You are here:
Mining a human transcriptome database for Nrf2 modulators
Citation:
Rooney, J. AND C. Corton. Mining a human transcriptome database for Nrf2 modulators. SOT, New Orleans, LA, March 13 - 17, 2016.
Impact/Purpose:
This study developed computational methods for mining a large human microarray database to find modulators of Nrf2 that can be used to populate adverse outcome pathway networks that link oxidative stress and a number of adverse outcomes.
Description:
Nuclear factor erythroid-2 related factor 2 (Nrf2) is a key transcription factor important in the protection against oxidative stress. We developed computational procedures to enable the identification of chemical, genetic and environmental modulators of Nrf2 in a large database of human microarray data (~38,000 transcript profiles). Multiple tissue-independent gene expression biomarkers were constructed from microarray experiments in which cancer cell lines derived from human hepatocyte, breast, and lung were exposed to chemicals (e.g., oltipraz, sulforaphane, CDDO-Im) known to activate Nrf2 or in which the Nrf2 suppressor Keap1 was knocked down. Biomarker genes were identified which had consistent directional expression changes across the comparisons and then were further filtered for those dependent on Nrf2 by removal of genes not affected by siRNA-mediated Nrf2 knockdown. The resulting 10 preliminary biomarkers were compared to 298 gene expression comparisons (biosets) from HepG2 cells or primary human hepatocytes exposed to 55 chemicals with known Nrf2 activity based on ToxCast and Tox21 high-throughput Nrf2 assays carried out in HepG2 cells. Comparisons were carried out using the Running Fisher algorithm, a statistical test of pair-wise similarity in expression. Balanced accuracies ranged from 73-82%. The most predictive biomarker consisted of 70 genes that included many well-known Nrf2 targets (e.g., AKR1B10, NQO1, TXNRD1, SXNRD1). In an examination of the biosets in the human microarray database, the biomarker was able to identify 1) many known Nrf2 chemical activators in diverse cell lines including aryl hydrocarbon receptor activators known to activate Nrf2 secondary to increases in CYP-mediated oxidative stress, 2) known environmental stressors including tobacco smoke, and 3) known (KEAP1, NFE2L2) and novel genes that when overexpressed or knocked down affect Nrf2 activation. In summary, we provide computational methods for mining a large human microarray database to find modulators of Nrf2 that can be used to populate adverse outcome pathway networks that link oxidative stress and a number of adverse outcomes. This abstract does not represent EPA policy.