Phase separation of FSP1 promotes ferroptosis – Nature

Chemicals

Lip-1 (Selleckchem, cat. no. S7699), doxycycline (Dox; Sigma, cat. no. D9891), RSL3 (Cayman, cat. no. 19288), BSO (Sigma, cat. no. B2515), iFSP1 (ChemDiv, cat. no. 8009-2626), icFSP1 (ChemDiv, cat. no. L892-0224 or custom synthesis by Intonation Research Laboratories), erastin (Merck, cat. no. 329600), ML210 (Cayman, cat. no. Cay23282-1), FIN56 (Cayman, cat. no. Cay25180-5), FINO2 (Cayman, cat. no. Cay25096-1), deferoxamine mesylate salt (DFO; Sigma, cat. no. 138-14-7), ferrostatin-1 (Fer-1; Sigma, cat. no. SML0583), zVAD-FMK (zVAD; Enzo Life Sciences, cat. no. ALX-260-02), necrostatin-1s (Nec-1s; Enzo Life Sciences, cat. no. BV-2263-5), MCC950 (Sigma, cat. no. 5381200001), olaparib (Selleckchem, cat. no. S1060), staurosporine (STS; Cayman, cat. no. 81590), recombinant human tumour necrosis factor (TNF; R&D Systems, cat. no. NBP2-35076), Smac mimic (BV-6; Selleckchem, cat. no. S7597), nigericin (Thermo Fisher, cat. no. N1495), IMP-1088 (Cayman, cat. no. Cay25366-1) and lipopolysaccharide (LPS; Sigma, cat. no. L2880) were used in this study.

Mice

Five- to six-week-old female C57BL/6J and athymic nude mice were obtained from Charles River, and 6- to 7-week-old mice were used for the experiments. All mice were kept under standard conditions with water and food provided ad libitum and in a controlled environment (22 ± 2 °C, 55% ± 5% humidity, 12-h light/12-h dark cycle) in the Helmholtz Munich animal facility under SPF-IVC standard conditions. All experiments were performed in compliance with the German Animal Welfare Law and were approved by the institutional committee on animal experimentation and the government of Upper Bavaria (ROB-55.2-2532.Vet_02-17-167).

Cell lines

TAM-inducible Gpx4-knockout mouse immortalized fibroblasts (referred to as Pfa1 cells) were reported previously¹⁸. Genomic Gpx4 deletion can be achieved by TAM-inducible activation of Cre recombinase using the CreER^T2/loxP system. HT-1080, 786-O, A375, NCI-H460 (H460), MDA-MB-436, HT-29, B16F10, and 4T1 cells were purchased from ATCC. THP-1 cells were obtained from DSMZ). Human PBMC cells were purchased from Tebu-bio (cat. no. 088SER-PBMC-F). SUDHL5, SUDHL6, DOHH2 and OCI-Ly19 cells were a gift from S. Hailfinger. Rat1 cells were a gift from Medizinische Hochschule Hannover. Pfa1, HT-1080, 786-O, A375, HT-29, Rat1 and B16F10 cells were cultured in high-glucose DMEM (4.5 g l^–1 glucose) with 10% FBS, 2 mM l-glutamine and 1% penicillin-streptomycin. H460, MDA-MB-436, THP-1, PBMC, SUDHL5, SUDHL6, DOHH2 and 4T1 cells were cultured in RPMI GlutaMax with 10% FBS and 1% penicillin-streptomycin. OCI-Ly19 cells were cultured in IMDM with 10% FBS and 1% penicillin-streptomycin. To generate cell lines with stable overexpression, appropriate antibiotics (1 µg ml^–1 puromycin, 10 µg ml^–1 blasticidin and 0.5–1.0 mg ml^–1 G418) were used. GPX4-knockout human A375 and H460 cells were cultured in the presence of 1 µM Lip-1 for maintenance. All cells were cultured at 37 °C with 5% CO₂ and verified to be negative for mycoplasma.

Cell viability assays

Cells were seeded on 96-well plates and cultured overnight. All cell number conditions for each cell line are described in Supplementary Table 1. The next day, the medium was changed to medium containing the following compounds: RSL3, ML210, erastin, FIN56, FINO2, BSO, iFSP1, icFSP1, Lip-1, DFO, Fer-1, zVAD, Nec-1s, MCC950, olaparib, STS, TNF, Smac mimic or nigericin at the indicated concentrations. For TAM and Dox treatment, cells were seeded with compounds at the same time. Cell viability was determined 1 h (for nigericin), 24–48 h (for RSL3, ML210, erastin, FIN56, FINO2, iFSP1, icFSP1, STS, TNF, Smac mimic and zVAD) or 72 h (for BSO, icFSP1, TAM and Dox) after the start of treatment using AquaBluer (MultiTarget Pharmaceuticals, cat. no. 6015) as an indicator of viable cells according to the manufacturer’s protocol. For apoptosis induction, HT-1080 cells were incubated with different concentrations of STS for 24 h. For necroptosis induction, HT-29 cells were incubated with different concentrations of TNF with Smac mimic (400 nM) and zVAD (30 µM) for 24 h. For pyroptosis induction, THP-1 cells stimulated with LPS (1 µg ml^–1, 2 h) were incubated with nigericin for 1 h. For ferroptosis induction, cells were incubated with ferroptosis inducers for 24–72 h.

As readout, fluorescence was measured at an excitation/emission wavelength of 540/590 nm using a SpectraMax M5 microplate reader with SoftMax Pro v.7 (Molecular Devices) after 4 h of incubation with AquaBluer in normal cell culture conditions. The relative cell viability (%) was calculated as follows: (fluorescence of samples – background)/(fluorescence of appropriate control samples – background) × 100.

LDH release assays

Cells were seeded on 96-well plates and cultured overnight. The next day, the medium was changed to medium containing compounds and cells were incubated for another 24 h. Cell death rates were measured using the Cytotoxicity Detection Kit (LDH) (Roche, cat. no. 11644793001) in principle following the manufacturer’s protocol. In brief, cell culture supernatant was collected as a medium sample. Cells were then lysed with PBS containing 0.1% Triton X-100 as a lysate sample. The medium and lysate samples were individually mixed with reagents on microplates, and the absorbance was measured at 492 nm using a SpectraMax M5 microplate reader after incubation for 15–30 min at room temperature. The cell death ratio was calculated by LDH release (%) as follows: (absorbance of medium samples – background)/((absorbance of lysate samples – background) + (absorbance of medium samples – background)) × 100.

Screening of FSP1 inhibitors

Wild-type and Gpx4-knockout Pfa1 cells stably overexpressing hFSP1 were seeded on separate 384-well plates (500 cells per well) and screened with a library of small molecule inhibitor compounds as reported previously⁵. Cell viability of the different cell lines was assessed 48 h after the start of treatment using AquaBluer. Compounds showing selective lethality in Gpx4-knockout Pfa1 cells stably overexpressing hFSP1 were then validated in cell viability assays and in vitro FSP1 enzymatic assays.

Lipid peroxidation assays

For lipid peroxidation assays, 100,000 cells per well were seeded on a 12-well plate 1 day before the experiments. The next day, cells were treated with 2.5 µM icFSP1 for 3 h and then incubated with 1.5 µM C11-BODIPY 581/591 (Invitrogen, cat. no. D3861) for 30 min in a 5% CO₂ atmosphere at 37 °C. Subsequently, cells were washed once with PBS, trypsinized and then resuspended in 500 µl PBS. Cells were passed through a 40-μm cell strainer and analysed with a flow cytometer (CytoFLEX, Beckman Coulter) with a 488-nm laser for excitation. Data were collected from the FITC detector (for the oxidized form of BODIPY) with a 525/40-nm bandpass filter and from the PE detector (for the reduced form of BODIPY) with a 585/42-nm bandpass filter using CytExpert v.2.4 (Beckman Coulter). At least 10,000 events were analysed per sample. Data were analysed using FlowJo software (FlowJo). The ratio of fluorescence of C11-BODIPY 581/591 (lipid peroxidation) (FITC/PE ratio (oxidized/reduced ratio)) was calculated as follows¹²: (median FITC-A fluorescence – median FITC-A fluorescence of unstained samples)/(median PE-A fluorescence – median PE-A fluorescence of unstained samples). An example gating strategy is shown in Supplementary Fig. 1.

Oxilipidomics analysis

Two million cells were seeded on 15-cm dishes 1 day before the experiments. The next day, cells were treated with 5 µM icFSP1 to induce lipid peroxidation. Five hours later, cells were collected, sampled to liquid nitrogen and stored at –80 °C. Lipids from cells were extracted using the methyl-tert-butyl ether (MTBE) method⁴⁰. In brief, cell pellets collected in PBS containing dibutylhydroxytoluene (BHT; 100 µM) and diethylenetriamine pentaacetate (DTPA; 100 µM) were washed and centrifuged. SPLASH LIPIDOMIX (Avanti Polar Lipids) was added (2.5 µl), and samples were incubated on ice for 15 min. After addition of ice-cold methanol (375 µl) and MTBE (1,250 µl), samples were vortexed and incubated for 1 h at 4 °C (orbital shaker, 32 rpm). Phase separation was induced by adding water (375 µl), and samples were vortexed, incubated for 10 min at 4 °C (orbital shaker, 32 rpm) and centrifuged to separate the organic and aqueous phases (10 min, 4 °C, 1,500g). The organic phase was collected, dried in a vacuum evaporator and redissolved in 100 µl isopropanol. Lipid extracts were transferred to glass vials for LC–MS analysis.

Reversed-phase LC was carried out on a Shimadzu ExionLC equipped with an Accucore C30 column (150 × 2.1 mm, 2.6 µm, 150 Å; Thermo Fisher Scientific). Lipids were separated by gradient elution with solvent A (1:1 (v/v) acetonitrile/water) and solvent B (85:15:5 (v/v/v) isopropanol/acetonitrile/water), both containing 5 mM NH₄HCO₂ and 0.1% (v/v) formic acid. Separation was performed at 50 °C with a flow rate of 0.3 ml min^–1 using the following gradient: 0–20 min, increase from 10% to 86% B (curve 4); 20–22 min, increase from 86% to 95% B (curve 5); 22–26 min, 95% isocratic; 26–26.1 min, decrease from 95% to 10% B (curve 5), which was followed by re-equilibration for 5 min at 10% B⁶. MS analysis was performed on a Sciex 7500 system equipped with an electrospray ionization (ESI) source and operated in negative-ion mode. Products were analysed in MRM mode monitoring transitions from the parent ion to the daughter ion, as described in Supplementary Table 1, with the following parameters: TEM, 500 °C; GS1, 40; GS2, 70; CUR, 40; CAD, 9; IS, −3,000 V.

The area under the curve for the parent to daughter ion transition was integrated and normalized by appropriate lipid species, PC(15:0/18:1(d7)) or PE(15:0/18:1(d7)), from the SPLASH LIPIDOMIX Mass Spec Standard (Avanti Polar Lipids). Normalized peak areas were further log transformed and auto-scaled in MetaboAnalyst online platform v.5.0 (https://www.metaboanalyst.ca)⁴¹. Zero values were replaced by 0.2 times the minimum value detected for a given oxidized lipid in the samples. Oxidized lipids showing a significant difference (ANOVA, adjusted P value (false discovery rate (FDR)) cut-off of 0.05) between samples were used for the heat maps. The heat maps were created in GraphPad Prism 9. The colour scheme corresponds to auto-scaled log-transformed fold change relative to the mean log value for the samples.

Cell lysis and immunoblotting

Cells were lysed in LCW lysis buffer (0.5% Triton X-100, 0.5% sodium deoxycholate salt, 150 mM NaCl, 20 mM Tris-HCl, 10 mM EDTA and 30 mM sodium pyrophosphate tetrabasic decahydrate) supplemented with protease and phosphatase inhibitor cocktail (cOmplete and phoSTOP; Roche, cat. nos. 04693116001 and 4906837001) and centrifuged at 20,000g for 1 h at 4 °C. The supernatant was sampled by adding 6× SDS sample buffer (375 mM Tris-HCl pH 6.8, 9% SDS, 50% glycerol, 9% β-mercaptoethanol and 0.03% bromophenol blue). After heating at 98 °C (55 °C for xCT) for 3 min, the samples were resolved on 12% SDS–PAGE gels (Bio-Rad, cat. no. 4568043 or 4568046) and subsequently electroblotted onto PVDF membrane (Bio-Rad, cat. no. 170-4156). The membranes were blocked with 5% skim milk (Roth, cat. no. T145.2) in TBS-T (20 mM Tris-HCl, 150 mM NaCl and 0.1% Tween-20) and then probed with primary antibodies, diluted in first antibody dilution buffer (TBS-T with 5% BSA and 0.1% NaN₃ (Sigma, cat. no. S2002)), against GPX4 (1:1,000; Abcam, cat. no. ab125066), valosin-containing protein (VCP; 1:1,0000; Abcam, cat. no. ab11433 or ab109240), Flag tag (1:5,000; Cell Signaling Technology, cat. no. 2368), HA tag (1:1,000; clone 3F10, homemade), hFSP1 (1:1,000; Santa Cruz, cat. no. sc-377120, AMID), mFSP1 (1:500; clone AIFM2 1A1, rat IgG2a), hFSP1 (1:10; clone 6D8, rat IgG2a), mFSP1 (1:5; clone AIFM2 14D7, rat IgG2b), human SLC7A11 (1:10; rat IgG2a monoclonal antibody against an N-terminal peptide of human xCT, clone 3A12-1-1, developed in house), mouse SLC7A11 (1:1,000; Cell Signaling Technology, cat. no. 98051), ACSL4 (1:1,000; clone A-5, Santa Cruz, cat. no. sc-271800) or β-actin-HRP (1:50,000; Sigma, cat. no. A3854) diluted in 5% skim milk in TBS-T overnight. After membranes were washed and probed with appropriate secondary antibodies diluted in 5% skim milk in TBS-T, antibody–antigen complexes were detected with the ChemiDoc Imaging System with Image Lab v.6.0 (Bio-Rad). Representative images are shown after adjustment to the appropriate brightness and angle using ImageJ/Fiji software (v.1.52 and v.1.53).

Expression and sgRNA plasmid construction

All plasmids for this study were constructed using standard molecular biology techniques and verified by sequencing as follows. A human FSP1 cDNA (NM_001198696.2, 1008:C>T) was cloned from previously reported vectors⁵. Codon-optimized sequences for Mus musculus (mouse) FSP1 (NP_001034283.1), Rattus norvegicus (rat) FSP1 (NP_001132955.1), Gallus gallus (chicken) FSP1 (XP_421597.1) and Xenopus laevis (frog) FSP1 (NP_001091397.1) were cloned into the p442 vector. Codon-optimized sequences for human FSP1 (NP_001185625.1) and mouse Fsp1 (NP_001034283.1) were synthesized by TWIST Bioscience and subcloned into 141-IRES-puro vector. To generate deletion mutants or perform subcloning, desired DNA sequences were first amplified using KOD One PCR master mix (Sigma, cat. no. KMM-201NV) or PrimeSTAR Max DNA polymerase master mix (Takara Bio, cat. no. R045A) and resulting PCR products were purified by Wizard SV Gel&PCR Clean-up System (Promega, cat. no. A9285) according to the manufacturer’s protocol. Ligation reactions of PCR products or single guide RNA (sgRNA) duplexes with digested vectors were performed using T4 ligase (NEB, cat. no. M0202L) or In-Fusion cloning enzymes (Takara Bio, cat. no. 639649 or 638948) according to the manufacturer’s protocol. Subsequently, reaction mixtures were transformed into stable competent cells (NEB, cat. no. C3040H). Plasmids were isolated using the QIAprep Spin Miniprep kit (Qiagen, cat. no. 27106) according to the manufacturer’s protocol; the correct inserts of plasmids were confirmed by sequencing.

Lentiviral production and transduction

HEK293T cells were used to produce lentiviral particles. The ecotropic envelope protein of murine leukaemia virus (MLV) was used for mouse-derived cells, while the amphitropic envelope protein VSV-G was used for human-derived cells. A third-generation lentiviral packaging system consisting of transfer plasmids, envelope plasmids (pEcoEnv-IRES-puro or pHCMV-EcoEnv (ecotropic particles) or pMD2.G (pantropic particles)) and packaging plasmids (pMDLg_pRRE and pRSV_Rev or psPAX2) was co-lipofected into HEK293T cells using transfection reagent (PEI MAX (Polysciences, cat. no. 24765) or X-tremeGENE HP reagent (Roche, cat. no. 06366236001)). Viral particle-containing cell culture supernatant was collected 48–72 h after transfection, filtered through a 0.45-µm PVDF filter (Millipore, cat. no. SLHV033RS) and then used for lentiviral transduction.

Cells were seeded on 12- or 6-well plates in medium supplemented with 10 µg ml^–1 protamine sulfate and lentivirus was incubated with cells overnight. The next day, the cell culture medium was replaced with fresh medium containing appropriate antibiotics, such as puromycin (Gibco, cat. no. A11138-03; 1 µg ml^–1), blasticidin (Invitrogen, cat. no. A1113903; 10 µg ml^–1) or G418 (Invitrogen, cat. no. 10131-035; 1 mg ml^–1)) and cells were cultured until non-transduced cells were dead.

CRISPR–Cas9-mediated gene knockout

sgRNAs were designed to target critical exons of the genes of interest, and gene knockout was confirmed by western blotting. sgRNAs were cloned into BsmBI-digested lentiCRISPRv2-blast, lentiCRISPRv2-puro and lentiGuide-neo vectors (Addgene, cat. nos. 98293, 98290 and 139449). All sequences for sgRNAs are listed in Supplementary Table 1.

To generate knockout cells, MDA-MB-436, 786-O, A375, H460, B16F10 and 4T1 cells were transiently co-transfected with the desired sgRNAs expressed from lentiCRISPRv2-blast and lentiCRISPRv2-puro using X-tremeGENE HP reagent as described previously^6,42. One day after transfection, selection was started with puromycin (1 µg ml^–1) and blasticidin (10 µg ml^–1). After selection for 2–3 days, single-cell clones were isolated, and knockout clones were validated by immunoblotting and sequencing of genomic DNA.

To generate Dox-inducible knockout cells, lentiviruses from pCW-Cas9-blast (Addgene, cat. no. 83481) were transduced into HT-1080 cells followed by selection and establishment of single-cell clones as described previously⁶. Lentiviruses generated from lentiGuide-neo vectors harbouring sgRNAs targeting FSP1 were transduced into HT-1080 pCW-Cas9-blast cells, followed by selection with G418 as above. After Dox induction, loss of FSP1 expression was confirmed by immunoblotting.

To generate Dox-inducible FSP1–EGFP-expressing cells, H460 FSP1-knockout cells were transduced with lentivirus (pCW-FSP1^WT-EGFP-blast or pCW-FSP1^Q319K-EGFP-blast). After Dox treatment of cells, scalable FSP1 expression was confirmed by immunoblotting.

Stable expression by transfection

Gpx4-knockout 4T1 cells and Gpx4 and Fsp1 double-knockout B16F10 cells were transfected with 141-IRES-puro, 141-hFSP1^WT or hFSP1^Q319K-IRES-puro and 141-mFsp1-IRES-puro vectors using X-tremeGENE HP reagent. One day after transfection, selection was started with puromycin (1 µg ml^–1) and Lip-1 was removed from the medium to select cells with stable FSP1 expression. To obtain cells with stable expression, cells were maintained under the selection condition.

Production and purification of FSP1 enzyme

Recombinant hFSP1 protein (non-myr-FSP1) was produced in BL21 E. coli and purified by affinity chromatography with a Ni-NTA system as described previously⁵.

For protein isolation by pulldown, HEK293T cells were seeded on 10- or 15-cm dishes and transfected with constructs encoding EGFP–Strep-tagged protein. After washing with PBS, cells were lysed in LCW lysis buffer supplemented with protease and phosphatase inhibitor cocktail and 1 mM dithiothreitol (DTT). Cell extracts were collected with a cell scraper and centrifuged at 20,000g for 1 h at 4 °C. Supernatants were incubated with MagStrep ‘type3’ XT beads (Biozol, cat. no. 2-4090-002) at 4 °C for 1–2 h. Beads were washed twice with washing buffer (100 mM Tris-HCl, 150 mM NaCl and 1 mM EDTA). EGFP–Strep-tagged proteins were eluted with elution buffer (100 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA and 50 mM biotin), followed by dilution to 3 µM with TBS (50 mM Tris-HCl and 150 mM NaCl). Protein concentration was estimated from the absorbance at 280 nm measured with a UV5Nano spectrophotometer (Mettler Toledo). The coefficient was calculated by using ExPASy ProtParam (https://web.expasy.org/protparam/).

Myristoylated protein was obtained by coexpressing N-myristoyltransferase 1 (hsNMT1) with FSP1. E. coli BL21 cells were transformed with constructs encoding hsNMT1 (petCDF vector, spectinomycin resistance) and FSP1 (FSP1–EGFP with a C-terminal His₆ tag, petM13, kanamycin resistance). Purification was performed as for wild-type FSP1 with a final step of gel filtration chromatography. Purified protein was obtained and confirmed by denaturing SDS gel. Confirmation of the presence of myristoylated protein was obtained with MS.

Mass spectrometry

Myristoylation was confirmed by LC–ESI–MS (Waters Synapt XS). Proteins were separated on an Acquity UPLC Protein BEH C4 column (0.4 ml min^–1; buffer A, 0.1% formic acid in water; buffer B, 0.1% formic acid in acetonitrile), and data were analysed using Masslynx v.4.2 (Waters).

FSP1 enzyme activity and inhibition assays

For resazurin assays, enzyme reactions were prepared in TBS (50 mM Tris-HCl and 150 mM NaCl) containing 50 nM non-myr-FSP1, 200 μM NADH and inhibitor (iFSP1 or icFSP1). After the addition of 100 μM resazurin sodium salt (Sigma, cat. no. R7017), fluorescence intensity (F; excitation/emission wavelength of 540/590 nm) was measured every 1 min using a SpectraMax iD5 microplate reader with SoftMax Pro v.7 (Molecular Devices) at 37 °C. Reactions with an equivalent volume of DMSO and without resazurin were used to calculate IC₅₀ values. Curve fitting and calculation of IC₅₀ values were conducted using GraphPad Prism 9.

For NADH consumption assays, enzyme reactions were prepared in PBS (Gibco, cat. no. 14190094) containing 25 nM non-myr-FSP1, 200 μM menadione (Sigma, cat. no. M5625) or 200 µM CoQ₀ (Sigma, cat. no. D9150), and inhibitor (iFSP1 or icFSP1). After the addition of 200 μM NADH, absorbance at 340 nm at 37 °C was measured every 30 s using a SpectraMax M5 microplate reader (Molecular Devices). Reactions without NADH and without enzyme were used to normalize the results. Curve fitting was conducted using GraphPad Prism 9.

In vitro FSP1 condensation assays

Purified EGFP–Strep or hFSP1–EGFP–Strep tagged protein was diluted in TBS supplemented with 1 mM DTT. Purified Strep-tagged proteins were then mixed with PEG 3350 (Sigma, cat. no. P3640) and/or icFSP1; final concentrations of the proteins, PEG and icFSP1 are indicated in each figure legend. For confocal microscopy analysis, samples were immediately transferred onto objective slides and EGFP signal was quickly captured using an LSM880 microscope with Zen Black software (v.2.3, ZWISS) with a ×63 water-immersion objective. For confocal microscopy analysis, recombinant C-terminally GFP-tagged FSP1 and myr-FSP1 were measured at a 15 μM protein concentration in PBS (pH 7.4; 300 mM NaCl) or at a 10 μM protein concentration in PBS (pH 7.4; 150 mM NaCl). Confocal fluorescence microscopy was performed at 255 °C on a Leica TCS SP8 confocal microscope using a ×63 water-immersion objective. Samples were excited with a 488-nm laser (GFP) and imaged at 498–545 nm.

To measure turbidity, different concentrations of non-myr-FSP1 and PEG were reconstituted in 10 µl in a 384-well plate and the absorbance at 600 nm was measured using a SpectraMax iD5 microplate reader (Molecular Devices). To show non-myr-FSP1 condensation in PCR tubes, images were acquired using a smartphone. Representative bright-field images of non-myr-FSP1 condensates on an objective slide were captured using an Eclipse Ts2 microscope (Nikon) with a ×40 objective.

For sedimentation assays, recombinant non-myr-FSP1 was mixed with the same amount of TBS with 0% or 20% PEG and 1 mM DTT and samples were centrifuged at 2,500g for 5 min. The supernatant was collected in a new tube and the pellet was resuspended in TBS supplemented with 1 mM DTT. Supernatant and resuspended non-myr-FSP1 were collected by adding 6× SDS sample buffer and subsequently resolved by SDS–PAGE. One gel was subjected to western blotting and probed with anti-FSP1 (1:1,000; Santa Cruz, cat. no. sc-377120, AMID). The other gel was immediately stained with Coomassie staining solution (1 mg ml^–1 Coomassie Brilliant Blue G-250 (Sigma, cat. no. 1154440025), 50% methanol and 10% acetic acid) for 15 min and then soaked in washing buffer (70% methanol and 7% acetic acid). The washing buffer was heated using a microwave, and the buffer was changed until protein bands gave clear signals.

In vitro saturation transfer difference experiments

Saturation transfer difference experiments were performed on a Bruker Avance III HD spectrometer at 600-MHz ¹H frequency using an H/N/C triple-resonance cryogenic probe. Spectra were recorded at 10 °C with 5 µM recombinant hFSP1 (mutant) and 100-fold molar excess of icFSP1 in PBS containing 150 mM NaCl, 1% (v/v) DMSO-d6 and 10% (v/v) D₂O for deuterium-lock. The saturation time was 2.5 s, and the on and off frequencies were 0.68 and −17 ppm, respectively. NMR spectra were processed using Topspin 4.0.6 (Bruker).

Immunocytochemistry

All confocal microscopy images were acquired using an LSM880 microscope (Zeiss) with a ×63 objective and the corresponding appropriate filter sets for fluorophores and analysed with Zen Blue software (v.3.2, ZWISS) or ImageJ/Fiji unless noted otherwise. Cells were seeded on µ-Slide 8-well slides (Ibidi, cat. no. 80826) 1 day before the experiments. The next day, the medium was changed to fresh cell culture medium supplemented with 2.5 µM icFSP1. After incubation for the indicated times, cells were fixed and stained according to the following procedure: fixation with 4% paraformaldehyde for 5–10 min; permeabilization and blocking for 15 min with 0.3% Triton X-100 and 10 mg ml^–1 BSA in PBS; and incubation at 4 °C overnight with primary antibodies or undiluted supernatant for anti-AIFM2 (FSP1; clone 14D7, homemade). Antibody dilutions were as follows: 1:10 for anti-YPYDVPDYA-tag (HA; clone 3F10) and 1:100 for anti-calnexin (Abcam, cat. no. ab22595), anti-GM130 (clone EP892Y, Abcam, cat. no. ab52649), anti-EEA1 (clone C45B10, Cell Signaling Technology, cat. no. 3288) and anti-LAMP1 (clone H4A3, Santa Cruz, cat. no. sc-20011)) in primary antibody dilution buffer. Cells were further stained with appropriate fluorophore-conjugated secondary antibodies (1:500 dilution) and DAPI (1:10,000 dilution) in TBS-T or PBS for 1–2 h at room temperature, avoiding light. Finally, all samples were mounted in Aqua-/PolyMount (Polysciences, cat. no. 18606-20) and dried at 4 °C overnight. Staining of mitochondria, lipid droplets and aggresomes was performed using MitoTracker Red CMXRos (20 nM; Invitrogen, cat. no. M7512), LipidSpot 610 (1:1,000; Biotium, cat. no. 70069-T) and the Proteostat detection kit (1:10,000; Enzo, cat. no. ENZ-51035-0025), respectively, according to the manufacturer’s protocol.

FRAP

Pfa1 cells (20,000 cells) were seeded on µ-Slide VI 0.4 slides (Ibidi, cat. no. 80606) 1 day before the experiments. The next day, the medium was changed to high-glucose DMEM supplemented with 10% FBS, 2 mM l-glutamine, 1% penicillin-streptomycin, 2.5 µM icFSP1 and 10 mM HEPES. After incubation with icFSP1 for 2–4 h, 2–5 rectangular areas that each contained more than three FSP1 condensates were selected as bleaching areas. One image acquired before bleaching was considered to correspond to time ‘0’. Subsequently, the selected areas were photobleached using the maximum intensity of the laser and FRAP was monitored at minimum intervals (~5 s) using an LSM880 microscope (Zeiss).

To quantify the FRAP rate, a region of interest (ROI) for each condensate (i) in the photobleached area and one condensate (c) in a non-photobleached area was determined using ImageJ/Fiji and the mean fluorescence intensity of condensate i at time t, F_i(t) was obtained. After determining each time of fluorescence value, F_i(t) was normalized by the value of F_i(0) to obtain the relative fluorescence (RF_i(t)) of each bleached condensate area. To reflect quenching effects during observation and photobleaching, each RF_i(t) value was normalized by relative fluorescence value at  time t of condensate (c) in  non-bleached condensate areas (RF_c(t)) as follows: F_i(t) = RF_i(t)/RF_c(t) = (F_i(t)/F_i(0))/(F_c(t)/F_c(0)). Finally, the FRAP rate (%) at time t in the particles was calculated as the mean of F_i(t) × 100 as described previously²¹.

Live-cell imaging

For co-staining or washout analyses, Pfa1 cells (15,000–30,000 cells) were seeded on µ-Dish 35-mm low dishes (Ibidi, cat. no. 80136) and incubated overnight. The next day, the cell culture medium was changed to FluoroBrite DMEM (Gibco, cat. no. A1896701) supplemented with 10% FBS, 2 mM l-glutamine and 1% penicillin-streptomycin. Live-cell microscopy was performed using the 3D Cell Explorer (Nanolive) with Eve v.1.8.2 software and the corresponding appropriate filter sets. During imaging, cells were maintained at 37 °C and 5% CO₂ using a temperature-controlled incubation chamber. For co-staining analysis, cells were pretreated for 1 h with 5 µM Liperfluo (Dojindo, cat. no. L248-10) and then changed to FluoroBrite DMEM containing 0.2 µg ml^–1 propidium iodide (Sigma, cat. no. P4170) and acquisition was started using Nanolive. After recording one image, 1 mM of icFSP1 in FluoroBrite DMEM was added to the dishes (final concentration of 10 µM) while continuing to record images. Images were acquired every 10 min for more than 4 h, and GFP, BFP and RFP filter sets were used to acquire signal. For washout experiments, high-glucose DMEM was changed to FluoroBrite DMEM before the experiments followed by data acquisition using Nanolive. After recording a few images, 0.25 mM of icFSP1 in FluoroBrite DMEM was added to the dishes (final concentration of 2.5 µM) and recording of images continued for 4 h. Thereafter, the dishes were carefully washed once with fresh FluoroBrite DMEM without icFSP1 and refilled with medium. Image acquisition was then immediately restarted. Images were recorded every 5 min for one more hour; that is, the total duration of data acquisition was around 5 h.

To determine the number of condensates in cells, Pfa1 cells (15,000–20,000 cells) were seeded on µ-Slide 8-well slides (Ibidi, cat. no. 80826) and incubated overnight. The next day, the medium was changed to high-glucose DMEM supplemented with 10% FBS, 2 mM l-glutamine, 1% penicillin-streptomycin, 2.5 µM icFSP1 and Hoechst. Immediately thereafter, the focus was adjusted and Hoechst and EGFP images were recorded using an Axio Observer Z1 imaging system with VisView v.4.0 (Visitron Systems, ZWISS) with a ×20 air objective and a CCD camera (CoolSnap ES2, Photometrics) with the corresponding filter sets. During imaging, cells were maintained at 37 °C and 5% CO₂ using a temperature-controlled incubation chamber. The imaging software ImageJ/Fiji was used for visualization, and CellProfiler (v.4.1.3, Broad Institute) was used to count the condensates in each cell.

Subcutaneous tumour models

All mice were obtained from Charles River. For syngeneic subcutaneous tumour experiments, Gpx4 and Fsp1 double-knockout B16F10 cells stably overexpressing hFSP1–HA (1 × 10⁶ cells in 100 µl PBS) were injected subcutaneously into the right flank of 7-week-old female C57BL/6J mice. After tumours reached approximately 25–50 mm³ in size, mice were randomized and treated with vehicle or icFSP1 (50 mg kg^–1, Intonation) by intraperitoneal injection twice daily for 4–5 days. To generate tumour samples for staining, Gpx4 and Fsp1 double-knockout B16F10 cells stably expressing hFSP1^WT–HA or hFSP1^Q319K–HA (1 × 10⁶ cells in 100 µl PBS) were injected subcutaneously into the right flank of 7-week-old female C57BL/6J mice. After tumours reached approximately 25 mm³ in size, mice were randomized and treated with vehicle or icFSP1 (50 mg kg^–1, Intonation) by intraperitoneal injection twice daily .

For xenograft subcutaneous tumour experiments, GPX4-knockout A375 cells (5 × 10⁶ cells in 100 µl PBS) were injected subcutaneously into the right flank of 7-week-old female athymic nude mice. After tumours reached approximately 25–100 mm³ in size, mice were randomized and treated with vehicle or icFSP1 (50 mg kg^–1, Intonation) by intraperitoneal injection twice daily for the first 4 days and once daily afterwards.

For xenograft subcutaneous tumour experiments, GPX4-knockout H460 cells (5 × 10⁶ cells in 100 µl PBS) were injected subcutaneously into the right flank of 6-week-old female athymic nude mice. After tumours reached approximately 100 mm³ in size, mice were randomized and treated with vehicle or icFSP1 (50 mg kg^–1, Intonation) by intraperitoneal injection twice daily.

icFSP1 was dissolved in 45% PEG E 300 (Sigma, cat. no. 91462-1KG) and 55% PBS (Gibco, cat. no. 14190094). Tumours were measured by calliper every day, and tumour volume was calculated using the following formula: tumour volume = length × width² × 0.52. When the tumour was greater than 1,000 mm³ in size at measurement or the tumour became necrotic, tumours were considered to have reached the humane endpoint. When tumours reached the humane endpoint, the experiment was stopped and no further study was conducted.

Tumour tissue staining

Dissected tissues were fixed in 4% paraformaldehyde in PBS overnight at 4 °C. For immunofluorescence staining, fixed tissues were incubated in 20% sucrose in PBS overnight at 4 °C, followed by embedding in OCT mounting compound (TissueTek, Sakura) on dry ice and storage at –80 °C. Frozen tissues were cut into 5-µm-thick sections using a Cryostat Microm HM 560 (Thermo Fisher Scientific) at –30 °C. Tissue sections were postfixed with 1% paraformaldehyde in PBS for 10 min and subsequently fixed with 67% ethanol and 33% acetic acid for 10 min. Sections were incubated with blocking solution (5% goat serum and 0.3% Triton X-100 in PBS) for 30 min and then incubated with primary antibodies (anti-HA (clone, 3F10; 1:10; developed in house), anti-4-HNE (JaICA, cat. no. HNEJ-2; 1:50) and anti-AIFM2 (FSP1, clone 14D7; undiluted; developed in house)) diluted in blocking solution overnight at 4 °C. The next day, sections were incubated with appropriate fluorophore-conjugated secondary antibodies (goat anti-rat Alexa Fluor 488 IgG (H+L) (1:500; A-11006, Invitrogen), goat anti-mouse IgG H&L Alexa Fluor 647 (1:500; ab150115, Abcam) and donkey anti-rat IgG Alexa Fluor 555 (1:500; ab150154, Abcam)) in secondary dilution buffer (1% BSA and 0.3% Triton X-100 in PBS) for 2 h at room temperature. DNA was visualized with DAPI staining for 5 min, and slides were mounted in Aqua-/PolyMount. Images were obtained using an LSM880 microscope (Zeiss) and analysed with Zen Blue or ImageJ/Fiji software.

Pharmacokinetics and metabolic stability analyses

All studies were performed by Bienta/Enamine Ltd.

Statistical analysis

All data shown are the mean ± s.e.m. or mean ± s.d., and the number (n) in each figure legend represents biological or technical replicates as specified. All experiments (except those described otherwise in the legend) were performed independently at least twice. For mouse experiments, at least three animals were included per group once or twice. Two-tailed Student’s t tests and one-way or two-way ANOVA followed by Bonferroni’s, Dunnett’s, Tukey’s or Sidak’s multiple-comparison tests were performed using GraphPad Prism 9 (GraphPad Software) (also see figure legends for more detail). The results of the statistical analyses are presented in each figure. P < 0.05 was considered to be statistically significant.

Reporting summary

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