CYP2E1 overexpression protects COS-7 cancer cells against ferroptosis

Ferroptosis is a recently described form of regulated cell death initiated by the iron-mediated one-electron reduction of lipid hydroperoxides (LOOH). Cytochrome P450 2E1 (CYP2E1) induction, a consequence of genetic polymorphisms or/and gene induction by xenobiotics, may promote ferroptosis by contributing to the cellular pool of LOOH. However, CYP2E1 induction also increases the transcription of anti-ferroptotic genes that regulate the activity of glutathione peroxidase 4 (GPX4), the main ferroptosis inhibitor. Based on the above, we hypothesize that the impact of CYP2E1 induction on ferroptosis depends on the balance between pro- and anti-ferroptotic pathways triggered by CYP2E1. To test our hypothesis, ferroptosis was induced with class 2 inducers (RSL-3 or ML-162) in mammalian COS-7 cancer cells that don’t express CYP2E1 (Mock cells), and in cells engineered to express human CYP2E1 (WT cells), and the impact on viability, lipid peroxidation and GPX4 was assessed. CYP2E1 overexpression protected COS-7 cancer cells against ferroptosis, evidenced by an increase in the IC50 and a decrease in lipid ROS in WT versus Mock cells after exposure to class 2 inducers. CYP2E1 overexpression produced an 80% increase in the levels of the GPX4 substrate glutathione (GSH). Increasing GSH in Mock cells protected cells against ferroptosis by ML-162. Depleting GSH, or inhibiting Nrf2 in WT cells reverted the protective effect mediated by CYP2E1, causing a decrease in the IC50 and an increase in lipid ROS after exposure to ML-162. These results show that CYP2E1 overexpression protects COS-7 cancer cells against ferroptosis, an effect probably mediated by Nrf2-dependent GSH induction.

Cells have evolved at least three defense mechanisms to suppress ferroptosis: a) glutathione peroxidase-4 (GPX4), which uses glutathione (GSH) to reduce and detoxify LOOH (Fig. 1); b) ferroptosis suppressor protein 1 (FSP1), which reduces ubiquinone to ubiquinol on the plasma membrane, which further traps and inactivates lipid peroxyl and alkoxyl radicals, and c) dihydroorotate dehydrogenase (DHODH), which carries out a similar reaction to that of FSP1 but on the inner mitochondrial membrane (Mao et al. 2021).
Increased activity of gene products that decrease the overall pool of LOOH, such as GPX4, γglutamylcysteine synthetase (γGCS, the rate-limiting enzyme for the synthesis of the GPX4 substrate GSH), and SLC7A11 (also known as xCT, a transporter that promotes cystine uptake and GSH biosynthesis) could decrease the rate of initiation of lipid peroxidation and could suppress ferroptosis (Sha et al. 2018; Koppula et al. 2018).
Several lines of evidence suggest that PORs are major contributors to ferroptosis initiation. For example, in several cancer cell lines susceptible to ferroptosis and lacking signi cant expression of LOXs, genetic POR depletion prevented ferroptosis triggered by class 1 (xCT inhibitors) and class 2 (GPX4 inhibitors) inducers, probably by preventing the cycling between Fe(II) and Fe(III) in the heme component of P450s Viability: Cell viability was determined by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay (Carmichael et al. 1987). Brie y, after the treatment period, the medium was aspirated, and replaced with fresh medium without FBS containing MTT at 1 mg/mL. Plates were incubated at 37°C in a humidi ed 5% CO 2 atmosphere for 1h. At the end of the incubation period, the medium was aspirated, and the formazan product was dissolved in 0.3 mL of n-propanol. Cell viability was evaluated by measuring the absorbance at the test wavelength of 570 nm, correcting it by the absorbance at the reference wavelength of 630nm, and converting this value to percentage of the control (de ned as cells without treatment). Cytotoxic effects were expressed as IC 50 (the concentration of drug that reduces viability by 50% of the non-treated control). The IC 50 was determined using the fourparameter logistic function y = D + (A-D)/1 + 10(x-logC) B, with parameter C representing the estimation of IC 50 , using the Sigma Plot 12.0 software program from Systat Software (Richmond, CA).
Lipid peroxidation: Lipid ROS were determined by the level of oxidation of the polyunsaturated butadienyl portion of C11BODIPY (Martinez et al. 2020). After the treatments, the medium was removed and replaced with 0% FBS-DMEM containing 2.5 µM C11BODIPY. After a 20-minute incubation at 37°C and 5% CO 2 , cells were washed with PBS, trypsinized, resuspended in 0% FBS-DMEM, and analyzed by ow cytometry using an Accuri C6 Flow Cytometer (BD Biosciences, San Jose, CA, USA) equipped with a 488 nm, 50 mW solid-state laser. C11BODIPY oxidation results in a shift of the uorescence emission peak from ~ 590nm to ~ 510nm. C11BODIPY oxidation was reported as the percentage of cells that exhibit high uorescence emission intensity in the FL1 uorescence channel (530 ± 30 nm).
Glutathione levels: After the treatments, cells were trypsinized, resuspended in PBS, and counted with a hemacytometer. Up to 8000 cells were dispensed at 50µl per well into 96-well chemiluminescence plates. Glutathione levels were measured using the GSH-Glo glutathione luminescent assay as per manufacturer's instructions (Promega, Madison, WI). Chemiluminescence was measured using a Synergy 2 multi-function microplate reader from BioTek (Winooski, VT), and normalized to the vehicle control. GSH detection by chemiluminescence was linear with cell density up to 8,000 cells/well in both Mock and WT cells (data not shown).
CYP2E1 activity: CYP2E1 activity in intact cells (in situ) or in CYP2E1 baculosomes (microsomes prepared from insect cells infected with recombinant baculovirus containing human CYP2E1 and POR, in vitro) was determined as the O-demethylation of 7 -methoxy-4-tri uoromethylcoumarin (7-MFC) to 7hydroxy-4-tri uoromethylcoumarin (7-HFC). For the in situ assay, 1 x 10 6 Mock or WT cells were plated in 1 mL phosphate buffered saline (PBS) supplemented with 5.5 mM glucose, 1 mM CaCl 2 and 1.8 mM Statistics: Data are expressed as means ± S.E. of the mean from 3-5 independent experiments. One-way analysis of variance with subsequent post hoc comparisons by Student-Newman-Keuls was performed. p < 0.05 was considered as statistically signi cant.

1) CYP2E1 overexpression in COS-7 cells limits cell death triggered by GPX4 inhibition.
Cell death was experimentally induced by irreversible inhibition of GPX4 with Ras-selective lethal small molecule 3 (RSL-3) via its reactive chloroacetamide moiety. The IC 50 of RSL-3 was higher in WT cells (COS-7 cells overexpressing human wild type CYP2E1) (0.082 µM) than in Mock cells (COS-7 cells not expressing any cytochrome P450) (0.039 µM) (Fig. 2a). ML-162 (a structurally different chloroacetamide GPX4 inhibitor) was also used in parallel experiments. ML-162 was used to verify the GPX4-dependence of RSL-3-mediated effects (Stockwell and Jiang 2020): the probability that similar results observed with RSL-3 and ML-162 are not caused by chloroacetamide-mediated GPX4 inhibition, but by an off-target effect, is minimal due to their signi cant structural dissimilarities. The IC 50 of ML-162 was higher in WT cells expressing human wild type CYP2E1 (0.089 µM) than in Mock cells not expressing any cytochrome P450 (0.035 µM) (Fig. 2b). In either cell type, the IC50 of RSL-3 was not signi cantly different to that of ML-162. Importantly, stable transfection of the same cell line with mutated CYP2E1 targeted to mitochondria (W23/30R cells) also increased the IC 50 of RSL-3 with respect to Mock cells (from 0.040 to 0.15 µM) (Fig. 2c), validating that CYP2E1 irrespective of its localization, and not an artifact of clonal selection, is responsible for the increased IC 50 of GPX4 inhibitors in WT cells.
2) GPX4 inhibition induces ferroptosis in both CYP2E1expressing and non-expressing COS-7 cells.  4a). O-demethylation of 7-MFC in vitro by human CYP2E1 baculosomes was signi cantly inhibited by diethyldithiocarbamate (a speci c CYP2E1 inhibitor), but not by the ferroptosis inhibitors mitoquinone, deferiprone, or ferrostatin, at the concentrations used in the cellular assays (Fig. 4b). PD146176, however, at the concentration used in the cellular assays, signi cantly inhibited CYP2E1 activity in vitro (Fig. 4b), suggesting that the inhibition by PD146176 of ML-162-induced ferroptosis in WT cells could be partially mediated by CYP2E1 inhibition and not only by 15-LOX inhibition. The ferroptosis inhibitors tested did not affect the uorescence of 7-HFC at the concentrations used in the cellular assays (data not shown).
3) CYP2E1 overexpression in COS-7 cells prevents the early increase in lipid ROS caused by GPX4 inhibition.
The intracellular oxidation of C11BODIPY is mainly caused by the oxygen-centered radicals (i.e. lipid peroxyl, alkoxyl) produced during lipid peroxidation (Michel et al. 2020). C11BODIPY oxidation is therefore an indirect manifestation of the lipid ROS ux, associated with the initiation and propagation of the lipid peroxidation that characterizes ferroptosis (Wiernicki et al. 2020). Therefore, an increase in C11BODIPY oxidation should be an early event (i.e. occurring under conditions prior to overt cell death) in the ferroptotic process. To evaluate lipid ROS ux, Mock and WT cells were treated with 0.01 µM ML-162 for 24h (prior to overt cell death), followed by the analysis of C11BODIPY oxidation. In Mock cells, but not in WT cells, ML-162 produced a signi cant 6-fold increase in the percentage of cells exhibiting high FL1 uorescence (M1%) with respect to untreated cells ( Fig. 5a and b). This increase was completely prevented by ferrostatin, con rming the speci city of the assay (Fig. 5b). No signi cant differences were observed in C11BODIPY oxidation between untreated Mock and WT cells (Fig. 5b).

4) Increased glutathione in CYP2E1-overexpressing COS-7 cells protects against ferroptosis.
Because GPX4 is the master regulator of ferroptosis due to its unique ability to directly reduce phospholipid hydroperoxides in membranes and prevent lipid peroxidation (Chen et al. 2021), we evaluated the possibility that CYP2E1 prevents the lipid peroxidation and ferroptosis induced by ML-162 by increasing GPX4 activity (phospholipid hydroperoxide reduction) or GPX4 substrate (GSH) levels. There were no signi cant differences in the activity of GPX4 between Mock and WT cells (Fig. 6a). A signi cant 1.8-fold increase in GSH levels was detected in WT cells with respect to Mock cells (Fig. 6b). An inhibitor of γGCS, buthionine sulfoximine (BSO), decreased GSH levels by more than 95% in WT cells, con rming the speci city of the assay (Fig. 6b). If the increased GSH levels in CYP2E1-overexpressing cells protects against ferroptosis, then: i) increasing GSH levels in Mock cells should prevent ferroptosis, and ii) decreasing GSH levels in WT cells promote ferroptosis. To test these possibilities, we increased GSH levels in Mock cells with the use of a cell-permeable form of GSH, glutathione ethyl ester (GSHee), and found that the IC 50 of ML-162 increased from 0.020 µM in the absence of GSHee to 0.052 µM in the presence of 0.2 mM GSHee (Fig. 7a). In addition, we decreased GSH levels in WT cells with the use of BSO, and found that the IC 50 of ML-162 decreased from 0.104 µM in the absence of BSO to 0.020 µM in the presence of 0.1 mM BSO (Fig. 7b). Because an increase in lipid ROS is an early event during ferroptosis (Wiernicki et al. 2020) we evaluated the effect of GSH depletion on lipid ROS levels in WT cells. C11BODIPY oxidation (expressed as percentage of cells exhibiting high C11BODIPY FL1 uorescence) in WT cells exposed to a non-lethal concentration of ML-162 was 6.7 ± 0.5% in the absence of BSO, a value that was not signi cantly different from that of untreated Mock or WT cells, or WT cells exposed to 0.1 mM BSO (Fig. 7c). C11BODIPY oxidation increased to 36.0 ± 4.0% in WT cells exposed to a non-lethal concentration of ML-162 in the presence of 0.1 mM BSO, a signi cant 5.4-fold increase (Fig.  7c).

5) CYP2E1-induced protection from ferroptosis in COS-7 cells is prevented by Nrf2 inhibition.
Nrf2 plays a key role in the adaptive response against increased oxidative stress caused by CYP2E1 in cultured cells, through up-regulation of γGCS and increase in GSH levels (Gong and Cederbaum 2006). Therefore, we tested the effect of a speci c chemical Nrf2 inhibitor (ML-385) on ferroptosis and GSH levels in WT cells exposed to ML-162. ML-385 by itself did not show signi cant toxicity in WT cells at the concentration used (Fig. 8a). The IC 50 of ML-162 in WT cells decreased from 0.088 µM in the absence of ML-385 to 0.035 µM in the presence of 5 µM ML-385 (Fig. 8a). The percentage of cells with high levels of lipid ROS (a pro-ferroptotic condition) increased from 9% in WT cells exposed to a non-lethal concentration of ML-162 to 38% in WT cells exposed to a non-lethal concentration of ML-162 in the presence of 5 uM ML-385, a signi cant 4-fold increase (Fig. 8b). To test if Nrf2 activity is required for the prevention of ferroptosis in CYP2E1-exprressing cells through the maintenance of GSH levels, we determined GSH levels in cells treated with ML-162 in the presence or absence of the Nrf2 inhibitor ML-385. Nrf2 inhibition decreased GSH levels in WT-cells exposed to ML-162 prior to overt cell death, suggesting that Nrf2 activity is required to maintain GSH levels in CYP2E1-overexpressing cells under ferroptosis conditions (Fig. 8c).

Discussion
We conclude that CYP2E1 overexpression protects COS-7 cancer cells against ferroptosis based on the following observations: a) the IC 50 of class 2 inducers (RSL-3 and ML-162) was higher in CYP2E1overexpressing (WT and W23/30R) than in non-CYP2E1 expressing (Mock) COS-7 cells, and b) ML-162 at 0.01 µM (prior to overt cell death in Mock cells) increased lipid peroxidation (central to the execution of ferroptosis) in Mock but not in WT cells. The mode of cell death by ML-162 was con rmed to be ferroptosis by its pharmacological ngerprint, which includes sensitivity to LOX, lipid peroxidation, iron and mitochondrial oxidant inhibitors. The fact that a speci c inhibitor of 15-LOX such as PD146176 prevented ferroptosis by ML-162 in Mock cells suggests that 15-LOX activity mediates LOOH accumulation and ferroptosis in non-CYP2E1 expressing (Mock) Cos-7 cells. PD146176 inhibited ferroptotic cell death in other cell lines exposed to class 2 inducers, also suggesting the involvement of show that PD146176 is also a CYP2E1 inhibitor in vitro, so through the use of PD146176 alone it is not possible to discard the contribution of CYP2E1 to ferroptosis in CYP2E1-overexpressing (WT) Cos-7 cells. PD146176 was recently identi ed as a cytochrome P450 epoxygenase inhibitor in human EA.hy926 endothelial cells (Du et al. 2022), so results with PD146176 should be interpreted with caution in cytochrome P450-expressing systems.
Our results suggest that the anti-ferroptotic response in CYP2E1-overexpressing cells is mediated by increased GSH levels. The experimental evidence to support this conclusion is: a) GSH levels were 80% higher in WT than in Mock cells; b) decreasing GSH levels in WT cells by inhibiting γGCS with BSO decreased the IC 50 of ML-162, blunting the protection induced by CYP2E1; c) exposing Mock cells to a cell-permeable form of glutathione (GSH-ee) increased the IC 50 of ML-162, indicating decreased toxicity under high GSH levels; d) decreasing GSH levels in WT cells by inhibiting γGCS with BSO increased lipid ROS (a trigger for ferroptosis) after exposing cells to non-lethal doses of ML-162, blunting the protection induced by CYP2E1. GSH levels directly modulates ferroptosis sensitivity in this and other cellular models. For example, GSH depletion promoted ferroptosis in human lens epithelial cells exposed to RSL3 CYP2E1 upregulates Nrf2 signaling in rat hepatocytes and HepG2 cells in an ROS-dependent manner, which induces the transcriptional activation of the rate limiting-enzyme in GSH synthesis (γGCS), and increases GSH levels (Gong and Cederbaum 2006). Our results suggest that this mechanism could mediate ferroptosis resistance in CYP2E1-overexpressing COS-7 cells: a) ML-385 (a chemical Nrf2 inhibitor) decreased the IC 50 of ML-162 in WT cells to levels comparable to that in Mock cells, blunting the protection induced by CYP2E1; b) lipid ROS (a trigger for ferroptosis) increased after exposing WT cells to non-lethal doses of ML-162 under Nrf2 inhibition; and c) ML-385 decreased GSH levels prior to cell death in WT cells exposed to ML-162. However, other Nrf2-dependent pathways potentially triggered by CYP2E1 could also in part mediate this increased resistance. For example, Nrf2-dependent microsomal glutathione S-transferase (MGST) expression was induced in CYP2E1-overexpressing HepG2 cells (Mari and Cederbaum 2001), and contributed to ferroptosis resistance in pancreatic cancer cells by inhibiting ALOX5 activity (Kuang et al. 2021). Also, Nrf2-dependent heme oxygenase 1 (HO-1) expression was induced in CYP2E1-overexpressing HepG2 cells (Gong and Cederbaum 2006), and contributed to ferroptosis resistance in hepatocellular carcinoma cells by modifying iron metabolism and lipid peroxidation (Sun et al. 2016). Lastly, xCT expression was induced in CYP2E1-overexpressing primary mouse hepatocytes (Choi et al. 2019), and contributed to ferroptosis resistance in Nrf2-knockdown F98 rat glioma cells (Fan et al. 2017).
Ferroptosis is a natural tumor-suppressor mechanism, and has been proven to be effective in anticancer therapy (Cai et (Bergheim et al. 2007), and in adjacent nontumor tissues from hepatocellular carcinoma patients (Ho et al. 2004), than in their normal counterparts. Based on the above, CYP2E1 induction could further promote carcinogenesis via inhibiting natural ferroptosis or enhancing resistance to ferroptotic treatments.
In conclusion, although POR-P450 activity has been associated with increased pro-ferroptotic generation of lipid hydroperoxides in other systems, competing anti-ferroptotic pathways shifted the pro-anti ferroptotic balance towards anti-ferroptosis in CYP2E1-overexpressing Cos-7 cells. Nrf2-dependent GSH induction is one factor that could mediate the net anti-ferroptotic effect of CYP2E1 overexpression.

Declarations
Funding This work was supported by the Arkansas INBRE, a grant from the National Institute of General Medical Sciences (NIGMS), P20 GM103429 from the National Institutes of Health.

Competing Interests
The authors have no relevant nancial or non-nancial interests to disclose.

Author Contributions
AAC contributed to the study conception and design. Material preparation, data collection and analysis were performed by DB, CG, WN, JP, RY, HS and AAC. The rst draft of the manuscript was written by AAC and all authors commented on previous versions of the manuscript. All authors read and approved the nal manuscript.

Data Availability
Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.
Ethics approval and consent to participate Not applicable.

Consent for publication
The manuscript has been seen and approved by all authors for publication.