The long-term objective of this research is to improve food safety by enabling the integration of in vitro data concerning mechanisms of toxicity to humans in chemical risk assessment. In this context, the goal of the proposed project is to establish a novel and practical systems-toxicology platform that integrates in vitro mechanistic information derived from characterization of cellular and molecular responses of human cells to chemicals with bioinformatics approaches, to predict liver injury from food-relevant chemicals. Food-relevant chemicals include pesticides, food additives, food contact materials and food processing products. Using non-alcoholic fatty liver disease (NAFLD) as an adverse outcome of human relevance, our central hypothesis is that quantification and integration of profiles of multiple cellular and molecular responses in human liver cells exposed to low doses of chemicals that induce NAFLD through complementary mechanisms can be used to identify patterns of toxicity relevant to NAFLD. We will test this hypothesis and use computational in vitro to in vivo extrapolation modelling of the acquired in vitro data to derive estimates of in vivo doses that correspond to the in vitro concentrations that trigger NAFLD-relevant adverse outcomes. These equivalent in vivo doses will be compared to human exposure estimates to predict and prioritize food-related chemicals as potential risk factors for NAFLD. We have routinely incorporated in our lab the use of a cell-based model for human liver responses (HepaRG) that is competent in critical metabolic activities that influence susceptibility to hepatotoxicity in humans in vivo. In addition, we have preliminary data supporting the technical feasibility of proteomics for characterization of cellular and molecular profiles in these cells. Our research team combines critical expertise in food toxicology, systems toxicology, proteomics, and bioinformatics. To test our hypothesis, we will address the following aims:
(1) Establish a multiplex assay for NAFLD-inducing chemicals in human liver cells. The working hypothesis is that chemicals known to induce NAFLD-related adverse outcomes in humans induce a quantifiable pattern of molecular responses in vitro that can be used as a predictive basis of their toxicological potency for NAFLD. The work of this aim will involve establishing and applying a multiplex flow cytometry-based assay to monitor multiple mechanism-based markers of known NAFLD key events in chemical-exposed metabolically-competent human liver cells (HepaRG). Experiments will be performed using a carefully selected group of training chemicals and data will be quantified, integrated, and validated against available in vitro and in vivo toxicity data for these chemicals to establish an in vitro-based predictive model for NAFLD, with which oral dose equivalents equal to the adverse-response-related in vitro concentration will be modeled.
(2) Identify convergent molecular targets of NAFLD-inducing chemicals as candidate quantifiable biomarkers. The working hypothesis is that human liver proteins that are common molecular nodes for the initiation of NAFLD are quantifiable candidate biomarkers for increased NAFLD risk. A proteomics-based workflow will be established and used to characterize the same training set addressed in Aim 1 studies. Known and previously unknown patterns of molecular responses identified in these studies will be used to delineate distinct responses between drugs that cause NAFLD through complementary mechanisms, thus refining the in vitro/in silico workflow of characterizing hepatotoxicants.
(3) Identify putative hepatotoxicants with food-related exposures and characterize them in vitro. The objective of this aim is to derive oral equivalent doses for chemicals predicted to induce NAFLD (or simple steatosis, as an NAFLD-susceptibility factor). The work of this aim will involve a combination of bioinformatics and experimentation whereby putative hepatotoxicants with exposures relevant to food will be ranked on the basis of publically available toxicological data and chemical structure-based modelling. A mechanism- and hypothesis-driven selection of compounds will be tested using a combination of cell sorting and measurement of candidate protein markers and integration of resulting data in the context of the predictive model developed in Aim 1.
The proposed project will provide proof-of-concept for mechanism-based quantitative risk assessment for NAFLD. This strategy will allow read across of large lists of compounds based on a combination of structural and mechanistic alerts in order to identify, group and prioritize food-related chemicals for further testing and for monitoring their use and associated human exposures, reducing the need for animal testing. The testing methodology of this project can be integrated into existing fit-for-purpose regulatory testing batteries to increase scientific confidence in the accuracy of hepatotoxicity risk prediction in food safety assessment. The resulting workflows will be transferrable to FSVO for in-house use and data handling know-how.
