Coenzyme A protects against ferroptosis via CoAlation of mitochondrial thioredoxin reductase

Abstract

The cystine-xCT transporter/glutathione/GPX4 axis is the canonical pathway protecting cells from ferroptosis. Whereas GPX4-targeting ferroptosis-inducing compounds (FINs) act independently of mitochondria, xCT-targeting FINs require mitochondrial lipid peroxidation, though the mechanism remains unclear. Because cysteine is also a precursor for coenzyme A (CoA) biosynthesis, here, we demonstrated that CoA supplementation selectively prevented ferroptosis triggered by xCT inhibition by regulating the mitochondrial thioredoxin system. Our data showed that CoA regulated the in vitro enzymatic activity of mitochondrial thioredoxin reductase-2 (TXNRD2) by covalently modifying the thiol group of cysteine (CoAlation) on Cys-483. Replacing Cys-483 with alanine on TXNRD2 abolished its enzymatic activity and ability to protect cells against ferroptosis. Targeting xCT to limit cysteine import and, therefore, CoA biosynthesis reduced CoAlation on TXNRD2. Furthermore, the fibroblasts from patients with disrupted CoA metabolism had increased mitochondrial lipid peroxidation. In organotypic brain slice cultures, inhibition of CoA biosynthesis led to an oxidized thioredoxin system, increased mitochondrial lipid peroxidation, and loss of cell viability, which were all rescued by ferrostatin-1. These findings identified CoA-mediated posttranslational modification to regulate the thioredoxin system as an alternative ferroptosis protection pathway with potential clinical relevance for patients with disrupted CoA metabolism.

Keywords: Amino acid metabolism; Cell biology; Cell stress; Metabolism; Mitochondria.

Figure 1. CoA is a specific ferroptosis inhibitor against xCT inhibitors.

(A and B) CoA supplementation increased the levels of intracellular CoA (A) and acetyl-CoA (B) as quantified by LC-MS/MS analysis. HT-1080 cells were treated with H₂O and 2 concentrations of CoA (30 μM and 100 μM) for 18 hours for LC-MS/MS analysis. (C) CoA inhibited erastin-induced ferroptosis. HT-1080 cells were treated with increasing doses of erastin, either alone or in combination with CoA (100 μM) or deferoxamine (80 μM), Fer-1 (10 μM), lipro (2 μM), or Trolox (100 μM). The cell viability was quantified by Cell-Titer Glo assay. (D and E) Erastin-induced lipid peroxidation (2 μM, 18 hours) in HT-1080 cells was inhibited by CoA treatment as determined by C11-BODIPY staining (D) and the quantification of the percentage of lipid peroxidation–positive cells (E). (F and G) CoA (100 μM) inhibited erastin-induced membrane rupture (erastin: 2.5 μM, 20 hours) in HT-1080 cells, as observed by CellTox Green under fluorescence microscope (F) and quantified by a plate reader (G). (H) CoA (100 μM) inhibited class I FIN–induced ferroptosis (via (sulfasalazine [SAS], 20 hours) in HT-1080 cells. (I and J) CoA (100 μM) inhibited 24 hours of cystine deprivation-induced ferroptosis as determined by Cell-Titer Glo assay (I) or CellTox Green assay (J). (K–M) CoA (100 μM) did not inhibit ferroptosis in HT-1080 cells induced by RSL3 (class II FIN, 20 hours) (K), FIN56 (class III FIN, 20 hours), and FINO2 (class IV FIN, 20 hours) (M). (A, B, E, and G) Data were analyzed by 1-way ANOVA and Tukey’s multiple comparisons. (C and H–M) Data were analyzed by 2-way ANOVA and Šídák’s multiple comparisons; n = 3 independent biological replicates. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Data represent mean ± SEM.

Figure 2. CoA regulates mitochondrial TXN system.

(A) HT-1080 cells were treated with CoA (100 μM) alone or in combination with the TXN inhibitor ferroptocide (2 μM), during erastin-induced ferroptosis (20 hours). Cell viability was quantified using the Cell-Titer Glo assay. (B) TXNRD activity was significantly reduced after erastin treatment (1.25 μM, 16 hours) in HT-1080 cell lysates, and this repression was restored by CoA (100 μM) supplementation. (C) Pooled siRNA knockdown of TXN2, but not TXN1, abolished CoA-mediated protection from ferroptosis, indicating a specific role for mitochondrial TXN2. siNC, negative control siRNA. (D) Erastin disrupted the interaction between TXNRD2 and TXN2, which was restored by CoA. HT-1080 cells overexpressing TXNRD2 and TXN2 were treated with erastin (2.5 μM, 18 hours), with or without CoA, and lysed in NEM buffer for coimmunoprecipitation analysis. (E) Western blot analysis revealed that erastin reduced the levels of PRDX3 monomers (reduced, active forms), which was reversed by CoA supplementation. (F) Monomer/dimer ratios of PRDX3 were quantified in cells treated with erastin and CoA. Data were analyzed by 1-way ANOVA and Tukey’s test. (A–C) Data were analyzed by 2-way ANOVA and Šídák’s test; n = 3 biological replicates. Data are reported as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Figure 3. Combined inhibition of GSH and CoA synthesis leads to synthetic lethality.

(A) Knockdown of COASY using 2 independent siRNAs decreased PRDX3 monomer levels, further linking mitochondrial redox regulation to CoA biosynthesis. (B) Quantification of PRDX3 monomer/dimer ratios upon COASY knockdown. siNC, negative control siRNA. (C) Stable COASY knockdown sensitized HT-1080 cells to BSO-induced ferroptosis, which was rescued by CoA (100 μM), 4′-PPT (100 μM), or Fer-1 (10 μM). (D) CoA inhibited ferroptosis induced by BSO and PANKi cotreatment. HT-1080 cells were exposed to PANKi (2.5 μM) and increasing doses of BSO with Fer-1, lipro (2 μM), Trolox (100 μM), or CoA (100 μM). (E and F) PANKi, but not BSO, increased mitochondrial lipid peroxidation, visualized with mitoPerOx (E) and quantified by mean fluorescence intensity (F). (G) COASY knockdown increased mitochondrial lipid peroxidation, which was rescued by CoA. (B and F) Data were analyzed by 1-way ANOVA and Tukey’s test. (C, D, and G) Data were analyzed by 2-way ANOVA and Šídák’s test; n = 3 biological replicates. Data are mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Figure 4. CoA supplement increased CoAlation of TXNRD2 and its interaction with TXN2.

(A) HT-1080 cells were cotransfected with expression vectors for TXNRD2 (V5-tagged) and TXN2 (Myc-tagged), followed by treatment with CoA (100 μM) for 18 hours. TXNRD2 was immunoprecipitated using anti-V5 beads and analyzed by nonreducing SDS-PAGE. CoAlation was detected with a pan-CoAlation antibody, showing enhanced conjugation upon CoA supplementation. Interaction between TXNRD2 and TXN2 was assessed by probing the V5 pulldown for TXN2 using an anti-Myc antibody. CoA treatment increased TXN2 association, suggesting CoAlation promotes TXNRD2-TXN2 complex formation. (B) Confocal microscopy showed that CoA supplement increased TXNRD2-TXN2 interaction in mitochondria. COX IV is a mitochondria marker. (C) Specificity of CoAlation detection was validated by DTT treatment, which abolished the signal. V5-purified TXNRD2 was incubated with oxidized CoA ± DTT, analyzed by nonreducing PAGE, and immunoblotted. (D) CoAlation enhanced TXNRD activity. Activity was measured using purified TXNRD2 with or without CoAlation or TXNRD inhibitor (auranofin), alongside background (bk) and TrxR(+) control. Data were analyzed by 2-way ANOVA and Šídák’s test; n = 3 biological replicates. **P < 0.01. Data presented as mean ± SEM.

Figure 5. CoAlation of Cys-483 on TXNRD2 protein regulates TXNRD activity.

(A) Tandem MS identified the CoA-modified peptide C*GASYAQVMR, localizing the modification to the N-terminal cysteine. (B) Relative quantification confirmed increased abundance of the modified peptide in oxidized CoA-treated versus DMSO samples, confirming C-483 CoAlation. (C) Mutation of Cys-483 to alanine (C483A) abolished most TXNRD2 CoAlation. Wild-type and mutant proteins were analyzed via nonreducing PAGE and Western blot. (D) The C483A mutation also abolished TXNRD activity, which could not be rescued by oxidized CoA. (E) Erastin treatment (2 μM, 16 hours) reduced CoAlation levels in HT-1080 cells overexpressing TXNRD2. (F) Cys-483 is essential for TXNRD2’s anti-ferroptotic function. Cells with TXNRD2 knockout were reconstituted with wild-type or C483A mutant, then treated with erastin (20 hours); only wild-type restored viability. (D and F) Data were analyzed by 2-way ANOVA and Šídák’s test; n = 3 biological replicates. *P < 0.05, **P < 0.01. Data presented as mean ± SEM. m/z, mass/charge ratio.

Figure 6. Disruption of CoA biosynthesis in PKAN fibroblasts and OBSC leads to mitochondrial lipid peroxidation.

(A) PANK1 knockdown sensitized HT-1080 cells to BSO treatment. HT-1080 cells with shRNA targeting PANK1, PANK2, or PANK3 were treated with BSO for 3 days for Cell-Titer Glo assay. Data were analyzed by 2-way ANOVA and Šídák’s multiple comparisons; n = 3 independent biological replicates. (B) PANK1, PANK2, and COASY knockdown in HT-1080 cells showed an increase in mitochondrial lipid peroxidation. (C and D) Primary fibroblasts from patients with PKAN and age- and sex-matched unaffected individuals were stained with mitoPerOx, a probe that detects mitochondrial lipid peroxides. Representative images of mitoPerOx staining in (C) demonstrate a visibly higher signal intensity in PKAN fibroblasts. Quantification of the percentage of mitoPerOx⁺ cells is presented in (D), confirming a significant increase in mitochondrial lipid peroxidation in fibroblasts from patients with PKAN. Data were analyzed by unpaired t test. (E and F) The cell death triggered by inhibiting CoA using PANKi in OBSC was rescued by Fer-1. OBSCs transfected with YFP were treated with PANKi (2.5 μM) or in combination with Fer-1 (2 μM, 24 hours). Neuronal viability was assessed by fluorescence microscopy: the number of YFP⁺ neurons remaining after treatment was quantified by fluorescence microscopy as a measure of cell survival (E and F). Representative images of YFP⁺ neurons are shown in (E), and quantification is presented in (F). (G and H) The elevated mitochondrial lipid peroxidation by PANKi was rescued by Fer-1. After 1 day of treatment with PANKi (2.5 μM) and Fer-1 (2 μM), OBSC was stained with mitoPerOx (G) for quantification (H). (I and J) PANKi treatment of brain tissue sections repressed the levels of reduced form of PRDX3, which was rescued by Fer-1. OBSC treated with PANKi (2.5 μM) or in combination with Fer-1 (2 μM) were blotted with PRDX3 (I) and quantification (J). (B, F, H, and J) Data were analyzed by 1-way ANOVA and Tukey’s multiple comparisons. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Data are mean ± SEM.