De-risking seizure liability: integrating adverse outcome pathways (AOPs), new approach methodologies (NAMs), and in silico approaches while highlighting knowledge gaps

Abstract

Animal studies are commonly used in drug development and in chemical and environmental toxicology to predict human toxicity, but their reliability, particularly in the central nervous system (CNS), is limited. For example, animal models often fail to predict drug-induced seizures, leading to unforeseen convulsions in clinical trials. Evaluating environmental compounds, such as pesticides, also poses challenges due to time and resource constraints, resulting in compounds remaining untested. To address these limitations, a government-industry collaboration identified 27 biological target families linked to seizure mechanisms by combining key events from adverse outcome pathways (AOPs) with drug discovery data. Over a hundred in vitro assay endpoints were identified, covering 26 of the target families, including neurotransmitter receptors, transporters, and voltage-gated calcium channels. A review of reference compounds identified 196 seizure-inducing and 34 seizure-negative chemicals, with 80% being tested in the in vitro assays. However, some target familes were more data-poor than others, highlighting significant data gaps. This proof-of-concept study demonstrates how mechanistic seizure liability can be assessed using an AOP framework and in vitro data. It underscores the need for expanded screening panels to include additional seizure-relevant targets. By integrating mechanistic insights into early drug development and environmental risk assessment, this approach enhances compound prioritization, complements animal studies, and optimizes resource use. Ultimately, this strategy refines CNS safety evaluation in drug development, improves public health protection to neurotoxicants, and bridges knowledge gaps.

Keywords: adverse outcome pathway (AOP); drug development; environmental exposure; in silico; new approach methodologies (NAMs); seizure.

Fig. 1.

Workflow to identify targets, assays (upper row), and reference compounds (lower row) related to seizures. Target families with potential to induce seizure (Table 2) were identified from an AOP network (Fig. 3) developed from 7 independent AOPs in the AOP-Wiki and the OFF-X database. Expressions of identified targets in the human and rat brain were evaluated using ArrayStudio (Fig. 4). In vitro assays annotated to these target families were identified in the Integrated Chemical Environment (ICE) and ChEMBL. Seizure-positive and seizure-negative reference compounds were identified in the literature (Fig. 5). Negative compounds were subsequently queried in PubChem. We then assessed the availability of data for the identified reference compounds in the in vitro assays represented in ICE and ChEMBL (Figs 6 and 7).

Fig. 2.

Automated AOP network of existing seizure-related AOPs. Seven distinct AOPs established independently by various research groups were identified from the AOP-Wiki when searching for an adverse outcome (AO) of “seizure,” “epilepsy,” and convulsion. The Biovista Vizit tool was used to integrate the AOPs and generate a schematic representation for evaluating overlap/commonalities. Purple nodes represent the AOP, whereas green nodes represent key events. https://aopkb.biovista.com/#!bv_gid=9fe865f17ee6cfe0a2daefa063c06af2

Fig. 3.

Manually integrated seizure AOP network. Simplified, harmonized, and curated terminologies from 7 AOPs were used to manually create an integrated seizure AOP network containing 20 MIEs/KEs/AO. Macromolecular events (green), cell/tissue-level events (orange), organ/system-level changes (red), and AO (white) are organized by level of biological complexity adhering to the AOP structure to allow mapping of trajectory to adversity.

Fig. 4.

Heatmap of mRNA expression for seizure-relevant targets. mRNA expression of 40 seizure-relevant targets (x-axis) in various tissues (y-axis) in humans and rats. Human expression data are derived from GTEx (n ≈ 1000 donors, across 53 non-diseased tissue types), where each row represents the average expression per tissue. Rat data represent pooled samples from n = 3 rats per sex (6 total) across 47 non-diseased tissues, with each row representing a specific tissue type. Notably, 12/40 targets exhibited relatively higher expression in the central nervous system (CNS) (highlighted in the black box), indicating potential cross-species translational relevance. The color gradient reflects log₂(FPKM + 0.1) values, ranging from low (blue, −5.5) to high (red, 5.5) expression.

Fig. 5.

Chemical diversity among reference compounds. Extended chemical fingerprints (ECFP6; panels A and D), MACCS keys (panels B and E), and simple chemical descriptors (SDC) (Chemaxon calculated molecular weight, AlogP, molecular fraction polar surface area (PSA), logD (pH 7.4), numbers of aromatic rings, hydrogen bond acceptor and donor, rings, and rotatable bonds; panels C and F) were evaluated and plotted by t-SNE. The negative compounds (blue) are just as diverse in structure and physiochemical properties as the positive compounds (green) (panels A to C). To confirm diversity, the distribution of reference compounds (blue) was overlaid with the Tox21 chemical inventory as a background (yellow) (panels D to F).

Fig. 6.

Heatmap summarizing data availability for identified reference compound testing across in vitro assays evaluating seizure-relevant targets. ChEMBL and ICE in vitro data were mined to identify assays that evaluated the molecular target families relevant for seizure. The plot shows which positive and negative seizure-liability compounds were tested (yellow; whether active or inactive) or not tested (dark purple) in assays mapped to the respective target families (y-axis).

Fig. 7.

Reference compounds with most testing data available for the identified seizure in vitro assays in ICE and ChEMBL. Relevant target families were searched to identify number of assay endpoints with testing data among the reference compound inventories. For ICE, an endpoint is defined as an AC50, whereas for CheMBL, an endpoint is defined as an IC50, EC50, or ki. These are visualized for the seizure-liability positive reference compounds (A) and for the negative reference compounds (B). For simplicity, only compounds with 10+ assay endpoints are included for the positive reference compound plot.