 across three replicates) larvae after treatment with three concentrations of soticlestat represented as a percent change in velocity from baseline. Increasing concentrations of soticlestat significantly increased velocity of WTs compared to DMSO-treated controls but did not alter the velocity of scn1lab larvae at any concentration. (B) Heatmap showing significant increases in average normalized swimming velocities of soticlestat treated WT larvae compared to scn1lab larvae. (C) Increasing concentrations of soticlestat treatment decreased survival rates of WT larvae but did not impact scn1lab larval survival or responsiveness. (D) LFP recording sample of seizure activity induced in WT larvae after soticlestat treatment along with (E) associated spectrogram. Note the high-frequency activity and electrodecremental LFP response following an ictal-like event. (F) Donut plots representing the percentage of fish showing type 0, 1 and 2 activity for both WT (top, N = 6 across three replicates) and scn1lab (bottom, N = 6 across three replicates) larvae treated with three concentrations of soticlestat with quantification of type 2 events for each seizing fish below each plot. Soticlestat induced type 1 and 2 activity in WT larvae at all concentrations but did not abolish seizure activity in scn1lab larvae unless at high concentrations. Unpaired t-test was performed for statistical analysis, P < 0.05 = *, P < 0.01 = **, P < 0.0001 = **.