Ferroptosis surveillance independent of GPX4 and differentially regulated by sex hormones

Abstract

Ferroptosis, a cell death process driven by iron-dependent phospholipid peroxidation, has been implicated in various diseases. There are two major surveillance mechanisms to suppress ferroptosis: one mediated by glutathione peroxidase 4 (GPX4) that catalyzes the reduction of phospholipid peroxides and the other mediated by enzymes, such as FSP1, that produce metabolites with free radical-trapping antioxidant activity. In this study, through a whole-genome CRISPR activation screen, followed by mechanistic investigation, we identified phospholipid-modifying enzymes MBOAT1 and MBOAT2 as ferroptosis suppressors. MBOAT1/2 inhibit ferroptosis by remodeling the cellular phospholipid profile, and strikingly, their ferroptosis surveillance function is independent of GPX4 or FSP1. MBOAT1 and MBOAT2 are transcriptionally upregulated by sex hormone receptors, i.e., estrogen receptor (ER) and androgen receptor (AR), respectively. A combination of ER or AR antagonist with ferroptosis induction significantly inhibited the growth of ER⁺ breast cancer and AR⁺ prostate cancer, even when tumors were resistant to single-agent hormonal therapies.

Keywords: MBOAT1; MBOAT2; androgen receptor; estrogen receptor; ferroptosis; phospholipid remodeling; sex hormone signaling.

Figure 1.. MBOAT2 is a GPX4/FSP1-independent ferroptosis suppressor.

(A) Schematic plot summarizing the workflow of CRISPR activation screen in HT1080 cells. (B) Top seven genes enriched in both RSL3 and cystine starvation conditions are highlighted. (C) Visualization of enrichment for sgRNAs targeting top seven genes in cystine starvation (top) and RSL3 (bottom) screen conditions. (D) Western blot confirming overexpression of MBOAT2 and PLA2G2F in HT1080 cells. (E, G) Viability analysis of HT1080 cells overexpressing indicated genes. Cells were treated with 0.3 μM IKE (E) or 0.1 μM RSL3 (G) for 24 h, in the absence or presence of ferrostastin-1 (Fer1, 10 μM). (F) Cell death time course of HT1080 cells overexpressing indicated genes. Ferroptosis was induced by cystine starvation. (H) Western blot confirming knockout of GPX4 and/or FSP1 in HT1080 cells. (I, J) Crystal violet staining of HT1080-GPX4^KO cells (I) or HT1080-GPX4/FSP1^DKO cells (J) overexpressing indicated genes cultured with or without Trolox for 72 h. Two independent experiments were performed, and representative images from one experiment are shown. Data are presented as mean ± SD, n = 3 biologically independent samples (E, G). Statistical analysis was performed using 1-way ANOVA. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (P value classification applies to all later figures). See also Figure S1.

Figure 2.. The ferroptosis-suppressing function of MBOAT2 requires either endogenous or exogenous MUFA.

(A) Viability analysis of HT1080-GPX4^KO cells overexpressing GCH1 or MBOAT2. Cells were treated for 48 h with TOFA (5 μM) in the presence or absence of Trolox (100 μM) as indicated. (B) (B, D) Viability analysis of HT1080-MBOAT2 cells. Cells were pretreated with indicated concentration of TOFA (B) or CAY10566 (D) for 24 h, followed by ferroptosis induction with RSL3 (0.1 μM) in the presence or absence of Trolox for another 24 h. (C) Viability assay of HT1080-Vector and HT1080-MBOAT2 cells. Cells were pretreated with or without TOFA (5 μM) as indicated for 24 h, followed by indicated concentrations of RSL3 for another 24 h. (E) Western blot analysis showing expression of SCD1 and MBOAT2 in indicated HT1080 cells. (F) Cell death time course of cells with indicated genetic background and with or without MBOAT2 overexpression. Cells were originally cultured in the presence of Trolox, and ferroptosis was initiated by removing Trolox (that was the starting point of the time course). (G) Viability analysis of HT1080-GPX4/SCD^DKO cells with or without MBOAT2 overexpression as indicated. Cells were pretreated with indicated concentration of Oleic Acid (OA) plus Trolox for 16 h, and ferroptosis was induced by removing Trolox but keeping same amount of OA, for another 24 h. (H) Western blot analysis confirming knockdown of MBOAT2 in HT1080 cells. (I-K) Viability analysis of HT1080 cells with control or MBOAT2 shRNA. Cells were pretreated with indicated concentration of OA for 16 h and followed by ferroptosis induction with 0.1 μM RSL (I), 1 μM ML210 (J), or 0.5 μM IKE (K) for 24 h. (L) Working model showing MBOAT2 utilizes endogenous or exogenous MFUA to suppress ferroptosis. Data are presented as mean ± SD, n=3 biologically independent samples (A-D, F, G, I-K). Statistical analysis was performed using 1-way ANOVA (B, D) or two-tailed t-test (A, I-K). See also Figure S2.

Figure 3.. MBOAT2 suppresses ferroptosis through phospholipid remodeling.

(A) Volcano plot showing upregulated lipid species (Red) and downregulated lipid species (Blue) by MBOAT2 overexpression in HT1080 cells. Cut-off: FC threshold = 2, P<0.01; two-tailed t-test. See also Table S2. (B-D) Quantification of most abundant PE-MUFAs (B), PC-MUFAs (C), and PE-AAs (D) in HT1080-vector and HT1080-MBOAT2 cells as indicated. (E, F) Stacked bars showing relative abundance of indicated PE-OA and PE-AA in HT1080-vector and HT1080-MBOAT2 cells. (G) Western blot showing the expression of MBOAT2 wildtype (WT) and H373A mutant in HT1080-GPX4^KO cells. (H) Cell death time course of HT1080-GPX4^KO cells overexpressing MBOAT2 WT or H373A mutant as indicated. Ferroptosis was initiated by removing Trolox from culture medium. (I) Western blot showing MBOAT2 knockdown in SUIT-2 cells. (J, K) Viability analysis of SUIT-2 cells expressing control or MBOAT2 shRNA. Ferroptosis was induced by indicated concentration of RSL3 for 24 h. Data are presented as mean ± SD, n=5 (B, C, D) or n=3 (J, K) biologically independent samples. Statistical analysis was performed using two-tailed t-test (B, C, D). See also Figure S3.

Figure 4.. AR signaling modulates ferroptosis sensitivity in prostate cancer cells through MBOAT2.

(A) MBOAT2 mRNA expression in cancer patient samples (TCGA PanCancer Atlas Studies, cBioPortal). MBOAT2 mRNA expression was batch normalized from Illumina HiSeq_RNASeqV2. Dash line indicated median. (B) Western blot analysis showing endogenous MBOAT2 expression in a panel of AR⁺ and AR⁻ PCa lines. AR⁺ PCa lines were treated with DMSO or 5 μM ENZ for 48 h. (C) Viability analysis of a panel of AR⁺ and AR⁻ PCa lines. Ferroptosis was induced by 3 μM RSL3 for 24 h. (D) Visualization of AR and FOXA1 binding to the MBOAT2 ARE region (normalized by RPKM, GSE37345). (E) ChIP-qPCR showing the occupancy of AR on the human MBOAT2 ARE region in LnAR cells with indicated treatment for 24 h. DHT: 100 nM; ENZ: 5 μM. (F) Western blot analysis showing MBOAT2 expression in LNCaP cells with indicated treatment for 48 h. DHT: 100 nM; ENZ: 5 μM. (G) Western blot analysis showing MBOAT2 expression in LnAR-ishAR cells with or without Dox (1 μg ml⁻¹) treatment for 48 h. (H) Western blot analysis showing MBOAT2 expression in PC3 and PC3-AR cells with or without DHT (100 nM) treatment for 48 h. (I) Western blot analysis confirming MBOAT2 knockout in LnAR-gMBOAT2 cells. (J) Viability analysis of LnAR-Cas9 and LnAR-gMBOAT2 cells with indicated RSL3 for 24 h. (K, L) Quantification of differential PL-PUFAs (K) and PL-MUFAs (L) in LnAR-shNT and LnAR-ishMBOAT2 cells. See also Table S3. Data are presented as mean ± SD, n = 3 (C, E, J) or n = 5 (K, L) biologically independent samples. Statistical analysis was performed using 2-way ANOVA (C), 1-way ANOVA (E) or two-tailed t-test (K, L). See also Figure S4.

Figure 5.. AR antagonist sensitizes AR⁺ prostate cancer cells to ferroptosis.

(A, C) Western blot showing MBOAT2 and AR expression in LnAR cells with indicated treatment for 48 h. (B, D) Bliss synergy score surface plots of LnAR cells with indicated combination treatment. Cells were pretreated with indicated ENZ (B) or ARV-110 (D) for 48 h, followed by treatment with indicated concentration of RSL3 for 24 h. Synergy scores and plots were generated by SynergyFinder 3.0. (E) Quantification of lipid peroxidation. LnAR cells were pretreated with DMSO control (Ctrl) or 5 μM ENZ for 48 h, followed by RSL3 (1 μM) for 3 h prior to labeling with BODIPY-C11. (F) Viability analysis of LnAR cells harboring vector or MBOAT2 overexpression. Cells were pretreated with DMSO (ctrl) or 5 μM ENZ for 48 h, followed by 1 μM RSL3 for 24 h. (G) Western blot showing expression of endogenous and ectopic MBOAT2 in LnAR cells harboring either vector control or MBOAT2-mCherry. Cells were treated with 5 μM ENZ as indicated for 48 h. (H) Viability of LnAR-shNT and LnAR-shMBOAT2 cells pretreated with DMSO or 5 μM ENZ for 48 h, followed by treatment with indicated concentration of RSL3 for 24 h. (I) Western blot showing MBOAT2 expression in LnAR-shNT and LnAR-shMBOAT2 cells treated with 5 μM ENZ as indicated for 48 h. (J, K) Viability analysis of LnAR cells pretreated with DMSO or 5 μM ENZ for 24 h, followed by incubation with indicated oleic acid (J) or linoleic acid (K) for 16h. Subsequently, ferroptosis was induced by 10 μM RSL3 (J) or 2 μM RSL3 (K) for 24 h. (L) Viability analysis of VCaP cells pretreated with DMSO or 5 μM ENZ for 24 h, followed by incubation with 10 μM LA for 16 h. Subsequently, ferroptosis was induced by indicated RSL3 for 24 h. (M) Viability analysis of 22Pc-EP cells pretreated with DMSO or 1 μM ARV-110 for 24 h, followed by incubation with 10 μM LA for 16 h. Subsequently, ferroptosis was induced by indicated RSL3 for 24 h. (N) Growth curves of xenograft tumors derived from LnAR-igGPX4 cells. Tumor bearing mice were randomly divided into 4 groups: Veh (normal diet + Vehicle), ENZ (normal diet + ENZ), DOX (Dox diet + Vehicle), and ENZ + DOX (Dox diet + ENZ). See detail in STAR methods. Mice were treated for 15 days. (O) Images and (P) weight of tumors with indicated treatment at day-15. (Q) Representative H&E and immunostaining images of GPX4, Ki67, MBOAT2 and PTGS2 are shown from sections of xenograft tumors with indicated treatment. Scale bar = 10 μm. Data are presented as mean ± SD, n=3 (E, F, H, J, K, L, M) or n=7 (N, P) biologically independent samples. Statistical analysis was performed using 1-way ANOVA (F, P), 2-way ANOVA (N), and two-tailed t-test (E). See also Figure S5.

Figure 6.. MBOAT1 suppresses ferroptosis and is regulated by ER signaling.

(A) Western blot showing ectopic expression of indicated genes in HT1080-GPX4^KO cells. (B) Viability analysis of HT1080-GPX4^KO cells ectopically overexpressing indicated genes. Cells were cultured in the presence of 100 μM Trolox, and ferroptosis was induced by removing Trolox for 24 h. (C, E) Viability analysis of HT1080-MBOAT1 cells. Cells were pretreated with indicated concentrations of TOFA (C) or CAY10566 (E) for 24 h, followed by ferroptosis induction with 0.1 μM RSL3 for another 24 h in the presence or absence of Trolox as indicated. (D) Viability of HT1080-Vector and HT1080-MBOAT1 cells with or without pretreatment of 5 μM TOFA for 24 h, followed by ferroptosis induction with indicated concentration of RSL3 for 24 h. (F, G) Quantification of PE-MUFAs (F), and PE-PUFAs (G) in HT1080-Vec and HT1080-MBOAT1 cells. (H, I) Stacked bars showing relative abundance of indicated PE-OA and PE-AA in HT1080-Vector and HT1080-MBOAT1 cells. (J) Quantification of lipid peroxidation. HT1080-Vector or HT1080-MBOAT1 cells were treated with 0.1 μM RSL3 for 3 h prior to labeling with BODIPY-C11. (K) MBOAT1 mRNA expression of different cancer patient samples (TCGA PanCancer Atlas Studies, cBioportal). MBOAT1 mRNA expression was batch normalized from Illumina HiSeq_RNASeqV2. Dash line indicates median. (L) Western blot showing endogenous MBOAT1 expression in a panel of ER⁺ and triple negative BCa lines. ER⁺ BCa lines were either treated with DMSO, 5 μM Tam or 0.5 μM Ful for 48 h. (M) Viability analysis of a panel of ER⁺ and triple negative BCa lines. Ferroptosis was induced by 3 μM RSL3 for 24 h. (N) Visualization of ER and FOXA1 binding to the MBOAT1 ERE region (normalized by RPKM, GSE72249). (O) ChIP-qPCR showing the occupancy of ER on the human MBOAT1 ERE region in MCF7 cells with indicated treatment for 24 h. Ful: 0.5 μM. (P, Q) MBOAT1 mRNA expression was detected by qRT-PCR in T47D cells treated with DMSO (Ctrl) or 100 nM E2 for 48 h (P); treated with DMSO control, 1 μM Tamoxifen (Tam) or 0.5 μM Fulvestrant (Ful) for 48 h (Q). (R) Western blot showing MBOAT1 and ER expression in T47D-shNT and T47D-shESR1 cells. (S) Western blot confirming MBOAT1 knockout in T47D-gMBOAT1 cells. (T, U) Viability analysis of T47D-Cas9 and T47D-gMBOAT1 cells with indicated RSL3 (T) or ML210 (U) for 24 h. (V, W) Quantification of differential PL-PUFAs (V) and PL-MUFAs (W) in T47D-Cas9 and T47D-gMBOAT1 cells. See also Table S4. Data are presented as mean ± SD, n = 3 (B, C, D, E, F, G, J, M, O, P, Q, T, U) or n = 5 (V, W) biological independent replicates. Statistical analysis was performed using 1-way ANOVA (B, C, E, Q), two-tailed t-test (F, G, J, O, P, V, W) and 2-way ANOVA (M). See also Figure S6.

Figure 7.. Fulvestrant sensitizes hormone therapy-resistant ER⁺ breast cancer to ferroptosis.

(A) Western blot showing MBOAT1 and ER expression in T47D cells with indicated treatment for 48 h. (B, C) Dose-response matrix (B) and Bliss synergy score surface plots (C) of T47D cells with a combination of Ful and RSL3 treatment. Cells were pretreated with indicated Ful for 48 h, followed by treatment with indicated concentration of RSL3 for 24 h. Matrix, synergy scores and plots were generated by SynergyFinder 3.0. (D) Viability of HCC1428 cells pretreated with or without 0.5 μM Ful for 48 h, followed by ferroptosis induction with indicated RSL3 for 24 h. (E) Western blot showing expression of endogenous and ectopic MBOAT1 in T47D cells harboring either vector control or MBOAT1-mCherry. Cells were treated with 0.5μM Ful as indicated for 48 h. (F) Viability analysis of T47D cells harboring vector or MBOAT1 overexpression. Cells were pretreated with DMSO (ctrl) or Ful for 48 h, followed by ferroptosis induction with indicated RSL3 for 24 h. (G) Viability of MCF7 cells pretreated with or without 0.5 μM Ful for 24 h, followed by incubation with indicated concentration LA for 24 h. Subsequently, ferroptosis was induced by 1 μM RSL3 for 24 h. (H) Quantification of lipid peroxidation in MCF7 cells pretreated with Ful for 24 h, followed by incubation with indicated concentration LA for 24 h. Subsequently, ferroptosis was induced by 1 μM RSL3 for 4 h prior to labeling with BODIPY-C11. (I) Western blot of MBOAT1 and ER-alpha in MCF7-Parental, MCF7-TamR⁺ and MCF7-FulR⁺ cells with or without 0.5 μM Ful treatment for 48 h. (J) Cell death analysis of MCF7-TamR⁺ and MCF7-FulR⁺ cells that pretreated with either DMSO, or 1 μM Ful, or 10 μM Tam for 48 h, and followed with either 10 μM IKE, or 10 μM IKE plus 100 μM Trolox for 24 h. (K) Growth curves of tumors derived from MCF7-FulR⁺ cells in xenograft mouse models. Tumors were grown to around 250 mm³, at which point mice were randomly divided into 4 groups: Vehicle, Ful (5 mg, SQ injection), IKE (40 mg/kg IP injection), and Ful + IKE. Mice were treated for 15 days. (L) Representative H&E and immunostaining images of ER, Ki67, MBOAT1 and PTGS2 are shown from sections of tumors derived from MCF7-FulR⁺ cells with indicated treatment. Scale bar = 10 μm. (M) Working model showing ER signaling upregulates MBOAT1 expression in ER⁺ cancer cells, while AR signaling upregulates MBOAT2 expression in AR⁺ cancer cells. MBOAT1/2 can utilize endogenous or exogenous MUFA to suppress ferroptosis through phospholipid remodeling. Data are presented as mean ± SD, n=3 (D, F, G, H, J) or n=5~6 (K) biologically independent samples. Statistical analysis was performed using 1-way ANOVA (H), 2-way ANOVA (K), and two-tailed t-test (J). See also Figure S7.