Main

Lipid components of cellular membranes are constantly exposed to free radical species that are competent to trigger their degradation through an oxygen-dependent process4. This process broadly known as lipid peroxidation is primarily dictated by the propagation rate constants (kp) of its lipidic elements, an intrinsic chemical feature unique to each of these components. The past few years have witnessed a surge of interest in understanding the cellular mechanisms that regulate lipid peroxidation as they have been associated as key determinants of a distinct non-apoptotic cell death modality, known as ferroptosis5.

Early works have established the central role of the enzymatic activity of the selenoprotein glutathione peroxidase 4 (GPX4)2,6 in suppressing the process of ferroptosis7,8,9. GPX4 is the sole enzyme in mammals capable of directly reducing a broad range of peroxidized lipids present in membranes10,11. GPX4 can be irreversibly inhibited by a series of alkylating small molecules, such as RSL3 and ML210 (ref. 8), leading to cell death in ferroptosis-sensitive cancer cell lines. The enrichment of phospholipids with polyunsaturated fatty acids (PUFAs) results in a marked dependency on GPX4 activity12,13. This high-PUFA state was shown to be largely dependent on the activity of the enzyme acyl-CoA-synthetase long-chain family 4 (ACSL4), which is required for the critical step of PUFA activation12. Accordingly, the inhibition of GPX4 in ferroptosis-prone cell lines leads to the characteristic oxidation fingerprint entailing the accumulation of peroxidized products of phosphatidylethanolamine (PE) containing arachidonic acid and adrenic acid14. It has been further demonstrated that the sole accumulation of peroxidized fatty acids is not sufficient to induce ferroptosis and a central role of the free radical-mediated propagation step has been unambiguously demonstrated15. The propagation step of lipid peroxidation was shown to contribute to the formation of pore-like structures of ill-defined identity16 that drive the osmotic lysis of the cells17.

The present study uncovered and characterized a role for 7-dehydrocholesterol reductase (DHCR7) in the ferroptotic process. DHCR7 catalyses the final step in cholesterol biosynthesis and its inhibition leads to the accumulation of 7-dehydrocholesterol (7-DHC). Others18 initially reported 7-DHC to accumulate in preputial gland tumours and whose function, at that time, was only assumed to be as a spare capacity for cholesterol synthesis. Subsequent studies characterized 7-DHC as the most oxidizable lipid ever reported and whose accumulation predisposes cells to lipid peroxidation19. By contrast, we now show that the accumulation of 7-DHC causes a paradoxical increased tolerance towards phospholipid peroxidation, thus providing a robust resistance to ferroptosis. Furthermore, the characterization of the protective effect of 7-DHC provided valuable insights into the distinction between lipid and phospholipid peroxidation in cell death processes. By demonstrating the accumulation of oxidatively truncated phospholipid species in ferroptotic cell death, we emphasize the crucial role of these species in the execution of ferroptosis. Together with the accompanying paper20, our findings suggest that manipulating this pathway could be exploited to increase ferroptosis resistance to suppress ferroptosis in acute settings but also exploited by cancer cells to evade ferroptosis.

DHCR7 is a proferroptotic gene

Spurred by the still incomplete understanding of the ferroptotic process and the development of next-generation single guide RNAs (sgRNAs)21, we performed a genome-wide reverse genetic CRISPR screen to identify genes that may confer robust protection against ferroptosis. To this end, the Pfa1 cell line6 was transduced with a CRISPR library covering 18,424 genes with a total representation of 90,230 sgRNAs followed by a stringent selection for 14 days using 200 nM of the GPX4 inhibitor (1S,3R)-RSL3 (in the following referred to as RSL3) (Fig. 1a). Consistent with the results of ours and others previous screens, Acsl4 emerged as the highest-scoring hit12,13,22,23,24. The second top-scoring gene was Dhcr7 (Fig. 1b). The identification of Dhcr7 as a potential proferroptotic gene was surprising in light of several studies indicating that loss or inhibition of DHCR7 is associated with an increased susceptibility to lipid peroxidation4,25, which, in principle, should lead to an increased susceptibility to ferroptosis26. Intrigued by this finding, we set out to explore the basis of this discovery. Using the bona fide ferroptosis fibrosarcoma cell line model HT1080, we generated polyclonal cultures of DHCR7-deficient cell lines using two independent sgRNAs. The successful loss of DHCR7 was validated by western blot and mirrored by the accumulation of its substrate 7-DHC (Fig. 1c,d) and impaired incorporation of C13-glucose into cholesterol (Fig. 1e). Notably, cholesterol depletion was less pronounced, suggesting that a substantial fraction is directly taken up from the serum. Importantly, knockout of DHCR7 did not concur in a marked alteration in the protein concentrations of known ferroptosis regulators (Fig. 1c) nor the phospholipid composition of cells (Extended Data Fig. 1a–c). Using these cellular models, we validated the screening results showing that DHCR7-deficient HT1080 cells present a marked resistance to ferroptosis (Fig. 1f). Similar results were obtained with three independent clonal cell lines derived from Pfa1, HT1080 and MDA-MB-435 cells, confirming the general impact of this system in specifically preventing ferroptosis (Extended Data Fig. 2a–d). Subsequent studies focused on the clonal cell line derived from the HT1080 DHCR7 knockout (KO) pool (a detailed characterization of the genetic modification of these cells is provided in Extended Data Fig. 3a–d). Thereby, we could unequivocally demonstrate the proferroptotic activity of DHCR7 because the genetic reconstitution of DHCR7 abolished 7-DHC concentrations and resensitized cells to ferroptosis without affecting the response of the cell to other cytotoxic agents (Extended Data Fig. 3e–g).

Fig. 1: Identification and impact of DHCR7 deficiency on ferroptosis.
figure 1

a, Schematic of the identification of Dhcr7 as a proferroptotic gene, using CRISPR-KO library and GPX4 inhibition. b, Volcano plot of sgRNA enriched in cells selected with RSL3 compared with untreated control cells. c, Immunoblot (IB) analysis of DHCR7 and key ferroptosis regulators, namely, FSP1, ACSL4 and GPX4 in cells expressing an sgRNA targeting DHCR7 and EGFP. Values represent mean ± s.d. of ratio of protein of interest in relation to β-actin, n = 3 independent experiments. d, Relative quantification of 7-DHC and cholesterol concentrations in HT1080 cell lines stably transduced with a vector expressing Cas9 and a sgRNA targeting DHCR7 and EGFP as a control. e, Assessment of de novo cholesterol biosynthesis, by means of the quantification of 13C-cholesterol originating from 13C-glucose in HT1080 cell expressing sgRNA targeting DHCR7 and EGFP as control. Data are the mean ± s.d. of n = 3 wells of a 6-well plate from one representative experiment (d,e). f, Dose-dependent toxicity of the ferroptosis inducers RSL3, ML210 and FIN56 in HT1080 cell lines stably transduced with a vector expressing Cas9 and an sgRNA targeting DHCR7 and EGFP as a control. Cell viability was monitored using Alamar blue after 48 h (f) and represented as the mean ± s.d. of triplicates from one representative of two independent experiments (f). *P < 0.05; two-way analysis of variance (ANOVA) (d–f).

Source Data

Full size image

7-DHC is an antiferroptotic metabolite

In the penultimate step of the cholesterol biosynthesis pathway, lathosterol, through lathosterol oxidase (SC5D), is converted to 7-DHC, which, in turn, is reduced to cholesterol by DHCR7 in the final step of the pathway (Fig. 2a and Extended Data Fig. 4a). Several previous studies have pointed to a toxic effect of 7-DHC accumulation through its inherent propensity to autoxidize and propagate radical chain reactions within the lipid bilayer25. To shed light on these seemingly paradoxical observations, we generated a DHCR7 SC5D double-mutant cell line to address whether 7-DHC accumulation mediates the protective effects induced by the loss of DHCR7. In agreement with a protective effect of 7-DHC, the loss of SC5D in the DHCR7 KO cell line completely abolished the resistance conferred by the single loss of DHCR7 (Fig. 2b). Similarly, pharmacological inhibition of upstream steps of cholesterol biosynthesis recapitulated this effect (Extended Data Fig. 4b). Accordingly, combined loss of DHCR7 and SC5D led to a detectable accumulation of lathosterol and suppressed 7-DHC accumulation (Fig. 2c). Subsequently, the serial reconstitution of DHCR7 and SC5D in a DHCR7 SC5D KO background demonstrated that the re-expression of SC5D resulted in substantial accumulation of 7-DHC as also validated by monitoring C13-labelled glucose incorporation into 7-DHC and cholesterol (Fig. 2d,e,f). This, in turn, resulted in a specific increased resistance to ferroptosis (Fig. 2g and Extended Data Fig. 4c). Using wild-type (WT), DHCR7 and DHCR7 SC5D-deficient cell lines in a series of sterol supplementation experiments, we further demonstrated that exogenous supplementation of 7-DHC protected all cell lines from ferroptosis; also, lathosterol only increased ferroptosis resistance in cell lines able to produce 7-DHC (Fig. 2h). Similar observations were made in the sgRNA expressing polyclonal cell population, where we could also demonstrate that squalene, a previously reported ferroptosis suppressor27, failed to inhibit cell death when supplemented exogenously (Extended Data Fig. 4d). Importantly, a similar protective effect of 7-DHC was observed in a genetic model of Gpx4 deficiency6 (Extended Data Fig. 4e). Curiously, free cholesterol blunted the protective effects in all genotypes (Fig. 2h, and Extended Data Fig. 4f). Building on this observation we could show that an enantiomer of cholesterol, which has an opposite three-dimensional structure but identical physical properties to cholesterol28, was markedly less efficient at blunting these protective effects (Extended Data Fig. 4g). Combined with the observed loss of 7-DHC in cells treated with free cholesterol (Extended Data Fig. 4h,i) our observations suggest an inhibitory effect on SREBP2 and biosynthetic activity of the mevalonate pathway.

Fig. 2: 7-DHC accumulation suppresses ferroptosis.
figure 2

a, Schematic of final steps of cholesterol biosynthesis. b, Dose-dependent toxicity of RSL3 in HT1080 Cas9 WT, DHCR7 and DHCR7 SC5D KO cell lines. c, Relative quantification of lathosterol, 7-DHC and cholesterol concentrations in HT1080 Cas9 WT, DHCR7 and DHCR7 SC5D KO cell lines. d, Immunoblotting for DHCR7 in HT1080 Cas9 expressing the indicated lentiviral constructs. β-ACTIN was used as a loading control. e, Relative quantification of 7-DHC and cholesterol concentrations in HT1080 Cas9 DHCR7 SC5D KO stably overexpressing SC5D and DHCR7 (SC5D/DHCR7), only SC5D (SC5D/mock), only DHCR7 (mock/DHCR7) and an empty vector (mock/mock). f, Assessment of de novo cholesterol biosynthesis, by means of the quantification of 13C-cholesterol and 13C-7-DHC originating from 13C-glucose, in HT1080 Cas9 in the indicated genotypes. g, Dose-dependent toxicity of the ferroptosis inducers RSL3 and ML210 in the HT1080 cell lines described in e,f. h, Effect of sterol supplementation [10 µM] on RSL3 toxicity in HT1080 Cas9 WT, DHCR7 and DHCR7 SC5D KO cell lines. Cell viability was assessed after 24 h (b) or 48 h using Alamar blue (g,h). Data are the mean ± s.d. of n = 3 wells of a 96-well plate (b,g,h) or a 6-well plate (c,e,f) from two (b,h) or three (g) or one (c,f) independent experiments. *P < 0.05; two-way ANOVA (b,c,e,g,h).

Source Data

Full size image

Given the suppressive function of cholesterol on the protective effect conferred by 7-DHC, we investigated the response to ferroptosis in settings where cholesterol supply is scarce. To accomplish this, we cultivated cells in delipidated fetal bovine serum (dlFBS), which effectively removes sterols from the culture medium (Extended Data Fig. 5a). In dlFBS, we noted a decline in total cholesterol concentrations and a concurrent increase in 7-DHC in DHCR7-deficient cells, indicating enhanced biosynthesis (Extended Data Fig. 5b). Under this experimental condition, we consistently observed similar responses, albeit with heightened sensitivity, which can probably be attributed to reduced expression of GPX4 resulting from the fumed silica treatment (Extended Data Fig. 5c,d). Notably, the loss of GPX4 seems to be independent of sterol concentrations and is probably due to selenium depletion (Extended Data Fig. 5e,f). To mitigate potential confounding factors, we investigated the impact of LDL-receptor (LDLR) KO (Extended Data Fig. 5g). As anticipated, the KO cells exhibited increased expression of SREBP2 target genes, no differences in GPX4 concentrations and an inability to efficiently internalize fluorescently labelled LDL (Extended Data Fig. 5h–k). Treatment with the EBP inhibitor Tasin-1 induced a substantial reduction in cholesterol concentrations in the LDLR KO cells, whereas the WT cells remained largely unaffected (Extended Data Fig. 5l). Using these models, we show that the loss of LDLR does not significantly affect ferroptosis under normal conditions but pretreatment with Tasin-1 markedly sensitizes LDLR KO cells to ferroptosis (Extended Data Fig. 5m). These findings substantiate the notion that 7-DHC plays a crucial role in cellular protection, particularly in conditions where biosynthesis is stimulated.

7-DHC blocks phospholipid peroxidation

The conjugated double-bond present in the sterol B-ring stands as the most prominent feature of 7-DHC, when compared to the other sterols. To probe the relevance of this feature in preventing ferroptosis we assayed the structurally related sterol ergosterol for its capacity to supress ferroptosis (Extended Data Fig. 6a) and showed that it has an equally potent antiferroptotic activity (Extended Data Fig. 6b). Given that ergosterol is the main sterol component in yeast and fungi, it was reasonable to assume that this lipid could be an important suppressor of cell death induced by PUFAs in these evolutionarily distant organisms. In fact, we could validate this hypothesis in yeast strains with targeted deficiencies of genes important for ergosterol biosynthesis (that is, erg2, erg3 and erg6)29 by revealing a hypersensitivity to PUFA supplementation in cells unable to generate sterol with the characteristic unsaturated B-ring structure (Extended Data Fig. 6c,d).

To investigate the impact of 7-DHC in a well-defined phospholipid autoxidation model, we prepared unilamellar liposomes of soy phosphatidylcholine (PC) loaded with 7-DHC (Fig. 3a). We used the recently developed FENIX assay to indirectly monitor in real time the process of phospholipid peroxidation30. The assay relies on the specific generation of lipid peroxyl radicals arising from the lipophilic radical generator di-tert-undecylhyponitrite (DTUN). A small amount of STY-BODIPY dye competes with PUFA for propagating lipid peroxyl radicals and the fluorescence of its oxidized product(s), STY-BODIPYox, can be monitored by fluorescence (Fig. 3a). Typical radical-trapping antioxidants inhibit autoxidation and thus retard STY-BODIPY oxidation until the radical-trapping antioxidant is consumed (Fig. 3a,b). Interestingly, 7-DHC-loaded liposomes resulted in a dose-dependent suppression of STY-BODIPY oxidation (Fig. 3b,c). As the suppression of STY-BODIPY oxidation could arise from dilution of the pool of autoxidizable phospholipids on supplementation of the liposomes with 7-DHC, similar experiments, wherein non-oxidizable dipalmitoyl PC (DPPC) was incorporated in place of 7-DHC, were performed, allowing us to demonstrate no difference from the native soy PC liposomes (Fig. 3b). Furthermore, because sterols alter membrane fluidity and may confer protection through dynamic parameters31 that could impact lipid peroxidation32, corresponding experiments were carried out on cholesterol-loaded liposomes (Fig. 3c). Yet again, there was no effect on the rate of STY-BODIPY oxidation—even beyond concentrations of 7-DHC used (Extended Data Fig. 7a)—suggesting that physical changes in the bilayer imparted by the sterol framework do not impact the oxidation rates in our model system, neither do they impact their integrity (Extended Data Fig. 7b). Given the indirect nature of the assay, we also directly measured the impact of 7-DHC on soy PC peroxidation, that is palmitoyl-linoleoyl PC (PLPC)-OOH, dilinoleoyl PC (DLPC)-OOH and DLPC-2OOH, by liquid chromatography with tandem mass spectrometry (LC–MS/MS) (Extended Data Fig. 7d,e). Although supplementation of the liposomes with DPPC (up to 32 mol%) had no effect on the rate of PLPC and DLPC oxidation, cholesterol (at 8 mol%) had only a modest effect on the accumulation of PLPC-OOH, DLPC-OOH and DLPC-2OOH. Entirely consistent with the FENIX results, 7-DHC supplementation led to a dose-dependent suppression in the rate of PLPC and DLPC oxidation (Extended Data Fig. 7d,e) which is in good agreement with previous reports in isotropic media33. To demonstrate that this suppression corresponded with the intervention of 7-DHC in the radical chain reaction, the consumption of 7-DHC was monitored spectrophotometrically through its characteristic absorbance (Extended Data Fig. 7f,g). These data thus suggest that the oxidation of 7-DHC in vitro is responsible for the inhibition of phospholipid peroxidation, a notion we could further validate in a model using iron/ascorbate as the source of oxidation (Extended Data Fig. 7h). Accordingly, we detected significant concentrations of the free radical-mediated oxidation product of 7-DHC, namely 3β,5α-dihydroxycholest-7-en-6-one (DHCEO) (Fig. 3d), during the liposomal oxidation under conditions where no phospholipid oxidation product was detectable (Fig. 3e). Hence, if our hypothesis was correct, 7-DHC oxidation should lead to the accumulation of these products during the course of ferroptosis and by doing so, it could spare phospholipids from oxidative damage. To assess whether 7-DHC oxidation products also accumulate on triggering ferroptosis in cells, we treated the HT1080 DHCR7 SC5D double-KO cell line expressing SC5D and an empty vector with the GPX4 inhibitor RSL3. Although no substantial loss in the total content of 7-DHC was noticeable (Extended Data Fig. 7i), the quantification of the main non-enzymatic oxidation products of 7-DHC, namely DHCEO, revealed a significant increase (Fig. 3f and Extended Data Fig. 7i). To demonstrate that the 7-DHC products originate from the peroxyl radical-mediated oxidation of 7-DHC, we further incubated these cells with Lip1 (ref. 9). In good agreement with the free radical-mediated formation of DHCEO25, Lip1 fully inhibited the formation of this product (Fig. 3f and Extended Data Fig. 7i). Therefore, these results firmly establish a unique role of unsaturated B-ring sterols in protecting cells from ferroptosis-like cell death by diverting the propagation of peroxyl radical-mediated damage from phospholipid components to its sterol core.

Fig. 3: 7-DHC acts to suppress (phospho)lipid peroxidation.
figure 3

a, Liposomes of soy PC (1 mM) supplemented with cholesterol, 7-DHC or DPPC were prepared and autoxidized using DTUN [200 nM]. STY-BODIPY co-autoxidations are monitored by fluorescence of oxidized STY-BODIPY. PMC, 2,2,5,7,8-pentamethyl-6-chromanol. b, Representative data from co-autoxidations of STY-BODIPY [1 μM] and liposomal soy PC. c, The rates of STY-BODIPYox formation plotted as a function of additive concentration. d, Schematic of 7-DHC oxidation highlighting DHCEO as the main detectable oxidation products. e, Quantification of 7-DHC and DHCEO in liposome with and without 7-DHC in the absence or presence of DTUN for 2 h. f, Quantification of DHCEO in HT1080 SC5D DHCR7 double-KO cells expressing empty vector (black) and SC5D (red) on [200 nM] RSL3 with and without 500 nM Lip1 for 6 h. Data are the representative of one independent experiment (b,c) or the mean ± s.d. of n = 3 technical replicates of a 6 cm plate from two independent experiments (e,f). *P < 0.05; two-way ANOVA (e,f).

Source Data

Full size image

Truncated phospholipids drive cell lysis

Following these results, we reasoned that the presence of 7-DHC in phospholipid bilayers generates a strong prosurvival effect by increasing the resistance of membranes to peroxidation-mediated permeabilization. Therefore, a model system was used that consists of 5(6)-carboxyfluorescein (CF) encapsulated in liposomes allowing for the detection of a fluorescent signal on membrane permeabilization (Extended Data Fig. 8a). Using the iron/ascorbate couple as an oxidation model, we showed that liposomes containing 7-DHC were markedly resistant to peroxidation-mediated membrane permeabilization (Extended Data Fig. 8b). To further support the relevance of this model system for ferroptosis, we could show that the process of vesicle rupture could be prevented by Lip1 (Extended Data Fig. 8c) and other molecules able to suppress ferroptosis, such as ergosterol, ubiquinone, α-tocopherol and squalene (Extended Data Fig. 8d), indicating that Lip1 and naturally occurring ferroptosis supressors could act similarly to prevent membrane permeabilization of cells.

Recent reports studying the relative contribution of different photosensitization mechanisms to membrane permeabilization suggested that truncated phospholipid species rather than phospholipid hydroperoxides are key in generating membrane pores and consequently mediating the loss of membrane integrity34. Therefore, we reasoned that a similar mechanism could be at play during iron-induced permeabilization and ferroptosis execution35. To establish a functional link between truncated lipids and ferroptosis execution, we initially assayed a panel of different truncated species (Extended Data Fig. 9a) regarding their capacity to destabilize membranes. Accordingly, all tested truncated lipids were able to permeabilize liposomal membranes and to induce cell death more efficiently than the parental lipid and the corresponding hydroperoxide (Extended Data Fig. 9b–e). Further validation was provided by using an orthogonal approach based on a photochemical probe (PhotoPC). Irradiation of the probe directly generates a truncated product (PhotoTrunc-PC) which does not rely on the presence of alkoxyl or peroxyl radicals intermediate (Extended Data Fig. 9f). Using this model we could demonstrate the higher membrane destabilizing capacity of the truncated product in vesicles and cells (Extended Data Fig. 9g,h). Although being highly supportive, it should be acknowledged that the truncated species were added exogenously and were performed using PC and not PE species12,14. To circumvent this issue, a system in which the species are formed in situ would be preferred. We took advantage of the cell’s own fatty acid incorporation machinery to achieve this goal. ACSL4-deficient cells have a profound loss of PUFA content in phospholipids12. The absence of PUFA containing phospholipids results in a marked resistance to ferroptosis because of the lack of oxidizable substrates. Sensitivity to ferroptosis in this setting can be regained by feeding exogenous PUFAs12. This feature can be leveraged to better control of the substrates used for ferroptosis execution. Using this model, we compared side-by-side the sensitization provided by α-linolenic acid (αLNN) and γ-linolenic acid (γLNN). Both fatty acids have an identical structure in length and number of double bonds leading to a similar propagation rate constant (kp), yet the position of the last double-bond determines the structure of the resulting truncated product. Analysis of the lipidomic changes of ACSL4 WT and KO cells treated with αLNN and γLNN confirmed that both lipids are directly and efficiently esterified into PE, thereby restoring the oxidizable pool of PUFA to a similar extent as in WT cells (Extended Data Fig. 10a,b). Remarkably, despite their equal abundance and propensity to undergo oxidation, γLNN seemed to be a superior ferroptosis-triggering substrate (Extended Data Fig. 10c,d), in line with its potential to generate shorter truncated phospholipid products. These results are remarkable because they indicate that the product formed determines cell death rather than solely its propensity to autoxidize. Supporting this notion, in-depth epilipidomics analysis indeed detected a substantial accumulation of PE and plasmalogen PE truncated products in cells undergoing ferroptosis (Fig. 4a). Notably, cell permeabilization, monitored as propidium iodide (PI)-positive cells, was only detectable in conditions where an increase in these oxidized and truncated species occurred (Fig. 4b). We further showed that Lip1 fully inhibited the formation of these species, thus confirming their origin from the autocatalytic lipid peroxidation process (Fig. 4a,b). In accordance, cells accumulating 7-DHC behaved similarly to Lip1-treated cells and the specific oxidation product of 7-DHC, DHCEO, accumulated in these cells (Fig. 4b). This demonstrates that 7-DHC is preferentially oxidized in cells, thereby sparing phospholipids and preventing the formation of oxidized and truncated species (Extended Data Fig. 9i). Supporting the proposed mechanism, 7-DHC did not affect permeabilization mediated by truncated phospholipid species (Extended Data Fig. 9e). Together, these observations provide compelling evidence for the role of truncated products in contributing to ferroptosis execution and that 7-DHC and other ferroptosis inhibitors such as Lip1, directly suppress their formation.

Fig. 4: Phospholipid truncated species contribute to ferroptosis execution.
figure 4

a, Epilipidomics analysis of HT1080 Cas9 DHCR7 SC5D double-KO cells overexpressing SC5D or an empty vector. Data are representative of one experiment performed five times independently (results from the other repetitions are in the Supplementary Information). b, Cell death (PI positive) and DHCEO values from matched experimental time points of samples depicted in a.

Source Data

Full size image

7-DHC accumulation increases cell fitness

Having characterized the molecular underpinnings by which 7-DHC prevents ferroptosis execution, we next asked whether this protective effect could have a potential role in supporting tumour growth under conditions in which ferroptosis inhibition is critical. To our initial surprise, DHCR7 mutations, despite being rare, have been described in people with Burkitt’s lymphoma (BL), with a reported 9.8% frequency of DHCR7 mutations as shown by ref. 36. Moreover, a recent report has also identified rare pathological mutations in DHCR7 in a cohort with neuroblastoma37. To gain insights into the topology of these mutations, we created a model for the DHCR7 structure using an homologous structure (PDB ID 4QUV, sequence identity 37%, similarity 51%) and identified that they are primarily located in the transmembrane domain of DHCR7 (Fig. 5a). Re-expression of DHCR7-Flag-tagged versions of the seven corresponding mutants in the DHCR7-deficient cells allowed us to validate these predictions experimentally. Figure 5b illustrates that, except for T93M, N274K and V353fs, all mutations were generally well expressed as compared to WT. We then addressed the functionality of these mutations: the A24S and L317V mutations seemed to be functional when overexpressed as they were able to (1) metabolize 7-DHC when overexpressed (Fig. 5c) and (2) to re-sensitize DHCR7-deficient cells to ferroptosis akin to the WT enzyme (Fig. 5d). On the other hand, all other assayed variants were dysfunctional and failed to metabolize 7-DHC (Fig. 5c) and were unable to restore sensitivity to ferroptosis (Fig. 5d). Further validation of the role of the DHCR7 was pursued by demonstrating that 7-DHC accumulation abolished the characteristic thiol-dependent growth of Burkitt’s lymphoma cell lines in the absence of thiol-donating compound (Extended Data Fig. 11a,b).

Fig. 5: Impact of 7-DHC accumulation on lymphoma growth.
figure 5

a, Homology model of the DHCR7 structure based on the sterol reductase from M. alcaliphilum. Mutations are indicated as ball and stick. b, Immunoblot analysis of the levels of expression of DHCR7, FLAG-tagged DHCR7 WT and mutant versions reported in patients with Burkitt’s lymphoma or neuroblastoma in a SK-N-DZ-DHCR7 KO cell line. c, Relative quantification of 7-DHC in a SK-N-DZ-DHCR7 KO cell line expressing FLAG-tagged DHCR7 WT and the indicated mutants. Data are the mean ± s.d. of triplicates from a 6-well plate from one representative experiment. d, Dose-dependent toxicity of the ferroptosis inducer ML210 in SK-N-DZ expressing FLAG-tagged DHCR7 WT and the indicated mutants. e, Left, dose-dependent toxicity of RSL3 in BL41 expressing two independent sgRNA targeting DHCR7 and EGFP. Right, relative quantification of 7-DHC concentrations in the BL41 cell line treated with RB38 and Tasin-1. Cell viability was assessed after 48 h using Alamar blue; data are the mean ± s.d. of n = 3 wells of a 96-well plate (d,e) from two (d) or three (e) independent experiments. *P < 0.05; two-way ANOVA (e). f, Schematic representation of the tail vein injection of control (DHCR7WT) or DHCR7-deficient (DHCR7KO) BL41 cell line. g, Kaplan–Meier plot displaying tumour-free survival for mice injected with DHCR7WT (blue, n = 5) or DHCR7KO (red, n = 5) BL41 cells, a log-rank (Mantel–Cox) test was conducted for statistical analysis; P value indicated. h, Tumour growth on implantation of DHCR7WT (blue, n = 5) or DHCR7KO (red, n = 5) of BL41 cell line. Data represent the mean ± s.e.m.; Mann–Whitney test one-tailed, P values are indicated. In each box, horizontal lines denote mean values, whereas the box contains the 25th to 75th percentiles of dataset and whiskers mark the 5th and 95th percentiles. i, Representative luminescence images from each group are shown (g,h). Images in f created with BioRender.com.

Source Data

Full size image

Next, the contribution of 7-DHC accumulation and tumour growth in a series of xenograft models was investigated. Initially, we deleted DHCR7 in two different cell lines: BL41 (a Burkitt’s lymphoma cell line) and MM1S (a multiple myeloma cell line). These cell lines showed distinct responses to ferroptosis when DHCR7 is deleted; whereas BL41 cells showed a significant increase in resistance against GPX4 inhibitors (Fig. 5e), the protective effect in MM1S cells was negligible (Extended Data Fig. 12a,b). Accordingly, when implanted into the tail vein of mice, the MM1S xenograft did not show any noticeable behavioural differences, in line with the absence of extra protective effect caused by DHCR7 loss (Extended Data Fig. 12c,d). By marked contrast, the DHCR7-deficient BL41 cell line showed a significantly more aggressive phenotype compared to its WT counterpart with marked decrease in overall survival of mice (±24 days versus ±60 days) (Fig. 5f–i).

To further substantiate whether this effect was attributed to increased resistance to ferroptosis, we performed an independent experiment where the growth of BL41 xenografts was compared in mice maintained on a selenium-adequate and -deprived diet (Extended Data Fig. 12e–h). This model mimics an in vivo ‘ferroptosis-prone’ condition38 by limiting the supply of selenium for the translation of selenoproteins, including GPX4 (Extended Data Fig. 12f). Under this proferroptotic conditions, an even more pronounced difference in tumour growth was observed (Extended Data Fig. 12g), strengthening the notion that ferroptosis-sensitive cancer cell lines benefit from the accumulation of 7-DHC and that its accumulation favours tumour growth by suppressing ferroptosis. In addition to this model, we also used an orthotopic neuroblastoma model (Extended Data Fig. 12i–p) using the ferroptosis-sensitive neuroblastoma cell line SK-N-DZ. Deletion of DHCR7 in SK-N-DZ cells provided a robust protection against GPX4 inhibitors (Extended Data Fig. 12i–l). Orthotopic implantation of these cells in the adrenal gland of mice led to a more aggressive phenotype, as indicated by the reduced survival of mice implanted with DHCR7-deficient cells (Extended Data Fig. 12m,n). Interestingly, and in agreement with the reduced survival, analysis of these mice showed a massive increase in the incidence of lung metastasis in the DHCR7-deficient neuroblastoma group (Extended Data Fig. 12o,p); we speculate that ferroptosis-sensitive tumour cell lines, such as SK-N-DZ and BL41, benefit from mechanisms that protect against ferroptosis in the bloodstream39. Collectively, our data indicate that in ferroptosis-sensitive cell lines, the extra survival advantage conferred by accumulating 7-DHC promotes a more aggressive phenotype in vivo.

Discussion

Our work introduces and characterizes an unforeseen role for DHCR7 in modulating ferroptosis. Although many reverse and forward genetic screens have been performed to identify regulators of ferroptosis, DHCR7 has not consistently emerged as a regulator—unlike ACSL4, for example13,40. Possible factors affecting DHCR7 inhibition and 7-DHC accumulation include defects in the cholesterol biosynthesis pathway and the impact of free cholesterol on the mevalonate pathway, a notion supported by our results. We now provide a comprehensive understanding of the protective role of B-ring-unsaturated sterols against phospholipid peroxidation and ferroptosis. Using 7-DHC and ergosterol as two representatives of this class of sterols, we show that the specific and robust protection against phospholipid peroxidation is a feature that is not limited to only mammalian cells but is shared between biologically distant organisms.

This discovery in itself poses a paradox: a lipid frequently reported to propagate radical chain reaction25,41 is capable of suppressing a cell death modality known to exclusively depend on these same biochemical events15,42. Initial studies have firmly established that, in cells undergoing ferroptosis, phospholipids furnished with PUFAs are the prime target for oxidation14. We now expand on this concept as we provide ample evidence supporting their role in ferroptosis execution through the formation of membrane destabilizing truncated phospholipid species. These observations thus imply that lipid peroxidation can be uncoupled from cell death as only products of phospholipid peroxidation generate efficient ferroptosis-inducing metabolites. By having established this interrelationship, these seemingly paradoxical findings can now be rationalized. Molecules able to suppress the peroxidation of fatty acids esterified into phospholipid species need to efficiently outcompete PUFAs during lipid peroxidation and stabilize radical chain propagating species. Mechanistically, in isotropic solution, PUFAs have reported propagation rate constants (kp) ranging from 62 in linoleic acid up to 197 M−1 s−1 for arachidonic acid, both of which can be easily outcompeted by 7-DHC given its extremely high (kp) of 2,260 M−1 s−1. This renders 7-DHC a superior phospholipid shield when compared to other sterols (cholesterol (kp) = 11 and lathosterol (kp) = 57 M−1 s−1). Despite 7-DHC and other B-ring-unsaturated sterols being principal contributors and targets of lipid peroxidation, radicals derived from these sterol metabolites are poor inducers of cell death, unlike radicals in phospholipids that can give rise to membrane-destabilizing truncated species.

Together with the accompanying paper20, we demonstrate that this process can be exploited to suppress ferroptosis in different settings. Specifically, inspired by the report that rare mutations in DHCR7 have been reported in both patients with Burkitt’s lymphoma and patients with neuroblastoma, we demonstrate that the accumulation of 7-DHC can lead to a more aggressive phenotype in xenograft models relevant to both entities, thus presenting a potential compensatory mechanism for their intrinsic sensitivity to ferroptosis. This recognition could be relevant as recent reports have indicated that amplification of MYC and MYCN increase sensitivity to ferroptosis43,44,45,46 and extra mechanism preventing ferroptosis in these oncogenic contexts could enhance cancer cell fitness. The ferroptosis-modulating activity of 7-DHC raises another noteworthy aspect; several recent screening studies have identified a series of FDA-approved drugs able to inhibit DHCR7 at nM concentrations47. For example, trazodone is prescribed more than 20 million times a year in the USA, sometimes off-label as a sleep aid; studies of patients on this drug have reported increased plasma concentrations of 7-DHC48. Epidemiological studies will be required to explore whether there are any groups of patients who regularly consume ferroptosis-modulating drugs and whether this has any impact on cancer incidence, metastasis occurrence or other public health-relevant aspects.

Interestingly, some organisms seem to have shifted away from this strategy. Specifically, DHCR7-like enzymes convert 5,7-unsaturated sterols to the less autoxidizable sterols, such as cholesterol, thus keeping the concentration of B-ring unsaturated sterols low. The replacement of 7-DHC with cholesterol in humans offers clear benefits; this is documented by the causative role of DHCR7 mutations and the developmental syndrome known as Smith–Lemli–Opitz syndrome. This syndrome is characterized by varying levels of neurodevelopmental defects depending on the severity of the mutation. However, our findings reveal a paradoxical aspect. Whereas previous studies have shown that oxidation products of 7-DHC are toxic to neuronal cells3 and can suppress key (neuro)developmental pathways like the Wnt/β-catenin49 and Hedgehog50 signal pathways, our study presents a contrasting perspective. We have observed the accumulation of 7-DHC oxidation formed during the process of preventing phospholipid peroxidation in cancer cells exposed to oxidants, such as conditions that induce ferroptosis. These findings emphasize the complex and context-dependent nature of 7-DHC and its oxidation products in different cellular contexts.

Ultimately, the mechanisms described here shed light in an unrecognized and primitive tolerance mechanism toward phospholipid peroxidation that could be highjacked by cancer cells to evade ferroptosis.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.