The NRF2-mediated oxidative stress response pathway is associated with tumor cell resistance to arsenic trioxide across the NCI-60 panel

Abstract

Background: Drinking water contaminated with inorganic arsenic is associated with increased risk for different types of cancer. Paradoxically, arsenic trioxide can also be used to induce remission in patients with acute promyelocytic leukemia (APL) with a success rate of approximately 80%. A comprehensive study examining the mechanisms and potential signaling pathways contributing to the anti-tumor properties of arsenic trioxide has not been carried out.

Methods: Here we applied a systems biology approach to identify gene biomarkers that underlie tumor cell responses to arsenic-induced cytotoxicity. The baseline gene expression levels of 14,500 well characterized human genes were associated with the GI50 data of the NCI-60 tumor cell line panel from the developmental therapeutics program (DTP) database. Selected biomarkers were tested in vitro for the ability to influence tumor susceptibility to arsenic trioxide.

Results: A significant association was found between the baseline expression levels of 209 human genes and the sensitivity of the tumor cell line panel upon exposure to arsenic trioxide. These genes were overlayed onto protein-protein network maps to identify transcriptional networks that modulate tumor cell responses to arsenic trioxide. The analysis revealed a significant enrichment for the oxidative stress response pathway mediated by nuclear factor erythroid 2-related factor 2 (NRF2) with high expression in arsenic resistant tumor cell lines. The role of the NRF2 pathway in protecting cells against arsenic-induced cell killing was validated in tumor cells using shRNA-mediated knock-down.

Conclusions: In this study, we show that the expression level of genes in the NRF2 pathway serve as potential gene biomarkers of tumor cell responses to arsenic trioxide. Importantly, we demonstrate that tumor cells that are deficient for NRF2 display increased sensitivity to arsenic trioxide. The results of our study will be useful in understanding the mechanism of arsenic-induced cytotoxicity in cells, as well as the increased applicability of arsenic trioxide as a chemotherapeutic agent in cancer treatment.

Figure 1

A range of susceptibilities to arsenic trioxide across the NCI-60 tumor cell panel. The GI₅₀ data of 59 tumor cell lines screened for arsenic trioxide-induced cell death are displayed. A total of nine tumor types were screened, including: breast, central nervous system (CNS), colon, leukemia, melanoma, non-small cell lung (NSCL), ovarian, prostate, and renal tumors. For the complete list of tumor cell lines refer to Additional File 1.

Figure 2

Potential gene biomarkers of tumor cell susceptibility to arsenic trioxide. A total of 209 genes (242 gene probes) were identified with significant expression association with tumor cell susceptibility to arsenic trioxide across 58 tumor cell lines (FDR < 0.05). Cell line numbers are displayed on the X-axis. For the complete list of tumor cell lines refer to Additional File 1. Gene expression values were mean centered and high relative expression is indicated in red and low relative expression indicated in blue.

Figure 3

Molecular interactomes and sub-networks associated with tumor cell susceptibility to arsenic trioxide. (A) A large arsenic-susceptibility interactome containing 317 proteins was identified. (B-D) The three most significant cancer and cell death enriched sub-networks within the large interactome were identified. Networks are displayed with symbols representing encoded proteins corresponding to their RNA transcripts that were either highly expressed in arsenic resistant cell lines (green symbols), highly expressed in arsenic sensitive cell lines (red symbols), or associated to the modified transcripts (white symbols). P-values representing the statistical significance of networks are shown.

Figure 4

Baseline expression levels of NRF2 target genes and tumor cell responses to arsenic trioxide. (A) Baseline gene expression levels of eight NRF2 target genes in the NCI-60 tumor cell panel. For each of the nine tumor types, the average gene expression level was calculated for all the tumor cell lines within this group. The cumulative gene expression levels of the eight target genes were calculated to represent the general gene expression level of that tumor type (Leu = Leukemia; CNS = Central Nervous System; Mel = Melanoma; Ov = Ovarian; NSCL = Non-small Cell Lung; Pro = Prostate). (B) Arsenic-specific log(GI₅₀) values of the NCI-60 tumor cell panel. For each of the nine tumor types, the average log(GI50) was calculated for all the tumor cell lines within this group to represent the general susceptibility of this tumor type to arsenic-induced cytotoxicity.

Figure 5

Decreased NRF2 expression alters tumor cell response to arsenic trioxide. Cells (A549 lung carcinoma) deficient for the expression of NRF2 (NRF2-KD), a scramble shRNA control (scramble), a turbo-GFP control (GFP) were generated using shRNAs and tested for their sensitivity to arsenic trioxide. (A) Cells expressing the NRF2-shRNA have decreased mRNA expression of NRF2 relative to the controls; (B) Cells expressing the NRF2-shRNA have decreased mRNA expression of NQO1, a well known NRF2 target gene; (C) Cells expressing the NRF2-shRNA show no alterations in mRNA expression of β-ACTIN; (D) Cells with decreased expression levels of NRF2 show increased sensitivity to arsenic trioxide induced killing.