ACSL4 dictates ferroptosis sensitivity by shaping cellular lipid composition

Abstract

Ferroptosis is a form of regulated necrotic cell death controlled by glutathione peroxidase 4 (GPX4). At present, mechanisms that could predict sensitivity and/or resistance and that may be exploited to modulate ferroptosis are needed. We applied two independent approaches-a genome-wide CRISPR-based genetic screen and microarray analysis of ferroptosis-resistant cell lines-to uncover acyl-CoA synthetase long-chain family member 4 (ACSL4) as an essential component for ferroptosis execution. Specifically, Gpx4-Acsl4 double-knockout cells showed marked resistance to ferroptosis. Mechanistically, ACSL4 enriched cellular membranes with long polyunsaturated ω6 fatty acids. Moreover, ACSL4 was preferentially expressed in a panel of basal-like breast cancer cell lines and predicted their sensitivity to ferroptosis. Pharmacological targeting of ACSL4 with thiazolidinediones, a class of antidiabetic compound, ameliorated tissue demise in a mouse model of ferroptosis, suggesting that ACSL4 inhibition is a viable therapeutic approach to preventing ferroptosis-related diseases.

Fig. 1. Identification of Acsl4 as an essential pre-requisite for ferroptosis execution

(a) Top, time-dependent increase of LDH release in Pfa1 cells and RSL3-resistant clones (cB1, cB3 and cB7) upon Gpx4 knockout induction. Below, Gpx4 immunoblot analysis 48 h after TAM treatment. (b) Heat maps of the 16 most down-regulated genes as detected by microarray analysis (Parental vs cB1/cB4/cB7 p<0.01). (c) Top 12 gRNAs identified in the population selected either with RSL3 (100 nM) or with ERA (1 μM). (d) Pfa1_Acsl4 KO (Acsl4−/−) cells are refractory to ferroptosis, which can be re-sensitized by Flag-tagged hACSL4 (ACSL4-FLAG) re-expression. Dose-dependent cytotoxicity of RSL3 was assessed 24 h after treatment using AquaBluer. Immunoblot depicts re-expression of ACSL4-FLAG in Pfa1_Acsl4 KO cells. (e) Time-dependent increase of LDH release was assayed in a 96 well plate using Pfa1_Acsl4 KO (Acsl4−/−) and Pfa1_Acsl4 WT (Acsl4 +/+) cells during KO induction. Immunoblot analysis of Gpx4 is shown at different time points after TAM treatment. Data shown represents the mean ± s.d. of n = 3 wells from a representative experiment performed independently two times (a), three times (d) or four times (e).

Fig. 2. Acsl4 specifically contributes to ferroptotic cell death

(a) Ultrastructural analysis revealed that upon RSL3 treatment Acsl4 WT cells unlike KO cells show outer membrane rupture (red arrows) as reported (Scale bars: 500 nm). (b) RSL3 induced lipid oxidation in WT Pfa1 cells (Acsl4 +/+) unlike in Pfa1_Acsl4 KO cells (Acsl4−/−) as determined by BODIPY 581/591 C11 and SPY-LHP oxidation at the indicated time points. (c) Knockout of Lpcat3 weakly protected against ferroptosis. Cell viability was determined 24 h after RSL3 treatment using AquaBluer. qPCR analysis indicates the relative expression of Lpcat3 in Pfa1 WT, Acsl4 −/− and Lpcat3 −/− cells. Time-dependent increase of LDH release in Pfa1_Acsl4 KO (Acsl4 −/−) and Pfa1_Acsl4 WT (Acls4 +/+) and Pfa1_Lpcat3 KO (Lpcat3 −/−) cells with or without TAM treatment. Data shown in and (c) represents the mean ± s.d. of n = 3 wells from a representative experiment performed independently three times.

Fig. 3. Acsl4 sensitizes to ferroptosis by specifically esterifying AA and AdA into PE

(a) Typical MS spectra of phosphatidylethanolamine (PE) from control (black) and Acsl4 −/− (red) Pfa1 cells. Inserts: MS spectra of PE molecular species (18:0/20:4) (left) and (18:0/22:4) (right) in the range of m/z 766.48–766.59 and 794.46–794.64, respectively. (b) Quantitative assessment of PE molecular species (18:0/20:4) (left panel) and (18:0/22:4) (right panel) in control WT and Acsl4 −/− cells in the absence and in the presence of AA. Cells were supplemented with arachidonic acid (AA) (2.5 μM, 16 hrs at 37°C). Data are mean ± s.d., n = 4. (c) Quantitative assessment of hydroperoxy-PE molecular species (18:0/20:4) (left panel) and (18:0/22:4) (right panel) in RSL3-exposed WT and Acsl4 −/− cells in presence and in the absence of AA. Cells were treated with AA for 16 hrs and exposed to RSL3 (100 nM) for 6 hrs at 37°C. Data represent the mean ± s.d., n = 4.

Fig. 4. Acsl4 predicts sensitivity in a panel of basal-like breast cancer cell lines

(a) Basallike breast cancer cells are susceptible to ferroptosis induction by RSL3 as determined by AquaBluer 48 h after treatment. For an extended panel of breast cancer cell lines please see Supplementary Fig. 4a. (b) Immunoblot analysis of ACSL4 and GPX4 in a panel of breast cancer cell lines. (c) Pfa1_Acsl4^−/− cells stably transfected with a doxycycline-inducible ACSL4-FLAG expression vector (pRTS1(ACSL4-FLAG)) (right) demonstrate that minute amounts of ACSL4 are required for RSL3-induced ferroptosis as assessed using the AquaBluer assay (left). (d) ACSL4 KO in MDA-MB-157 confers resistance to RSL3-triggered ferroptosis. Single cell clones selected after transient transfection with Cas9 and a gRNA targeting ACSL4 were tested for their sensitivity to RSL3 and KO was confirmed by immunoblotting. (e) MDA157_Acsl4 KO cells are refractory to RSL3-triggered lipid peroxidation (100 nM) using BODIPY 581/591 C11 staining after 3h.(f) Expression of FLAG-tagged human ACSL4 (ACSL4-FLAG) in breast cancer cells devoid of endogenous ACSL4 sensitizes to RSL3-induced ferroptosis (g). (h) Cells expressing Flag_hACSL4 show increased levels of BODIPY 581/591 C11 oxidation upon treatment with RSL3 (100 nM – 4h). Data shown represents the mean ± s.d. of n ≥ 3 wells from representative experiments performed independently two times (d), three times (c, g) or four times (a).

Fig. 5. Thiazolidinediones protect from ferroptosis through inhibition of Acsl4

(a) Top: the thiazolidinedione, troglitazones (TRO), rosiglitazone (ROSI) and pioglitazone (PIO), protect from RSL3-induced cell death. Bottom: ROSI treatment phenocopies Acsl4 knockout measured by LDH release upon Gpx4 deletion. Data shown represents the mean ± s.d. of n = 3 wells of a 96-well plate from a representative experiment performed independently 2 times. (b) Thiazolidinediones prevent BODIPY 581/591 C11 oxidation induced by 125 nM RSL3 for 3h. (c) MS spectra of PE species at the indicated conditions. Highlighted in red are the molecular species of PE decreased by the treatment with ROSI and Acsl4 KO. (d) Heatmap of all major PE species with hierarchical clustering of the groups WT, WT+ROSI, Acsl4 KO, Acsl4 KO+ROSI. Each PE species was normalized to the corresponding mean value. (e) Effects of ROSI and Acsl4 KO on two major molecular species of PE(18:0/20:4) and PE(18:0/22:4) representing substrates for oxygenation during ferroptosis. Data represents the mean ± s.d. of n = 4 samples, *p<0.05 vs WT. (f) Principal component analysis (PCA) score plot of the first two principal components of PE between four groups. PE data were mean-centered and UV-scaled. The red circle highlights the close similarity of the effects of ROSI and Acsl4 KO on the resulting PE profiles. (g) ROSI treatment delays mortality due to acute renal failure in TAM-inducible Gpx4 knockout mice. Drinking water treatment started 1 week prior to TAM injection.