Screening And Prediction Of Chemical Toxicity Using Functional Genomics Approaches

Chemicals have become the most important source of risk on global ecological safety and human health.In recent years,thousands of registered chemicals lack toxicity information,meanwhile the concentrations of environmental pollutants such as pesticides,antibiotics and flame retardants are ever increasing,which has raised great concerms on their adverse effects on environment and health.Additionally,tens of thousands of chemicals exist in environmental medium,causing complex mixture pollution.However,traditional whole animal-based toxicity testing that is expensive and time-consuming cannot fill the gap of toxicity data of so many chemicals.It is urgent to develop cost-effective and high-throughput predictive approaches,which can identify chemically-induced perturbations on biological pathways.Omics tools can promote the investigation of perturbed biological pathways on whole genome scale.However,due to the high cost of current omics technology,it is not allowed to measure thousands of chemicals and wide range of concentrations by omics.Moreover,traditional omics approaches mainly focus on passive gene expression to chemicals,which cannot reflect direct associations between chemicals and genes.This study based on two kinds of gene functions to developed two novel omics approaches for assessing environmental chemicals.Firstly,based on CRISPR-Cas9 gene deletion technology,a CRISPR-Cas9 functional genomic screening approach was developed,which can identify critical sensitive genes of environmental chemicals via screening on genome-deleted human cell lines.Secondly,based on target-sequencing transcriptional expression technology,a reduced human transcriptomics(RHT)approach that measure only a small set of key human genes was developed,which can be alternative to whole transcriptomics for cost-effective and highly dynamic assessment of transcriptional expression of chemicals.A streamlined bioinformatics pipeline based on RHT testing was proposed for quantitative characterization of biological pathways,which was further extended to develop a novel framework for chemicals classification by dose-dependent omics data.For assessing effects of mixtures,dose-dependent RHT was evaluated to be able to validly characterize bioactivity of mixtures either by whole mixtures or by components.Finally,RHT approach was successfully applied into benchmarking water quality of water samples from waste water to drinking water.The specific research content and conclusions are as follows:(1)A CRISPR-Cas9 functional genomic screening approach was developed for identification of gene targets of environmental chemicals.By using whole-genome knockout HepG2 cell lines,resistant genes to triclosan(TCS)at highly cytotoxic concentrations of IC50 were identified to be mainly associated with signaling pathways,followed by individual knockout validation of key resistant genes including FTO and MAP2K3.Sensitive genes to TCS at lowly cytotoxic concentrations of IC10 and IC20 were examined to be mainly involved in immune response pathways,which were consistent with transcriptomics profiles of TCS at low concentrations.Combination of CRISPR?Cas9 functional genomic profiles of TCS and existing "gene-chemical" or"gene-disease" database can be explored to provide potential associations between TCS and human disease,such as breast cancer and obesity.The CRISPR-Casg functional genomic screening can offer a novel sight into identifying key gene targets of environmental chemicals.(2)A standardized reduced human transcriptomics(RHT)approach that measure 1200 human key genes was developed.Bioinformatics was performed to validate that the 1200 RHT genes can cover at least 90%of known biological pathways,can qualitatively simulate profiles of whole transcriptomics for grouping chemicals,and can quantitatively benchmark point of departure(POD)derived from whole transcriptomics data.Further,based on RHT testing,a bioinformatics pipeline for quantitative characterization of biological pathways was developed,which can use dose-response models to derive chemically-related POD.The repeatability of RHT was evaluated by conducting dose-dependent RHT testing on three successive passages of HepG2,which showed robustly reproducibility of RHT-derived POD within 1 magnitude,and consistent sensitive pathways related to immune response.Finally,compared with the profiles of ToxCast,RHT approach can derive POD of TCS more sensitively than ToxCast.(3)A pathway-based chemical classification approach using dose-dependent RHT data was proposed.A case study was conducted on 17(non-)genotoxic chemicals,whose bioactivity was characterized by dose-dependent RHT testing in HepG2 cells.Three typical pathway analysis methods,PODA,ORA and GSEA,were individually evaluated on their performance for chemical classification using RHT profiles of 17(non-)genotoxic chemicals.PODA can provide POD values for quantitatively discrimination of chemicals.GSEA can provide enrichment scores of pathways to indicate the situation of perturbed pathways.ORA heavily relied on statistical significance,which failed to provide useful information for pathway enrichment analysis.Thus,PODA and GSEA were proposed as an integrative approach(PODA+GSEA)for chemical classification.Total ten indexes were used,including POD derived by PODA and enrichment scores of each 9 categories of biological processes identified by GSEA,which were submitted to weight-of-evident hierarchical clustering using ToxPi software.The PODA+GSEA approach successfully distinguished 17 genotoxic or non-genotoxic chemicals.(4)Application of dose-dependent RHT testing into characterizing bioactivity of artificial mixtures composed of 12 on environmental chemicals.Dose-dependent RHT data in MCF7 can validly capture profiles of perturbed pathways of 12 individual chemicals,which were consistent with known MOAs of corresponding chemicals.Further,using CA and IA models,bioactivity of individual components can predict observed bioactivity of artificial mixtures,whose difference was within 2-fold.Moreover,for 9 categories of specific biological processes,component-based approach can still predict observed bioactivity of artificial mixtures within 1 magnitude.Finally,by comparing the results of RHT and previous in vitro bioassays,significantly positive correlation can be observed for pathways associated with biomarkers including AhR,Nrf2,AR and ER.(5)Application of dose-dependent RHT testing into benchmarking water quality.Dose-dependent RHT was applied for profiling bioactivity of 10 water samples ranging from wastewater to drinking water in human cells,HepG2 and MCF7.Dose-response models were used to identify dose-responsive genes(DRGs)and to calculate POD of DRGs,which could be ranked to investigate low dose response.Furthermore,molecular pathways disrupted by different samples were evaluated by Gene Ontology(GO)enrichment analysis.The ability of RHT for representing bioactivity utilizing both HepG2 and MCF7 was shown to be comparable to the results of previous in vitro bioassays.Finally,the relative potencies of the mixtures indicated by RHT analysis were consistent with the chemical profiles of the samples.RHT analysis with human cells provides an efficient and cost-effective approach to benchmarking mixture of micropollutants and may offer novel insight into the assessment of mixture toxicity in water.