You are here:
Identification of potential chemical carcinogens in compendia of gene expression profiles.
Citation:
Corton, C. Identification of potential chemical carcinogens in compendia of gene expression profiles. SOT, San Antonio, Texas, March 12 - 15, 2018.
Impact/Purpose:
Chemicals induce cancer through partially characterized adverse outcome pathways (AOPs) that include molecular initiating events (MIEs) and downstream key events (KEs). We developed computational strategies to identify chemicals in large microarray compendia that activate cancer MIEs/KEs.
Description:
Chemicals induce cancer through partially characterized adverse outcome pathways (AOPs) that include molecular initiating events (MIEs) and downstream key events (KEs). We developed computational strategies to identify chemicals in large microarray compendia that activate cancer MIEs/KEs. Predictive biomarkers were built from genes regulated in response to knockdown or overexpression of the transcription factor (TF) and/or those consistently responding to chemicals known to modulate the TF. Biomarkers were compared to gene expression profiles using a correlation-based method which determines the significance of the overlapping gene expression profiles allowing for prediction of modulation of that MIE/KE. We used predictive biomarkers that can accurately identify chemicals that modulate a number of MIEs (genotoxicity assessed by p53-dependent targets, the endocrine disruptor targets estrogen receptor (ER) and androgen receptor (AR), as well as xenobiotic receptors aryl hydrocarbon receptor (AhR), constitutive activated/androstane receptor (CAR), peroxisome proliferator-activated receptor α (PPARα)) or downstream KEs (cell proliferation and oxidative stress (assessed indirectly through Nrf2 activation)). A number of examples will be given of applications of our approach including screening in 1) ER positive breast MCF-7 cells, 2) AR positive prostate LAPC-4 cells, and 3) human primary hepatocytes. Our screen identified many unique chemicals previously unknown to affect these pathways. In the MCF-7 compendium of 1153 unique compounds, 75 compounds were found to activate ERα and 39 suppressed ERα. ERα activators included those that are known ERα agonists as well as previously unrecognized activators. Chemicals were identified that increased cell proliferation either dependent or independent of ER/AR activation. Our approach highlights the value of using transcript profiling to populate cancer AOPs and identify chemicals that have the potential to cause cancer. This does not represent EPA policy.