You are here:
Computational toxicology and in silico modeling of embryogenesis
Citation:
KNUDSEN, T. B., N. SIPES, N. KLEINSTREUER, M. ROUNTREE, N. C. BAKER, R. M. SPENCER, K. J. CHANDLER, S. J. PADILLA, T. TAL, AND E. S. HUNTER. Computational toxicology and in silico modeling of embryogenesis . Presented at 40th Meeting of the European Teratology Society, ETS-EUSAAT Symposium, Johannes Kepler University, Linz, AUSTRIA, September 02 - 05, 2012.
Impact/Purpose:
By utilizing signaling networks and gradients for the system being studied, a cellular-ABM can run a morphogenetic series of events to analyze complex relationships, such as interactions between varied cell types, and predict tissue-level outcomes from in vitro HTS data. Progress has been demonstrated in this regard for vasculogenesis and limb-bud morphogenesis. A ‘virtual embryo’ toolbox provides a means to integrate HTS data with relevant empirical knowledge to generate AOPs useful in assessing early life-stage specific susceptibility.
Description:
High-throughput screening (HTS) is providing a rich source of in vitro data for predictive toxicology. ToxCast™ HTS data presently covers 1060 broad-use chemicals and captures >650 in vitro features for diverse biochemical and receptor binding activities, multiplexed reporter gene activation, gene expression profiles, stress-response indicators, cellular states/functions, zebrafish embryogenesis and mouse embryonic stem (mES) cell differentiation. Predictors of prenatal developmental toxicity in the first-generation predictive models (based on 309 chemicals) were assays for retinoic acid, G-protein coupled receptors, and TGF-beta signaling in pregnant rats; and for CCL2-chemokine, interleukins, and TGF-beta signaling in pregnant rabbits. These findings suggest that the predictive power of in vitro profiling, coupled with biological plausibility can inform chemical prioritization in the absence of associated in vivo data on developmental toxicity. Biological insight is, however, necessary to extend these domain-specific model structures into mechanistically-based models. For example, an a priori ToxCast signature for redox-disruption correlated with inhibition of cardiomyocyte differentiation in mES cells. This finding is consistent with knowledge that cellular redox control is important for embryogenesis. An a priori ToxCast signature derived for potential vascular disruption correlated with developmental toxicity in pregnant rats and rabbits. This finding supports the concept that disruption of embryonic vascular development functions as an Adverse Outcome Pathway (AOP) for developmental toxicity. AOPs are conceptual models that link molecular targets to organelle dysfunction, cellular injury, tissue disruption, organ pathophysiology and clinical phenotype. Computer simulations may clarify AOPs by coupling events at various levels of biological organization to cellular and/or subcellular dynamics. This can be accomplished with multi-scale cellular agent-based models (ABMs) that integrate diverse biological information. By utilizing signaling networks and gradients for the system being studied, a cellular-ABM can run a morphogenetic series of events to analyze complex relationships, such as interactions between varied cell types, and predict tissue-level outcomes from in vitro HTS data. Progress has been demonstrated in this regard for vasculogenesis and limb-bud morphogenesis. A ‘virtual embryo’ toolbox provides a means to integrate HTS data with relevant empirical knowledge to generate AOPs useful in assessing early life-stage specific susceptibility. This abstract does not necessarily reflect US EPA policy.