QSAR at the estrogen and mineralocorticoid receptors

The presence of hormonally active compounds in the biosphere, as well as the adverse health effects due to endocrine disruption have lately received considerable public attention. Numerous substances that are associated with toxic effects, such as dioxins, DDT and related compounds, phthalates, had been suspected of acting through endocrine disruption. This public concern led to the formulation of regulations and legislation, both in Europe and in the United States. Endocrine-disrupting chemicals are compounds that interfere with the endocrine system, alter its functions, and consequently may trigger adverse health effects in man and wildlife. Many of these compounds are resistant to biodegradation and often persist in the environment over longer periods of time. The identification of the toxic potential of drugs and chemicals is widely performed with techniques such as in vitro. In addition, a safe in silico identification is also highly desirable to regulatory bodies, the pharmaceutical, and the chemical industry.
Nuclear receptors regulate biological functions such as cell growth and differentiation, metabolic processes, reproduction and development, intracellular signaling and can be involved in carcinogenesis through the control of gene expression. Chemicals that disrupt the endocrine system interfere with the function of nuclear receptors, alter their functions and cause adverse health effects.
In this thesis, the development and validation of three-dimensional in silico models for the mineralocorticoid and the estrogen receptors _, both belonging to the nuclear receptor superfamily, are presented. These models aim at the screening of drug candidates for mineralocorticoid and estrogen _ activity as well as of environmental chemicals for potential endocrine-disrupting activity.
Different in silico tools and protocols were used to quantify receptor-ligand interactions. Those included, molecular dynamics simulations allowing gaining an insight into the dynamical character of the protein-ligand interactions. Multi-dimensional QSAR models were built, using two different technologies, and validated by employing external validation sets, and scramble tests. The models have been added to the VirtualToxLab - a technology for the in silico identification of the toxic (endocrine-disrupting) potential of drugs and environmental chemicals.
A Cambridge Structural database (CSD) search was performed in order to obtain an insight on the existence and the nature of intermolecular interactions involving halogen atoms. Both geometry and topology of such interactions were analyzed and quantified. An algorithm was developed to ensure the high-throughput analysis of the 3D structures, which were used as input for the QSAR study and the VirtualToxLab. The input for this program is the coordinate file and the name of the compound, and the output is a flag (verified/not verified) along with detailed information about this decision file.
The benefit of this thesis on the society, is that an in silico approach can be used, which is faster and less expensive when comparing to in vitro and in vivo experiments. Such an approach may contribute to the well being of man and wildlife since no chemicals are used and the natural resources are retained.