Cancer is one of the most frequent pathologies worldwide; according to the World Health Organization (WHO) statistics, there were 18.1 million new cases and 9.6 million deaths related to this disease in 2018 (https://www.who.int/news-room/fact-sheets/detail/cancer). Cancers are difficult to treat because they employ multiple molecular mechanisms to evade different types of cell death, such as apoptosis, due to their overexpression of antiapoptotic proteins such as Bcl-2 and Bcl-xL and low expression of proapoptotic factors such as Bax, Bim and Puma ( Figures 1A, B ) (1). At the same time, it is known that the low efficacy of apoptosis induction with conventional therapies is due to the robust antioxidative defenses of tumor cells (2). Among the main antioxidants that confer apoptosis resistance on tumor cells is glutathione (GSH) (3, 4).

Due to the high resistance of tumors to apoptosis, the induction of necroptosis was postulated to be a potential therapeutic approach (5, 6). In contrast to apoptosis, which does not generate an inflammatory response, necroptosis induces death by cellular explosion, which generates a microenvironment of proinflammatory signals that could favor tumor death (5, 6). Thus, necroptosis, a form of regulated necrosis dependent on RIPK1, RIPK3 and MLKL, was postulated as a potential therapy for cancer ( Figures 1A, B ) (7–9). Unfortunately, several tumor cells evade necroptosis efficiently by inhibiting the expression of RIPK3 via epigenetic control mechanisms (10–12).

In line with this idea, new and emerging forms of regulated cell death with characteristics of necrotic disintegration have been described and postulated as treatments for cancer; among these, ferroptosis is highlighted (13–19). Here, we describe the main pharmacological targets for the induction of ferroptosis with emphasis on the treatment of brain tumors.

System x c − is an antiporter that imports cystine and exports glutamate from the cell in a 1:1 ratio ( Figures 1A, B ). System x c − is composed of 2 subunits: the SLC7A11 subunit (also called xCT), with a transport function and solute carrier family 3 member 2 (SLC3A2; also called CD98hc or 4F2hc), a chaperone with a plasma membrane anchoring function (20–23). For the purposes of this review, we refer only to the SLC7A11 subunit, given the importance of cystine transport to the cell ( Figures 1A, B ). In this context, the uptake of cystine into the cell is essential to maintain the redox state, since the reduced form of this amino acid (nonessential) is necessary for the biosynthesis of the main intracellular antioxidant, glutathione (GSH) ( Figure 1A ). Interestingly, most cancer cells overexpress SLC7A11 (22), suggesting a strong dependence on GSH to maintain the levels of controlled reactive oxygen species (ROS) ( Figure 1B ); thus, SLC7A11 is an potential therapeutic target. Interestingly, in 2012, it was determined that the small molecule erastin (13, 24) targeted SLC7A11 for inhibition, which led to depletion of GSH, inducing a type of death dependent on iron and lipid ROS, called ferroptosis (14). This type of cell death was inhibited by radical trapping antioxidants (RTAs) such as Ferrostatin-1 (Fer-1), lipophilic antioxidants such as vitamin E or iron chelators such as Desferoxamine (DFO) (18, 25), with potential application in the treatment of cancer and other pathologies ( Figure 2B ) (26).

Thus, when tumor cells are incubated with erastin, cell death is induced independent of caspases (13) or mitochondrial oxidative stress but in a manner dependent on iron, ROS and lipid ROS (14). Even though there is evidence that mitochondria could be involved, regulating the “avidity” for ferroptosis induction (27–29), they are not necessary for activation of this pathway (30). Inhibition of system x c − results in intracellular depletion of cysteine ​​ because extracellular cystine (Cys₂) is imported through SLC7A11 and reduced intracellularly to cysteine ( Figure 2B ) (16, 31). Intracellular cysteine ​​is necessary for the biosynthesis of GSH (16, 32). In turn, GSH is a cofactor for the selenoprotein GPX4, a hydroperoxidase responsible for detoxifying toxic hydroperoxides to alcohols (15). Therefore, erastin triggers indirect inhibition of GPX4 activity mediated by GSH depletion ( Figure 2B ).

Despite this apparent dependence of cells on system x c − , animals with knockout of the slc7a11 gene are fertile and develop completely normally (33), which prompted the consideration of SLC7A11 inhibition as an eventual cancer therapy with few adverse effects ( Figure 2A ).

Thus, although many tumor cells can evade apoptosis and necroptosis due to their low expression of key genes for the activation of these pathways ( Figure 1B ) (1, 10, 11), RNA-seq data show that most cancer cells have high expression levels of SLC7A11 and GPX4 (https://portals.broadinstitute.org/ccle). Similarly, tumor cells are “addicted” to iron because they have decreased expression of ferroportin (FPN), the iron efflux pump, and overexpress the transferrin receptor (TfR1), the iron importer ( Figure 1B ) (34–37). Indeed, excess iron contributes to both tumor initiation and tumor growth (34). These observations indicate that SLC7A11, GPX4, iron and ferroptosis are potential therapeutic targets for cancer ( Figures 2B , 3 ). However, there are cancer cells that do not express FPN (MCF-7 cells, among others) and therefore accumulate excess intracellular iron but are still resistant to ferroptosis (38, 39). An explanation for this phenomenon is the recent finding that in addition to GPX4 and iron, acyl-CoA synthetase long-chain family member 4 (ACSL4) is another component that dictates sensitivity to ferroptosis (39). Reinforcing this concept, ACSL4 is a key protein because it incorporates long polyunsaturated fatty acids (PUFAs) into membranes, which allows lipid peroxidation to proceed and ferroptosis to be carried out (39–41). In another context, the erastin analog imidazole ketone erastin (IKE) has been shown to be metabolically stable and a potent inducer of ferroptosis in tumor cells in vivo (19, 42). Thus, induction of ferroptosis in tumor cells through inhibition of SLC7A11 may be a promising treatment for use in patients.

Interestingly, high doses of glutamate can inhibit system x c − , emulating the effects induced by erastin (14, 43, 44). However, it is known that the responses to glutamate treatment are diverse and can induce cell death by apoptosis or necroptosis (45, 46) and eventually by other pathways of regulated necrosis. Thus, although high doses of glutamate can inhibit system x c − , they are not necessarily a specific inducer of ferroptosis in tumor cells but could induce ferroptosis in normal tissues under pathophysiological conditions (44, 47–49).

Normal adult neurons are equipped to survive because they express low levels of proapoptotic proteins and high levels of antiapoptotic proteins (57, 58). Furthermore, it has been shown that as neurons mature, they lose chemosensitivity to staurosporine and doxorubicin (58). This evidence suggests that brain tumors would be highly resistant to conventional antineoplastic agents, given the preconditioning of this type of cell to efficiently evade apoptosis. At the same time, it has been shown that tumor cells of astroglial origin (T98G, U251 and A172) efficiently evade necroptosis induced by chemotherapeutic agents because they do not express RIPK3 due to epigenetic modifications (10, 11).

Thus, the development of new therapies for the treatment of brain tumors that do not involve the induction of apoptosis or necroptosis as the main strategy is urgently needed. In this sense, in recent years, the induction of ferroptosis has gained great relevance as a possible therapeutic approach to induce cell death in brain tumors (17, 50, 59, 60). Considering this concept in the following sections, we focus on the induction of ferroptosis in neuroblastoma (NB) and glioblastoma multiforme (GBM).

Because NB generally presents resistance to Erastin and RSL-3, it is necessary to search for new ferroptosis inducers. To this end, it was recently determined that treatment with the natural compound withaferin A (WA) can eradicate high-risk NB (17) by inducing ferroptosis through the canonical pathway, this means with GPX4 as a direct target. On the other hand, via a noncanonical pathway, where keap1 is the target, thus favoring an increase in labile iron pool (LIP) (17). This dual behavior of WA, similar to that of a mixture of FIN56 and FINO2 ( Figures 4 , 5 ) (52, 53) (compounds that have not been tested in NB models), seems to render it a promising drug therapy for NB, since to date, it is unknown whether in vivo application of FINO₂ is possible (40). Fortunately, WA has been shown to be effective in promoting the eradication of NB in vivo (17, 50). Interestingly, even though WA has been shown to induce iron-dependent lipid peroxidation and GPX4 depletion, Fer-1 treatment does not completely rescue NB cells from cell death (17). This suggests two alternatives; the first is that WA induces other types of ferroptosis-independent death in NB. However, WA induces lipid peroxidation, which is completely inhibited by treatment with DFO and partially inhibited by Fer-1, suggesting a strong iron dependence (17).

In this context, as a second alternative, the authors suggest that WA could eventually favor an overload of lipid ROS of various origins that may not necessarily be inhibited by Fer-1, such as lipid ROS generated by H₂O₂ (17). It is important to note that Fer-1 does not inhibit death induced by H₂O₂ treatment (14) or by extracellular H₂O₂ production mediated by pharmacological doses of ascorbic acid (68), because these treatments preferentially induce conventional necrosis. However, it has recently been determined that NADPH-cytochrome P450 reductase (POR) favors the induction of ferroptosis due to the cytoplasmic production of H₂O₂ (69), which is inhibited by Fer-1 or by the intracellular expression of catalase (69) but thus far, this finding is limited to cervical cancer cells (HeLa).

In this context, and considering the particularities of NB cells, it is likely that it is possible to classify the cell death induced in this type of tumor (or others) as ferroptosis, even when it is not inhibited by Fer-1, if it has other hallmarks, such as lipid peroxidation and iron dependence. Indeed, it has recently been determined that inhibition of lipid peroxidation mediated by liproxstatin-1 treatment is not sufficient to rescue SLC7A11 KO melanoma cells from death (70). Furthermore, it has also been shown that there is strong induction of lipid peroxidation during the activation of noncanonical pyroptosis that is not necessarily related to the direct execution of this death pathway (71). This evidence could limit lipid peroxidation as an exclusive hallmark of ferroptosis, driving the definition of ferroptosis toward a type of death dependent on lipid peroxidation (72).

In another context and emphasizing that MYCN is a protein overexpressed in NB, recent advances have been achieved to determine that MYCN favors an increase in intracellular iron per se, which could favor the pharmacological sensitization of NB to ferroptosis induction (66, 67). Thus, the authors determined that inhibition of SLC7A11 with sulfasalazine (14) or inhibition of GSH synthesis with Buthionine sulfoximine (BSO) favors the induction of ferroptosis in models of NB with MYCN amplification (67). This in vitro evidence from patient samples is closely related to an eventual clinical application, since the toxicity of BSO has been evaluated in a phase I clinical trial, as a possible treatment for NB in conjunction with other drugs (73). Despite being relatively well tolerated, the treated patients presented vomiting/nausea as adverse effects (73). However, there is also evidence indicating that the administration of BSO can trigger kidney failure in animal models (74) and patients (75). Thus, special precautions must be taken when trying to directly extrapolate in vitro findings to in vivo models or patients.

Curiously, some of the NB cell lines used in this study show partial resistance to death induced by the SLC7A11 inhibitor and GSH depletor erastin (17), which leads to an intriguing question: why are some NB cell lines resistant to erastin but sensitive to inhibition of GSH synthesis or inhibition of SLC7A11 mediated by sulfasalazine? In this scenario, it is important to highlight that in lung adenocarcinoma cells, it was recently determined that the SLC7A11 inhibitor HG106 preferentially induces GSH depletion and cell death by apoptosis, which is inhibited by the recovery of cysteine ​​uptake, but without eventual induction of ferroptosis, since DFO treatment does not prevent cell death (76). This evidence suggests that although HG106 has the same target as erastin (SLC7A11), there are other off-targets that favor the induction of one type of death over another (apoptosis or ferroptosis) or the production of particular ROS that trigger differential cellular responses (22). Despite these pharmacological dichotomies, which induce different types of death even when the target is the same, or which have differential action mechanisms in response to treatment with SAS, erastin (IKE) or HG106, the message that remains the same: SLC7A11 is a potent therapeutic target for cancer ( Figure 2B ).

It was shown that NB cells with MYCN amplification are particularly sensitive to the induction of death mediated by treatment with auranofin (a rheumatoid arthritis drug) (67). Although the authors attribute the effect of auranofin to the induction of ferroptosis, treatment with Fer-1 only partially rescues cells from cell death, even when there is an increase in lipid peroxidation, and treatment with DFO effectively prevents cell death and ROS production (67). Again, this finding leads us to conclude that apparently in NBs, the lipid ROS generated are specific to this tumor type or there are parallel mechanisms of cell death, since Fer-1 is not capable of completely inhibiting cell death, even when the evidence points to iron and lipid ROS dependency. Accumulating evidence, the literature indicates that iron accumulation and increased LIP are strong candidates for exploiting the pharmacological sensitivity of high-risk NB to ferroptosis induction ( Figure 5 ) (17, 50, 66, 67, 77). Thus, the use of compounds that promote the mobilization or uptake of iron in this type of tumor, in combination with ferroptosis inducers, could exploit the vulnerabilities of this tumor to favor its eradication. However, further studies are still needed to determine the potential lethal effects on nervous tissue and to assess whether these types of therapeutic agents can penetrate the blood–brain barrier.

Although there are several inducers of ferroptosis, the potential use of these drugs as cancer treatments is limited because they have little bioavailability for action in vivo. However, with the development of IKE, an avenue was opened for ferroptosis induction as an in vivo treatment by targeting system x c − (19). To date, evidence suggests that inhibition of system x c − could be a safe therapeutic approach as a tumor suppressor. Unfortunately, several types of tumors, including NB and GBM, are resistant to system x c − inhibition for ferroptosis induction (13, 17, 92). Thus, combination therapies of other antineoplastic drugs with IKE may represent an option for the treatment of cancers highly resistant to cell death. However, ferroptosis dogma dictates that GPX4 is the key protein (15, 106); thus, all efforts have been focused on the development of new drugs for its inhibition. Although there are direct GPX4 inhibitors, such as RSL3 (15), they have little application in vivo (40), and GPX4 deletion in some types of cancer is not lethal (51), suggesting that there may be other mechanisms in addition to GPX4 inhibition to suppress lethal lipid peroxidation. FSP1, GCH1 and BH4/BH2 are proteins with the ability to inhibit ferroptosis independently of GPX4 and GSH levels (54, 55, 65), and FSP1 is a druggable protein (54). In line with this idea, a new avenue has been opened for the development of drugs that include SLC7A11, GPX4 and FSP1 inhibitors with potential in vivo application as a cancer treatment.

LF, MB, KS, AG and FN conceived the ideas and concepts. LF wrote the article. AG, KS, MV and FN critically revised the manuscript. MB generated the scientific illustrations. All authors approved the final version of the manuscript. All authors contributed to the article and approved the submitted version.

This work was supported by a Fondecyt regular grants 1181243 and 1221147, ANID PIA ECM-12 grant (to FN), Fondecyt iniciacion 11200335 (LF), Fondecyt postdoctorado 3210076 (MB) and U.S. Department of Defense W81XWH-12-1-0341 (AG).

The authors declare that the research was conducted in the absence of any commercial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.