Chaperone-mediated autophagy is involved in the execution of ferroptosis

Triterpenoid CDDO Inhibits both Necroptosis and Ferroptosis.

To identify new regulators of necroptosis, we performed a chemical screen with a library of ∼600 known bioactive compounds. The screen was carried out using a hippocampal-derived HT-22 cell line treated with a combination of TNF-α, zVAD.fmk, and the small molecule inhibitor of apoptotic proteins antagonist SM-164 (5). In this screen, we identified the triterpenoid compound CDDO (also known as bardoxolone) (SI Appendix, Fig. S1A) as a potent inhibitor of necroptosis in a time-dependent manner (Fig. 1A). CDDO inhibited necroptosis of HT-22 cells dose-dependently, with an EC₅₀ of 10 μM (Fig. 1B). We tested the ability of CDDO to block necroptosis in additional cell models and found that CDDO effectively inhibited necroptosis of mouse neural cell line 661W cells and human colon cancer cell line HT-29 cells induced by TNF-α/zVAD/SM-164, as well as necroptosis of human leukemia FADD-deficient Jurkat cells induced by TNF-α only, TNF-α/SM-164, or TNF-α/cycloheximide (CHX) (SI Appendix, Fig. S1 B–D).

We next investigated the effect of CDDO on other forms of cell death, including ferroptosis induced by glutamate and erastin, TNF-α–induced apoptosis, DNA damage-induced apoptosis, and oxidative stress-induced necrosis. Interestingly, we found that CDDO could also inhibit ferroptosis of HT-22, 661W, and human HT-1080 cells induced by glutamate and erastin, with EC₅₀s of 5.5 μM and 6.8 μM, respectively (Fig. 1 C–E and SI Appendix, Fig. S1 E and F). The inhibitory effect of CDDO on both ferroptosis and necroptosis was further confirmed using a propidium iodide (PI) uptake assay (Fig. 1F). However, the addition of CDDO sensitized HT-22 cells to apoptosis induced by TNF-α, TNF-α/SM-164, and staurosporine, while it had no effect on apoptosis induced by various agents that included TNF-α–mediated RIPK1-independent apoptosis induced by TNF-α/CHX, DNA damage induced apoptosis by doxorubicin or etoposide, ER stress-induced apoptosis by thapsigargin or tunicamycin, and necrosis induced by oxidative stress using tert-butyl hydroperoxide (t-BOOH) or H₂O₂ (SI Appendix, Fig. S1G). Thus, CDDO can inhibit both necroptosis and ferroptosis, two necrotic cell death pathways that have no known common regulatory mechanism, but not apoptosis.

Since ferroptosis is sensitive to cellular redox status (6), we next explored if the effect of CDDO on cell death was due to its intrinsic antioxidant potential. We performed a 2, 2-diphenyl-1-picrylhydrazyl (DPPH) assay to monitor the antioxidant activity of CDDO. We found that unlike other inhibitors of ferroptosis such as ferrostatin-1 (Fer-1), which is known to display a strong antioxidant effect with high free radical scavenging activity (33), CDDO did not demonstrate any antioxidant activity (SI Appendix, Fig. S1H). Thus, CDDO is a chemical inhibitor of ferroptosis and necroptosis but not an antioxidant, and it may be useful as a probe to investigate the mechanism that is in common between necroptosis and ferroptosis.

CDDO Inhibits the Activation of RIPK1 Kinase Activity in Necroptosis.

Next, we characterized the effects of CDDO on the biochemical hallmarks of necroptosis and ferroptosis. We first analyzed the effect of CDDO on specific phosphorylation events on RIPK1, RIPK3, and MLKL, including p-S166–RIPK1, p-T231/S232–RIPK3, and p-S345–mMLKL/p-S358–hMLKL, the biomarkers of their activation, during necroptosis (34–36). We found that the addition of CDDO strongly inhibited the induction of p-S166–RIPK1, p-RIPK3, and p-MLKL of HT-22 cells and HT-29 cells during necroptosis induced by TNF-α/SM-164/zVAD.fmk (Fig. 2A and SI Appendix, Fig. S2A).

Since RIPK1 is an upstream activator of RIPK3 and MLKL, we next tested if the formation of complex IIb, defined by the interacting complex of RIPK1 and RIPK3, was disrupted by CDDO. As shown in Fig. 2B, CDDO strongly inhibited the interaction of RIPK1 and RIPK3 in HT-22 cells stimulated by TNF-α/SM-164/zVAD.fmk. These results suggest that CDDO inhibits the activation of RIPK1 kinase, which, in turn, blocks the formation of complex IIb, as well as the phosphorylation of RIPK3 and MLKL.

To explore the mechanism by which CDDO inhibits RIPK1 kinase activity, we next investigated the effect of CDDO in TNFR1 signaling complex (TNF-RSC, complex I) in TNF-α–stimulated cells. RIPK1 was rapidly recruited into TNF-RSC within 5 min of TNF-α stimulation, which was not affected by the cotreatment with CDDO (SI Appendix, Fig. S2B). Consistent with the report that CDDO could inhibit the activity of inhibitor of NF-κB kinase subunit β (IKKβ) (37), we found that the treatment with CDDO slightly reduced the phosphorylation of IKK-α/β in complex I (SI Appendix, Fig. S2B) and the activation of NF-κB, as the phosphorylation and degradation of inhibitor of NF-κB (IκBα) were also only slightly inhibited (SI Appendix, Fig. S2C). These results suggest that CDDO does not have a strong effect on the activation of NF-κB.

Since CDDO was highly effective in inhibiting the activation of RIPK1, we next tested if CDDO may inhibit the activation of RIPK1 directly. Interestingly, we found that CDDO markedly inhibited the phosphorylation of S166 RIPK1, a biomarker for activated RIPK1 (8, 38), on overexpressed RIPK1 in 293T cells (Fig. 2C). Moreover, we found that CDDO inhibited RIPK1 kinase activity in HT-22 cells induced by different necroptosis inducers (Fig. 2D). However, CDDO could not inhibit RIPK1 activity in an in vitro kinase assay, different from R-7-Cl-O-Nec-1 (Nec-1s) (SI Appendix, Fig. S2D). Based on these results, we concluded that CDDO was an indirect inhibitor of RIPK1 kinase activity.

CDDO Inhibits Ferroptosis by Blocking GPX4 Degradation.

To explore the mechanism by which CDDO inhibits ferroptosis, we analyzed the effect of CDDO on the biomarkers of ferroptosis, including reactive oxygen species (ROS) accumulation, lipid peroxidation, and GPX4 degradation. We first characterized the effect of CDDO on ROS levels in the HT-22 cell lines using 5-(and-6)-carboxy-2′,7′-dichlorodihydrofluorescein diacetate (carboxy-H₂DCFDA), a cytosolic ROS sensor. Interestingly, the induction of ROS in ferroptosis by erastin was significantly inhibited in the presence of CDDO (Fig. 3A).

Lipid peroxidation has been identified to be directly involved in mediating necrosis and ferroptosis (17). We next investigated whether CDDO affected the production of malondialdehyde (MDA; an end product of lipid peroxidation) in ferroptosis induced by glutamate and erastin. We found that CDDO was highly effective in inhibiting the production of MDA, similar to that of positive control Fer-1 (Fig. 3B). These results suggested that CDDO acted upstream of lipid peroxidation.

Degradation of GPX4, a critical inhibitor of lipid peroxidation, is important for promoting lipid peroxidation in ferroptosis (39). We next investigated the effect of CDDO on GPX4 degradation during ferroptosis. We found that the treatment of HT-22 cells with ferroptosis inducers such as erastin and glutamate led to a reduction in the levels of GPX4 protein, which was blocked in the presence of CDDO (Fig. 3C and SI Appendix, Fig. S3A). In addition, inhibition of GSH production by L-buthionine-sulfoximine (BSO) can, in turn, impair GPX4 activity and promote ferroptosis (40). The addition of CDDO was also able to protect ferroptosis and inhibit the degradation of GPX4 induced by BSO (Fig. 3 D and E). On the other hand, ferroptosis can also be induced directly by knocking down GPX4 (39). However, unlike Fer-1, CDDO could not inhibit ferroptosis induced by direct knockdown of GPX4 (Fig. 3F and SI Appendix, Fig. S3B). Since the activity of GPX4 is crucial for preventing ferroptosis, CDDO cannot inhibit type 2 ferroptosis induced by RSL3, a covalent inhibitor of GPX4, even though the protein level of GPX4 was also restored in the presence of CDDO (SI Appendix, Fig. S3 C and D). Thus, restoring the levels of GPX4 when its activity is inhibited by a covalent inhibitor, RSL3, is not sufficient to block ferroptosis. This result suggests that CDDO protects erastin/glutamate-induced ferroptosis by preserving physiological levels of active GPX4.

To determine if CDDO affects the transcription level of GPX4, we measured the levels of GPX4 mRNA and confirmed that induction of ferroptosis by glutamate/erastin had no effect on the levels of GPX4 mRNA. Moreover, the addition of CDDO also had no influence on GPX4 mRNA levels in HT-22 cells with or without glutamate/erastin (SI Appendix, Fig. S3E).

Taken together, we conclude that CDDO is a potent inhibitor of ferroptosis that can block GPX4 degradation, lipid peroxidation, and ROS accumulation.

CDDO Targets HSP90 to Inhibit both Necroptosis and Ferroptosis.

CDDO has been shown to inhibit multiple proteins, including PPAR-γ, ErbB2, Akt, STATs, Keap1, and HSP90 (41). We next evaluated the relevance of these targets in necroptosis and ferroptosis using their inhibitors directly. Among all of the inhibitors we tested, we found that only Keap1 inhibitor tertiary butylhydroquinone (t-BHQ), which can activate the nuclear factor-erythroid 2 p45-related factor 2 (Nrf2)–ARE system, and HSP90 inhibitor tanespimycin (17AAG) could present good inhibitory effect on both necroptosis and ferroptosis (Fig. 4A). T0070907 (PPAR-γ antagonist), troglitazone (PPAR-γ agonist), TPCA-1 (IKK-β inhibitor), JAK_i (Jak inhibitor), STAT3i (STAT inhibitor), AV-412 (EGFR inhibitor), and MK2206 (Akt inhibitor) inhibited ferroptosis but not necroptosis (SI Appendix, Fig. S4A).

Nrf2, a basic leucine zipper redox-sensitive transcription factor and a master regulator of cellular antioxidant response, has been shown to be a target for CDDO (42). However, in cells with Nrf2 knockdown, we found that CDDO could still protect necroptosis and ferroptosis (Fig. 4 B and C). Since Nrf2 is a transcription factor, we used transcription and translation inhibitors to block protein synthesis, and found that CDDO still protected cell death induced by TNF-α/SM-164/zVAD.fmk (SI Appendix, Fig. S4B). We also evaluated Keap1, the cofactor of Nrf2, and found that other Keap1 inhibitors such as sulforaphane (SFN) and dimethyl fumarate (DMF) inhibited necroptosis but not ferroptosis (SI Appendix, Fig. S4C). Moreover, Keap1 inhibitors cannot protect necroptosis in HT-29 cells (SI Appendix, Fig. S4 D and E). In addition, as Keap1 inhibition leads to the accumulation of polyubiquitylated Nrf2, we checked the Nrf2 induction effect of these compounds. In SI Appendix, Fig. S4F, we found that CDDO is not a potent Nrf2 inducer compared with other Keap1 inhibitors. Thus, we conclude that Nrf2 is not the common regulatory nodal for CDDO’s effect on necroptosis and ferroptosis.

HSP90 is a highly evolutionarily conserved chaperone protein and plays a critical role in the maturation, stability, and activation of a number of diverse client proteins. Since HSP90 is known to be associated with RIPK1 (29, 30), we next considered the role of HSP90 as the target of CDDO in blocking both necroptosis and ferroptosis. As shown in Fig. 4 D and E, knockdown of HSP90 in HT-22 cells inhibited necroptosis induced by TNF-α/SM-164/zVAD.fmk and TNF-α/CHX/zVAD.fmk and ferroptosis induced by glutamate and erastin. Thus, HSP90 is involved in mediating both necroptosis and ferroptosis.

We next examined if CDDO could inhibit HSP90 in cells by testing several biomarkers of HSP90 inhibition. Consistent with an inhibitor of HSP90, we found that CDDO induced remarkable increases in the levels of HSP70 and reduction in the levels of a well-known HSP90 client protein, EGFR, in a dose-dependent manner (Fig. 4F), which were both known to occur upon the inhibition of HSP90 (43). Furthermore, we found that CDDO significantly disrupted the interactions between HSP90 and its client proteins, including RIPK1 (Fig. 4G).

These data suggest that CDDO inhibits HSP90 activity in cells and HSP90 may be a common regulatory nodal for CDDO to block both necroptosis and ferroptosis.

CDDO Inhibits RIPK1 Kinase Activity Through HSP90.

A previous study reported that treatment with HSP90 inhibitors for long periods of time (>10 h) reduced levels of RIPK1 (44). While treatment with CDDO for long time (>10 h) also reduced the levels of RIPK1 in HT-22 cells (SI Appendix, Fig. S5A), shorter treatment with CDDO (3–5 h), as used in our experimental conditions, led to no obvious difference in the levels of RIPK1 (Fig. 2). Thus, we next examined whether HSP90 could regulate RIPK1 kinase activity. We characterized the effects of two HSP90 inhibitors, geldanamycin (GA) and 17AAG, in an RIPK1 overexpression assay system. We found that the activation of RIPK1 kinase, marked by p-S166, upon its overexpression in 293T cells was dose-dependently inhibited by CDDO as well as by GA and 17AAG, which were the most commonly used HSP90 inhibitors, while the levels of RIPK1 protein were not affected (Fig. 5A).

To test if CDDO might inhibit RIPK1 through HSP90, we constructed an HT-29 cell line that overexpressed HSP90. Consistent with HSP90 being a positive regulator of RIPK1 kinase activity, the cells overexpressing HSP90 showed enhanced RIPK1 kinase activity upon induction of necroptosis by TNF-α/SM-164/zVAD.fmk (Fig. 5B). Cotreatment of CDDO with TNF-α/SM-164/zVAD.fmk effectively inhibited the activation of RIPK1 with or without overexpression of HSP90, suggesting that CDDO plays an inhibitory role through HSP90 on RIPK1 kinase activity (Fig. 5B).

Since HSP90 functions by promoting the structural stability of its client proteins, inhibition of HSP90 may lead to their misfolding and degradation (43). Consistently, we found that the treatment with CDDO reduced the protein levels of RIPK1 in the Nonidet P-40 soluble fraction and increased its presence in the Nonidet P-40 insoluble fraction (Fig. 5C), suggesting that inhibition of HSP90 may lead to conformation changes in RIPK1 that lead to inhibition of its kinase activity.

As reported, treatment with TNF-α/SM-164 induces RIPK1-dependent apoptosis (RDA), which can be inhibited by Nec-1s (45). We then investigated the effect of CDDO on RDA. Treatment with CDDO also blocked the activation of RIPK1 kinase induced by treatment of TNF-α/SM-164 and TNF-α/5z7 (Fig. 5D). However, treatment with CDDO sensitized cells to RDA induced by TNF-α/SM-164 or TNF-α/5z7 (SI Appendix, Figs. S1G and S5 B and C) and increased the cleavage of caspase-8 (Fig. 5D). Since inhibition of HSP90 is known to induce apoptosis (46) and RIPK1 is cleaved by caspase-8 (47), blocking the activation of RIPK1 by HSP90 inhibitors is not sufficient to inhibit RDA.

Degradation of GPX4 by CMA Pathway During Ferroptosis.

Degradation of GPX4 protein is a pivotal event in ferroptosis, which, in turn, promotes ROS and irreversible lipid peroxidation and, consequently, cell death (48). Since we found that CDDO blocked the degradation of GPX4 in ferroptosis induced by glutamate or erastin, we next investigated the mechanism by which HSP90 regulated the degradation of GPX4 using several protein degradation inhibitors for different types of protein degradation, including proteasome inhibitors (MG-132 and PS-341) and lysosome inhibitors [chloroquine (CQ), NH₄Cl, and Baf-A1]. We found that only the lysosome pathway inhibitor CQ, NH₄Cl, or Baf-A1 could inhibit both cell death and GPX4 degradation induced by glutamate or erastin (Fig. 6 A and B).

Since HSP90 regulates stability and activity of its client proteins, we next used mass spectrometry to determine which client protein(s) of HSP90 might be involved in the degradation of GPX4. We generated a line of 661W cells stably expressing Flag-GPX4. Using this cell line, we analyzed the binding proteins of GPX4 by mass spectrometry. We found that among the proteins binding with GPX4, the levels of both HSC70 and HSP90 were increased after erastin treatment (Fig. 6C). The interactions of HSC70 with HSP90 and GPX4 with HSC70 were confirmed using coimmunoprecipitation (Fig. 6 D and E). Furthermore, we found that the interaction of HSC70 and HSP90 was increased in cells treated with erastin, which was reduced by CDDO (Fig. 6D).

HSC70 is known to be associated with Lamp-2a to mediate CMA (49). We next examined the effect of HSC70 and Lamp-2a in ferroptosis. We found that knockdown of Lamp-2a and HSC70 significantly reduced ferroptosis and restored GPX4 protein levels in cells treated with erastin (Fig. 6 F and G and SI Appendix, Fig. S6). As Lamp-2a and HSC70 are required for the protein degradation through CMA in the lysosome compartment, these results are consistent with lysosome inhibitors inhibiting the GPX4 degradation induced by glutamate or erastin.

Overexpression of Lamp-2a can promote CMA (50). Consistently, we found that overexpression of Lamp-2a sensitized cells to ferroptosis induced by glutamate and erastin (Fig. 6H). Consistent with the role of CMA in promoting the degradation of GPX4, we found that the level of GPX4 protein in cells with overexpressed Lamp-2a was lower than that of GFP-expressing cells upon erastin treatment (Fig. 6I).

We next measured the effect of CMA on ferroptosis using a photoactivatable (PA) reporter fused with a CMA-targeting motif (KFERQ-PA-mCherry) (51). Activation of CMA mobilizes this mCherry-tagged CMA substrate from the cytosol to lysosomes, which can be tracked as a change in the reporting fluorescence from diffuse signals to a punctate pattern. Treatment with erastin markedly increased the signals of this CMA reporter in HT1080 cells (Fig. 6J). Thus, CMA is activated in cells undergoing ferroptosis.

CDDO Protects Ferroptosis by Inhibiting CMA Pathway.

HSC70 is known to recognize and bind with substrates to deliver to the lysosome membrane, while Lamp-2a acts as a receptor to transport CMA substrates into the lysosome matrix for degradation (49, 50). We found that induction of ferroptosis by erastin, glutamate, and RSL3 all led to increases in the levels of Lamp-2a in different cell lines (Fig. 7 A–C and SI Appendix, Fig. S7 A–C), and that the increased levels of Lamp-2a were inhibited by the addition of CDDO (Fig. 7B and SI Appendix, Fig. S7 A and B). HSP90 has been shown to associate with Lamp-2a at the lysosomal membrane and to regulate the functional dynamics of the Lamp-2a complexes for CMA activation (52). Consistently, the interaction between HSP90 and Lamp-2a was induced by erastin and reduced by treatment with CDDO, suggesting that erastin may affect the interaction of HSP90 and Lamp-2a (Fig. 7D). HSP90 may also regulate the stability of Lamp-2a, as the treatment of CDDO also reduced the levels of Lamp-2a in control cells (SI Appendix, Fig. S7D).

HSC70 and Lamp-2a interact with CMA substrates to promote their degradation (52). We found that the interactions between GPX4 and HSC70 or Lamp-2a were markedly increased in the Flag-GPX4-661W cells after erastin treatment (Fig. 7C). CMA substrates are known to contain a KEFRQ-like motif that can be recognized and bound with HSC70 (53). Indeed, we found that GPX4 harbored several pentapeptide sequences (¹²⁴NVKFD¹²⁸, ¹⁶⁹LIDKN¹⁷³, and ¹⁸⁷QVIEK¹⁹¹) that were consistent with HSC70 recognition motifs. To functionally determine the importance of these motifs in GPX4 for its degradation during the ferroptosis process, we generated GPX4 mutants with a 2-aa mutation in the hypothetical CMA motifs (¹²⁴NVKFD¹²⁸ to ¹²⁴AAKFD¹²⁸, ¹⁶⁹LIDKN¹⁷³ to ¹⁶⁹LIDAA¹⁷³, and ¹⁸⁷QVIEK¹⁹¹ to ¹⁸⁷AAIEK¹⁹¹). We found that GPX4 with mutated motifs ¹²⁴NVKFD¹²⁸ or ¹⁸⁷QVIEK¹⁹¹, which were marked as NVKFD* or QVIEK*, showed increased resistance to erastin-induced degradation (Fig. 7E). To further confirm the role of the KEFRQ-like motif, we generated a double-motif mutation. As shown in Fig. 7F, when both the ¹²⁴NVKFD¹²⁸ and ¹⁸⁷QVIEK¹⁹¹ motifs were mutated, degradation of this GPX4 double mutant was significantly impaired after erastin treatment. Moreover, this GPX4 double mutant failed to bind to HSC70 when overexpressed together in 293T cells (Fig. 7G).

To further verify the degradation of GPX4 in the lysosomes, we isolated lysosome-enriched fractions of cells treated with erastin. A portion of GPX4 was associated with lysosomal factions under the control condition. After treatment with erastin, the lysosomal fractions of HT-22 cells showed a marked enrichment of GPX4 and an increased amount of Lamp-2a (Fig. 7H), demonstrating the lysosomal targeting of GPX4 upon CMA activation.

Since the increased levels of Lamp-2a are predicted to promote CMA (49, 50), we tested the possibility that ferroptosis may promote the degradation of other CMA substrates. Indeed, we found that the protein levels of GAPDH, one of the well-established CMA substrates (54), was also reduced in cells undergoing ferroptosis induced by erastin. Furthermore, the treatment with CDDO significantly restored the protein levels of GAPDH (Fig. 7I).

Taken together, we conclude that activation of ferroptosis leads to increases in the levels of Lamp-2a to promote CMA, which, in turn, mediate the degradation of multiple CMA substrates, including GPX4. Thus, the activation of CMA is involved in mediating ferroptosis (Fig. 7J).

Reagents and Antibodies.

Recombinant human TNF-α (catalog no. C008) was purchased from Novoprotein Scientific, Inc. The 17AAG, GA, and erastin were purchased from TargetMol. MG-132 and Fer-1 were purchased from Selleckchem. Carboxy-H₂DCFDA and 5Z-7-oxozeaenol were purchased from Sigma–Aldrich. CDDO, TPCA-1, t-BHQ, MK2206, AV-412, T0070907, and troglitazone were purchased from MedChemExpress. Nec-1s, zVAD.fmk, and SM-164 were custom-synthesized. The following commercial antibodies were used in this study: RIPK1 (3493; Cell Signaling Technology); p-S166–RIPK1 (Biolynx); p-S345–MLKL (ab196436; Abcam); p-S358–MLKL (91689; Cell Signaling Technology); TNFR1 (AF-425-PB; R&D Systems); Myc (C3956; Sigma–Aldrich); GPX4 (ab125066; Abcam); HSP90-α (13171-1-AP; Proteintech); HSP90-β (ADI-SPA-845-F; Enzolife); Lamp-2a (51-2200; Invitrogen); HSC70 (10654-1-AP; Proteintech); and p-T231/S232–RIPK3, IκBα, p-IκBα, IKK-α, IKK-β, cl-caspase-8, and Nrf2 (Cell Signaling Technology). All of the siRNA oligos were ordered from GenePharma.

Cell Culture.

HT-22, 661W, HT-1080, and HEK293T cells were cultured in DMEM (Gibco) with 10% (vol/vol) FBS (Gibco) and 100 units/mL penicillin/streptomycin. Jurkat cells were cultured in RPMI-1640 (Gibco) with 10% (vol/vol) FBS (Gibco) and 100 units/mL penicillin/streptomycin. HT-29 cells were cultured in McCoy’s 5A (Gibco) with 10% (vol/vol) FBS (Gibco) and 100 units/mL penicillin/streptomycin. All cells were cultured at 37 °C with 5% CO₂.

CellTiter-Glo Assay and PI Staining for Cell Survival.

The CellTiter-Glo luminescent cell viability kit was from Promega. The viability of cells was normalized to that of control cells treated with vehicle alone. For the PI staining, 1 μM PI and 10 μg/mL Hoechst stain were added to cells for 10 min at room temperature and imaged using a Leica fluorescence microscope. All cell survival assays were performed in triplicate.

Immunoprecipitation.

Cells were lysed with Nonidet P-40 buffer [150 mM NaCl, 50 mM Tris⋅HCl (pH 7.4), 1 mM EDTA, 1 mM EGTA, 0.5% Nonidet P-40, 5% glycerol] supplemented with 1 mM PMSF, 1× protease inhibitor mixture (Roche), 10 mM β-glycerophosphate, 5 mM NaF, and 1 mM Na₃VO₄. Debris was precipitated, and the lysate was incubated with an antibody overnight at 4 °C. The immunocomplex was captured by Protein A/G Agarose (Life Technologies) with the appropriate antibodies for 3–5 h at 4 °C. Beads were washed four times, and the immunocomplex was eluted from beads by loading buffer.

In Vitro Kinase Assays.

HEK293T cells were transfected with Flag-tagged RIPK1 expression plasmids for 24 h and immunoprecipitated with anti-Flag antibody. The immunoprecipitated complexes were divided into equal parts and incubated in 30 μL of kinase assay buffer [20 mM Hepes (pH 7.5), 2 mM DTT, 10 mM MgCl₂] supplemented with 50 μM cold ATP for 30 min at 30 °C in the presence of different compounds. The reaction was stopped by adding SDS/PAGE loading buffer and heating at 95 °C for 5 min.

MDA Assay.

The relative MDA concentration in cell lysates was assessed using an MDA Assay Kit (Beyotime) according to the manufacturer’s instructions.

DPPH Assay.

The stable radical DPPH was dissolved in methanol to a final working concentration of 0.05 mM. One milliliter of DPPH solution was added to a small volume (<5 μL) of each test compound dissolved in DMSO. The final concentration of each test compound was 0.05 mM. Samples were inverted several times and allowed to incubate at room temperature for 30 min. Samples were then aliquoted to white, 96-well, solid-bottomed dishes (Corning), and the absorbance at 517 nm was recorded. All values were normalized to background (methanol only).

KFERQ-PA-mCherry1 Reporter Assay.

The pLVX-KFERQ-PA-mCherryN1 plasmid was a gift from Guoqiang Chen and Qian Zhao of Shanghai Jiao Tong University School of Medicine, Shanghai, China (51). HT-1080 cells were stably infected with pLVX-KFERQ-PA-mCherryN1 lentivirus. For CMA activity analysis, cells were photoactivated by a 405/20-nm LED array, followed by treatment, and were then fixed and costained with DAPI. The images were acquired using a Leica SP8 confocal microscope. Activation of CMA activity was calculated as the number of cells that present bright puncta under a 20× or 63× objective and were quantified using ImageJ software (NIH).

Lysosome Isolation.

Lysosome fractions of cultured cells were obtained using homogenization and sequential centrifugation. Lysosomes were isolated using a Lysosome Enrichment Kit (Pierce) according to the kit protocol.

Mass Spectrometry.

The 661w-Flag-GPX4 cells were treated with 10 μM erastin for 18 h and harvested with 0.5% Nonidet P-40 buffer. The cell lysates were immunoprecipitated with anti-Flag antibody. The proteins were trypsin-digested and analyzed on an Orbitrap Q Exactive mass spectrometer (Thermo Scientific). The protein identification and label-free quantification were done using MaxQuant software. The tandem mass spectra were searched against the UniProt mouse protein database and a set of commonly observed contaminants. The precursor mass tolerance was set as 20 ppm, and the fragment mass tolerance was set as 0.1 Da. The cysteine carbamidomethylation was set as a static modification, and the methionine oxidation was set as a variable modification. The false discovery rates at the peptide spectrum match level and protein level were controlled below 1%.

Statistics.

Data are expressed as the mean ± SEM. Error bars indicate the SEM. Pairwise comparisons between two groups were performed using the Student’s t test. Differences were considered statistically significant if *P < 0.05, **P < 0.01, ***P < 0.001, or ****P < 0.0001, or as not significant. At least three independent biological repeats were included in each data point. Each experiment was repeated at least three times.