A targetable antioxidant defense mechanism to EZH2 inhibitors enhances tumor cell vulnerability to ferroptosis

Abstract

Epigenetic changes are present in all human cancers and are responsible for switching on or off genes, thus controlling tumor cell transcriptome. These changes occur through DNA methylation, histone modifiers and readers, chromatin remodelers, and microRNAs. The histone H3 methyl-transferase EZH2 gene is overexpressed in several cancer types, including adrenocortical carcinoma (ACC), a rare cancer still lacking a targeted therapy. EZH2 inhibitors (EZH2i) have been tested in several clinical trials, but their effectiveness was limited by the toxic effects of the therapeutic doses. We tested several EZH2i on ACC cells, and observed a significant reduction in cell growth only with doses much higher than those required to prevent H3 methylation. We found that all tested EZH2i doses affected lipid metabolism genes, ROS, and glutathione production. Transcript changes correlated with metabolic data, which suggested the effects of EZH2i on ferroptosis. We found that EZH2i dose-dependently increased SLC7A11/glutathione axis and glutathione peroxidase-4 (GPX4), required to counteract lipid peroxidation and ferroptosis. A GPX4 inhibitor synergized with EZH2i, making low doses - which otherwise do not affect cell viability - able to significantly reduce ACC cell growth in vitro and in vivo. Importantly, we found that the anti-ferroptosis defense mechanism induced by EZH2i is a common response for several aggressive tumor phenotypes, uncovering a general co-targetable mechanism that could limit EZH2i effectiveness. Correcting this antioxidant response by ferroptosis inducers may be a new combination therapy that will easily find clinical applications.

Fig. 1. EZH2i affect cell viability and lipid content in ACC cells.

ACC cells treated with DZNep, GSK126, GSK343, and Tazemetostat (TAZ). A–D, H–K Cell viability of H295R (A–D) and MUC-1 (H–K) was evaluated by MTT assay at 24, 48, and 72 h. (n = 3. Data were expressed as means ± SEM. *p < 0.05; **p < 0.01; ****p < 0.0001. E, F Total protein lysates from 48 h treated H295R cells were immunoblotted for H3K27me3. Actin was used as a loading control. G, L Confocal images of neutral lipids and free fatty acids (FFA) in 48 h treated H295R (G) and MUC-1 (L) cells. Nuclei were stained by DAPI (Scale bar 50 µm).

Fig. 2. EZH2i change the metabolic profile of ACC cells.

A, B Heatmap for mRNA expression of lipid metabolism genes in H295R cells after 48 h of GSK126 treatment. Data were the values from three separate RNA samples presented as log10 of fold change. LDs lipid droplets; FAO fatty acids oxidation, TF transcription factors. C–K QPCR graphs and western blots of representative genes from (A) (n = 3, means ± SEM *p < 0.05; **p < 0.01; ****p <0.0001). L Untargeted metabolomics analysis was performed on H295R cells treated with GSK126 (5 and 25 µM) for 48 h. Venn diagram represents statistically significant differentially abundant metabolites across 5 and 25 µM GSK126-treated cells. M Pie chart depicting the proportions of metabolite classes in the 108 common metabolites. N Schematic representation of linoleic acid oxidation pathways.

Fig. 3. EZH2i activate antioxidant genes inversely correlated with ACC patients’ survival.

H295R cells were treated for 48 h with GSK126. A Glutathione (GSH) content normalized to the number of cells. (n = 3, means ± SEM, ****p < 0.0001). B–H mRNA expression of genes related to ferroptosis. (n = 3, means ± SEM, **p < 0.01; ****p < 0.0001). I Unsaturated FA (MUFA and PUFA) quantification normalized to the number of cells. J Total protein lysates from 48 h treated H295R cells were immunoblotted for GPX4 and ACSL4. Actin was used as a loading control. K, L Overview of the pathway enrichment analysis based on both transcripts and metabolites with statistically significant differences in GSK126-treated H295R cells. Only pathways altered in both treatments and with –log10(p) >2 are shown. M Differential expression analysis of selected genes among normal adrenal (NC), adrenocortical adenomas (ACA), and adrenocortical carcinomas (ACC) from microarray cohort [61]. N Overall survival depicted by Kaplan–Meier plots for each of the selected genes in the ACC-TCGA dataset [63]. “Low” and “High” level expression groups are indicated by blue and red colors, respectively. Log-rank p values are shown at the bottom of each plot.

Fig. 4. Low doses of EZH2i synergize with ferroptosis inducers.

A, B ACC cells treated for 48 h with GSK126 (5 μM) alone or in combination with inhibitors for lipid metabolism enzymes: Orlistat (ORLI 100 μM), FASNi (G28UCM, 1 μM), ACLYi (SB204990, 50 μM), ATGLi (ATGLstatin, 50 μM), SCD1i (A959572, 5 nM), ACATi (Avasimibe, 2.5 μM), CD36i (Sulfosuccinimidyl oleate, 10 μM), and CPT1i (Etomoxir, 10 μM). C, D ACC cells treated for 48 h with GSK126 (5 μM) in the presence or absence of RSL3 (1 µM, for the last 4 h) and Ferrostatin (1 µM). (n = 3. Data were expressed as means ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. ns not significant). E, F Confocal images of lipid peroxidation detected by BODIPY-C11 fluorescent dye in 2D (E) and 3D H295R cultures (F) (scale bar 50 µm). G Electron micrographs. Nucleus (N), nucleolus (n), mitochondria (black arrows), endoplasmic reticulum (black arrowheads), Golgi apparatus (G), electron transparent vesicles and vacuoles (v), lipid-like vacuoles of low electron density (*), lipid droplets (LD) (scale bars: 5 µm, 2 µm and 500 nm).

Fig. 5. In vitro and in vivo synergistic effects of tazemetostat and ferroptosis inducers.

A, B mRNA expression of GPX4 and SLC7A11 in ACC cell lines treated for 48 h with tazemetostat (TAZ 5 µM). C, D Cell viability of tumor cell lines treated for 48 h with TAZ (5 μM) and RSL3 (1 µM, for the last 4 h) and their combinations. E Schematic representation of xenograft experiment setup. F, G Images of explanted xenograft tumors (F) and relative tumor weight graphs (G). H, I mRNA expression of GPX4 and SLC7A11 in xenografts tumor samples. Data were expressed as means ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001 ****p < 0.0001.

Fig. 6. Tazemetostat combined with ferroptosis inducers is an effective treatment for different tumor types.

A–C mRNA expression of GPX4 and SLC7A11 in cell lines from different tumor types treated for 48 h with tazemetostat (TAZ 5 µM). D–F Cell viability of tumor cell lines treated for 48 h with TAZ (5 μM) and RSL3 (1 µM, for the last 4 h) and their combinations. G–I Representative images of explanted MDA-MB231 xenografts (G) and relative graphs of tumor volume (H) and weight (I). J, K mRNA expression of GPX4 and SLC7A11 in xenografts tumor samples. L Protein expression and relative quantification of GPX4 in tumor samples. Data were expressed as means ± SEM. *p < 0.05; **p < 0.01, ***p < 0.001;  ****p < 0.0001.