Thrombin induces ACSL4-dependent ferroptosis during cerebral ischemia/reperfusion

Abstract

Ischemic stroke represents a significant danger to human beings, especially the elderly. Interventions are only available to remove the clot, and the mechanism of neuronal death during ischemic stroke is still in debate. Ferroptosis is increasingly appreciated as a mechanism of cell death after ischemia in various organs. Here we report that the serine protease, thrombin, instigates ferroptotic signaling by promoting arachidonic acid mobilization and subsequent esterification by the ferroptotic gene, acyl-CoA synthetase long-chain family member 4 (ACSL4). An unbiased multi-omics approach identified thrombin and ACSL4 genes/proteins, and their pro-ferroptotic phosphatidylethanolamine lipid products, as prominently altered upon the middle cerebral artery occlusion in rodents. Genetically or pharmacologically inhibiting multiple points in this pathway attenuated outcomes of models of ischemia in vitro and in vivo. Therefore, the thrombin-ACSL4 axis may be a key therapeutic target to ameliorate ferroptotic neuronal injury during ischemic stroke.

Fig. 1

Arachidonic acid on the side chain of phospholipid is released after ischemia/reperfusion. a t-Distributed Stochastic Neighbor Embedding (tSNE) analysis shows remarkable stability of the instrument and good reproducibility of samples. b Heatmap analysis of the different expression (DE) lipids in contralateral (Contra) group and ipsilateral (Ipsi) group. The DE lipids were selected by VIP > 1, which was calculated by OPLS-DA analysis, and p value <0.05 (t test). c Different expression lipids in the Ipsi group. Values are shown as log2 fold change relative to the Contra group. Each dot denotes one lipid type, and the dot color represents lipid class, and the dot size shows significance (t test). d–g The abundance of PC (36:4, 16:0/20:4), PE (42:8, 22:4/20:4), PE (38:6, 18:2/20:4), and PE (40:7, 18:1/22:6) are significantly downregulated in Ipsi group. Data are mean ± SEM, n = 6. t test was performed. h The correlation among the 69 significantly changed lipids between Contra and Ipsi. The color of a node represents the ratio (Ipsi/Contra) of the lipids. The wathet edge shows a negative, and the orange edge indicates a positive correlation. i The ion characteristic fragments for the four reduced lipids. j The statistics of the fatty acid side chain of the four reduced lipids. k AA contents of the contralateral and ipsilateral hippocampus of mice were assayed 6 h after MCAO/R. Data are means ± SEM, n = 5. t test was performed

Fig. 2

Thrombin is upregulated after acute cerebral ischemia/reperfusion. a The unimodal distributions of the protein intensities suggests no obvious degradation in samples. b Distribution of log2-transformed intensity of identified proteins in 6 samples. Black presents Contra, and red denotes Ipsi. c Principal-component analysis shows a clear separation between Contra (red) and Ipsi (water blue). d Sample volcano plot for MCAO mice model showing –log10 (p value) and logFC values for all proteins with highlighting for those that are significantly upregulated (red dots) or downregulated (blue dots) after MCAO, and the most changed protein - prothrombin was labeled. Proteins in black are not significantly changed after MCAO. e Reactome enrichment for the 75 upregulated (ratio Ipsi/Contra >2, and p < 0.05, t test) and 111 downregulated (ratio Ipsi/Contra <0.5, and p < 0.05, t test) proteins, based on the Metascape. f Protein-protein interaction (PPI) analysis for the 75 upregulated proteins. The size and color for the node represent the ratio of Ipsi/Contra. g Levels of prothrombin assayed by mass spectrometry in the contralateral and ipsilateral hippocampus of mice 6 h after MCAO/R. Data are means ± SEM, n = 3. t test was performed. h Thrombin protein levels were examined from the contralateral and ipsilateral hippocampus of mice 6 h after MCAO/R. Western blots were analyzed with Image J and normalized to β-actin expression. Data are means ± SEM, n = 5. t test was performed

Fig. 3

Thrombin induces neuronal ferroptosis. a Thrombin cytotoxicity in N27 cells. Data are means ± SEM, n = 6 wells from one representative of 3 independent experiments. b AA content was assayed from the N27 cells 24 h after thrombin (0.5 U/mL) treatment. Data are means ± SEM, n = 5 wells from one representative of 3 independent experiments. t test was performed. c Lipid ROS in N27 cells treated with thrombin for 24 h (representative histogram plot for fluorescence of oxidized BODIPY-C11). d Relative lipid ROS is expressed as the ratio of oxidized to reduced BODIPY-C11 mean fluorescence intensity in N27 cells treated with thrombin for 24 h. Data are means ± SEM, n = 3 wells from one representative of 3 independent experiments. t test was performed. e Transmission electron microscopy (TEM) of N27 cells treated with thrombin (0.5 U/mL) for 24 h. Yellow arrows indicate shrunken mitochondria. f Detection of intracellular Fe²⁺ in N27 cells treated with thrombin (0.5 U/mL) for 6 h using FerroOrange. Data are means ± SEM, n = 3 wells from one representative of 3 independent experiments. t test was performed. g Cell viability of N27 cells 24 h after thrombin (0.5 U/mL) and Liproxstatin-1 (Lip-1) of concentration gradient co-treatment. Data are means ± SEM, n = 6 wells from one representative of 3 independent experiments. h, i Cell viability of N27 cells 24 h after thrombin (0.5 U/mL), with NAC (h) or Darapladib (i) co-treatment. Data are means ± SEM, n = 6 wells from one representative of 3 independent experiments. j, k Cell viability of MDA-MB-231 cells 24 h after thrombin (0.5 U/mL), with Fer-1 (j), Lip-1 (k) co-treatment. Data are means ± SEM, n = 5 wells from one representative of 3 independent experiments

Fig. 4

Oxygen-dependent ferroptosis in cerebral reperfusion injury. a LPO of the hippocampus and cortex of mice were assayed following MCAO/R for 24 h. Data are means ± SEM, n = 5 animals per group. Two-way ANOVA with post hoc Sidak test was performed. b MDAs of the hippocampus and cortex of mice were assayed following MCAO/R for 24 h. Data are means ± SEM, n = 5 animals per group. Two-way ANOVA with post-hoc Sidak test was performed. c MDA was detected in the hippocampus of mice treated with RSL3 following MCAO/R for 24 h. Data are means ± SEM, n = 5 animals per group. One-way ANOVA with post-hoc Tukey test was performed. d Neurological score (higher numbers indicating more severe impairment) was performed at 0, 6, and 24 h after MCAO/R. Data are means ± SEM. Sham, n = 5; MCAO + DMSO, n = 7; MCAO + RSL3, n = 5. Two-way ANOVA with post-hoc Tukey test was performed. e Representative 2,3,5-triphenyl tetrazolium chloride (TTC)-stained serial brain sections of mice 24 h after MCAO/R, where viable tissue stains red. Quantification of infarction volume indicated by TTC staining using Image J. Data are means ± SEM. Sham, n = 5; MCAO + DMSO, n = 7; MCAO + RSL3, n = 5. One-way ANOVA with post-hoc Tukey test was performed. f Neurological score was performed at 1, 6, and 24 h after permanent-MCAO. Data are means ± SEM, n = 6 animals per group. Two-way ANOVA with post-hoc Tukey test was performed. g Representative TTC-stained serial brain sections of mice 24 h after permanent-MCAO, where viable tissue stains red. Quantification of infarction volume indicated by TTC staining using Image J. Data are means ± SEM, n = 6 animals per group. One-way ANOVA with post hoc Tukey test was performed. h Neuronal cell viability after OGD 2 h/reoxygenation 18 h. Data are means ± SEM, n = 10 wells from one representative of 4 independent experiments. t test was performed. i Detection of intracellular Fe²⁺ in N27 cells using FerroOrange at 2 h of OGD. Data are means ± SEM, n = 6 wells from one representative of 3 independent experiments. t test was performed. j Lipid ROS in N27 cells treated with OGD for 2 h (representative histogram plot for fluorescence of oxidized BODIPY-C11). k Relative lipid ROS is expressed as the ratio of oxidized to reduced BODIPY-C11 mean fluorescence intensity (MFI) in N27 cells treated with OGD for 2 h. Data are means ± SEM, n = 6 wells from one representative of 3 independent experiments. t test was performed. l Cell viability of N27 cells 24 h after OGD and Liproxstatin-1 (Lip-1) of concentration gradient co-treatment. Data are means ± SEM, n = 6 wells from one representative of 3 independent experiments

Fig. 5

Platelet activation after ischemic stroke. a The statistic of Spearman coefficients for 59 samples in two technical workflow replicates. b The unimodal distributions of the protein intensity suggest no palpable degradation in the serum sample. c The distribution of peptides for quantified proteins. d The distribution of quantified protein numbers in samples. e tSNE analysis of the proteomics for 59 serum samples, and no batch effects were observed. healthy controls (HC), ischemic stroke (IS). f A volcano plot of the 362 proteins, quantified in > 50% samples, and imputed by random forest. g The significantly changed (p < 0.05 and FDR < 0.25) pathway in ischemic stroke patients, based on gene set enrichment analysis (GSEA). h, i GSEA analysis of common pathway of fibrin clot formation (h) and immunoregulatory interactions between a lymphoid and a non-lymphoid cell (i) gene sets enriched among 362 proteins. j A heatmap of GSEA enriched proteins in the common pathway of fibrin clot formation gene set from 362 proteins. k Protein–protein analysis of the 22 upregulated proteins in stroke. l Using mass spectrometry to detect plasma prothrombin abundance in clinical ischemic stroke (IS) patients and healthy controls (HC). Data are means ± SEM. HC, n = 27; IS, n = 32. t test was performed

Fig. 6

Thrombin induces the downregulation of ACSL4 after acute cerebral ischemia/reperfusion. a ACSL4 protein level was examined in N27 cells treated with thrombin (0.5 U/mL) for 24 h. Western blots were analyzed with Image J and normalized to β-actin expression. Data are means ± SEM, n = 4. t test was performed. b ACSL4 protein levels were examined from the ischemic ipsilateral hippocampus of rats that underwent MCAO after 0, 6, or 24 h of reperfusion. Western blots were analyzed with Image J and normalized to β-actin expression. Data are means ± SEM, n = 3 animals per group. One-way ANOVA with post-hoc Tukey test was performed. c Degeneration of CA1 pyramidal neurons following I/R. Hippocampal sections from sham, R3h (3 h after MCAO/R), and R6h (6 h after MCAO/R) rats were stained with NeuN (red) and Fluoro-Jade (green). Representative Immunofluorescence staining for ACSL4 from adjacent brain tissue sections spaced 4 μm apart. d Quantification of survival and death neuron numbers in a region of interest (ROI = 0.1 mm²) Data are means ± SEM, n = 3. Two-way ANOVA with post-hoc Tukey test was performed. e The intensity of ACSL4 immunofluorescence in an ROI was quantified using Image J. Data are means ± SEM, n = 3 animals per group. One-way ANOVA with post-hoc Tukey test was performed. f Thrombin protein levels were examined from the ischemic ipsilateral hippocampus of rats 0.5, 1, and 1.5 h after MCAO. Western blots were analyzed with Image J and normalized to β-actin expression. Data are means ± SEM, n = 3 animals per group. One-way ANOVA with post-hoc Tukey test was performed. g Cell viability of WT, ACSL4 KO, and ACSL4 OE N27 cells 24 h after thrombin (0.5 U/mL) treatment. Data are means ± SEM, n = 6 wells from one representative of 3 independent experiments. One-way ANOVA with post-hoc Tukey test was performed

Fig. 7

Modulating ACSL4 expression altered the outcomes of acute ischemic brain injury. a Schematic of the experimental paradigm. b The location of AAV injection and ACSL4 expression in the brain of mice were verified. c Representative images of TTC staining of EGFP and ACSL4 OE mice 24 h after 30 min of MCAO. Quantification of infarction volume indicated by TTC staining using Image J. Data are means ± SEM, n = 5 animals per group. t test was performed. d The neurological score was performed at 0 h, 6 h, 24 h, 3 days, and 5 days after MCAO/R. Data are means ± SEM. EGFP, n = 7; ACSL4 OE, n = 9. Two-way ANOVA with post-hoc Sidak test was performed. e The performance on the rotarod test was analyzed at 1 day, 3 days, and 5 days after MCAO surgery. Data are means ± SEM. Sham, n = 6; EGFP, n = 7; ACSL4 OE, n = 9. Two-way ANOVA with post-hoc Tukey test was performed. f Representative images of TTC staining of EGFP and ACSL4 KO mice 24 h after 60 min of MCAO. Quantification of infarction volume indicated by TTC staining using Image J. Data are means ± SEM, n = 5 animals per group. t test was performed. g Neurological scoring was performed at 0 h, 6 h, 24 h, 3 days, and 5 days after MCAO/R. Data are means ± SEM, n = 6 animals per group. Two-way ANOVA with post-hoc Sidak test was performed. h The performance on the rotarod test was analyzed at 1 day, 3 days, and 5 days after MCAO surgery. Data are means ± SEM, n = 6 animals per group. Two-way ANOVA with post-hoc Tukey test was performed. i Neurological scoring was performed at 1, 6, and 24 h after permanent-MCAO. Data are means ± SEM. EGFP, n = 5; ACSL4 KO, n = 6. Two-way ANOVA with post-hoc Sidak test was performed. j Representative images of TTC staining of EGFP and ACSL4 KO mice 24 h after permanent-MCAO. Quantification of infarction volume indicated by TTC staining using Image J. Data are means ± SEM. EGFP, n = 5; ACSL4 KO, n = 6. t test was performed. k Representative images obtained from PeriCam PSI System. The brighter area indicates higher blood perfusion. l Cortical blood flow changes before and after MCAO/R in rats. m The neurological score was performed at 0, 6, and 24 h after MCAO/R in rats. Data are means ± SEM. EGFP, n = 9; ACSL4 KO, n = 12. Two-way ANOVA with post-hoc Sidak test was performed. n Representative TTC-stained serial brain sections of rats 24 h after MCAO/R, where viable tissue stains red. Quantification of infarction volume indicated by TTC staining using Image J. Data are means ± SEM. EGFP, n = 9; ACSL4 KO, n = 12. t test was performed

Fig. 8

Schematic hypothesis. Cerebral ischemia leads to an unexpected increase in thrombin within neurons, promotes the mobilization of PE and PC in the phospholipid membrane of neuronal cells through cPLA2α, and accelerates the production of AA. Under the catalytic action of ACSL4, AA is esterified and made available as a ferroptotic fuel. This injury process can be blocked by thrombin inhibitors, cPLA2α inhibitors, and ACSL4 inhibitors. Concurrently, iron accumulates during I/R, which also contributes to ferroptosis and can be blocked by an iron chelator