The ferroptosis inducer erastin irreversibly inhibits system xc− and synergizes with cisplatin to increase cisplatin’s cytotoxicity in cancer cells

System xc− was recently described as the most upstream node in a novel form of regulated necrotic cell death, called ferroptosis. In this context, the small molecule erastin was reported to target and inhibit system xc−, leading to cysteine starvation, glutathione depletion and consequently ferroptotic cell death. Although the inhibitory effect of erastin towards system xc− is well-documented, nothing is known about its mechanism of action. Therefore, we sought to interrogate in more detail the underlying mechanism of erastin’s pro-ferroptotic effects. When comparing with some well-known inhibitors of system xc−, erastin was the most efficient inhibitor acting at low micromolar concentrations. Notably, only a very short exposure of cells with low erastin concentrations was sufficient to cause a strong and persistent inhibition of system xc−, causing glutathione depletion. These inhibitory effects towards system xc− did not involve cysteine modifications of the transporter. More importantly, short exposure of tumor cells with erastin strongly potentiated the cytotoxic effects of cisplatin to efficiently eradicate tumor cells. Hence, our data suggests that only a very short pre-treatment of erastin suffices to synergize with cisplatin to efficiently induce cancer cell death, findings that might guide us in the design of novel cancer treatment paradigms.

neurological diseases, such as Parkinson's disease, Alzheimer's disease, and amyotrophic lateral sclerosis 14 . In addition, system x c − has recently emerged as a potential target in the context of cancer therapy 15 . In fact, many reports have demonstrated that inhibition or down-regulation of system x c − function attenuates proliferation, invasion, and metastasis of cancer cells in vitro and in vivo 16 . Therefore, exploitation of specific and potent inhibitors of system x c − is considered to be of potentially great benefit for cancer chemotherapy. In this regard, many compounds have been found as inhibitors of system x c −17,18 . Among these, erastin (named for eradicator of RAS and ST-expressing cells) was first identified by synthetic lethal high-throughput screening by Stockwell's group as a small molecule compound efficiently killing human tumor cells without affecting their isogenic normal cell counterparts 19 . Then, the same group discovered that erastin is a potent and selective inhibitor of system x c − causing a novel iron-dependent form of non-apoptotic cell death, designated as ferroptosis 20,21 .
Yet, the mode of the inhibition of system x c − by erastin has remained unclear. In the present study, we have investigated the inhibitory characteristics of erastin on the activity of system x c − and intracellular glutathione levels, and found that erastin has a persistent inhibitory effect, which appears to be entirely different from other system x c − inhibitors.

Results
Specificity of the inhibitory effects of erastin on system x c − activity. To confirm that erastin specifically inhibits the activity of system x c − , we measured the activity of the uptake of arginine, leucine and serine in addition to cystine in the presence or absence of 10 μM erastin in xCT-overexpressing MEF (Fig. 1). No inhibition was detectable for arginine uptake (system y + ), leucine uptake (system L), and serine uptake (system ASC), whereas cystine uptake was strongly impaired by erastin in xCT-overexpressing MEFs. These data unequivocally show that erastin selectively inhibits system x c − and that it has no impact on other amino acid transport systems.

Comparison of inhibitory efficiency of xCT inhibitors.
There are a number of small molecules inhibitors known that inhibit the uptake of cystine or glutamate into cells. We compared the inhibitory efficiency of these inhibitors in xCT-overexpressing MEF by determining the uptake of 0.05 mM cystine into cells. As shown in Fig. 2, sulfasalazine (SAS), (S)-4-carboxyphenylglycine (CPG), and quisqualic acid (QA) inhibited the cystine uptake as efficiently as erastin at the concentration of 100 μM. Yet, at lower concentrations erastin was by far more potent in inhibiting the uptake of cystine as compared to the other inhibitors tested. The calculated IC 50 values of erastin, CPG, QA and SAS were 1.4 µM, 4.4 µM, 15.4 µM and 26.1 µM, respectively. L-glutamate is one of the physiological substrates of system x c − and is known to competitively inhibit the uptake of cystine. At 100 µM glutamate only inhibited the cystine uptake by approximately 50%, indicating that the affinity of erastin towards xCT is much higher compared to physiological substrates.

Comparison of persistent effects of xCT inhibitors.
At 100 µM most of the xCT inhibitors showed a similar inhibitory effect on the uptake of cystine. Therefore, we have further investigated the characteristics of the mode of inhibition of these reagents towards xCT. xCT-overexpressing MEF were exposed to 100 μM of each inhibitor for 5 min. After 5 min incubation, the cell culture medium was removed, cells were rinsed once with 1 ml of fresh medium and incubated with 2 ml of fresh medium lacking any inhibitors for 24 h. After 24 h, the cystine uptake was measured in the absence of the inhibitors. At the same time, intracellular glutathione levels were determined. As shown in Fig. 3A, the inhibitory effects of CPG, QA, SAS, and glutamate towards xCT were completely abolished 24 h after exposing cells for 5 min to these compounds, while in stark contrast the uptake of cystine was still inhibited dramatically in the cells which had been exposed for just 5 min with erastin. Accordingly, only the cells treated with erastin presented a very low level of intracellular glutathione which persisted even 24 h after washing out the inhibitor (Fig. 3B). Next, we performed time course experiments to further explore the mechanism of inhibition and its impact on cellular glutathione levels. As illustrated in Fig. 4A, a 5 min exposition of cells to erastin sufficed to obtain a strong inhibitory effect towards xCT which persisted over the entire observation time. This strong and persistent inhibitory effect was followed by a massive drop of intracellular glutathione concentrations which reached its lowest levels as early as 6 h upon erastin treatment (Fig. 4B). It was, however, not possible to explore if the inhibitory effect of erastin would have lasted for longer time periods as the cells started to die after 24 h under the routine cell culture conditions due to strongly decreased intracellular glutathione concentrations. Sensitivity to erastin by sulfhydryl modification of each cysteine residue of xCT. The results obtained above suggested that erastin irreversibly binds to xCT protein. If this is the case, it might be that erastin covalently reacts with sulfhydryl group of cysteine residues. To this end, xCT mutant forms were engineered by replacing the individual six endogenous cysteine residues with serine as described in Materials and Methods. All six mutant forms of xCT were expressed in xCT-deficient MEF and no differences of proliferation rates were detectable between the different cell lines (not shown). As shown in Fig. 5, xCT-deficient MEF transfected with each cysteine-mutant showed somewhat different activity of cystine uptake in the absence of erastin. However, the cystine uptake activity in all of these cell lines could be equally and potently inhibited by erastin, indicating that erastin does not inhibit cystine uptake by reacting with a single sulfhydryl group of any of these cysteine residues.  . Persistent effects of various xCT inhibitors on the activity of cystine uptake and intracellular total glutathione in xCT-overexpressing MEF exposed for a short time. xCT-overexpressing MEF were cultured for 24 h and then exposed to the inhibitors at the concentration of 100 µM each for 5 min. After exposure, cells were washed with 1 mL of fresh medium once. 2 mL of fresh medium was added and cells were allowed to grow for another 24 h. Then, the activity of cystine uptake (A) and intracellular glutathione levels (B) were measured. Bars represent the mean of percentages ± S.D. (n = 8 for control; n = 4 for the inhibitors). P values were obtained by one-way ANOVA followed by Tukey's multiple comparison test. *** P = 1 × 10 −6 .

Effect of erastin on proliferation of cisplatin-resistant human ovarian cancer cells.
We previously investigated the activity of cystine uptake and intracellular glutathione in human ovarian cell line (A2780) and its cisplatin (CDDP)-resistant counterpart (A2780DDP), and found that the activity of cystine uptake and intracellular glutathione in A2780DDP cells were significantly higher than those in A2780 cells 22 . Using both cell lines, we investigated the pre-treatment effect of erastin. After 5 min incubation with 10 or 100 μM erastin, the cell culture medium was removed, cells were rinsed once with 1 ml of fresh medium and incubated with 2 ml of fresh medium lacking any inhibitors for 24 h. After 24 h, the cystine uptake was measured in the absence of erastin. At the same time, intracellular glutathione levels were determined. As shown in Fig. 6A, the uptake of cystine was still inhibited in a dose-dependent manner in both types of cells which had been exposed for just 5 min with erastin. Accordingly, the cells treated with erastin presented a low level of intracellular glutathione which persisted even 24 h after washing out the inhibitor (Fig. 6B). Then, we performed a side by side comparison study and explored the sensitivity of cisplatin in these cells which had been exposed to erastin for a short time (Fig. 7). When A2780 cells were cultured for 48 h with 10 μM cisplatin alone, the cell number was decreased in xCT-overexpressing MEF exposed to erastin for a short time. xCT-overexpressing MEF were cultured for 24 h and subsequently exposed to 100 µM erastin for 5 min. After exposure, cells were washed with 1 mL fresh medium once. Two mL of fresh medium was added (regarded as 0 h) and cells were allowed to grow for 24 h.  to approximately 36% of control cells (no additives), whereas A2780DDP cells showed significant resistance to 10 μM cisplatin, and the cell number was decreased to only approximately 73% of control cells. When the cells were exposed to 10 μM erastin for 5 min and the cells were allowed to grow for another 48 h after the short erastin treatment, cell numbers of A2780 and A2780DDP cells were approximately 82% and 74% of cell number of untreated control cells, respectively. On the other hand, when the cells were exposed to 10 μM erastin for 5 min, and allowed to grow for another 48 h in the presence of 10 μM cisplatin alone after removing erastin by washing out from the medium, cell numbers of A2780 and A2780DDP cells were reduced to approximately 2% and 42% of the cell number of control cells, respectively. These results indicate that most of A2780 cells died and the survival of A2780DDP cells was significantly suppressed under these conditions. When the cells were exposed to 100 μM  erastin for 5 min and allowed to grow for another 48 h in the presence of 10 μM cisplatin alone after removing erastin, the sensitivity of the cells to cisplatin was drastically increased. Under these conditions, cell numbers of A2780 and A2780DDP cells were only 0.02% and 8% of the cell number of control cells. These results show that A2780 cells almost completely died and only a minority of A2780DDP cells survived this combination treatment. Taken together, the present results indicate that the cell death inducing effects of cisplatin can be strongly potentiated by a short bolus treatment of cells with erastin.

Discussion
Erastin was first reported as a small molecule compound efficiently killing human tumor cells without killing their isogenic normal cell counterparts 19 . Dixon and colleagues further found that erastin causes a novel iron-dependent form of regulated necrotic cell death, designated as ferroptosis 20 . They also showed that erastin is a potent and selective inhibitor of system x c −21 . In addition to their data that erastin has no impact on system L-mediated phenylalanine uptake, our present study now shows that erastin does not affect system ASC-mediated serine uptake and system y + -mediated arginine uptake (Fig. 1). We further demonstrate that a short time exposure of erastin is sufficient to confer the inhibitory effect of erastin towards cystine uptake via system x c − inhibition, and consequently deprivation of intracellular glutathione (Fig. 3). Cystine uptake via xCT is competitively blocked by substrates of system x c − , such as L-glutamate, α-aminoadipate, L-homocysteate, and L-cystathionine 8 . In addition to these substrates, several non-substrate inhibitors, such as (S)-4-carboxyphenylglycine were described by Patel and colleagues 23 . They also analyzed analogues of amino methylisoxazole propionic acid as the inhibitor of system x c −24 . On the other hand, Gout, et al., found that the anti-inflammatory drug sulfasalazine is a potent and specific inhibitor of this transport system 25 . Compared to these known inhibitors, erastin inhibits cystine uptake much more efficiently with an IC 50 value of around 1.4 µM (Fig. 2). Moreover, we provide strong evidence that the inhibitory effect of erastin on cystine uptake via system x c − lasts at least for 24 h after removing erastin from the cell culture medium. Probably due to the continuous inhibition against cystine uptake, intracellular glutathione concentrations were found to be sharply decreased and remained at very low levels for 24 h under these conditions (Fig. 4). No other known xCT inhibitor has ever shown such characteristics. These results strongly suggest that erastin irreversibly binds to and thereby inactivates xCT.
Jiménez-Vidal, et al., performed cysteine scanning accessibility studies to unravel the mechanism of the inhibitory effects of mercurial reagents, such as pCMB and pCMBS, which also potently inhibit system x c − activity. Their results indicated that these mercurial reagents react with Cys327 of human xCT, because an xCT mutant with a cysteine to serine substitution at Cys327 had a similar activity of glutamate uptake as wild-type xCT in the presence of pCMB or pCMBS 26 . Therefore, we hypothesized that if erastin reacts with a certain part of the xCT protein in a manner like the mercurial reagents, it might be that the chlorobenzene part of erastin may be replaced with sulfhydryl groups of certain cysteine residues of xCT. Yet, as shown in Fig. 5, a strong inhibitory effect of erastin was observed in the xCT-deficient cells regardless which mutant form of xCT they expressed. These findings suggest that erastin does not react with any of the sulfhydryl group of cysteine residue in the xCT protein, although the possibility that erastin reacts with more than one cysteine residue to inhibit cystine uptake cannot be formally ruled out. Yagoda, et al., previously showed that erastin acts by binding directly to mitochondrial voltage-dependent anion channels (VDACs) 27 . Thus, this compound seems to have the property of binding membrane proteins somehow, although the molecular mechanism of how erastin interacts with xCT needs further investigations. Interestingly, when the C86S construct is expressed in the cells, the activity of cystine uptake was significantly decreased, as compared with the wild-type construct. Cys86 is located in the middle part of transmembrane domain 2 by topology predictions 26 , and a similar tendency was observed in the case of human xCT 26 . The sulfhydryl group of this cysteine residue may thus be involved in the permeation pathway of the substrates.
Recent studies have suggested that xCT is involved in cancer pathology 13 . Gout, et al., showed that inhibiting cystine uptake via system x c − arrests proliferation of lymphoma cell lines in vitro 28 . In many human cancer cell lines, high expression of xCT is observed 29 . Chung, et al., demonstrated that sulfasalazine inhibits the growth of transplanted glioma cells in mice 30 , and that glutamate released via system x c − acts as an autocrine and/or paracrine signal for promoting glioma cell invasion 31 . Savaskan, et al., also reported that xCT confers a crucial role in glioma-induced neurodegeneration and brain edema 32 . On the other hand, xCT is stabilized by interacting with the splice variant of CD44, one of surface maker molecules associated with cancer stem cells, causing enhanced glutathione level and an increased defense against reactive oxygen species in human gastrointestinal cancer cells 33 . Recently, the major tumor suppressor gene, p53, has been reported to inhibit cystine uptake and sensitize cells to ferroptosis by repressing expression of xCT in human cancer cell lines 34 . Therefore, potent and highly specific inhibitors of xCT have been regarded as useful drugs for cancer chemotherapy. From this point of view, a variety of compounds have been investigated, including sulfasalazine analogs 35 . However, the clinical trial of sulfasalazine for patients with gliomas was unsuccessful due to the lack of clinical response and severe side effects 36,37 . Recently, Roh, et al., have demonstrated that inhibition of xCT function enhances cisplatin cytotoxicity of cisplatin-resistant head and neck cancer cells 38 . Their results are consistent with those of our present study. Here, we demonstrate that a brief erastin exposure of tumor cells is sufficient to dramatically enhance cisplatin's cytotoxicity in cisplatin-resistant ovary cancer cells. It is likely that a compound like erastin, which persistently blocks cystine uptake, might show beneficial and synergistic effects in cancer treatment regimens by applying it just before the onset of other anticancer drugs. Dixon, et al., have performed a structure activity relationship analysis of erastin and described some improved analogs with increased inhibitory effects towards system x c −21 . In case these compounds confer the same persistent inhibitory effect on xCT similarly like erastin, they may be expected to be used at lower concentrations before treating patients with anticancer drugs to increase chemotherapy efficacy. Cell culture. xCT-overexpressing mouse embryonic fibroblasts (MEF) were generated by stable transfection of cells with pCAG-3SIP-based vector containing the coding region of murine xCT 11,39 in a manner similarly as described by Lewerenz,et al. 40 . pCAG-3SIP-based vector contain a strong hybrid promoter, consisting of the chicken β-actin promoter and the CMV enhancer, which is highly active in many cell types including fibroblasts 41 . xCT-overexpressing MEF and xCT-deficient MEF 42 were cultured routinely in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, penicillin (50 U/mL) and streptomycin (50 µg/mL) at 37 °C in 5% CO 2 and 95% air. xCT-overexpressing MEF were seeded on 35 mm dishes (2 × 10 5 cells/dish), cultured for 24 h, and then measured for amino acid uptake activities in the presence or absence of xCT inhibitors at the concentrations indicated. When the cells were exposed to xCT inhibitors for a short time (5 min, 24 h after seeding xCT-overexpressing cells on the dishes), the medium was removed, and the cells were washed with fresh medium once, and then 2 ml of fresh medium was added. After culturing the cells at the time periods indicated, the activity of cystine uptake was measured in the absence of inhibitors. Under the same conditions, intracellular glutathione was measured.

Determination of intracellular total glutathione (GSH and the oxidized form of GSH (GSSG)).
Cells were rinsed three times in ice-cold phosphate-buffered saline (pH 7.4). Intracellular glutathione was extracted with 5% trichloroacetic acid and then treated with ether to remove the acid. Total glutathione content in the aqueous layer was measured using an enzymatic method described previously, which is based on the catalytic action of GSH in the reduction of 5,5′-dithiobis(2-nitrobenzoic acid) by the GSH reductase system 44 .
Site-directed mutagenesis. Mouse xCT cDNA cloned in pEF-BOS 45 was used as template. Six endogenous cysteine residues of mouse xCT were individually mutated to serine using PrimeSTAR ® Mutagenesis Basal Kit (TaKaRa Bio, Inc, Shiga, Japan) according to the manufacturer's instructions. pEF-BOS was kindly donated by Kazuichi Sakamoto (University of Tsukuba, Ibaraki, Japan). Primer sets used for the site-directed mutagenesis studies are shown in Table 1.
Transfection. After seeding 1.0 × 10 5 cells/35-mm dish and culturing for 24 h, xCT-deficient cells were transfected with 5 µg of each mutated xCT plasmid using Lipofectamine ® 3000 Transfection Reagent (Life Technologies, Carlsbad, USA) according to the manufacturer's instructions. After 48 h incubation, the activity of cystine uptake was measured as mentioned above. Determination of cell numbers. Cell viability was measured by trypan blue staining. After washing cells with PBS three times, cells were trypsinized and re-suspended in 1 ml of fresh medium. Then, an equal volume of 0.1% trypan blue solution was added to 50 µL of the cell suspension and viable cell numbers were counted with a hemocytometer.
Statistical analyses. Statistical significances of the differences were determined by Student's t test. For multiple comparison procedures, one-way ANOVA followed by Tukey's multiple comparison test was used.