Concentration–response evaluation of ToxCast compounds for multivariate activity patterns of neural network function

Abstract

The US Environmental Protection Agency's ToxCast program has generated toxicity data for thousands of chemicals but does not adequately assess potential neurotoxicity. Networks of neurons grown on microelectrode arrays (MEAs) offer an efficient approach to screen compounds for neuroactivity and distinguish between compound effects on firing, bursting, and connectivity patterns. Previously, single concentrations of the ToxCast Phase II library were screened for effects on mean firing rate (MFR) in rat primary cortical networks. Here, we expand this approach by retesting 384 of those compounds (including 222 active in the previous screen) in concentration-response across 43 network activity parameters to evaluate neural network function. Using hierarchical clustering and machine learning methods on the full suite of chemicalparameter response data, we identified 15 network activity parameters crucial in characterizing activity of 237 compounds that were response actives ("hits"). Recognized neurotoxic compounds in this network function assay were often more potent compared to other ToxCast assays. Of these chemical-parameter responses, we identified three k-means clusters of chemical-parameter activity (i.e., multivariate MEA response patterns). Next, we evaluated the MEA clusters for enrichment of

Predictive modeling of biological responses in the rat liver using in vitro Tox21 bioactivity: Benefits from high-throughput toxicokinetics

“…This data imbalance presents a challenge for training a machine-learning model to predict activity vs. inactivity, since the model has relatively few examples of activity from which to learn. The problem of imbalanced data has been pervasive throughout research aimed at incorporating in vitro screening into predictive biology applications [ 14 , 23 , 62 , 63 ], and is a widely-recognized issue in many other applications of machine learning [ 64 , 65 ]. Here, we addressed this limitation through the application of an algorithm, namely, SMOTE, which was selected based on its ability to allow improved characterization of in vitro activity response distributions; though other approaches could be applied in future investigations.…”

A human stem cell-derived test system for agents modifying neuronal N-methyl-d-aspartate-type glutamate receptor Ca2+-signalling

“…Most cell systems developed for neurotoxicity testing have until recently been based on rodent primary cultures (Alépée et al 2014 ; Hogberg et al 2011 ; Hondebrink et al 2016 ; Kosnik et al 2020 ; Kreir et al 2018 ; Strickland et al 2018 ; Suñol et al 2008 ; Zurich et al 2013 ; Zwartsen et al 2018 ). The human iPSCs based systems that have undergone first evaluations rely on commercially available cells with costs that prevent their routine use in academic laboratories.…”

“…One of the most comprehensive approaches to identify potential functional neurotoxicity is the use of neuronal networks on MEA (Kosnik et al 2020 ; Strickland et al 2018 ; Vassallo et al 2017 ). The hitherto most robust MEA data have been generated with rat primary neurons.…”

Culture of Rat Primary Cortical Cells for Microelectrode Array (MEA) Recordings to Screen for Acute and Developmental Neurotoxicity

“…Finally, the user can check network synchronicity by looking at number of network bursts, (network) burst percentage, area under the normalized cross-correlation, and full width at half height of the normalized cross-correlation. For more information on MEA parameters, see Cotterill et al (2016); Kosnik et al (2020), Frank, Brown, Wallace, Mundy, & Shafer (2017; Mack et al (2014).…”