Abstract
Providing an unbiased and comprehensive view of the DNA damage response in cells is critical in genotoxicity screening to identify substances that cause diverse types of DNA damage. Considering that S. cerevisiae is one of the most well-characterized model organisms in molecular and cellular biology, we created a map of the DNA damage response network containing the reported signaling pathways in yeast cells programmed to constitutively respond to DNA damage. A collection of GFP-fused S. cerevisiae yeast strains treated with typical genotoxic agents illuminated the cellular response to DNA damage, thereby identifying 15 protein biomarkers encompassing all eight documented DNA damage response pathways. Three statistical and one deep learning models were proposed to interpret the quantitative molecular toxicity end point, i.e. protein effect level index (PELI), by introducing weights of 15 biomarkers in genotoxicity assessment. The method based on standard deviation exhibited the best performance, with an R (2) of 0.916 compared to the SOS/umu test and an R (2) of 0.989 compared to the comet assay. The GFP-fused yeast-based proteomic assay has minute-level resolution of pathway activation data. It provides a concise alternative for fast, efficient, and mechanistic genotoxicity screening for various environmental and health applications.