Citation:

Frank, C., J. Brown, K. Wallace, W. Mundy, I. Shah, AND Tim Shafer. Defining toxicological tipping points using microelectrode array recordings of developing neural networks. North Carolina Society of Toxicology, RTP, NC, October 25, 2016.

Impact/Purpose:

This abstract presents analysis of data from microelectrode array recordings of developing neural networks. This approach may be useful for screening compounds for potential developmental neurotoxicity and prioritizing them for additional testing. The data analysis approaches presented are a novel approach to these data and are necessary advances to make the approach useful for compound screening and prioritization.

Description:

Only a small fraction of environmental compounds humans come in contact with have been evaluated for developmental neurotoxicity (DNT) hazard. Resource and time constraints make DNT guideline animal studies untenable for large numbers of compounds. The EPA is therefore developing more efficient methods to screen and prioritize the thousands of chemicals with undefined DNT hazard. Recent efforts have focused on key neurodevelopmental processes, like synaptogenesis, neurite outgrowth, or glial proliferation to identify adverse outcomes originating from diverse putative molecular initiating events. However, the major end product of neurodevelopment is fully-functional neural networks, capable of complex communication with both excitatory and inhibitory input. To fill this gap, we developed an assay for DNT hazard using in vitro microelectrode array (MEA) recordings of developing cortical neuron networks. A complex culture is isolated from postnatal day zero rats that shows a characteristic increase in spontaneous network activity and synchrony as neurons form and strengthen connections on the MEA platform (Figure 1). By 12 days in vitro, mature networks of cortical neurons are established, enabling examination of chemical effects on network formation and maintenance.Figure 1. A-C: 20x images of DIV 12 cortical culture on the MEA shows (A) dense culture around electrodes with (B) extensive dendrites, astrocyte bed, and (C) excitatory and inhibitory synapses. Scale bar is 100 µm. D-F: Untreated culture development of mean firing rate, burst rate, and correlation network parameters over time (n=250).We exposed developing cortical neuron networks to a library of 70 chemicals across a range of concentrations and quantified 16 different network parameters for 4 timepoints, resulting in a dataset comprised of 31,360 endpoints. To prioritize library compounds for additional testing, novel computational methods were required. I developed two complimentary methods to quantify compound potency in producing adverse network effects over developmental time. First, an area-under-the-curve metric was applied to network parameter values over time to simplify concentration-response modeling without loss of developmental delay effects. This method also allowed for estimation of the concentration where 50% of network activity is lost (EC50) for each network parameter. Comparison of AUC EC50 values to two terminal cell viability assays multiplexed with the MEA recordings was used to provide insight into the selectivity of network effects relative to cell killing effects for 40 compounds (Figure 2).Figure 2. Summary ranking of 40 positive compounds with impact on network formation by potency (AUC-derived EC50 values) and selectivity (based on lower EC50 of LDH and CTB cytotoxicity assays), both averaged across 16 network parameters. Compounds circled in red exhibit cortical network effects at concentrations below those that caused overt cell death.Second, to integrate data across network parameters and model adaptive responses to chemical exposures, I calculated the total scalar perturbation across time. Each network parameter was converted to a z-score metric of deviation from untreated control range and 8 key parameters (Figure 3A) were aggregated into a single value representing total scalar perturbation of the system (Figure 3B). This allowed for estimation of system velocities indicating whether a given concentration causes network failure or allows for recovery by adaptive response (Figure 3C). Critical concentrations (i.e. tipping points) that mark the point at which toxicity begins to overwhelm the developing system were defined from the system velocities of 35 compounds. Comparison of these network formation tipping points to network and cell viability EC50 estimates suggested tipping points are often more sensitive than individual netw

Record Details: