Computational Toxicology Research
v-Embryo™ research is focused on improving the predictive capabilities of in vitro toxicity testing through the application of in silico models with sufficient depth and detail for predictive modeling of developmental defects. The project is currently focused on the following areas:
The developing eye has been chosen as a prototype developing organ. This morphogenic system has many attributes which makes it a good model.
- Well studied
- Good literature base
- Developmental ontology
- Human relevance
- Genetic/environmental susceptibility
- Range of ocular phenotypes
- Variety of signaling pathways involved
- Reciprocal tissue interactions
- Conservation of cell signaling
- Mitochondrial dependency
Eye development is simple enough to capture a great deal of experimental detail for computer simulation and yet complex enough to warrant predictive in silico modeling for developmental toxicity.
Our current approach for informing our in silico models is through predictive/statistical associations between the ToxRefDB in vivo database and ToxCast™ in vitro database, and through experimental data gathered from our Virtual Tissues Knowledgebase (VT-KB).
Predictive/statistical associations – Machine learning tools can help build signatures of associations between disrupted cellular signaling and pathways and ocular defects. These associations can help guide us to important cellular pathways that ocular teratogens disrupt, which may lead to ocular defects.
Predictive associations can only give us a limited and defined scope of the cellular processes disrupted of ocular teratogens. The soon to be developed Virtual Tissues Knowledgebase will scout the literature for all of the cellular pathways and processes important during normal and disrupted ocular teratogenesis to help build our model and give context to our predictive/statistical associations. Our VT-KB will scout the literature for all of the cellular pathways and processes important during normal and disrupted ocular teratogenesis to help build our model and give context to our predictive/statistical associations. An initial simulation model of early eye development was designed using CompuCell3D, an Agent Based Modeling (ABM) technique using the cell as the smallest agent. This simulation shows the morphogenesis of lens invagination to lens vesicle detachment. The simulation is able to recapitulate some of the cellular processes driving lens invagination including changes in cell shape, growth, and differential adhesion. Additional cellular processes, cell signaling, and gene regulation during normal and disrupted eye development derived from our VT-KB and predictive/statistical associations will be important parameters in making this model more realistic for predictive modeling efforts.
EPA's Virtual Embryo project is building in silico models of vascular morphogenesis to test mechanistic hypotheses by simulating developmental toxicity effects.
- Deepen the current understanding of developmental vascular biology
- Identify biologically significant perturbations in potential toxicity pathways that may lead to vascular disruption and downstream consequences
- Develop and implement new testing strategies with greater specificity and predictive power
Abnormal vascular development alters maternal placental oxygen and nutrient delivery and interrupts morphogenetic gradients that affect the distribution of nutrients, metabolic products and/or chemicals that can lead to serious impact on embryo development and organogenesis.
Due to the centralized role and complexity of the signaling networks underlying blood vessel formation and remodeling and their centralized role across all embryonic systems, developmental defects mediated by disruption of the vascular network can be manifested in diverse ways and attributed to a variety of factors.
Vascular-associated severe developmental effects include:
- embryonic lethality
- neurological deficits
- limb defects
Traditional prenatal animal testing is uninformative for resolving and attributing the inherent phenotypic ambiguity that arises due to various vascular network perturbations, prompting a new paradigm for toxicity testing as set forth by the National Research Council in 2007.
ToxCast data from 467 assays and 309 high priority pesticides/chemicals shows a number of developmental toxicants were found to also disrupt in vitro assays for specific targets or cellular processes important to vasculogenesis and angiogenesis. The predictive signature built from ToxCast data for several in vivo developmental endpoints from EPA’s ToxRefDB database includes a pro-inflammatory/anti-angiogenic chemokine network, elements of the vascular endothelial growth factor (VEGF) signaling pathway, and the plasminogen activating system (PAS) of enzymes and growth factors mediating matrix remodeling and local signaling during blood vessel growth.
We hypothesize that embryonic microvascular networks are targets for certain environmental compounds with teratogenic potential, and have identified a group of putative Vascular Disruptor Compounds (VDCs) from the ToxCast Phase I dataset.
Limb bud development is another embryonic system that is affected by chemical disruption and is currently being modeled at NCCT. We hope this in silico model can provide us with an integrative systems approach in understanding normal and disrupted cellular processes that are important in the developing limb bud, and we plan to use the process developed from this project.
To this end, we have developed a preliminary model using CompuCell3D, an agent-based cellular automata model, to recapitulate the normal developing limb bud spatial morphology gradient control. In the limb bud, ectodermal tissue is signaled by a region of cells, called the ZPA, to bunch up and form an AER (apical ectodermal ridge). The AER is maintained by the signal from the ZPA and the AER produces a growth factor that maintains the ZPA. This growth factor also encourages cellular proliferation and chemotaxis in the direction of the growth factor in the SAM (sub apical region) which is just below the AER. As this SAM proliferates, the AER is forced away from the developing embryo which results in a bulging of tissue and a separation of the ZPA and AER. When these two regions are far enough apart, they are no longer able to maintain each other and so the growth process of the AER is suspended. At the point of suspension, the limb bud is fully developed, having vasculature, preliminary bone formation, and preliminary digit specification.
The initial computational model shows the formation of the ZPA and AER and the interactions between them, and the control the growth factors from these regions exert over the cells in the region of influence. Additionally, each cell has a pre-defined gene regulatory network (defined from the literature) that is influenced by neighboring cells actions (such as secreting morphogens). Gene expression changes in the cells as a consequence of each individual cells local environment physical and signaling information. Linking these expression patterns, as opposed to individual expression, to specific cellular control mechanisms directs cell behavior leads to the formation of the limb. The cellular interactions developing as the simulation runs, may give rise to emergent phenotypes resembling true biological morphologies, and give us a better idea of how cellular crosstalk plays a role in limb bud development.
Stem cell biology is an exciting field of research with nearly 40,000 articles published in the past five years. Embryonic stem cells are an excellent model system to identify toxicity pathways because they are amenable to high-throughput screening and genetic manipulation. In addition, ES cell differentiation has been shown to mimic early development suggesting critical pathways are active in this model system. Using both ToxCast™ assay data and ToxRefDB endpoints for 309 mostly food-use pesticides, we are building associations with ES cell endpoints and predictive models of perturbed ES cell differentiation to develop hypotheses regarding chemical perturbations of our model system. Furthermore, a computational model of ES cell differentiation will allow us to test chemical effects in silico. The goals of this project are to:
- Identify toxicity pathways.
- Build an ES cell state map with coordinating marker genes.
- Use chemical perturbations to phenocopy genetic effects.