Abstract:
The widespread use of chemicals in today's world has led to significant concerns about their impact on human health and the environment. The traditional use of animal testing to establish safe levels for chemicals is impractical due to cost and time constraints, animal ethics and concerns about the relevance of animal data to humans. In response, there is a growing call for a paradigm shift towards new approach methodologies (NAMs) that are animal-free and include in silico and in vitro methods. In vitro bioassays using reporter gene cell lines are a key component of the 3R (Replacement, Refinement, Reduction) strategy, as they offer a promising, cost-effective and automatable alternative with high-throughput capabilities. To generate reliable in vitro data, the planning, execution and evaluation of the bioassays must be carried out with the utmost care. Chemicals are subject to various loss processes in the bioassay, which can lead to a deviation between the dosed concentration and the actual bioavailable concentration. These processes include reversible distribution and binding to media components and plastic, but also irreversible loss processes due to volatilization or abiotic or biotic degradation processes. If the latter loss processes remain unnoticed, this can lead to an incorrect interpretation of the bioassay results and an underestimation of the chemical hazard. The primary objective of this thesis was to improve the use of high-throughput cell-based bioassays used for single chemical screening. The study aimed to identify potential challenges and limitations, especially considering the influence of chemical transformation processes on bioassay results. Baseline toxicity, the minimal toxicity of a chemical, is caused by accumulation in the cell membrane and can be used to classify chemical toxicity. Chemicals with higher measured toxicity than baseline toxicity may have a specific mode of toxicity and lower experimental toxicity may indicate experimental artifacts and loss processes. A novel baseline toxicity model was developed based on a critical membrane burden derived from freely dissolved effect concentrations of charged and hydrophilic chemicals to consider distribution processes to media components and to make the model applicable to a wide range of chemicals. The measured cytotoxicity of 94 chemicals in three bioassays with different cell lines (AREc32, ARE-bla, and GR-bla) were compared with baseline toxicity by calculating the toxic ratio (TR). Between 44 and 50 chemicals could be identified as baseline toxicants and 22 to 28 chemicals showed a specific toxicity mechanism (TR ≥ 10). However, seven chemicals showed TR < 0.1, which could be an indication of possible artifacts or loss processes. To identify abiotic transformation processes of chemicals in in vitro bioassays, a high-throughput workflow was developed based on 22 potentially unstable chemicals. Chemical stability was assessed in different bioassay media, buffer solutions (pH 4, 7.4 and 9) and solutions of bovine serum albumin and glutathione to examine the influence of hydrolysis and covalent reactions with proteins. Photodegradation and abiotic oxidative degradation were also investigated, but were found to be less relevant for in vitro bioassay conditions. To assess the degradation kinetics of the chemicals, a high-throughput solid-phase microextraction (SPME) workflow using a BioSPME 96-Pin Device was established for extracting chemicals from the exposure solutions. The results indicated that the main contributors to the depletion of test chemicals in the bioassay media were reactions with hydroxide ions and covalent interactions with proteins. In silico models predicting the half-life of the hydrolytic degradation of chemicals in the environment and qualitative models based on structural features predicting reactivity towards proteins were compared with the experimental results. Since these models were not tailored to the bioassay conditions, there were deviations from the experimental results but the models provided a useful initial estimate of stability. The reactivity of the chemicals with glutathione could not reflect the stability in the bioassay medium but gave indications of the possible reactive toxicity of the chemicals. This relationship was further investigated using ten (meth)acrylamides by comparing their measured cytotoxicity and activation of oxidative stress response with their reactivity towards glutathione. Notably, there was a linear relationship between the reactivity of the tested acrylamides and the toxicity and activation of the oxidative stress response, while methacrylamides did not react with glutathione and acted as baseline toxicants. The differences in reactivity were explained by the lower electrophilicity of methacrylamides caused by their different chemical structure. The metabolic activity was found to be different in all three cell lines (AREc32, ARE-bla, and GR-bla) and ARE-bla showed the highest metabolic activity. Cytochrome P450 enzymes could be induced by xenobiotic chemicals in ARE-bla and AREc32. The effect concentrations of 94 chemicals measured in the three cell lines were compared and none of the cell lines showed significantly higher or lower toxicity, which implies that the differences in metabolic activity had no influence on the bioassay results. In summary, this work contributes significantly to refining the interpretation of bioassay data by providing a new baseline toxicity model and developing an experimental approach to assess chemical stability. The knowledge gained from this work on the high-throughput testing of chemicals improved the understanding of potential confounding factors in bioassay results and laid the foundation for improved risk assessment methods using in vitro bioassays.
Bioassay