SBIR Phase I: Quantum mechanical predictive tools for identification and redesign of skin sensitizing chemicals

This Small Business Innovation Research (SBIR) Phase I project addresses the timely need to develop new and robust in silico tools for the prediction of skin sensitization potency of commercial chemicals. Currently, there are no reliable and broadly applicable alternatives to the costly and ethically unfeasible in vivo screening methods for skin sensitization testing. Skin sensitization response is triggered by covalent modification of key biochemical targets by the xenobiotic. These targets have been identified as glutathione, lysine-containing peptides and surface cysteine residues of the Keap1 protein. The objectives of the proposed research are to (i) build quantum-mechanical models representative of covalent interactions between the xenobiotic and biochemical targets, and (ii) incorporate supporting models for skin bioavailability and identification of potential metabolites to compose a comprehensive model of skin sensitization potency. First, reaction energetics associated with binding of skin sensitizers to the biochemical targets will be calculated. Next, property-based linear models will be developed to account for skin bioavailability. Potential metabolites will be identified using existing validated models. The final in silico tool is anticipated to predict skin sensitization potency based on a chemical structure and inform design of safer alternative chemicals based on chemical reactivity and physical properties. The broader impact/commercial potential of this project is to provide the cosmetics, consumer chemicals, pharmaceuticals, textile and petroleum industries with a means of identifying chemicals in their product lines that induce human skin sensitization response and replacing them with safer alternatives. Successful completion of this project will also bring immeasurable benefits to the scientific community. A precedent will be set for applying techniques from the computational chemist's toolbox for building predictive toxicology models capable of informing the design of safer chemicals. This novel synergistic approach to computational toxicology and computational chemistry places our in silico tool in a unique position in the market with no direct competitors. Current approaches for toxicity predictions in silico are dominated by statistical models that rely solely on structural descriptors and/or properties, and subsequently suffer from limited and often uncertain applicability domains. The proposed research alleviates these deficiencies by including chemical reactivity data obtained from direct modeling of molecular interactions. The societal impact of the proposed research is in the safer product delivered to the consumer and in the reduced dependence on animal testing required by in vivo screening methods.