Development of computer models to predict chemicals interference with thyroid hormones

Marianne Dybdahl

There is a growing concern about chemicals that may interfere with the body’s hormone systems and cause harmful effects. So far most focus has been on chemicals that alter the balance of sex hormones, but interference with thyroid hormones is now receiving increasing attention. Besides their function in energy metabolism, thyroid hormones play an important role during early brain development. Even moderate and transient reductions in maternal hormone levels during pregnancy can adversely affect the child’s neurological development.

Chemicals may interfere with the thyroid hormones in many different ways, such as altering the production of the hormones, or altering their transport and metabolism in the body. A battery of different test methods is therefore needed to study the potential effects of chemicals, and this is a very resource-consuming task for the many thousands of untested chemicals.

To facilitate this task, computer models have become an important tool to screen and prioritize chemicals for further experimental testing, thereby reducing costs and number of test animals. At DTU Food we are using computer models to predict health effects of chemicals based on their molecular structure. The models are so-called QSARs (quantitative structure-activity relationships).

The aim of the project is to develop computer models for some of the many mechanisms underlying interference with thyroid hormones. To build the models, we will e.g. use experimental test data obtained through collaboration with the US Environmental Protection Agency.

The developed models will be used to screen more than 600,000 chemicals, including around 70,000 chemicals available at the European market. The resulting predictions will be made freely available in an online database. Such a virtual screening may in a fast and cost-efficient way identify potential thyroid hormone disrupting chemicals in our food, environment and consumer products. The new models can also contribute to future design of safer chemicals and drugs.

Project status at January 2017

Two of the three QSAR models have now been developed and validated. The first model can predict whether chemicals activate a receptor with an effect on hormone turnover. This work has just been accepted for publication in Computational Toxicology (Development and Profiling of 72,524 REACH Substances for PXR Activation and CYP3A4 Induction. Rosenberg, S. A., Xia, M., Huang, R., Nikolov, N. G., Wedebye, E. B., and Dybdahl, M.).

The other model can predict chemicals’ inhibition of an enzyme that is instrumental in forming the hormones. We are currently typing up the results of this work for a scientific paper. The latter model is being developed and can be used to predict chemicals’ activation of another receptor which is also involved in hormone conversion.

Sign up for our newsletter - don't miss information about our annual symposium etc.