Jump to main content.

Data collection and computational modeling for the "virtual embryo"


This task focuses on the predictive toxicology of children’s health and development following prenatal or lactational exposure to environmental chemicals. The research is motivated by a computational framework of developmental toxicity to formally guide the generation, assessment, and evaluation of data, tools, and approaches focused on embryonic growth, morphogenesis, and differentiation. Research outcomes will be improved understanding of the molecular pathways and cellular processes leading to adverse pregnancy outcomes, and better ways to assess the impacts of prenatal and postnatal exposure to chemicals at various stages of development and scales of biological organization.

Rationale and Research Approach:

The developing embryo is a complex adaptive system in which the susceptibility or resilience to chemical disruption varies by organ system and stage. The individual molecules and cells have their own developmental trajectories and form complex networks of interactions that regulate morphogenesis and differentiation. Whereas genetic approaches have identified these critical pathways, little is known about the pathways of toxicity leading to adverse developmental outcomes. Biomechanical forces, signaling gradients, and genetic oscillations that pattern the embryo act with remarkable precision (robustness) but also must react quickly to perturbation (flexibility). The combination of robustness and flexibility is characteristic of an adaptive system at the edge of chaos.

This task aims to build computational models of development that can be used to analyze these complex interactions and predict chemical interactions with core functions such as molecular clocks, spatial gradients, molecular machines, growth mechanics, and differentiation pathways. We propose that an array of Virtual Embryo models representing key aspects of embryonic development will effectively capture the flow of molecular information across discrete cellular networks, and simulate mechanistic relationships leading to adverse developmental outcomes.

To test this hypothesis, we will build cell-based computer models that simulate a morphogenetic series of events (e.g., blood vessel development, limb-bud morphogenesis, reproductive system, neural differentiation). The models will be hypothesis-based and incorporate vast information from the literature and high throughput screening data from ToxCast to simulate emergent systems-level behaviors. The simulations will be qualified against experimental data and the models deployed for toxicological assessment. To do this, we propose a cell-agent based model (ABM) approach in which every cell is capable of autonomous decisions. Each simulated cell, like a biological cell, processes local cues from its environment and behaves according to its own blueprint or history. Higher-order responses of the system emerge from the collective cellular behavior in the simulation. This type of model is ideal for predictive toxicology because it integrates information across different biological scales: molecular information such as internal clocks, biochemical gradients, and gene regulatory networks; cellular properties such as growth, adhesion, and differentiation; and tissue-level properties such as homeostasis, morphogenesis, and repair.

The cell-agent-based models in Virtual Embryo will be run from a computer program coding using the Python and XML scripting languages. Specific rules for molecular pathways and cellular behaviors are captured from intramural and extramural scientific literature in a virtual tissue knowledgebase (VT-KB), and the software to implement multicellular interactions is an open access tissue simulation environment (http://www.compucell3d.org/). We will collect experimental data to profile the consequences of chemical exposure, using assays with targets specific to developmentally important adverse outcome pathways (AOPs). We will conduct research to identify important toxicity pathways, develop assays to evaluate those pathways, and link in vitro signatures to phenotype. For example, a phenotype integral to proper embryogenesis is vascular development, and disruption of that process can have serious developmental consequences. Because the pathways active in angiogenesis are similar across vertebrates it is possible to use an alternative species (zebrafish) to define the chemical effects on the process, thereby providing experimental data for profiling the consequences of chemical exposures. Furthermore, targets in angiogenesis pathways can be identified via an -omics approach for the identification of diagnostic biomarkers to facilitate model development and prioritizing chemicals based on modes of action. A life-stage based risk assessment paradigm requires information on absorption, distribution, metabolism, and excretion (ADME) and unique life-stage and/or species susceptibilities that can be attributed to maternal-fetal exchange.

There is little experimental data regarding the placental and lactational transfer of chemicals from mother to infant. We will perform research to reduce uncertainty regarding the role of maternal body burden in the exposure of the fetus and breastfeeding infant to environmental chemicals. We will develop exposure models and tools to better characterize and define the trans-placental and trans-lactational exposure pathways. For example, novel lipid-producing cells that model the mammary epithelium will be evaluated for their use in determining ADME, kinetic, and dosimetry parameters with respect to chemical storage, release, and partitioning. In conjunction with this cell-based approach, research will be conducted to develop predictive models for estimating the concentrations of environmental chemicals in breast milk, which will facilitate the utilization of available biomonitoring data to predict potential infant exposures. These models can be integrated with physiologically-based PK models to aid in life-stage specific exposure estimation and chemical risk assessment.

MED Scientists:

Sig Degitz



Expected Products:




Sep 30, 2013

(2) Implementation of cell-agent based models linked to complex embryological phenomena (limb-bud morphogenesis, reproductive tract development). This product will deliver computer models imputing knowledge of signaling networks, tissue induction, spatial patterning, and molecular clocks to recapitulate morphogenesis and predict points of departure for dysmorphogenesis. The models would be trained with 20 reference compounds in ToxCast and applied to 20 data-poor chemicals in TSCA21.

Sig Degitz
Sep 30, 2015 (5) Expansion and integration of the Virtual Embryo toolbox. This product will plug-in other key systems to the toolbox, including neural crest and somite development. Information from in silico models will be corroborated by in vitro work (stem cells, zebrafish and Xenopus embryos, cultured organs) and targeted in vivo studies. Product development will be guided by cell-agent based models and cell-imaging data, and implemented for 10 chemicals in EDSP21/TSCA21 with relevant activities in ToxCast. Sig Degitz
Sep 30, 2012 (1) Integration of angiogenesis information: A cell-agent based systems model developed from empirical data and biological knowledge of blood vessel development. The angiogenesis model will be trained with compounds showing anti-angiogenic properties, assessed in a forward validation for predictive developmental toxicity among 1,000+ ToxCast chemicals in pregnant rats/rabbits, and tested for vascular disruption in zebrafish embryos and embryonic stem cell assays for 30+ chemicals in EDSP21/TSCA21/OW21/OPP21. Sig Degitz

Top of page

Local Navigation

Jump to main content.