Skip to main page content
U.S. flag
See More

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2020 Sep;585(7826):603-608.
doi: 10.1038/s41586-020-2732-8. Epub 2020 Sep 16.

Plasticity of ether lipids promotes ferroptosis susceptibility and evasion

Yilong Zou #  1   2 , Whitney S Henry #  3 , Emily L Ricq #  4   5 , Emily T Graham  4 , Vaishnavi V Phadnis  3 , Pema Maretich  6 , Sateja Paradkar  3 , Natalie Boehnke  7 , Amy A Deik  4 , Ferenc Reinhardt  3 , John K Eaton  4 , Bryan Ferguson  4 , Wenyu Wang  4 , Joshua Fairman  3 , Heather R Keys  3 , Vlado Dančík  4 , Clary B Clish  4 , Paul A Clemons  4 , Paula T Hammond  7   8 , Laurie A Boyer  6   9 , Robert A Weinberg  10 , Stuart L Schreiber  11   12
Affiliations

Plasticity of ether lipids promotes ferroptosis susceptibility and evasion

Yilong Zou et al. Nature. 2020 Sep.

Abstract

Ferroptosis-an iron-dependent, non-apoptotic cell death process-is involved in various degenerative diseases and represents a targetable susceptibility in certain cancers1. The ferroptosis-susceptible cell state can either pre-exist in cells that arise from certain lineages or be acquired during cell-state transitions2-5. However, precisely how susceptibility to ferroptosis is dynamically regulated remains poorly understood. Here we use genome-wide CRISPR-Cas9 suppressor screens to identify the oxidative organelles peroxisomes as critical contributors to ferroptosis sensitivity in human renal and ovarian carcinoma cells. Using lipidomic profiling we show that peroxisomes contribute to ferroptosis by synthesizing polyunsaturated ether phospholipids (PUFA-ePLs), which act as substrates for lipid peroxidation that, in turn, results in the induction of ferroptosis. Carcinoma cells that are initially sensitive to ferroptosis can switch to a ferroptosis-resistant state in vivo in mice, which is associated with extensive downregulation of PUFA-ePLs. We further find that the pro-ferroptotic role of PUFA-ePLs can be extended beyond neoplastic cells to other cell types, including neurons and cardiomyocytes. Together, our work reveals roles for the peroxisome-ether-phospholipid axis in driving susceptibility to and evasion from ferroptosis, highlights PUFA-ePL as a distinct functional lipid class that is dynamically regulated during cell-state transitions, and suggests multiple regulatory nodes for therapeutic interventions in diseases that involve ferroptosis.

PubMed Disclaimer

Conflict of interest statement

Competing interests

S.L.S. serves on the Board of Directors of the Genomics Institute of the Novartis Research Foundation (“GNF”); is a shareholder and serves on the Board of Directors of Jnana Therapeutics; is a shareholder of Forma Therapeutics; is a shareholder and advises Kojin Therapeutics, Kisbee Therapeutics, Decibel Therapeutics and Eikonizo Therapeutics; serves on the Scientific Advisory Boards of Eisai Co., Ltd., Ono Pharma Foundation, Exo Therapeutics, and F-Prime Capital Partners; and is a Novartis Faculty Scholar. Kojin Therapeutics in particular explores the medical potential of cell plasticity related to ferroptosis. Other authors declare no conflict of interest relevant to this study.

Figures

Extended Data Figure 1.
Extended Data Figure 1.. CRISPR screens identify peroxisome components as contributors of sensitivity to ferroptosis.
a. STRING Protein Network analysis of the top OVCAR-8 screening hits. b. STRING Protein Network analysis of the top 786-O screening hits. c. Schematic diagram showing simplified illustration of the principle and work flow for the Gene-list Network Enrichment Analysis (GeLiNEA) method. d. Table showing the top pathways enriched in the 786-O screen hits using GeLiNEA. Overlap is the number of genes in common between a gene set and the screening hit list, nConnections is the number of connections between a gene set and the screening hit list, p-values are computed using equation (1), and q-values are p-values adjusted for multiple testing using the Benjamini-Hochberg correction method. e. Table showing the top pathways enriched in OVCAR-8 screen hits using GeLiNEA. f. Venn diagram showing the overlapped CRISPR screen hits in OVCAR-8 and 786-O cells. g. Volcano plots showing the top hits in OVCAR-8 and 786-O genome-wide CRISPR screens. For presentation purposes, only genes that are enriched in the RSL3 or ML210 treated condition for ≥ 1.5 fold (log2FC ≥ 0.585) are plotted. See Methods for data analysis methods.
Extended Data Figure 2.
Extended Data Figure 2.. Peroxisomes contribute to ferroptosis sensitivity in renal and ovarian carcinomas cells.
a. Fluorescent imaging analysis of peroxisome abundances, reported by PTS1-GFP signal, in OVCAR-8 (top) and 786-O (bottom) cells expressing sgNC, or PEX3, PEX10-targeting sgRNAs. Scale bar, 50 µm. Peroxisome quantitations are shown on the right as violin plots. OVCAR-8-sgNC, n=295, PEX3-sg1, n=318, PEX10-sg1, n=301; 786-O-sgNC, n=604, PEX3-sg1, n=617, PEX10-sg1, n=543. Lines in violin plots indicate median and quartiles. b. Fluorescent imaging analysis of peroxisome abundances, reported by PTS1-GFP signal, in OVCAR-8 and 786-O cells expressing sgNC, or PEX12-targeting sgRNAs. Scale bar, 50 µm. Peroxisome quantitations are shown on the right as violin plots. OVCAR-8-sgNC, n=1032, PEX12-sg1, n=259, PEX12-sg2, n=444; 786-O-sgNC, n=1139, PEX12-sg1, n= 623, PEX12-sg2, n=1326. Lines in violin plots indicate median and quartiles. c. Immunoblot analysis showing the PEX10 protein levels in OVCAR-8 cells expressing sgNC or PEX10-targeting sgRNAs. COX-IV was used as a loading control. Representative result of experiment performed in duplicate. See Supplementary Information for uncropped immunoblot images. d. Viability curves of 786-O cells expressing negative control (sgNC) or sgRNAs targeting PEX3, PEX10 or PEX12 treated with indicated concentrations of ML210 or RSL3 for 48h. n=4 biologically independent samples. Representative result of experiment performed in triplicate. e. Viability curves of PEX3-sg1-expressing OVCAR-8 cells rescued with sgRNA-resistant mouse Pex3 cDNA respectively and being treated with indicated concentrations of ML210 or RSL3. In this experiment, cellular viability was measured at 24 hours of treatment, at which time point the PEX3-sg1-only cells were not yet dying. n=4 biologically independent samples. Data of experiment performed once. h. Viability curves of PEX10-sg1-expressing OVCAR-8 cells rescued with sgRNA-resistant mouse Pex10 cDNA respectively and being treated with indicated concentrations of ML210 or RSL3. In this experiment, cellular viability was measured at 24 hours of treatment, at which time point the PEX10-sg1-only cells were not yet dying. n=4 biologically independent samples. Data of experiment performed once. For viability curves, data center and error bars indicate mean±s.d..
Extended Data Figure 3.
Extended Data Figure 3.. Peroxisomes contribute to ferroptosis sensitivity via the ether lipid biosynthesis pathway.
a. Schematic diagram of the known functional pathways involved in lipid metabolism and reduction of reactive oxygen species (ROS) in the peroxisome. Abbreviations, VLCFA, very long-chain fatty acids; BCFA, branched-chain fatty acids; DHAP, dihydroxyacetone; AGPS, alkylglycerone phosphate synthase. b. Volcano plots showing the lipidomic analysis of 786-O cells expressing sgNC or PEX3-targeting sgRNAs. n=3 biologically independent samples. Two tailed Student’s T-test. Multiple-testing adjustment was performed using the Benjamini-Hochberg method. Abbreviations: ePE, ether-linked phosphatidylethanolamine; ePC, ether-linked phosphatidylcholine; PUFA-, polyunsaturated fatty acyl-; TAG, triacylglycerol; PL, phospholipids. c. Viability curves of OVCAR-8 cells expressing negative control (sgNC), FAR1-targeting sgRNAs (left) or AGPS-targeting sgRNAs (right) and treated with indicated concentrations of RSL3 for 72 h. n=3 biologically independent samples. Representative results of experiment performed in triplicate. d. Fluorescent imaging showing nuclear staining by Hoechst 33342 in OVCAR-8 cells with the indicated genetic perturbations and treated with vehicle (DMSO) or indicated concentrations of ML210 for 5 days. Representative images from experiment performed once, and each condition has three biological replicates. e. Relative growth rates measured by areas of live cell coverage in OVCAR-8 cells expressing sgNC, AGPS-sg2 or FAR1-sg2 and treated with indicated concentrations of ML210 for 5 days. n=2 or 3 biologically independent samples. Data of experiment performed once. f. Immunoblot analysis of AGPS protein levels in 786-O cells expressing negative control (sgNC) or AGPS-targeting sgRNAs. g. Viability curves of 786-O cells expressing sgNC or AGPS-targeting sgRNAs treated with indicated concentrations of ML210 or RSL3 for 48h. n=4 biologically independent samples. Representative result of experiment performed in triplicate. h. Immunoblot analysis of FAR1 protein levels in HuH-7 cells expressing negative control (sgNC) or AGPS-targeting sgRNAs. β-Actin was used as a loading control. i. Viability curves of 786-O cells expressing sgNC or FAR1-targeting sgRNAs treated with indicated concentrations of ML210 or RSL3 for 48 h. n=4 biologically independent samples. Representative result of experiment performed in triplicate. j. Immunoblot showing AGPS protein levels in OVCAR-8 cells expressing sgNC or AGPS-sg2, and AGPS−/− single cell clone (SCC)#9. k. Viability curves of OVCAR-8 cells expressing sgNC or AGPS-sg2, and AGPS−/− SCC9 treated with ML210 for 72 h. n=3 biologically independent samples. Representative result of experiment performed in duplicate. l. Immunoblot showing FAR1 protein levels in OVCAR-8 cells expressing sgNC or FAR1-sg1, and FAR1−/− single cell clone (SCC)#9. m. Viability curves of OVCAR-8 cells expressing sgNC or FAR1-sg1, and FAR1−/− SCC9 treated with ML210 for 72 h. n=3 biologically independent samples. Representative result of experiment performed in duplicate. Immunoblots are representative data of experiments performed twice. See Supplementary Information for uncropped immunoblot images. β-Actin or GAPDH was used as a loading control. For viability curves, data center and error bars indicate mean±s.d..
Extended Data Figure 4.
Extended Data Figure 4.. AGPS/FAR1-depletion blocks ether phospholipid synthesis and lipid peroxidation.
a. Volcano plots showing the lipidomic analysis of OVCAR-8 and 786-O cells expressing sgNC or FAR1-targeting sgRNAs. n=3 biologically independent samples. Two tailed Student’s T-test. Multiple-testing adjustment was performed using the Benjamini-Hochberg method. Abbreviations: ePE, ether-linked phosphatidylethanolamine; ePC, ether-linked phosphatidylcholine; PUFA-, polyunsaturated fatty acyl-; TAG, triacylglycerol; PL, phospholipids. b. Heatmap showing the relative abundances of free fatty acids in wildtype and AGPS, FAR1, or PEX10-depleted OVCAR-8 cells. n=3 biologically independent samples. Highlighted in red are polyunsaturated fatty acids that are enriched in response to AGPS or FAR1 knockout. c. Volcano plots showing free fatty acid lipidomic analysis in wildtype and AGPS, FAR1, PEX3 or PEX10-depleted 786-O cells. n=3 biologically independent samples. Highlighted in red are free polyunsaturated fatty acids upregulated in the knockout cells. Two tailed Student’s T-test. Multiple-testing adjustment was performed using the Benjamini-Hochberg method. d. Representative gating strategy used in the flow cytometry analysis of BODIPY-C11 oxidation levels. e. Histogram showing the lipid peroxidation levels reported by the ratio between oxidized and reduced BODIPY-C11 levels in the indicated OVCAR-8 cells treated with DMSO or ML210 for 2 h. Plot of experiment performed once.
Extended Data Figure 5.
Extended Data Figure 5.. The ether lipid biosynthesis pathway but not other peroxisomal pathways contributes to ferroptosis susceptibility.
a. Immunoblot analysis of AGPS protein levels in cells expressing the indicated constructs. Plot of experiment performed once. b. Viability curves of AGPS-sg2-expressing OVCAR-8 cells rescued with sgRNA-resistant mouse Agps cDNA and being treated with indicated concentrations of RSL3 for 24 h. n=4 biologically independent samples. Plot of experiment performed once. c. Viability curves of FAR1-sg2-expressing OVCAR-8 cells rescued with sgRNA-resistant mouse Far1 cDNA and being treated with indicated concentrations of RSL3 for 24 h. n=4 biologically independent samples. Plot of experiment performed once. d. Viability curves for OVCAR-8 cells expressing empty vector or cDNAs of mouse Agps, Far1, Pex3 or Pex10 and treated with indicated concentrations of ML210 or RSL3. In this experiment, cellular viability was read at 24 hours of treatment (instead of normally at 72 hours), at which time point the control cells were not yet dying. n=4 biologically independent samples. Plot of experiment performed once. e. Viability curves of 786-O cells expressing non-targeting negative control shRNA (shNC) or FAR1-targeting shRNAs treated with indicated concentrations of ML210 or RSL3 for 48 h. n=4 biologically independent samples. Representative data of experiment performed twice. f. Viability curves of 786-O cells expressing shNC or GNPAT-targeting shRNAs treated with indicated concentrations of ML210 or RSL3 for 48 h. n=4 biologically independent samples. Representative data of experiment performed twice. g. Percentage of remaining DPPH levels in in vitro DPPH assay system containing indicated concentrations of ZINC-69435460 or ferrostatin-1 (Fer-1). n=3 biologically independent samples. Data center and error bars: mean±s.d. DPPH, 2,2-diphenyl-1-picrylhydrazyl. ZINC-69435460 vs Fer-1 both at 1mM, p=6.05x10e-8. h. Relative viability of OVCAR-8 cells pre-treated with AGPS inhibitor ZINC-69435460 (ZINC) for 24h, followed by ML210 treatment for another 72 h. n=3 biologically independent samples. Data center and error bar, mean±s.e.m. Representative data of experiment performed in triplicate. For 0.25 µM ML210 conditions, 0 µM vs 150 µM ZINC, p=0.000076; vs 250 µM ZINC, p=0.000073; vs 350 µM ZINC, p=0.00011; vs 500 µM ZINC, p=0.00045. For 0.35 µM ML210 conditions, 0 µM vs 150 µM ZINC, p=0.0017; vs 250 µM ZINC, p=0.000011; vs 350 µM ZINC, p=7.55x10e-9; vs 500 µM ZINC, p=0.00015. i. Immunoblot analysis of Catalase (CAT) protein levels in 786-O cells expressing sgNC or CAT-targeting sgRNAs. Plot of experiment performed once. j. Immunoblot analysis of Superoxide Dismutase 1 (SOD1) protein levels in 786-O cells expressing sgNC or SOD1-targeting sgRNAs. Plot of experiment performed once. k. Viability curves of 786-O cells expressing sgNC or SOD1- or CAT-targeting sgRNAs treated with indicated concentrations of ML210 or RSL3 for 48 h. n=4 biologically independent samples. Plots of experiment performed once. β-ACTIN was used as a loading control in immunoblots. See Supplementary Information for uncropped immunoblot images. For viability curves, data center and error bars indicate mean±s.d.. P-values were calculated using two-tailed Student’s T-test.
Extended Data Figure 6.
Extended Data Figure 6.. Peroxisomes and the ether lipid biosynthesis pathway contribute to ferroptosis in liver, endometrial and kidney cancers.
a. Fluorescent imaging analysis of peroxisome abundances, reported by PTS1-emGFP signal, in HuH-7 cells expressing the indicated sgRNAs. Scale bar, 25 µm. Representative images from experiment performed once, and each condition has three biological replicates. b. Violin plots showing the quantitation of peroxisomes in HuH-7 cells expressing indicated sgRNAs. HuH-7-sgNC, n=250, PEX3-sg1, n=231, PEX3-sg2, n=256, PEX10-sg1, n=600, PEX10-sg2, n=455, PEX12-sg1, n=295, PEX12-sg2, n=278, AGPS-sg1, n=321, AGPS-sg2, n=304, FAR1-sg1, n=411. Lines in violin plots indicate median and quartiles. c. Immunoblot analysis of AGPS protein levels in HuH-7 cells expressing negative control (sgNC) or AGPS-targeting sgRNAs. β-Actin was used as a loading control. Plot of experiment performed once. d. Viability curves for HuH-7 cells expressing the indicated sgRNAs treated with indicated concentrations of ML210 or RSL3 for 48 h. n=4 biologically independent samples. Representative results from experiment performed twice. e. Viability curves for SNU-685 cells expressing the indicated sgRNAs treated with indicated concentrations of ML210 or RSL3 for 48 h. n=4 biologically independent samples. Representative results from experiment performed once. f. Volcano plots showing the lipidomic analysis (Saito et al. 2016) results comparing 49 pairs of clear-cell renal cell carcinoma (ccRCC) tumor and adjacent normal kidney tissues. n=49 tumor samples, n=49 normal samples. Two-tailed Student’s T-test. Multiple-testing adjustment was performed using the Benjamini-Hochberg method. Abbreviations: PC, phosphatidylcholine; PE, phosphatidylethanolamine; ePE, ether-phosphatidylethanolamine; ePC, ether-phosphatidylcholine. See Supplementary Information for uncropped immunoblot images. For viability curves, data center and error bars indicate mean±s.d..
Extended Data Figure 7.
Extended Data Figure 7.. AGPAT3 contributes to polyunsaturated ether phospholipid synthesis downstream of peroxisomes.
a. Immunoblot analysis of ACSL4 and LPCAT3 protein levels in 786-O cells expressing the indicated sgRNAs. Plot of experiment performed once. b. Immunoblot analysis of ACSL4 and LPCAT3 protein levels in OVCAR-8 cells expressing the indicated sgRNAs. Plot of experiment performed once. c. Immunoblotting of ACSL4, AGPS and FAR1 protein levels in the indicated OVCAR-8 cell lines. Representative result of experiment performed in duplicate. d. Viability curves of OVCAR-8 cells expressing sgNC or sgRNAs targeting the gene of interest (GOI) as indicated on top of each graph, and transduced with doxycycline (dox)-inducible ACSL4-sgRNA1TetOn construct. These cells were pre-treated with vehicle or dox, and then treated with indicated concentrations of ML210, or ML210+Fer-1 for 72 h. n=3 biologically independent samples. Representative results from experiment performed twice. e. Volcano plot showing the changes in phospholipid levels in sgNC or AGPAT3-targeting sgRNA expressing 769-P cells. n=3 biologically independent samples. Two tailed Student’s T-test. Multiple-testing adjustment was performed using the Benjamini-Hochberg method. f. Viability curves for 769-P cells expressing sgNC or AGPAT3-targeting sgRNAs treated with indicated concentrations of ML210 or RSL3 for 48 h. n=4 biologically independent samples. Representative results from experiment performed in triplicate. g. Viability curves for 786-O and OVCAR-8 cells expressing sgNC or AGPAT3-targeting sgRNAs treated with indicated concentrations of RSL3 for 48 h. n=3 (OVCAR-8) or n=4 (786-O) biologically independent samples. Representative results from experiment performed in triplicate. h. Fluorescent imaging showing nuclear staining by Hoechst 33342 in OVCAR-8 cells with the indicated genetic perturbations and treated with vehicle (DMSO) or indicated concentrations of ML210 for 5 days. Representative images from experiment performed once, and each condition has three biological replicates. i. qRT-PCR analysis of relative AGPAT3 mRNA expression in 786-O and 769-P cells expressing non-targeting negative control shRNA (shNC) or AGPAT3-targeting shRNAs. B2M was used as a loading control. n=3 biologically independent samples. Two-tailed student’s T-test; ns, not significant. j. Viability curves for 786-O, and 769-P cells expressing shNC or AGPAT3-targeting shRNAs treated with indicated concentrations of ML210 or RSL3 for 48 h. n=4 biologically independent samples. Representative results from experiment performed in triplicate. k. Nucleotide traces in sanger sequencing analysis showing the point mutation introduced in the mouse Agpat3E176A cDNA construct. l. Viability curves of AGPAT3-sg1-expressing OVCAR-8 cells rescued with sgRNA-resistant, wildtype (WT) mouse Agpat3 or Agpat3E176A mutant cDNA and being treated with indicated concentrations of RSL3 for 24 h. n=4 biologically independent samples. Representative data of experiment performed twice. β-Actin or GAPDH was used as a loading control. See Supplementary Information for uncropped immunoblot images. For viability curves and bar graphs, data center and error bars indicate mean±s.d..
Extended Data Figure 8.
Extended Data Figure 8.. Polyunsaturated ether lipid nanoparticles increase cellular sensitivity to ferroptosis.
a. Schematic of the plasmalogen biosynthesis pathway. Genes marked in red highlight pro-ferroptotic genes identified from the CRISPR screens. b. Strategy used to deliver synthetic phospholipids to OVCAR-8 cells using nanoparticles, and the chemical structures of synthetic phospholipids used. c. Viability curves of OVCAR-8 cells expressing the indicated sgRNAs pre-treated with vehicle (sterilized water) or the specified PE nanoparticles, and then treated with indicated concentrations of ML210 or RSL3 for 72 h. n=4 biologically independent samples. Veh, vehicle used to package the phospholipids. d. Viability curves of OVCAR-8 cells expressing the indicated sgRNAs pre-treated with vehicle or the specified PC nanoparticles, and then treated with indicated concentrations of ML210 or RSL3 for 72 h. n=4 biologically independent samples. Veh, vehicle used to package the phospholipids. P value for comparing the relative viabilities of NP5: C18:0-C20:4PC treated cells (blue) with that of NP6: C18 (plasm)-C20:4 PC (red): for ML210=0.0391 µM, p=0.00741; for ML210=0.0781 µM, p=0.00095; for ML210=0.156 µM, p=0.00177; for ML210=0.3125 µM, p=0.0101; for RSL3=0.0156 µM, p=0.0000468; for RSL3=0.03125 µM, p=0.00241; for RSL3=0.0625 µM, p=0.00104. Two-tailed Student’s T test. e. Viability curves of 786-O cells expressing the indicated sgRNAs pre-treated with vehicle or the specified PC nanoparticles, and then treated with ferroptosis inducers. n=4 biologically independent samples. For viability curves, data center and error bars indicate mean±s.d..
Extended Data Figure 9.
Extended Data Figure 9.. Polyunsaturated plasmalogens promote lipid peroxidation in GPX4-inhibited cells.
a. Quantification of time-lapse imaging of lipid peroxidation levels reported by BODIPY-C11 oxidation in 786-O cells co-treated with ML210 and indicated PE nanoparticles or Lip-1. n=3 biologically independent samples. b. Quantification of time-lapse imaging of lipid peroxidation levels reported by BODIPY-C11 oxidation in 786-O cells co-treated with ML210 and indicated PC nanoparticles or Lip-1. n=3 biologically independent samples. c. Quantification of time-lapse imaging of lipid peroxidation levels reported by BODIPY-C11 oxidation in 786-O cells expressing the indicated sgRNAs treated with ML210 and indicated PE nanoparticles or Lip-1. n=2 biologically independent samples. Nanoparticles were added at the same time as ML210. d. Quantification of time-lapse imaging of lipid peroxidation levels reported by BODIPY-C11 oxidation in 786-O cells treated with ML210 and indicated PE nanoparticles or Lip-1. n=3 biologically independent samples. Nanoparticles were added 1 h after ML210 administration (indicated by the arrow, note a 10 min time is deduced for reagent and equipment handling). e. Quantification of time-lapse imaging of lipid peroxidation levels reported by BODIPY-C11 oxidation in 786-O cells treated with ML210 and indicated PC nanoparticles or Lip1. n=3 biologically independent samples. Nanoparticles were added 1 h after ML210 administration (indicated by the arrow, note a 10 min time is deduced for reagent and equipment handling). f. Volcano plots showing the lipidomic analysis of 786-O cells treated with ML210 or DMSO for 90 min. n=3 biologically independent samples. Two tailed Student’s T-test. Multiple-testing adjustment was performed using the Benjamini-Hochberg method. For viability curves and lipid peroxidation time-lapse imaging quantification, data center and error bars indicate mean±s.d..
Extended Data Figure 10.
Extended Data Figure 10.. Polyunsaturated ether phospholipid down-regulation is associated with acquired ferroptosis resistance in vivo.
a. Immunoblotting analysis of GPX4, AGPS, FAR1 and ACSL4 protein levels of GPX4+/+ and GPX4−/− OVCAR-8 cells expressing sgNC or sgRNAs targeting each of AGPS, FAR1, and PEX3. β-tubulin was used as a loading control. Representative result of experiment performed in duplicate. b. Tumor growth curves from mice implanted with 786-O cells expressing sgNC or sgRNAs targeting each of AGPS, FAR1, PEX3 and AGPAT3. n=5 mice per group, each mouse was injected with two tumors. Plot of experiment performed once. For day-42 tumor sizes, sgNC vs AGPS-sg1, p=0.496; sgNC vs FAR1-sg2, p=0.899; sgNC vs PEX3-sg1, p=0.066; sgNC vs AGPAT3-sg, p=0.54. Two tailed Student’s T-test. c. Tumor images showing the relative sizes of the xenograft tumors formed by 786-O cells expressing sgNC or sgRNAs targeting each of AGPS, FAR1, PEX3 and AGPAT3 and dissected at day 42. Results from experiment performed once. d. Relative viability of OVCAR-8 cells expressing sgNC, AGPS-sg2 or FAR1-sg1, and AGPS−/− and FAR1−/− single cell clones (SCC) over a 3-day time course. n=3 biologically independent samples. Representative results of experiments performed in triplicates. For day-3 viability, sgNC vs AGPS-sg2 bulk, p=0.872; vs AGPS-sg2 SCC9, p=0.172; vs FAR1-sg1 bulk, p=0.151; vs FAR1-sg1 SCC9, p=0.01. Two tailed Student’s T-test. e. Relative sizes (left) and weights (right) of xenograft tumors dissected from immunocompromised mice injected with OVCAR-8 cells with the indicated genetic background. sgNC, n=3 tumors, FAR1-sg1 bulk, n=4 tumors, FAR1-sg1 SCC9, n=5 tumors, AGPS-sg2 bulk, n=4 tumors, AGPS-sg2 SCC9, n=5 tumors. Two tailed Student’s T-test., ns, not significant (p>0.05). Data of experiment performed once. f. Volcano plot showing the global lipidomic analysis comparing GPX4−/− FR2#a cells and GPX4+/+ wildtype (WT) 786-O cells isolated from xenograft tumors. Two tailed Student’s T-test. Multiple-testing adjustment was performed using the Benjamini-Hochberg method. TAG, triacylglycerol; PUFA-, polyunsaturated fatty acyl-. n=6 biologically independent samples. g. Volcano plot showing polar metabolomic analysis using HILIC-positive method and comparing GPX4−/− FR2#a (left) or FR2#d (right) cells and GPX4+/+ wildtype (WT) 786-O cells. n=6 in each group. Two tailed Student’s T-test. Multiple-testing adjustment was performed using the Benjamini-Hochberg method. h. Volcano plot showing the free fatty acid lipidomic analysis comparing GPX4−/− FR2#a cells and GPX4+/+ wildtype (WT) 786-O cells. n=6 biologically independent samples in each group. Two tailed Student’s T-test. Multiple-testing adjustment was performed using the Benjamini-Hochberg method. S1P: sphingosine-1-phosphate. i. Representative fluorescent images (left) of peroxisomes reported by PTS1-GFP expression and quantifications (right) in GPX4+/+ (WT), GPX4−/− FR2#a and FR2#d 786-O cells. Scale bar indicates 50 µm. WT-L, n=1320, WT-R, n=863, FR2#a-L, n=533, FR2#a-R, n=512, FR2#d-L, n=876, FR2#d-R, n=1019. Lines in violin plots indicate median and quartiles. Data of experiment performed once. j. Volcano plots showing the relative mRNA expression (RNA-seq) of 87 peroxisome and ether-lipid biosynthesis-related genes comparing GPX4+/+ and GPX4−/− FR2#d cells. n=4 biologically independent samples. See Supplementary Information for statistical methods used. k. Heatmap showing the relative mRNA expression of indicated genes GPX4+/+ (WT), GPX4−/− FR2#a and FR2#d 786-O cells analyzed by RNA-seq. l. Immunoblotting analysis of AIFM2/FSP1 protein levels in GPX4+/+ (WT), GPX4−/− FR2#a and FR2#d 786-O cells. β-ACTIN was used as a loading control. Results from experiment performed once. See Supplementary Information for uncropped immunoblot images. For cell and tumor growth curves and bar graphs, data center and error bars indicate mean±s.d..
Extended Data Figure 11.
Extended Data Figure 11.. ER-resident enzyme plasmanylethanolamine desaturase/TMEM189 is not required for ferroptosis sensitivity in cancer cells.
a. Top genes that show co-dependency with TMEM189 (left) or AGPS (right) using the Cancer Dependency Map (DepMap) database. PCE: Pearson correlation coefficient. b. Immunoblotting analysis of TMEM189 protein levels in 786-O cells expressing sgNC or TMEM189-targeting sgRNAs. Representative results of experiment performed twice. c. Viability curves for 786-O cells expressing sgNC or TMEM189-targeting sgRNAs and treated with indicated concentrations of ML210 or RSL3 for 48 h. n=4 biologically independent samples. Representative results of experiment performed twice. d. Immunoblotting analysis of TMEM189 protein levels in 786-O cells expressing doxycycline (dox)-inducible TMEM189-shRNAs. β-ACTIN was used as a loading control. Results from experiment performed once. e. Viability curves for 786-O, OVCAR-8 and HuH-7 cells expressing dox-inducible TMEM189-targeting shRNAs, pretreated with vehicle (DMSO) or dox, and then treated with indicated concentrations of ML210 or RSL3. n=4 biologically independent samples. Representative results of experiment performed twice. f. Immunoblotting analysis of TMEM189 protein levels in 786-O cells expressing empty vector (EV) or human TMEM189 cDNA construct. β-ACTIN was used as a loading control. Results from experiment performed once. g. Viability curves for 786-O cells expressing empty vector or TMEM189 cDNA and treated with indicated concentrations of ML210 or RSL3 for 48 h. n=4 biologically independent samples. Representative results of experiment performed twice. β-ACTIN was used as a loading control in immunoblots. See Supplementary Information for uncropped immunoblot images. For viability curves, data center and error bars indicate mean±s.d..
Extended Data Figure 12.
Extended Data Figure 12.. Neurons and cardiomyocytes acquire increased ether-phospholipid levels and elevated sensitivity to ferroptosis.
a. Scheme showing the experimental strategy for neuronal differentiation of SH-SY5Y cells, and representative images showing the cell morphology at indicated stages. RA, retinoic acid; BDNF, brain-derived neurotrophic factor; FBS, fetal bovine serum. b. Immunoblot analysis showing the protein expression levels of relevant neuronal markers including MAP2 (microtubule associated protein 2), tyrosine hydroxylase, NeuN (neuronal nuclei antigen) and β-3-tubulin. Arrows indicate the band for the indicated full length protein, * indicates non-specific bands. β-Actin was used as a loading control. Representative results of experiment performed twice. c. Viability curves of SH-SY5Y parental cells and cells at day 6 of neuronal differentiation under the treatment of indicated concentrations of RSL3 for 48 h. n=2 biologically independent samples. Representative results of experiment performed twice. d. Fluorescent images showing lipid peroxidation levels reported by BODIPY-C11 oxidation. Representative images of experiment performed in duplicate. e. Volcano plot showing free fatty acid lipidomics analysis in parental or differentiated SH-SY5Y cells. n=3 biological replicates for the parental condition, n= 4 biological replicates for the day 6 and day 12 differentiation condition. Two tailed Student’s T test. Multiple-testing adjustment was performed using the Benjamini-Hochberg method. f. Immunofluorescence images showing the expression of cardiac troponin T, a marker of differentiated human cardiomyocytes, and NKX2.5, a cardiac-lineage specific marker in the cardiac progenitor (CP) cells and cardiomyocytes (CM). Scale bars indicate 15 µm. Representative results of experiment performed twice. g. Viability curves of iPS cells and differentiated cardiomyocytes (CM) treated with indicated concentrations of ML210 or RSL3 for 24 h. n=4 biologically independent samples. Results from experiment performed once. h. Bright field images showing cardiomyocytes treated with ML210 undergoing cell death. Scale bars indicate 30 µm. Representative results of experiment performed twice. i. Bar plots showing relative viability of CMs treated with ML210 and indicated cell death inhibitors. z-VAD, z-VAD-FMK; Nec-1, necrostatin-1. n=3 biologically independent samples. Representative results of experiment performed twice. 0 µM vs 0.25 µM ML210 treated conditions with additional DMSO treatment only, p=0.00011. For 0.25 µM ML210 treated conditions, DMSO vs z-VAD, p=0.089; DMSO vs Nec-1, p=0.125; DMSO vs Fer-1, p=0.00061; DMSO vs Lip-1, p=0.00008. Two tailed Student’s t-Test. j. Volcano plot showing free fatty acid lipidomics analysis in cardiac progenitors (CP) or differentiated cardiomyocytes (CM). n=2 biological replicates for CP, n=4 biological replicates for CM. Two tailed Student’s t-Test. Multiple-testing adjustment was performed using the Benjamini-Hochberg method. k. qRT-PCR analysis showing the relative abundances of PEX3 (left) or AGPS (right) mRNAs in cardiomyocytes treated with the indicated siRNAs. n=2 biologically independent samples. See Supplementary Information for uncropped immunoblot images. For viability curves and bar graphs, data center and error bars indicate mean±s.d..
Figure 1.
Figure 1.. Genome-wide CRISPR screens identify peroxisome components as contributors for ferroptosis susceptibility.
a. Schematic diagram summarizing the CRISPR screens in OVCAR-8 and 786-O cells to identify ferroptosis regulators. HGSOC, high grade serous ovarian carcinoma; ccRCC, clear-cell renal cell carcinoma; ML210 and RSL3, covalent small molecule inhibitors of glutathione peroxidase 4 (GPX4) and inducers of ferroptosis. b. Volcano plots showing the top genes in 12-day RSL3-treated OVCAR-8 (left) and 8-day ML210-treated 786-O (right) cells. For presentation purposes, only genes enriched in the RSL3 or ML210 treated condition for ≥ 1.5 fold (log2 fold change ≥ 0.585) are plotted. Blue highlights lipid synthesis genes; red highlights peroxisome genes. See Methods for data analysis methods. c. Viability curves of OVCAR-8 cells expressing a non-targeting negative control sgRNA (sgNC) or sgRNAs targeting PEX10, PEX3, or PEX12 and being treated with indicated concentrations of ML210 or RSL3 for 72 h. n=3 biologically independent samples. Data center and error bars indicate mean±s.d.. Representative results of experiments performed in triplicate.
Figure 2.
Figure 2.. The polyunsaturated ether lipid biosynthesis pathway mediates the pro-ferroptotic roles of peroxisomes.
a. Schematic diagram showing the distinct structures of diacyl- and two subtypes of ether-linked phospholipids. Bottom, the chemical structure of an example plasmalogen, C18(plasm)-C20:4 PE. b. Volcano plots showing the lipidomic analysis of OVCAR-8 and 786-O cells expressing sgNC or PEX10-targeting sgRNAs. n=3 biologically independent samples. c. Immunoblot showing FAR1 protein levels in OVCAR-8 cells expressing negative control (sgNC) or FAR1-targeting sgRNAs. d. Viability curves of OVCAR-8 cells expressing non-targeting negative control sgRNA (sgNC) or FAR1-targeting sgRNAs treated with ML210 for 72 h. n=3 biologically independent samples. e. Immunoblot showing AGPS protein levels in OVCAR-8 cells expressing sgNC or AGPS-targeting sgRNAs. Arrow indicates the AGPS protein band. f. Viability curves of OVCAR-8 cells expressing sgNC or AGPS-targeting sgRNAs treated with ML210 for 72 h. n=3 biologically independent samples. g. Volcano plots showing the lipidomics results of OVCAR-8 and 786-O cells expressing sgNC or AGPS-targeting sgRNAs. n=3 biologically independent samples. h. Volcano plots showing the changes in the phospho-lipidome in OVCAR-8 and 786-O cells expressing sgNC or AGPAT3-targeting sgRNA. n=3. i. Viability curves of AGPAT3-depleted cells treated with ML210. n=3 (OVCAR-8) or n=4 (786-O) biologically independent samples. Immunoblots in c, e are representative data from experiments performed twice. COX-IV was used as a loading control. See Supplementary Information for uncropped immunoblot images. Viability curves are representative data from experiments performed in triplicate, and data center and error bars indicate mean±s.d.. Two-tailed Student’s T test was used in calculating p values in volcano plots. Multiple-testing adjustment was performed using the Benjamini-Hochberg method. Abbreviations: ePE, ether-linked phosphatidylethanolamine; ePC, ether-linked phosphatidylcholine; PUFA-, polyunsaturated fatty acyl-; TAG, triacylglycerol; PL, phospholipids.
Figure 3.
Figure 3.. Cancer cells initially dependent on GPX4 downregulate polyunsaturated ether phospholipids to evade ferroptosis.
a. Relative viability of indicated OVCAR-8 cells following ferrostatin-1 (Fer-1) withdrawal. n=8 biologically independent samples. GPX4+/+-sgNC (black dash curve) was analyzed independently and included as a reference. P values for 90-hour viability of GPX4−/− derivatives: GPX4−/−-sgNC vs GPX4−/−-AGPS-sg1, p=9.07x10e-8; vs -AGPS-sg2, p=9.66x10e-15; vs -FAR1-sg2, p=3.79x10e-5; vs -PEX3-sg1, p=5.9x10e-12; vs -PEX3-sg2, p=2.92×10e-17; vs -PEX10-sg1, p=1.99x10e-16; vs -PEX10-sg2, p=3.53x10e-13. b. Tumor growth rates of indicated OVCAR-8 xenografts. GPX4−/−-sgNC: n=8 mice, other conditions: n=5 mice. P values for day-37 tumor sizes: GPX4+/+ vs GPX4−/−-sgNC, p=4.03x10e-7. GPX4−/−-sgNC vs -AGPS-sg2, p=0.000115; vs -FAR1-sg2, p=0.0446; vs -PEX3-sg1, p=0.0000213; vs -PEX10-sg1, p=0.0197. c. Tumor growth rates of indicated 786-O xenografts. n=5 mice. For day-27 tumors, GPX4+/+ vs GPX4−/−, p=1.4×10e-8. d. Immunoblot showing GPX4 protein levels in GPX4+/+ 786-O, the original ferroptosis-sensitive (FS) GPX4−/− clone, and the first round of ferroptosis-resistant (FR1) cells. e. Tumor growth curves of indicated 786-O xenografts. n=5 mice. P values for day-30 tumors, GPX4−/− FR1#a vs FS, p=1.96x10e-8; GPX4−/− FR1#d vs FS, p=3.42×10e-7. f. Immunoblot showing GPX4 protein levels in FR2 cells. FR2#a: from a FR1#a tumor; FR2#d: from a FR1#d tumor. L/R, left/right tumor. g. Volcano plot showing the lipidomic analysis comparing GPX4−/− FR2#d and GPX4+/+ 786-O cells. LPE, lysophosphatidylethanolamine; LPC, lysophosphatidylcholine; TAG, triacylglycerol; PUFA-, polyunsaturated fatty acyl-. n=6 biologically independent samples. h. Volcano plot showing the free fatty acid lipidomic analysis comparing GPX4−/− FR2#d and GPX4+/+ 786-O cells. n=6 biologically independent samples. i. Volcano plots showing the relative mRNA expression (RNA-seq) of peroxisome and ether-lipid metabolism-related genes comparing indicated conditions. n=4 biologically independent samples. See Methods for data analysis methods. j. Immunoblotting showing TMEM189 and AGPS protein levels in indicated cells. For immunoblots, β-ACTIN was used as a loading control. a,c,d,f,j, representative results of experiments performed twice. See Supplementary Information for uncropped immunoblot images. Data center and error bars: mean±s.d.. P-values were calculated using two-tailed Student’s T-test. Multiple-testing adjustment was performed using the Benjamini-Hochberg method. Two tumors were injected per mouse in animal experiments.
Figure 4.
Figure 4.. Neurons and cardiomyocytes acquire increased polyunsaturated ether phospholipids and gain sensitivity to ferroptosis during differentiation.
a. Fluorescent images showing the immunofluorescence analysis of neuronal marker β-3-tubulin expression in day 0 (parental) and day 12 (differentiated) SH-SY5Y cells. b. Viability curves of SH-SY5Y cells at each stage treated with ML210 or ML210+liproxstatin-1 (Lip-1) for 48 h. n=2 or 4 biologically independent samples. c. Volcano plots showing the lipidomic profiling results of day 0, 6 and 12 differentiating SH-SY5Y cells. n=3 biologically independent samples for parental and n=4 for day 6 or day 12 differentiation conditions. d. Bar plots showing relative viability of human iPSC-derived cardiac progenitors (CP) and mature cardiomyocytes (CM) treated with ML210 for 24h. n=6 biologically independent samples. e. Volcano plot showing the phospho-lipidome of CP (n=2 biologically independent samples) and CM (n=4 biologically independent samples). f. Bar plots showing the cardiomyocytes transiently transfected with siRNAs targeting control (si-NC), PEX3, or AGPS. n=3. g. Schematic diagram summarizing the peroxisomes and the polyunsaturated ether phospholipid biosynthesis pathway and their contribution to ferroptosis susceptibility in cell-state transitions. Data center and error bars: mean±s.d.. P-values in volcano plots and growth curves were calculated using two-tailed Student’s T-test. Multiple-testing adjustment was performed using the Benjamini-Hochberg method. Scale bars, 100 µm. a,b,d,e,f: representative data from experiments performed in duplicates. Abbreviations: ePL, ether-phospholipids; FA, fatty acids; LPA, lysophosphatidic acid; PUFA, polyunsaturated fatty acids; G3P, glycerol-3-phosphate; AGP, 1-O-alkyl G3P; ePE, ether-phosphatidylethanolamine; ePC, ether-phosphatidylcholine; CE, cholesterol esters; DAG, diacylglycerol; LOOH, phospholipid hydroperoxides.
See this image and copyright information in PMC

Comment in

  • Ferroptosis: the Good, the Bad and the Ugly.
    Aldrovandi M, Conrad M. Aldrovandi M, et al. Cell Res. 2020 Dec;30(12):1061-1062. doi: 10.1038/s41422-020-00434-0. Cell Res. 2020. PMID: 33149249 Free PMC article. No abstract available.
  • Peroxisome: the new player in ferroptosis.
    Tang D, Kroemer G. Tang D, et al. Signal Transduct Target Ther. 2020 Nov 24;5(1):273. doi: 10.1038/s41392-020-00404-3. Signal Transduct Target Ther. 2020. PMID: 33235217 Free PMC article. No abstract available.

References

    1. Stockwell BR et al. Ferroptosis: A Regulated Cell Death Nexus Linking Metabolism, Redox Biology, and Disease. Cell 171, 273–285 (2017). - PMC - PubMed
    1. Matsushita M et al. T cell lipid peroxidation induces ferroptosis and prevents immunity to infection. J. Exp. Med 212, 555–568 (2015). - PMC - PubMed
    1. Zou Y et al. A GPX4-dependent cancer cell state underlies the clear-cell morphology and confers sensitivity to ferroptosis. Nat. Commun 10, 1617 (2019). - PMC - PubMed
    1. Viswanathan VS et al. Dependency of a therapy-resistant state of cancer cells on a lipid peroxidase pathway. Nature 547, 453–457 (2017). - PMC - PubMed
    1. Hangauer MJ et al. Drug-tolerant persister cancer cells are vulnerable to GPX4 inhibition. Nature 551, 247–250 (2017). - PMC - PubMed

Additional References

    1. Schilder RJ et al. Metallothionein gene expression and resistance to cisplatin in human ovarian cancer. Int. J. Cancer 45, 416–422 (1990). - PubMed
    1. Cholody WM et al. Derivatives of fluorene, anthracene, xanthene, dibenzosuberone and acridine and uses thereof. US Patent (2012).
    1. Shimada K et al. Global survey of cell death mechanisms reveals metabolic regulation of ferroptosis. Nat. Chem. Biol 12, 497–503 (2016). - PMC - PubMed
    1. Paynter NP et al. Metabolic Predictors of Incident Coronary Heart Disease in Women. Circulation 137, 841–853 (2018). - PMC - PubMed
    1. Wang T et al. Identification and characterization of essential genes in the human genome. Science 350, 1096–1101 (2015). - PMC - PubMed

Publication types

MeSH terms