Propafenone facilitates mitochondrial-associated ferroptosis and synergizes with immunotherapy in melanoma Authors: Qian Zhou, Yating Dian, Yi He, Lei Yao, Hui Su, Yu Meng, Yuming Sun, Daishi Li, Yixiao Xiong, Furong Zeng, Xiaowei Liang, Hong Liu, Xiang Chen, Guangtong Deng Published: 2024 ABSTRACT AbstractBackgroundDespite the successful application of immunotherapy, both innate and acquired resistance are typical in melanoma. Ferroptosis induction appears to be a potential strategy to enhance the effectiveness of immunotherapy. However, the relationship between the status of ferroptosis and the effectiveness of immunotherapy, as well as viable strategies to augment ferroptosis, remains unclear.MethodsA screening of 200 cardiovascular drugs obtained from the Food and Drug Administration-approved drug library was conducted to identify the potential ferroptosis sensitizer. In vitro and in vivo experiments explored the effects of propafenone on ferroptosis in melanoma. Animal models and transcriptomic analyses evaluated the therapeutic effects and survival benefits of propafenone combined with immune checkpoint blockades (ICBs). The relationship between propafenone targets and the efficacy of ICBs was validated using the Xiangya melanoma data set and publicly available clinical data sets.ResultsThrough large-scale drug screening of cardiovascular drugs, we identified propafenone, an anti-arrhythmia medication, as capable of synergizing with ferroptosis inducers in melanoma. Furthermore, we observed that propafenone, in combination with glutathione peroxidase 4 inhibitor RSL3, collaboratively induces mitochondrial-associated ferroptosis. Mechanistically, propafenone transcriptionally upregulates mitochondrial heme oxygenase 1 through the activation of the Jun N-terminal kinase (JNK)/JUN signaling pathway under RSL3 treatment, leading to overloaded ferrous iron and reactive oxygen species within the mitochondria. In xenograft models, the combination of propafenone and ferroptosis induction led to nearly complete tumor regression and prolonged survival. Consistently, propafenone enhances immunotherapy-induced tumorous ferroptosis and antitumor immunity in tumor-bearing mice. Significantly, patients exhibiting high levels of ferroptosis/JUN/HMOX1 exhibited improved efficacy of immunotherapy and prolonged progression-free survival.ConclusionsTaken together, our findings suggest that propafenone holds promise as a candidate drug for enhancing the efficacy of immunotherapy and other ferroptosis-targeted therapies in the treatment of melanoma.Keywords: Melanoma, Immunotherapy WHAT IS ALREADY KNOWN ON THIS TOPIC Immunotherapies, particularly immune checkpoint blockades (ICBs), have led to clear improvements in outcomes for patients with melanoma, yet face challenges due to the development of both innate and acquired resistance. Ferroptosis, a form of regulated cell death driven by excess iron-dependent accumulation of lethal membrane-localized lipid peroxides, has recently been identified as a strategy to enhance the efficacy of immunotherapy. However, the relationship between the status of ferroptosis and the effectiveness of immunotherapy, as well as viable strategies to augment ferroptosis, remains unclear. Immunotherapies, particularly immune checkpoint blockades (ICBs), have led to clear improvements in outcomes for patients with melanoma, yet face challenges due to the development of both innate and acquired resistance. Ferroptosis, a form of regulated cell death driven by excess iron-dependent accumulation of lethal membrane-localized lipid peroxides, has recently been identified as a strategy to enhance the efficacy of immunotherapy. However, the relationship between the status of ferroptosis and the effectiveness of immunotherapy, as well as viable strategies to augment ferroptosis, remains unclear. WHAT THIS STUDY ADDS This research identified propafenone as a potent ferroptosis sensitizer through a screening of 200 Food and Drug Administration-approved cardiovascular drugs. Mechanistically, propafenone transcriptionally upregulates mitochondrial heme oxygenase 1 (HMOX1) via activation of the Jun N-terminal kinase (JNK)/JUN signaling pathway, thereby sensitizing melanoma to ferroptosis inducers and immunotherapy. Notably, the targets of propafenone (HMOX1 and JUN) are associated with poor efficacy of ICBs in both the Xiangya melanoma data set and publicly available clinical data sets. This research identified propafenone as a potent ferroptosis sensitizer through a screening of 200 Food and Drug Administration-approved cardiovascular drugs. Mechanistically, propafenone transcriptionally upregulates mitochondrial heme oxygenase 1 (HMOX1) via activation of the Jun N-terminal kinase (JNK)/JUN signaling pathway, thereby sensitizing melanoma to ferroptosis inducers and immunotherapy. Notably, the targets of propafenone (HMOX1 and JUN) are associated with poor efficacy of ICBs in both the Xiangya melanoma data set and publicly available clinical data sets. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY These findings uncover a novel mechanism by which propafenone sensitizes melanoma to ferroptosis inducers and ICBs, providing a promising strategy to enhance the effectiveness of ICBs and other ferroptosis-targeted therapies in melanoma treatment. These findings uncover a novel mechanism by which propafenone sensitizes melanoma to ferroptosis inducers and ICBs, providing a promising strategy to enhance the effectiveness of ICBs and other ferroptosis-targeted therapies in melanoma treatment. Introduction The breakthroughs in immunotherapeutic and targeted interventions have led to clear improvements in outcomes for patients with melanoma. However, the efficacy of these therapies is often limited by innate and acquired resistance, highlighting the need for innovative therapeutic approaches. Ferroptosis is a form of regulated cell death driven by excess iron-dependent accumulation of lethal membrane-localized lipid peroxides. Melanoma cells with mesenchymal or dedifferentiated properties, which typically exhibit resistance to apoptosis-mediated traditional therapy, surprisingly display vulnerability to ferroptosis.1,3 Additionally, several studies have shown that ferroptosis contributes, at least in part, to the tumoricidal effects of immunotherapy.4 5 Thus, targeting ferroptosis holds great potential for melanoma treatment. However, melanoma cells have evolved various anti-ferroptotic mechanisms to promote tumor progression and resistance to treatment. For instance, the lymphatic environment contains higher levels of glutathione and oleic acid, as well as lower levels of free iron, which confer resistance to ferroptosis on metastatic tumor cells.6 Furthermore, melanoma-derived circulating tumor cells upregulate sterol regulatory element binding transcription factor 2 to induce transcription of the iron transport protein transferrin, thereby reducing intracellular iron levels and inhibiting ferroptosis to achieve metastasis.7 Notably, TYRO3 and MER proto-oncogene tyrosine kinase (MerTK) can render tumor resistant to immunotherapy by limiting vulnerability to ferroptosis,8 9 highlighting the potential of ferroptosis induction as a strategy to enhance the efficacy of immunotherapy. Therefore, clarifying the relationship between ferroptosis status and the efficacy of immunotherapy, and exploring effective strategies to enhance ferroptosis hold significant promise for individual or immunotherapy-related combinational therapies of melanoma. Given the high costs, failure rates, and lengthy testing periods for the development of new medicines, repurposing drugs that have already entered in clinic represents a faster and more cost-effective approach for the clinical implementation of ferroptosis-enhancing therapies. Ferroptosis has been revealed to be involved in several pathological processes, including neoplastic and degenerative diseases, as well as cardiovascular diseases. Interestingly, cardiovascular diseases and tumors share several risk factors.10 For example, the risk of cancer increases in patients with myocardial infarction.10 Several cardiovascular drugs, such as aspirin, have been reported to reduce the risk of tumors and decrease mortality rates in patients with cancer.11 Therefore, we hypothesize that cardiovascular drugs might serve as the greatest potential clinically applicable drugs implicated in regulating melanoma ferroptosis. In our study, we conducted a drug screening of 200 Food and Drug Administration (FDA)-approved cardiovascular drugs and identified the anti-arrhythmia drug propafenone as synergistic with ferroptosis inducers in melanoma. We further demonstrated that the combination of propafenone and glutathione peroxidase 4 (GPX4) inhibitor RSL3 collaboratively triggers mitochondrial-associated ferroptosis. Mechanistically, propafenone transcriptionally upregulates mitochondrial heme oxygenase 1 (HMOX1) via activation of Jun N-terminal kinase (JNK)/JUN signaling under RSL3 treatment, leading to overloaded ferrous iron accumulation and reactive oxygen species (ROS) in mitochondria. In xenograft models, this combination resulted in near-complete tumor regression and extended survival. Consistently, propafenone enhanced immunotherapy-induced tumorous ferroptosis and antitumor immunity in tumor-bearing mice. Significantly, patients with high-ferroptosis/JUN/HMOX1 exhibited better efficacy of immunotherapy and longer progression-free survival. These findings suggest that propafenone can be a promising candidate drug for improving the efficacy of immunotherapy and other ferroptosis-involved therapies in the treatment of melanoma. Results Identification and validation of propafenone to synergize with ferroptosis inducers in melanoma To systematically test whether clinically applicable cardiovascular drugs impact ferroptotic sensitivity of melanoma, we conducted a screening of 200 cardiovascular drugs obtained from the FDA-approved drug library for ferroptosis assay in the two different melanoma cell lines (figure 1A). Propafenone, an antiarrhythmic drug, emerged as one of the most promising candidates for enhancing RSL3-induced cell death in both A375 and SK-MEL-28 melanoma cells (figure 1B). To determine whether propafenone synergizes with RSL3, a comprehensive set of screening experiments, structured in a 6×6 matrix format, was conducted, and the average Combination Index (CI) was far less than 1, indicating a strong synergy (figure 1C). Next, live-cell time-lapse imaging was performed to observe morphological changes following drug treatment. We observed that propafenone seemed to exhibit no detrimental effects on A375 and SK-MEL-28 cells. However, melanoma cells co-treated with propafenone and RSL3 showed increased cell demise, manifesting distinct morphological features indicative of necrosis, including cell rounding, swelling, and plasma membrane rupture, compared with those treated with dimethyl sulfoxide (DMSO), propafenone, and RSL3 alone, as observed after 3 hours (figure 1D,E). Consistently, the clonogenic assays arrived at a similar conclusion (figure 1F). RSL3 primarily functions by binding to and inactivating the peroxidase activity of GPX4, and genetic inactivation of GPX4 renders melanoma cells susceptible to ferroptosis.2 12 We therefore constructed GPX4 knockout melanoma cell lines (figure 1G), and observed that propafenone induced cell death in a dose-dependent manner in these cells (figure 1H). We further sought to determine the role of propafenone in enhancing cell death induced by various other ferroptosis inducers across multiple melanoma cell lines. As anticipated, propafenone synergizes with IKE, RSL3, ML210, ML162, and FINO2, in a panel of melanoma cell lines (WM35, A375, A2058, SK-MEL-2, SK-MEL-5, and SK-MEL-28) (figure 1I), which further bolster the reliability of our findings regarding the synergistic lethal effects of propafenone in combination with ferroptosis inducers in melanoma. Identification and validation of propafenone to synergize with ferroptosis inducers in melanoma. (A) Schematic of the identification of clinically applicable cardiovascular drugs that sensitize melanoma to RSL3. (B) Scatter plot shows the viability ratio of A375 and SK-MEL-28 cells treated with single-drug versus dual-drug treatment. The red plot shows propafenone (PPF). (C) Percentage of cell viability rate was presented in a series of 6×6 screening experiments in A375 and SK-MEL-28 cells. Synergy was evaluated by the Chou-Talalay Combination Index (CI) for PPF and RSL3 across the indicated cell lines. The x-axis of CI plots represents the fraction affected. (D–E) Cell morphological features (D) and cell viability (E) of A375 and SK-MEL-28 cells at different time points of treatment with DMSO, PPF, RSL3, or the combination of PPF and RSL3 (COM). PPF, 5 µM; RSL3, 2.5 µM. Images were taken at 200× magnification. (F) Clonogenic assay of A375 and SK-MEL-28 cells treated with PPF, RSL3 or a combination as indicated. (G) GPX4 protein levels were quantified by western blotting in control (sgCtrl) and GPX4 deficient (sgGPX4) cells. (H) Cell viability of sgCtrl or sgGPX4 cells treated with different concentrations of PPF for 12 hours. (I) Heatmap shows the cell viability of WM35, A375, A2058, SK-MEL-2, SK-MEL-5 and SK-MEL-28 cell lines treated with PPF, various ferroptosis inducers IKE, RSL3, ML210, ML162, and FINO2, or a combination as indicated. COM, combination; CCK-8, Cell Counting Kit-8; DMSO, dimethyl sulfoxide; FDA, Food and Drug Administration; GPX4, glutathione peroxidase 4. Propafenone boosts GPX4 inhibition-induced ferroptosis in vitro and in vivo To ascertain the types of cell death induced by the combination of propafenone and RSL3, we co-treated the cells with a range of well-known inhibitors of cell death. Pharmacological or genetic inhibition of apoptosis (using Z-VAD-FMK), necroptosis (using necrostatin-1s), and autophagic cell death (using chloroquine and knockdown autophagy-related 5 (ATG5)) failed to rescue cell death induced by the combination therapy. Conversely, the toxic effects of the combined treatment were completely abrogated by ferroptosis inhibitors, including ferrostatin-1 (Fer-1) and deferoxamine (DFO) (figure 2A, online supplemental figure S1A–D). We further used propidium iodide (PI) to validate corresponding cell death under indicated treatment by the fluorescence microscopy and flow cytometry analysis and showed that co-administration of propafenone and RSL3 resulted in marked PI-positive staining in A375 and SK-MEL-28 cells, which could be almost completely blocked by Fer-1 (figure 2B,C), suggesting that the combination triggers ferroptosis of melanoma. To further reinforce the idea, several markers of ferroptosis were tested to determine if ferroptosis was indeed induced by co-treatment with propafenone and RSL3. First, lipid peroxidation, a hallmark of ferroptosis, was significantly elevated after co-treatment compared with single-drug treatment, and this effect was completely abolished by Fer-1 (figure 2D–F). Second, messenger RNA (mRNA) levels of prostaglandin-endoperoxide synthase 2 (PTGS2) and contents of lipid peroxidation end products, malondialdehyde (MDA), were also notably increased after combined treatment (online supplemental figure S1E,F). Third, the content of glutathione (GSH), a critical intracellular cofactor for GPX4 to counteract ferroptosis, decreased after dual-drug treatment (figure 2G). Additionally, the cell death in GPX4-deficient melanoma cells treated with propafenone could also be reversed by Fer-1 (figure 2H). Taken together, our data strongly suggest that the combination of propafenone and RSL3 could synergistically induce melanoma ferroptosis. Propafenone boosts glutathione peroxidase 4 inhibition-induced ferroptosis in vitro and in vivo. (A) Indicated melanoma cells were treated with PPF (5 µM), RSL3 (2.5 µM), or a combination of both drugs with or without cell death inhibitors (Z-VAD-FMK, 5 µM; necrostatin-1s (Nec-1s), 10 µM; chloroquine (CQ), 10 µM; deferoxamine (DFO), 100 µM; ferrostatin-1 (Fer-1), 4 µM) for 6 hours, and cell viability was assessed. (B–C) Cell death of A375 and SK-MEL-28 cells induced by the indicated treatment for 6 hours was shown by fluorescence microscopy and quantified by PI-staining coupled with flow cytometry. Representative images are presented at 200× magnification. (D–F) Lipid peroxidation was quantified with BODIPY-C11 using flow cytometry. Cells were treated as indicated for 3 hours. (G) The relative levels of glutathione were assayed in A375 and SK-MEL-28 cells after indicated treatment for 3 hours. (H) Cell viability of sgCtrl or sgGPX4 cells treated with different concentrations of PPF for 12 hours in the absence or presence of Fer-1 (4 µM). (I–K) BALB/c nude mice were injected with SK-MEL-28 cells and treated with PPF (10 mg/kg, every day) or RSL3 (50 mg/kg, intratumorally, every 3 days) as indicated when the tumor size reached approximately 100 mm3 (I). Tumor volume (J), and body weight (K) in the indicated groups. (L) Survival curves of SK-MEL-28 tumor-bearing mice treated as indicated. (M–O) BALB/c nude mice were injected with sgCtrl or sgGPX4 A375 cells and treated with PPF (10 mg/kg) and vehicle intraperitoneally every day as indicated. Lip-1 was given 10 mg/kg intraperitoneally every day (M). Tumor weight (N), and tumor volume (O) in the indicated groups. n=5 per group. P values were calculated using one-way analysis of variance analysis in A–G and N, or mixed effects models in J and O, or log-rank test in L. ns, p>0.05; *p<0.05; **p<0.01; ***p<0.001. COM, combination; DMSO, dimethyl sulfoxide; GPX4, glutathione peroxidase 4; GSH, glutathione; i.t, intratumorally; i.p, intraperitoneally; Lip-1, liproxstatin-1; PI, propidium iodide; PPF, propafenone. To validate our in vitro findings and assess the potential of combined propafenone and ferroptosis induction in preclinical models, nude mice implanted with SK-MEL-28 cells were treated with propafenone, the GPX4 inhibitor RSL3, or both. Consistent with prior studies, the GPX4 inhibitor RSL3 was found to slow melanoma growth in vivo.13 Moreover, the combination treatment of propafenone and RSL3 significantly suppressed tumor growth and prolonged mouse survival compared with monotherapy without apparent changes in body weight (figure 2I–L). To extend this finding, GPX4 knockout or control A375 cells were inoculated into the right flank of nude mice to generate subcutaneous xenograft models. On reaching a tumor size of 50–100 mm3, tumor-bearing mice were randomly assigned to groups and treated with vehicle, propafenone, or a combination of propafenone and ferroptosis inhibitor liproxstatin-1 (Lip-1) (figure 2M). As anticipated, GPX4 knockout had minimal impact on A375 cell survival and tumor development in vivo.2 12 However, combining GPX4 knockout with propafenone treatment, rather than either alone, led to marked tumor regressions (figure 2N,O, online supplemental figure S1G). Melanoma suppression was nearly completely blocked by treatment with the ferroptosis inhibitor Lip-1 (figure 2N,O). These results collectively demonstrate that propafenone could boost GPX4 inhibition-induced ferroptosis to effectively inhibit melanoma both in vitro and in vivo. Propafenone combined with RSL3 collaboratively triggers mitochondrial-associated ferroptosis Propafenone inhibits sodium channels or β-adrenergic receptors to exert its antiarrhythmic effects,14 and we, therefore, wonder whether propafenone facilitates melanoma ferroptosis through similar mechanisms. However, treatment with three different inhibitors of sodium channels (carbamazepine, mexiletine HCl, procainamide HCl) or/and β-adrenergic receptors (sotalol HCl, betaxolol, timolol maleate) at various concentrations failed to boost RSL3-mediated ferroptosis in melanoma (online supplemental figure S2A–C). This suggests that propafenone-mediated enhancement of ferroptotic sensitivity is not dependent on its classic antiarrhythmic mechanisms. We then investigated the alterations in classic ferroptosis-related proteins on propafenone treatment. However, supplementation of propafenone did not significantly affect the status of these ferroptosis-related proteins and mRNA levels, including NRF2, GPX4, AIFM2/FSP1, DHODH, LPCAT3, and ACSL4, with or without RSL3 treatment (online supplemental figure S3A–C). Also, co-treatment with RSL3 and propafenone did not further inhibit GPX4 activity compared with RSL3 treatment alone (online supplemental figure S3D). Propafenone has been reported to inhibit esophageal carcinoma by triggering mitochondrial dysfunction.15 Mitochondria, the main source of ROS, play a crucial role in ferroptosis, and multiple proteins might operate in mitochondria to control lipid peroxidation and ferroptosis.16 17 We thus sought to explore the role of propafenone in mitochondria-associated ferroptosis. Morphologically, transmission electron microscopy revealed that propafenone, in combination with RSL3, induced mitochondrial shrinkage, reduced cristae, and increased membrane density within cells (figure 3A). Functionally, co-treatment with propafenone and RSL3 significantly reduced mitochondrial membrane potential, as indicated by reduced aggregate JC-1, and increased mitochondrial ROS production, assessed by the fluorescent probe Mito-SOX (figure 3B,C). ROS produced by mitochondria has been shown to induce ferroptosis.18 To collaborate this finding, mito-2,2,6,6-tetramethylpiperidin-1-oxyl (TEMPO), a mitochondria-targeted antioxidant, significantly attenuated propafenone-enhanced RSL3-induced cell death compared with the non-mitochondria-targeted version (TEMPO) (figure 3D). Similarly, mito-TEMPO, but not TEMPO, significantly reduced enhanced mitochondrial lipid peroxidation and increased PTGS2 mRNA levels induced by co-treatment with propafenone and RSL3 (figure 3E,F). Propafenone combination with RSL3 collaboratively triggers mitochondrial-associated ferroptosis. (A) Transmission electron microscopy images of A375 cells after the indicated treatment for 3 hours. PPF, 5 µM; RSL3, 2.5 µM. Scale bar, upper, 2 µm; lower, 500 nm. (B–C) Mitochondrial membrane potential (B) or mitochondrial ROS (C) was determined in A375 and SK-MEL-28 cells after indicated treatment for 3 hours by flow cytometry using JC-1 staining or Mito-SOX red fluorescence probe, respectively. (D) Dose response of RSL3-induced death of PPF-treated A375 and SK-MEL-28 cells in the presence of TEMPO (20 µM) or Mito-TEMPO (10 µM) for 6 hours. (E–F) Mito-lipid peroxidation (E) and PTGS2 mRNA (F) were measured in A375 and SK-MEL-28 cells after the indicated treatment for 3 hours. PPF, 5 µM; RSL3, 2.5 µM. TEMPO, 20 µM; Mito-TEMPO, 10 µM. (G) Western blotting showing GPX4 protein levels in sgGPX4 A375 cells that overexpress the cytosolic or mitochondrial GPX4. (H) Cell viability measurement in sgGPX4 A375 cells that express the indicated GPX4 constructs treated with different doses of PPF for 6 hours. (I) Relative levels of overall cellular ferrous iron were assessed in A375 and SK-MEL-28 cells after the indicated treatment for 3 hours. (J) Mitochondrial ferrous iron levels were assessed by fluorescence microscopy using Mito-FerroGreen (green) in A375 cells after the indicated treatment for 1 hour. PPF, 5 µM; RSL3, 2.5 µM; DFO, 100 µM. Mitochondria stained with Mito-Tracker (red). Nuclei counterstained by Hoechst (blue). Scale bar, 100 µm. (K) Mitochondrial ferrous iron levels were assessed by flow cytometry using Mito-FerroGreen in A375 and SK-MEL-28 cells after the indicated treatment for 1 hour. Then the mean Mito-FerroGreen fluorescence intensity of each cell was quantified. P values were calculated using one-way analysis of variance analysis. ns, p>0.05; ***p<0.001. COM, combination; DFO, deferoxamine; DMSO, dimethyl sulfoxide; GPX4, glutathione peroxidase 4; mRNA, messenger RNA; PPF, propafenone; PTGS2, prostaglandin-endoperoxide synthase 2; ROS, reactive oxygen species; TEMPO, 2,2,6,6-tetramethylpiperidin-1-oxyl. GPX4 includes several isoforms with distinctive subcellular localization, including cytosol-localized and mitochondria-localized GPX4.19 Overexpression of mitochondrial GPX4, as opposed to cytosolic GPX4, can rescue mitochondrial lipid peroxidation and mitigate mitochondrial-associated ferroptosis.17 Notably, we found that restoring mitochondrial GPX4, rather than cytosolic GPX4, in GPX4 knockout A375 cells rescued ferroptosis under treatment with propafenone and RSL3 (figure 3G,H). These results indicate that propafenone sensitizes melanoma to mitochondrial-associated ferroptosis. Mitochondrial ferrous iron levels impact mitochondria-associated ferroptotic sensitivity.16 20 We thus examined the level of cytosol-ferrous and mitochondria-ferrous iron. As expected, co-administration of propafenone and RSL3 exerted no effect on cytosol ferrous iron (figure 3I, online supplemental figure S3E). Contrarily, after the co-treatment of propafenone and RSL3, we observed a notable increase in mitochondrial ferrous iron levels, as indicated by Mito-FerroGreen. This increase could be completely negated by iron chelator DFO (figure 3J,K). Overall, these findings suggest that propafenone in combination with RSL3 collaboratively triggers mitochondria-associated ferroptosis in melanoma by promoting mitochondrial ferrous iron accumulation and subsequent mitochondrial ROS overload. HMOX1 upregulation contributes to the mitochondria-associated ferroptosis To elaborate the underlying mechanism for the mitochondria-associated ferroptosis under the co-treatment of propafenone and RSL3, the A375 cells treated with propafenone and/or RSL3 underwent high-throughput RNA sequencing. Co-treatment with propafenone and RSL3 resulted in substantial alterations in intracellular expression profiles compared with cells treated with DMSO or either drug alone (figure 4A, online supplemental figure S4A). Not unexpectedly, Gene Set Enrichment Analysis (GSEA) showed significant activation of the ferroptosis pathway in the combination group compared with other groups (online supplemental figure S4B), which further supports our previous findings. Propafenone facilitates mitochondria-associated ferroptosis through synergistic upregulation of HMOX1 when combined with RSL3. (A) Volcano plots of RNA sequencing data showing the most differentially expressed genes in combination versus DMSO (left), combination versus PPF (median), and combination versus RSL3 (right). Significantly expressed genes above and below 1.5-fold are shown in red and blue, respectively; PPF, 5 µM; RSL3, 2.5 µM. (B) Venn diagram showing the overlap of common differentially expressed upregulated genes and mitochondrial iron metabolism-related genes identified from GeneCards. (C) Heatmap represents the normalized expression of five genes in each group. (D) Real-time PCR analysis of HMOX1 mRNA levels in A375 and SK-MEL-28 cells after the indicated treatment for 3 hours. (E–F) Western blotting showing HMOX1 protein levels in whole cell and mitochondrial fractions from A375 (E) and SK-MEL-28 cells (F) after indicated treatment for 3 hours. PPF, 5 µM; RSL3, 2.5 µM. (G) Gene Set Enrichment Analysis showing the enrichment of response to endoplasmic reticulum stress pathway (blue) and heme metabolism pathway (red) between co-treatment group and RSL3 group in RNA sequencing. (H) HMOX1 was positively correlated with WP_ferroptosis score using Spearman’s correlation in cell lines from the Cancer Cell Line Encyclopedia database. (I) Relative cell viability of indicated cells treated with DMSO, 20 µM or 30 µM hemin in the presence of RSL3 for 6 hours. (J) Mitochondrial ferrous iron levels were assessed by fluorescence microscopy using Mito-FerroGreen (green) in A375 cells after the indicated treatment for 1 hour. PPF, 5 µM; RSL3, 2.5 µM; ZnPP, 10 µM. Mitochondria stained with Mito-Tracker (red). Nuclei counterstained by Hoechst (blue). Scale bar, 100 µm. (K) Mitochondrial ferrous iron levels were assessed by flow cytometry using Mito-FerroGreen in A375 and SK-MEL-28 cells after the indicated treatment for 1 hour. (L–M) Mito-lipid peroxidation (L) and lipid peroxidation (M) were measured by flow cytometry in A375 and SK-MEL-28 cells after the indicated treatment for 3 hours. (N) Dose-response of RSL3-induced death of DMSO or PPF-treated A375 and SK-MEL-28 cells in the absence or presence of ZnPP for 6 hours. (O) Cell viability of shCtrl and shHMOX1 A375 and SK-MEL-28 cells following PPF and RSL3 co-treatment for 3 hours. P values were calculated using one-way ANOVA analysis in D,I, k, L and M or two-way ANOVA analysis in O. *p<0.05; ***p<0.001. ANOVA, analysis of variance; COM, combination; DMSO, dimethyl sulfoxide; HMOX1, heme oxygenase 1; PPF, propafenone; TPM, transcripts per million; ZnPP, zinc protoporphyrin IX. Given the accumulation of mitochondrial ferrous iron caused by the co-administration of propafenone and RSL3, we intersected the upregulated or downregulated differentially expressed genes with mitochondrial iron metabolism genes, respectively, to investigate the underlying mechanism of this synergy (figure 4B, online supplemental figure S4C). We identified five upregulated differentially expressed genes associated with mitochondrial iron metabolism, including HMOX1, JUN, GFAP, CXCL8, and FN1 (figure 4C). Among these genes, HMOX1 showed the most significant upregulation (figure 4A–C), which was further confirmed by quantitative real-time PCR analysis and western blot (figure 4D–F). Interestingly, gene signatures associated with HMOX1-related endoplasmic reticulum stress and heme metabolism were significantly activated after co-treatment with propafenone and RSL3 compared with RSL3 treatment alone (figure 4G).16 21 HMOX1 is one of the heme oxygenase enzymes that catalyze the hydroxylation of heme to produce carbon monoxide, ferrous iron, and biliverdin in mitochondria,16 and it has been demonstrated to promote ferroptosis in cancer cells.22,24 To further support this, we conducted gene set variation analysis (GSVA) and found that HMOX1 expression positively correlated with the activation of the ferroptosis pathway (figure 4H, online supplemental figure S4D). To verify that HMOX1 is responsible for the mitochondria-associated ferroptosis triggered by the propafenone and RSL3 co-treatment, we first used the HMOX1 activator, heme analog hemin, to mimic the pro-ferroptotic effect of propafenone in melanoma. We found that hemin strongly potentiated RSL3-induced cell death (figure 4I), but did not enhance it further in the presence of propafenone (online supplemental figure S4E). Conversely, pharmacological inhibition of HMOX1 with its competitive inhibitor zinc protoporphyrin IX (ZnPP) could significantly block the accumulation of mitochondrial ferrous iron, mitochondrial lipid peroxidation and total lipid peroxidation, and inhibit the upregulation of PTGS2 mRNA level, and rescue the mitochondria-associated ferroptosis induced by co-treatment with propafenone and RSL3 (figure 4J–N, online supplemental figure S4F–H). Likewise, marked inhibition of ferroptosis in response to co-treatment was observed after knockdown of HMOX1 in A375 and SK-MEL-28 cells (online supplemental figure S4I, figure 4O). These findings indicate that co-administration of propafenone and RSL3 synergistically induces mitochondria-associated ferroptosis by upregulating mitochondrial HMOX1, leading to mitochondrial iron accumulation and ROS overload. JNK/JUN signaling-mediated HMOX1 upregulation promotes mitochondria-associated ferroptosis To further elucidate the upstream regulatory mechanism of HMOX1 in melanoma cells treated with propafenone and RSL3, we examined the mRNA level of HMOX1 in the presence of cycloheximide (CHX), a protein synthesis inhibitor that acts specifically on the 60 S subunit of eukaryotic ribosomes.25 We found that propafenone failed to increase the mRNA level of HMOX1 in melanoma cells under RSL3 treatment when exposed to CHX (figure 5A), indicating that propafenone upregulated the mRNA level of HMOX1 by triggering de novo protein synthesis of a potential HMOX1 positive regulator. NRF2, a key negative regulator of ferroptosis, has been reported to transcriptionally regulate HMOX1 expression.16 22 However, pharmacological inhibition of NRF2 with trigonelline and ML385 could not rescue the combination-induced cell death (online supplemental figure S5A,B), suggesting that NRF2 was not involved in the HMOX1-induced mitochondria-associated ferroptosis. JNK/JUN signaling-mediated HMOX1 upregulation promotes mitochondria-associated ferroptosis. (A) Relative mRNA levels of HMOX1 in A375 and SK-MEL-28 cells treated with RSL3 alone or in combination with PPF, compared with the DMSO group, were assessed in the presence of CHX (5 µg/mL). (B) Venn diagram showing the putative upstream transcription factors of HMOX1 predicted by JASPAR, PROMO, and hTFtarget databases. (C) Correlation between JUN and HMOX1 using Spearman’s correlation in cell lines from the CCLE database. (D) Correlation between JUN and WP_ferroptosis score using Spearman’s correlation in cell lines from the CCLE database. (E) JUN mRNA level in A375 and SK-MEL-28 cells after indicated treatment. (F) A375 and SK-MEL-28 cells were treated with PPF, RSL3 or a combination for 3 hours and subjected to immunofluorescence microscopy with antibodies against JUN. Nuclei counterstained by DAPI (blue). Scale bar, 50 µm. (G) GSEA shows the enrichment of the JNK cascade pathway (blue) and MAPK pathway (red) between the co-treatment group and the RSL3 group in RNA sequencing. (H–I) The relative viability of A375 and SK-MEL-28 cells treated with the combination of PPF and RSL3 with or without compounds from the kinase drug library for 6 hours. Specific compounds are annotated. (J) Cell viability of A375 and SK-MEL-28 cells at different time points after co-treatment of PPF and RSL3 in the absence or presence of tanzisertib (5 µM, 10 µM or 15 µM). (K–M) Cell death of A375 and SK-MEL-28 cells induced by the indicated treatment for 6 hours was shown using fluorescence microscopy and quantified by PI-staining coupled with flow cytometry. Representative images are presented at 200× magnification. PPF, 5 µM; RSL3, 2.5 µM; tanzisertib, 10 µM. (N) Mitochondrial ferrous iron levels were assessed by flow cytometry in A375 and SK-MEL-28 cells after the indicated treatment for 1 hour. (O) Genome browser view of binding peaks on the HMOX1 promoter in JUN ChIP and input control of MDA-MB-231 cells and MDA-MB-231–1833 cells based on data from GSE112444. P values were calculated using one-way analysis of variance analysis. ***p<0.001. CCK-8, Cell Counting Kit-8; CCLE, Cancer Cell Line Encyclopedia; ChIP, chromatin immunoprecipitation; CHX, cycloheximide; COM, combination; DMSO, dimethyl sulfoxide; GSEA, Gene Set Enrichment Analysis; HOMX1, heme oxygenase 1; JUN/JNK, Jun N-terminal kinase (JNK)/JUN; MDA, malondialdehyde; mRNA, messenger RNA; NES, normalized enrichment score; PI, propidium iodide; PPF, propafenone; TPM, transcripts per million. To identify the upstream transcription factor (TF) controlling HMOX1 expression, we used publicly accessible databases JASPA, PROMO, and hTFtarget.26 19 potential TFs that could bind to the HMOX1 promoter were identified (figure 5B). Among these TFs, the expression profiles of JUN and ETS1 resembled that of HMOX1 after corresponding treatment and were positively correlated with HMOX1 expression (figure 5C, online supplemental figure S5C,D). Notably, JUN, but not ETS1, showed a positive association with ferroptosis (figure 5D, online supplemental figure S5E), suggesting that JUN might be the potential TF of HMOX1 to induce mitochondria-associated ferroptosis under the co-treatment of propafenone and RSL3. Real-time PCR and immunofluorescence staining demonstrated that JUN expression was upregulated and translocated into the nucleus after combined treatment (figure 5E,F). JUN is usually transported to the nucleus after phosphorylation by JNK to exert its transcriptional activation activity.27 Interestingly, GSEA also demonstrated that MAPK and JNK pathways, but not ERK1/2 and P38 MAPK pathways, were significantly activated in melanoma cells treated with propafenone and RSL3 compared with RSL3 alone (figure 5G, online supplemental figure S5F–G), further supporting our aforementioned findings. Consistent with this, a large-scale kinase inhibitor screening showed that the JNK inhibitor tanzisertib could most significantly rescue the combined drugs-induced inhibition of melanoma cell viability (figure 5H,I), and the rescue effect was dose-dependent and time-dependent (figure 5J, online supplemental figure S5H). Similarly, tanzisertib significantly abrogated dual drug-induced cell death (figure 5K–M) and blocked the increased mitochondrial ferrous iron (figure 5N). An analysis of published chromatin immunoprecipitation sequencing (ChIP-seq) data showed a strikingly enhanced JUN‐binding peak on the HMOX1 promoter (figure 5O).28 Collectively, these findings suggested that combined treatment with propafenone and RSL3 could activate JNK/JUN signaling to promote the transcriptional regulation of mitochondrial HMOX1, thereby inducing mitochondria-associated ferroptosis. Enhanced efficacy of immunotherapy combined with propafenone in vivo Cancer immunotherapy, particularly immune checkpoint blockades (ICBs), represents an important ferroptosis-associated pathological model in vivo,5 29 30 yet further investigations are necessary to comprehend the connection between ferroptosis status and immunotherapy outcomes. To address this, we therefore analyzed an immunotherapy data set comprising 437 samples from patients treated with ICB to assess the correlation between ferroptosis levels and immunotherapy responses.31 The ferroptosis score for each sample was evaluated using single-sample GSEA (ssGSEA) with the WP_ferroptosis gene set.32 To validate the reliability of the ferroptosis scoring model, we assessed enrichment scores for 20 metabolic pathways related to ferroptosis using the same method,33 and found that ferroptosis score is positively associated with these ferroptosis-related metabolic pathways such as ROS accumulation, GSH metabolism, and fatty acid synthesis signaling pathways (figure 6A), underscoring the robustness of the scoring model for characterizing ferroptosis status. Notably, we found a significant elevation in ferroptosis levels in the ICB-response group compared with the non-response group (figure 6B). Intriguingly, propafenone-activated downstream genes (HMOX1 and JUN) were also upregulated in the ICB-response group compared with the non-response group (figure 6B). Based on these three indicators, we combined them into a two-dimensional index. We observed that patients with high ferroptosis/JUN/HMOX1 exhibited better therapeutic outcomes of immunotherapy and longer progressional-free survival than those with low-ferroptosis/JUN/HMOX1 (figure 6C,D). Consistently, samples with high ferroptosis/JUN/HMOX1 tended to have significantly higher levels of cytolytic activity, T-cell receptor and B-cell receptor abundance, interferon (IFN)-γ response, CD274 levels, T-cell inflammatory gene expression profile, CD8+ T-cell infiltration levels, and cytotoxic T lymphocytes, along with lower immunosuppressive cell characteristics such as myeloid suppressor cells, tumor-associated macrophages M2 subtype, and T-cell rejection scores (figure 6E). These results were further validated in the Xiangya melanoma data set and phs000452.V.3 data set34 35 (online supplemental figure S6A–D), suggesting that propafenone may have potential to enhance the therapeutic effect of immunotherapy in melanoma. We, therefore, implemented a combinative therapeutic approach of propafenone and anti-programmed cell death 1 (PD-1) antibody in a mouse model (figure 6F). Immunocompetent C57BL/6 mice implanted with YUMM1.7 melanoma cells received intraperitoneal injections of anti-PD-1 treatment or IgG control with or without propafenone on the indicated day after the tumor reached approximately 100 mm3. We observed that anti-PD-1 single treatment exhibited a moderate tumor inhibitory effect, but combined treatment with propafenone and an anti-PD-1 antibody significantly reduced tumor growth and prolonged mouse survival compared with monotherapy, without significant changes in body weight (figure 6G–I). Therefore, propafenone represents a promising candidate drug to synergize with immunotherapy in the treatment of melanoma. Enhanced efficacy of immunotherapy combined with propafenone in vivo. (A) Heatmap showing the WP_ferroptosis score and normalized ferroptosis-related pathway enrichment scores of each sample calculated by single-sample Gene Set Enrichment Analysis (Ribas’s cohort, n=437). (B) VlnPlot showing the differences of WP_ferroptosis, JUN and HMOX1 between responses to the immunotherapy group and non-responses to the immunotherapy group. (C) The proportion of patients with different responses to immunotherapy in the melanoma immune checkpoint inhibitor (ICI) cohort (PRJEB23709). (D) Kaplan-Meier curves compare progression-free survival (PRJEB23709; ICI cohorts) between the high WP_ferroptosis/JUN/HMOX1 and low WP_ferroptosis/JUN/ HMOX1 groups. (E) The differences in immunotherapy outcome-related scores between high WP_ferroptosis/JUN/HMOX1 and low WP_ferroptosis/JUN/ HMOX1 groups. (F) Schedule for administration of PPF (10 mg/kg) or anti-PD-1 (100 µg/per mouse) in YUMM1.7 tumor-bearing C57BL/6 mice. (G–I) Tumor volume (G), body weight (H), and survival curves (I) in the indicated groups. n=5 per group. Statistical significance was assessed using the Fisher’s exact test/χ2 test in C, or log-rank test in D and I, or mixed-effects models in G. Wilcoxon rank-sum test was used in E. ns, p>0.05; *p<0.05, **p<0.01, and ***p<0.001. BCR, B-cell receptor; COM, combination; CTL, cytotoxic T lymphocytes; CYT, cytolytic activity; GEP, gene expression profile; HMOX1, heme oxygenase 1; IFN, interferon; i.p, intraperitoneally; JUN, Jun proto-oncogene; mAb, monoclonal antibody; PD-1, programmed cell death 1; PPF, propafenone; MDSC, myeloid-derived suppressor cells; TAM, tumor-associated macrophages; TCR, T-cell receptor. Propafenone enhances the efficacy of immunotherapy through ferroptosis To determine whether propafenone enhances the efficacy of immunotherapy via ferroptosis, we additionally established B16F10 tumors in mice and treated them with propafenone, anti-PD-1 antibody, or the combination of both with or without Lip-1. Again, synergistic effects were observed in these mice with the combined treatment of propafenone and anti-PD-1 antibody (figure 7A). Inhibiting ferroptosis by Lip-1 significantly alleviated the antitumor effect of the combination (figure 7B–D, online supplemental figure S7A). This suggests a crucial role of ferroptosis in the enhanced immunotherapy efficacy of propafenone. However, the mechanism by which anti-PD-1 and propafenone exert their synergistic effects in vivo is not yet clear. Therefore, we performed bulk transcriptomic sequencing on the tumor tissue samples. To investigate the alterations in the ferroptosis pathway and immune cell infiltration among the four groups, we applied the ssGSEA algorithm to estimate ferroptosis activity and immune infiltration levels for each sample based on the WP_ferroptosis gene set and gene sets related to immune infiltration level (online supplemental table S1).36,38 Our analysis revealed that the combination therapy group exhibited a significant increase in ferroptosis scores (figure 7E,F), along with generally elevated levels associated with immune infiltration (figure 7E). Furthermore, the combination therapy group showed trends of increased CD8+ T-cell infiltration levels, T-cell inflammatory gene expression profile (GEP), cytolytic activity (CYT), and effector memory T cell infiltration levels (figure 7F). To elucidate the key pathways modulated by the combined treatment, we conducted cancer hallmark pathway enrichment analyses comparing the combination group to the control group (online supplemental figure S7B), the propafenone monotherapy group (online supplemental figure S7C), and the anti-PD-1 monotherapy group, respectively (figure 7G). Consistently, we found that the combination treatment significantly enhanced immune-related pathways such as IFN-α response, IFN-γ response, and inflammatory response. Subsequently, we performed the immunohistochemistry and immunofluorescence staining of tumors. As expected, the combination treatment significantly enhanced the expression of the ferroptosis marker 4-HNE, as well as the percentages of CD8+ and GZMB+ cells in tumors, which were nearly completely reversed by Lip-1 (figure 7H, online supplemental figure S7D). Collectively, these data suggest that propafenone combined with anti-PD-1 therapy activates the immune environment in vivo and enhances the effect of immunotherapy through ferroptosis. Propafenone enhances the efficacy of immunotherapy through ferroptosis. (A) Schedule for administration of PPF (10 mg/kg), anti-PD-1 (100 µg/per mouse) or Lip-1 (10 mg/kg) in B16F10 tumor-bearing C57BL/6 mice. (B–D) Tumor volume (B), body weight (C), and survival curves (D) in the indicated groups. n=5 per group. (E) Heatmap showing the normalized WP_ferroptosis score and immune infiltration pathway enrichment scores of each sample calculated by single-sample Gene Set Enrichment Analysis (mice tumor, n=20). (F) WP_ferroptosis pathway scores, CD8+ cell infiltration levels, T-cell inflammatory gene expression profile, cytolytic activity, and effector memory T-cell infiltration levels in the indicated groups. (G) Bar plot showing the enrichment level of 50 cancer hallmark gene sets between the combination group and anti-PD-1 group. (H) Representative images of immunohistochemistry and immunofluorescence staining of 4-HNE, CD8+ and GZMB+ cells in B16F10 tumors after indicated treatment. Scale bars=50 µm. (I) A schematic diagram about the mechanism of co-treatment with PPF and RSL3 in mitochondria-associated ferroptosis induction. Statistical significance was assessed using mixed-effects models in B, or log-rank test in D, or Bonferroni correction in F. ns, p>0.05; *p<0.05; **p<0.01; ***p<0.001. COM, combination; GEP, gene expression profile; GPX4, glutathione peroxidase 4; HMOX1, heme oxygenase 1; i.p, intraperitoneally; JNK, Jun N-terminal kinase; JUN, Jun proto-oncogene proto-oncogene; Lip-1, liproxstatin-1; mAb monoclonal antibody; NES, normalized enrichment score; PD-1, programmed cell death 1; PPF, propafenone; Tem, effector memory T cell; ZnPP, zinc protoporphyrin IX. Discussion Ferroptosis is an intracellular iron-dependent form of cell death that plays a crucial role in cancer progression and therapy.30 39 40 In this study, to search for clinically applicable cardiovascular drugs that synergize with ferroptosis inducers, we performed a drug library screening and identified propafenone as a potential candidate. Mechanistically, propafenone transcriptionally upregulates mitochondrial HMOX1 via the activation of JNK/JUN signaling under RSL3 treatment, leading to enhanced ferrous iron accumulation and ultimately triggering mitochondria-associated ferroptosis (figure 7I). These findings suggest that propafenone holds promise as a candidate drug for enhancing the efficacy of immunotherapy and other ferroptosis-involved therapies in the treatment of melanoma. Propafenone is a well-tolerated class I antiarrhythmic agent and has been reported to be able to suppress esophageal squamous cell carcinoma in vivo and in vitro by inducing mitochondrial dysfunction with a decreased mitochondrial membrane potential and Bcl-XL and Bcl-2 protein expression.15 Additionally, a recent large-scale screening of 1,347 drugs identified that propafenone can effectively inhibit conjunctival melanoma cell viability and proliferation by impairing DNA repairing-related homologous recombination pathway.41 However, the precise role of propafenone in melanoma and its relation to ferroptosis remains unclear. In this study, we revealed that propafenone alone does not significantly induce mitochondrial dysfunction and inhibit melanoma development, while its combination with ferroptosis inducers RSL3 could result in fragmented and condensed mitochondria, reduced mitochondrial membrane potential and elevated ROS levels, ultimately leading to mitochondrial dysfunction. Furthermore, targeting the removal of mitochondrial ROS significantly inhibits co-induced cell death and mitochondrial lipid oxidation. Restoring GPX4mito, but not GPX4cyto, in GPX4 knockout A375 cells rescued cellular sensitivity to propafenone-induced cell death. These findings establish a causal connection between mitochondria and propafenone-induced ferroptosis under RSL3 treatment. High levels of mitochondrial ROS and ferrous iron, as well as starvation due to the decomposition of mitochondrial glutamine, can contribute to mitochondria-associated ferroptosis.1842,44 However, while glutamine metabolism is essential for cysteine deprivation-induced ferroptosis, it is not required for GPX4 inhibition-induced ferroptosis.44 Hence, we focused our attention on mitochondrial ROS and ferrous iron. Given the crucial role of iron in mitochondrial ROS and ferroptosis,45 we hypothesized that propafenone, combined with GPX4 inhibition, triggers mitochondria-associated ferroptosis by modulating mitochondrial iron metabolism and increasing mitochondrial ROS levels. The elevated levels of mitochondrial ROS and ferrous iron in the combination group confirmed this hypothesis. A previous study has shown that O-GlcNAcylation-mediated ferritinophagy leads to the accumulation of labile iron in mitochondria, triggering mitophagy and ferroptosis.43 However, pharmacological and genetic inhibition of autophagy failed to block the cell death induced by the combination of propafenone and RSL3, suggesting an autophagy-independent mitochondria-associated ferroptosis involved in it. We thus performed RNA sequencing to identify differentially expressed genes related to mitochondrial iron metabolism after propafenone treatment in the presence of RSL3 and found that HMOX1 was significantly upregulated. Pharmacological and genetic inhibition of HMOX1 attenuated mitochondrial ferrous iron levels and ferroptosis susceptibility in melanoma treated with propafenone and RSL3, indicating the crucial role of HMOX1 in propafenone-facilitated mitochondria-associated ferroptosis. HMOX1 is a mitochondrial enzyme responsible for catalyzing the hydroxylation of heme to produce ferrous iron, carbon monoxide, and biliverdin by opening the iron-protoporphyrin ring, consuming O2 and nicotinamide adenine dinucleotide phosphate hydride as electron donors.16 46 47 HMOX1 has a dual role in regulating ferroptosis.22,2448 Initially, it may exert an anti-ferroptotic function through antioxidant activity-mediated cytoprotective mechanisms. Its product, biliverdin, can be rapidly converted by biliverdin reductase into a potent antioxidant, bilirubin. However, excessive activation of HMOX1 can lead to an enhanced generation of ferrous iron beyond the buffering capacity of ferritin. Consequently, free ferrous ions can instigate ROS production through the Fenton reaction, ultimately culminating in cellular ferroptosis. Our study displayed an elevation in the mRNA levels of FTL and FTH1 in melanoma cells following co-treatment with propafenone and RSL3, corroborating this mechanism. Consistent with our findings, several studies have also highlighted the role of HMOX1 activation in inducing ferroptosis.22,24 Various genes, including NRF2, HIF1α, p53, MAFG, and BACH1, have been reported to modulate HMOX1-mediated ferroptosis.1649,53 However, pharmacological inhibition of NRF2 failed to rescue the combination-induced cell death, suggesting that NRF2, a key regulator of ferroptosis, was not involved in this process. The upregulation of JUN expression and its nuclear translocation, along with the activation of the upstream JNK pathway following co-treatment with propafenone and RSL3, indicates that this combination can promote HMOX1 expression through activation of the JNK/JUN axis. Pharmacological inhibition of JNK signaling, resulting in the suppression of mitochondrial ferrous ions and cell death, further supports our hypothesis. Immunotherapy exerts its antitumor function in part by inducing ferroptosis.4 5 54 However, the association between ferroptosis vulnerability and immunotherapy outcomes remains elusive. In this study, we investigated this association using an immunotherapy data set comprising 437 samples and found that ferroptosis levels were elevated in the ICB-response group compared with the non-response group, suggesting a potential link between ferroptosis susceptibility and effective immunotherapy. Previous research has demonstrated that enhancing the susceptibility of immunotherapy-resistant TYRO3high tumors to ferroptosis by inhibiting the TYRO3-mediated AKT/NRF2 pathway can re-sensitize them to anti-PD-1 therapy.8 Similarly, another recent study revealed that MerTK renders hepatocellular carcinoma resistant to anti-PD-1/programmed cell death 1 ligand 1 (PD-L1) therapy by counteracting ferroptosis through upregulating solute carrier family 7 member 11 via the ERK/SP1 pathway. Disrupting MerTK-mediated ferroptosis resistance and the immunosuppressive tumor microenvironment enhances the efficacy of checkpoint inhibitors.9 Targeting bromodomain containing 4 was demonstrated to be able to potentiate melanoma ferroptosis and immunotherapy through aldo-keto reductase family 1 member C2 inhibition.29 Consistent with these findings, our study showed that propafenone significantly improved immunotherapeutic efficacy by facilitating ferroptosis. Patients with high levels of ferroptosis/JUN/HMOX1 exhibited better therapeutic outcomes with immunotherapy and longer progression-free survival compared with those with low levels of ferroptosis/JUN/HMOX1. All these findings indicate that increasing the ferroptosis vulnerability of tumors serves as a promising approach to overcoming anti-PD-1/PD-L1 resistance. In summary, we have identified a clinically applicable cardiovascular drug, propafenone, that dramatically enhances melanoma vulnerability to ferroptosis. Mechanistically, propafenone facilitates mitochondria-associated ferroptosis in combination with RSL3 by transcriptionally upregulating mitochondrial HMOX1 via activation of the JNK/JUN signaling pathway. Moreover, propafenone can synergize with immunotherapy in melanoma by facilitating ferroptosis. Taken together, our findings may offer potential therapeutic strategies for the amplification of the desirable ferroptosis and provide preclinical evidence for the application of ferroptosis-based therapy. Materials and methods Cell culture The human melanoma WM35, A375, A2058, SK-MEL-2, SK-MEL-5, and SK-MEL-28 cell lines as well as HEK293T cell lines were obtained from the American Type Culture Collection (ATCC) and cultured in Dulbecco’s Modified Eagle’s Medium (Biological Industries) supplemented with 10% fetal bovine serum (Biological Industries) and 1% penicillin-streptomycin solution (Beyotime Biotechnology) at 37°C in a humidified atmosphere with 5% CO2. The murine melanoma B16F10 cell line, also obtained from ATCC, was cultured in RPMI1640 medium (Biological Industries), while the murine melanoma YUMM1.7 cell line, obtained from Yale University,55 was cultured in Dulbecco’s Modified Eagle Medium/Nutrient Mixture F-12 (CellMax). Chemicals RSL3 (HY-100218A), IKE (HY-114481), ML210 (HY-100003), ML162 (HY-100002), FINO2 (HY-129457), chloroquine (HY-17589A), necrostatin-1s (HY-14622A), Z-VAD-FMK (HY-16658B), DFO (HY-B1625), Fer-1 (HY-100579), TEMPO (HY-W001187), Mito-TEMPO (HY-112879), hemin (HY-19424), ZnPP (HY-101193), CHX (HY-12320), tanzisertib (HY-15495), trigonelline (HY-N0414) and ML385 (HY-100523) were purchased from MedChemExpress. Propafenone hydrochloride (HCl) (S2500), Lip-1 (S7699), carbamazepine (S1693), mexiletine HCl (S4225), procainamide HCl (S4294), sotalol HCl (S2509), betaxolol (S1827), timolol maleate (S4123) were obtained from Selleck China. RNA extraction and quantitative real-time PCR Total RNA was extracted using Magzol reagent (Magen, R4801), followed by reverse transcription with the HiScript Q RT SuperMix kit (Vazyme, R223-01). Real-time PCR was performed using SYBR Green Master Mix (Bimake, B21703) in the Applied Biosystems QuantStudio 3 Real-Time PCR System (Thermo Fisher Scientific). Differences in mRNA expression were calculated using the ΔΔCt method with GAPDH as an internal reference control. Primer sequences used for quantitative real-time PCR analysis are provided in online supplemental table S2. Western blot analysis Total protein was extracted from cells using RIPA lysis buffer (Beyotime, P0013C). Mitochondrial fractions were isolated using the Mitochondria Isolation Kit for Cultured Cells (Thermo Fisher Scientific, 89874) according to the manufacturer’s instructions. Western blot analysis was performed following the protocols outlined in our previous study.12 Images were captured using the Odyssey Fc Imaging System (LI-COR Biosciences, Lincoln, Nebraska, USA). The primary antibodies used were as follows: LPCAT3 (Abcam, ab232958), GPX4 (CST, 52455), NRF2 (CST, 12721), DHODH (CST, 26381), ACSL4 (Santa Cruz, sc-365230), TOMM20 (Abcam, ab186735), FSP1 (CST, 24972), ATG5 (CST, 12994T), HMOX1 (Proteintech, 10701–1-AP), GAPDH (Proteintech, 60004–1-Ig), and β-ACTIN (Proteintech, 66009–1-Ig). Cell viability and death analysis Cell viability was determined using the Cell Counting Kit-8 (CCK-8) assay (Bimake, B34302), as previously described.12 Cell death was detected using 5 mg/mL PI (Invitrogen, P1304MP). For imaging, cells (30,000/well) were seeded in a 24-well plate and treated as indicated. After treatment, the medium was removed, and the cells were incubated for 1 min after adding 500 µL of phosphate-buffered saline (PBS) containing 5 µg/mL PI to each well. Images were captured using a fluorescence microscope (Nikon, Japan). For quantitative flow cytometry analysis, cells (300,000/well) were seeded in a 6-well plate and treated accordingly. Both floating and adherent cells were collected, and cell pellets were resuspended in 200 µL of PBS containing 5 mg/mL PI for analysis by flow cytometry (BD, USA). Screening of cardiovascular and kinase drug library A375 or SK-MEL-28 cells (6,000/well) were seeded onto 96-well plates. Following overnight incubation, cells were subjected to treatment with DMSO, RSL3 (2.5 µM), 200 FDA-approved cardiovascular drugs (Selleck Chemicals, online supplemental table S3), or combinations for a duration of 12 hours. For kinase drug library screening, cells in the 96-well plates were treated with DMSO, RSL3 (2.5 µM), propafenone (5 µM), and the combination of propafenone and RSL3 with or without compounds from the kinase drug library (MedChemExpress, online supplemental table S4) for a period of 6 hours. Cell viability was evaluated using the CCK-8 assay. Treatment combination analysis A375 and SK-MEL-28 melanoma cells were treated with propafenone and RSL3 at indicated concentrations. The combination effect was assessed using the CI values, which were calculated by CompuSyn software applying the Chou-Talalay method, as described previously. Log (CI) values greater than 0, equal to 0, or less than 0 indicate antagonistic, additive, or synergistic effects, respectively. Colony formation assay Cells were seeded at a density of 5,000 cells per well in 6-well plates and allowed to grow for 2–3 days before treatment with the specified reagents. Fresh medium containing the indicated drugs was replenished every 3 days. Following a 10-day treatment period, the cells were fixed in 4% paraformaldehyde and subsequently stained with 0.5% crystal violet for 20 min. Images were captured using a camera. DNA constructs, lentiviral transduction and RNA interference Single-guide RNAs targeting GPX4 were cloned into the pLenti-CRISPR-V2 plasmid. The sgGPX4 sequences were as follows: 5′-GGTGAAGCGCTACGGACCCA-3′; Flag-tagged cytoplasmic or mitochondrial GPX4 overexpression plasmids, along with control empty vectors, were obtained from Youze Biotechnology. Human short hairpin RNA (shRNA) targeting HMOX1 and control empty shRNA were purchased from Heyuan Biotechnology. All constructs were confirmed by DNA sequencing. Human ATG5 small interfering RNA (siRNA) (5′-GGACGAAUUCCAACUUGUUTTAACAAGUUGGAAUUCGUCCTT-3′) and control empty siRNA were purchased from GenePharma. Transfection with siRNA, shRNA, and complementary DNA was conducted using TurboFect (Thermo Fisher Scientific, R0532) following the manufacturer’s protocol, as previously described.12 To select transduced cells, 2 µg/mL puromycin (Basalmedia, S250J0) was added. Additionally, flow cytometry was employed to screen for successful transfection of flag-tagged control, cytosolic, or GPX4-expressing plasmids in GPX4 KO A375 cells. Lipid peroxidation, mitochondrial lipid peroxidation and ROS assay To evaluate lipid peroxidation, mitochondrial lipid peroxidation, or mitochondrial ROS levels, cells were seeded in 6-well plates and treated with the specified compounds. Subsequently, the collected cells were stained with 1 µM C11-BODIPY 581/591 (Invitrogen, D3861), 1 µM Mito-PerOx (Cayman, 18798), or 2.5 µM Mito-SOX Red (Invitrogen, M36008) at 37°C for 20 min, respectively. After staining, cells were washed twice with PBS and resuspended in 200 µL of PBS for flow cytometry analysis. The flow cytometry data were analyzed using FlowJo software (V.10.4). GSH, glutathione peroxidase, and MDA assay GSH, GPX4 activity, or MDA were measured using GSH and GSSG assay (Beyotime Biotechnology, S0053), Glutathione Peroxidase Assay Kit (Abcam, ab102530), or Lipid Peroxidation Assay Kit (Merck, MAK085), respectively, according to the manufacturer’s instructions, as previously described.12 Transmission electron microscopy A375 cells (2×106/dish) were seeded in 10 cm dishes. After the specified treatment, the cells were fixed with 2.5% glutaraldehyde (Coolaber, SL1770), followed by incubation in 1% osmium tetroxide for 1 hour. Subsequently, they underwent dehydration, embedding, sectioning, and staining as described in previous studies for transmission electron microscopy using the JEM-1400FLASH electron microscope (JEOL, Japan).12 Ferrous iron and mitochondrial ferrous iron assay The relative ferrous iron levels were assessed using the Iron Assay Kit (Abcam, #ab83366) following the manufacturer’s protocol, as previously described.12 Mitochondrial ferrous iron was evaluated using Mito-FerroGreen (Dojindo, # M489) probes. For fluorescence imaging, cells were treated as indicated and washed three times with serum-free culture medium. They were then incubated with 5 µmol/L Mito-FerroGreen and 200 nmol/L Mito-Tracker Red (Beyotime, C1035) at 37°C for 30 min, followed by three washes with serum-free culture medium. Subsequently, cells were incubated with Hoechst (Beyotime, 33258) for 10 min, washed three times with serum-free culture medium, and finally added 500 µL of serum-free culture medium for fluorescence imaging using the BioTek Cytation 10 Cell Imaging Multifunctional Detection System (USA). Measurement of mitochondrial membrane potential Mitochondrial membrane potential was assessed using the enhanced JC-1 assay kit (Beyotime, C2003S). Cells collected after the indicated treatment were washed three times with JC-1 staining buffer, incubated with JC-1 solution at 37°C for 25 min, and then analyzed by flow cytometry (BD, USA). Immunohistochemistry Tumor tissues were harvested, fixed, embedded, and sectioned. For immunohistochemistry staining, the slides were stained with 4-HNE (Abcam, ab46545) antibodies according to the methods previously described.56 Images were obtained at 200× magnification using the microscope and quantified by Image J software. Immunofluorescence Cells were seeded onto sterilized glass coverslips in a 24-well plate and fixed with cold methanol at room temperature for 10 min after the indicated treatment. The cells were permeabilized with PBS containing 0.1% Triton X-100 (Sangon Biotech, A600198) for 5 min and blocked with goat serum (ZSGB-Bio, ZLI-9056). Next, the cells were incubated overnight at 4°C with a c-JUN antibody (Santa Cruz, sc-74543, 1:100). For immunofluorescence staining on tumor tissue sections, sections were prepared following routine protocols for immunohistochemistry and were stained with anti-CD8α (Abcam, ab217344, 1:1000) and GZMB (Abcam, ab255598, 1:200). Then, cells and tumor tissue were incubated with the secondary antibodies conjugated with Alexa Fluor 488 (Invitrogen, A21202, 1:1,000) or Alexa Fluor 594 (Invitrogen, A21207, 1:1,000) at room temperature for 1 hour. Finally, samples were mounted using a fluoroshield mounting medium with DAPI (Abcam, ab104139). Images were captured using the BioTek Cytation 10 Cell Imaging Multifunctional Detection System (USA). Animal experiments All experimental protocols and procedures involving mice were approved by the Ethical Review of the Animal Care and Use Committee of Central South University (no.202312144, Changsha, Hunan, China). Female BALB/c nude and C57BL/6 mice, aged 4–6 weeks, were obtained from GemPharmatech and were maintained under specific-pathogen-free conditions in the Animal Facility of Central South University. To investigate the combination effect of propafenone and ferroptosis induction, 2×106 GPX4 KO or control A375 cells per mouse were suspended in 100 µL of PBS and inoculated subcutaneously into the right back flanks of the mice. When tumor volumes reached approximately 100 mm3, mice were randomly allocated into five subgroups: (1) injected with control A375 cells and treated with vehicle (intraperitoneally, daily), (2) injected with control A375 cells and treated with propafenone (10 mg/kg, intraperitoneally, daily), (3) injected with GPX4 KO A375 cells and treated with vehicle, (4) injected with GPX4 KO A375 cells and treated with propafenone, (5) injected with GPX4 KO A375 cells and treated with propafenone and Lip-1 (10 mg/kg, intraperitoneally, daily). Additionally, 2×106 SK-MEL-28 cells were inoculated subcutaneously into the BALB/c nude mice to establish another tumor model. Once the tumor volume reached approximately 100 mm3, the mice were treated with propafenone, RSL3 (50 mg/kg, intratumorally, every 3 days), or both. For the propafenone and immunotherapy combination therapy assay, YUMM1.7 cells (1×106 cells per mouse) or B16F10 cells (5×105 cells per mouse) were injected subcutaneously into immunocompetent C57BL/6 mice. Tumor-bearing mice were randomly divided into the following subgroups: (1) vehicle plus isotype IgG2α (Bio X Cell, #BE0089), (2) vehicle plus checkpoint inhibitor anti-PD-1 antibody (Bio X Cell, #BE0146), (3) propafenone plus isotype IgG2α, (4) propafenone plus anti-PD-1 antibody with or without Lip-1. The isotype IgG2α or anti-PD-1 antibody was administered at a dose of 100 µg per mouse intraperitoneally every 3 days. The administration route and dose of propafenone or Lip-1 were the same as above. The tumor volumes and body weight of all mice were monitored every 3 days. The tumor volumes were calculated as 0.5×length×width2 based on caliper measurements. To assess the survival of animals, mice bearing tumors exceeding 2,000 mm3 were defined as “dead.” Bioinformatics analysis RNA sequencing and data processing A375 melanoma cell lines were treated with DMSO, propafenone, RSL3, or propafenone combined with RSL3 for RNA sequencing analysis. Experiments were performed in biological duplicates. RNA sequencing reads were adaptor trimmed and the data quality was assessed with the FastQC software (https://github.com/s-andrews/FastQC). After obtaining the BAM files, HISAT2 software was used to map reads to the human reference genome (GRCh38.p12 assembly). The Stringtie software and genome annotation file (hg38_ucsc.annotated.gtf) were used to assemble the plotted reads into transcripts or genes. The differential genes between the groups were calculated using DESeq2. Significant genes were identified by the criterion: |fold change|≥1.5, adjust p value<0.05. Data were presented in online supplemental table S5 (related to figures4AC, G and 5G). Melanoma data sets collection The melanoma data set was collected from different data sets, mRNA expression, protein expression, and clinical data of skin cutaneous melanoma (SKCM) samples were downloaded from The Cancer Genome Atlas (TCGA) data portal. Gene-Expression Omnibus (GEO) SKCM cohorts were downloaded from GEO (https://www.ncbi.nlm.nih.gov/geo/). Cell Line mRNA expression was obtained from the Cancer Cell Line Encyclopedia (CCLE) (https://sites.broadinstitute.org/ccle/). The transcriptome dataset of 437 patients with advanced melanoma treated with immune checkpoint inhibitors (ICI) agents was obtained from Ribas et al.31 The Gide cohort (PRJEB23709: anti-PD-1 monotherapy and anti-PD-1/anti-cytotoxic T lymphocyte-associated protein-4 (CTLA-4) combined therapy) was downloaded from the SRA database (https://www.ncbi.nlm.nih.gov/bioproject/).57 The Liu/VanAllen cohort (phs000452.v3: anti-PD-1/CTLA4-treated metastatic melanoma) was downloaded from the dbGaP database (https://www.ncbi.nlm.nih.gov/gap/).58 59 The richness of T-cell receptor/B-cell receptor among TCGA-SKCM samples was available at https://gdc.cancer.gov/about-data/publications/panimmune. The data was analyzed using the R (V.4.2.0) and R Bioconductor software packages (related to figuresS4H, 5C,D, 6AE and online supplemental figure S6A–D). Calculating the ferroptosis-related pathway scores The ferroptosis score was calculated using the WP_ferroptosis gene set obtained from MsigDB (https://www.gsea-msigdb.org/gsea/msigdb). The gene set for ferroptosis-related pathways was available in a published article.33 ssGSEA was performed to calculate the enrichment scores for ferroptosis-related pathways in each sample, using the “GSVA” function in R (related to figures6A 7E,F). Calculating the immunotherapy outcome-related scores To calculate the immunotherapy outcome-related scores, T cell-inflamed GEP level in each sample was calculated based on the gene signatures from Ayers et al by performing GSVA.38 We calculated the geometric mean of gene expression of two cytolytic markers (GZMA and PRF1) to assess CYT level.60 Other immune features, such as IFN-γ response, microsatellite instability score, the infiltration of CD8+ T cells, cytotoxic T lymphocytes, myeloid-derived suppressor cells, M2 subtype of tumor-associated macrophages, and T-cell exclusion score in each melanoma sample was obtained based on the tumor immune dysfunction and exclusion computational framework (related to figure 6E, online supplemental figure S6C,D).61 Transcription factor prediction Three publicly accessible databases, including hTFtarget (http://bioinfo.life.hust.edu.cn/hTFtarget), PROMO (http://alggen.lsi.upc.es), and JASPAR (http://jaspardev.genereg.net/), were used to predict candidate transcription factors that may bind to the HMOX1 promoter. The confidence scores were set above 10 in hTFtarget. Dissimilarity was set under 15% in PROMO. The sequence-matching score was set above 10 in JASPAR (related to figure 5B). RNA sequencing for mice tumor Immunocompetent C57BL/6 mice implanted with B16F10 melanoma cells received intraperitoneal injections of anti-PD-1 treatment or IgG control with or without propafenone on the indicated day after the tumor reached approximately 100 mm3 and tumors were excised for RNA sequencing on day 16. Data quality was assessed using FastQC software (https://github.com/s-andrews/FastQC). After obtaining the BAM file, paired-end clean reads were aligned to the GRCm38 reference genome (mm10) using HISAT2 (V.2.2.1). We further processed the sequencing alignments using SAMTools (V.1.9), which facilitated sorting, indexing, and filtering for RNA sequencing data preparation. StringTie (V.2.2.1) was then used to calculate gene counts and transcripts per million for downstream analyses (related to figure 7E–G). Statistical analysis The data were presented as mean±SD and analyzed using GraphPad Prism software (V.8.0.1). A two-tailed unpaired Student’s t-test was used to compare the two groups. For comparisons among multiple groups, one-way or two-way analysis of variance analysis, or mixed-effects models with Tukey’s multiple comparison test were employed. The Wilcoxon rank-sum test was applied to compare the differences in immunotherapy outcome-related scores between the high ferroptosis/HMOX1/JUN group and the low ferroptosis/HMOX1/JUN group. Fisher’s exact test/χ2 test was performed to compare the proportion of patients with different responses to immunotherapy between the ferroptosis/HMOX1/JUN group and the low ferroptosis/HMOX1/JUN group. Kaplan-Meier comparative survival analysis was used to analyze the prognosis, and the log-rank test was used to determine the significance. P values of <0.05 were considered statistically significant and represented as *p<0.05, **p<0.01, and ***p<0.001. supplementary material Acknowledgements We are very grateful to Gene Expression Omnibus and The Cancer Genome Atlas database for providing the transcriptome and clinical information. Additionally, we thank Biorender (https://www.biorender.com/) for the assistance with the illustration. Footnotes Funding: This work was supported by the National Natural Science Foundation of China (Grant Nos. 82103183 to FZ, 82272849 to GD, 82403139 to YL, 82403618 to YX), Natural Science Foundation of Hunan Province (Grant Nos. 2022JJ40767 to FZ), Natural Science Fund for Outstanding Youths in Hunan Province (2023JJ20093 to GD), Huxiang Youth Talent Program (Grant Nos. 2023RC3072 to GD), and Postdoctoral Fellowship Program of CPSF (Grant Nos. GZB20240875 to QZ). Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: Not applicable. Data availability free text: All data will be made available on request. The raw and processed bulk RNA sequencing data of tumor tissues in this paper are available through the Gene Expression Omnibus (accession numbers GSE279924). Contributor Information Qian Zhou, Email: zhouqian_103288@outlook.com. Yating Dian, Email: dianyating@outlook.com. Yi He, Email: yihe0808@csu.edu.cn. Lei Yao, Email: leiyao@csu.edu.cn. Hui Su, Email: suhui8@mail2.sysu.edu.cn. Yu Meng, Email: Meng.yuyu@outlook.com. Yuming Sun, Email: 962881298@qq.com. Daishi Li, Email: dermlidaishi@163.com. Yixiao Xiong, Email: xiongyixiao@tjh.tjmu.edu.cn. Furong Zeng, Email: zengflorachn@hotmail.com. Xiaowei Liang, Email: xiaowei.liang@csu.edu.cn. Hong Liu, Email: hongliu1014@csu.edu.cn. Xiang Chen, Email: chenxiangck@126.com. Guangtong Deng, Email: dengguangtong@csu.edu.cn. Data availability statement Data are available upon reasonable request. References Associated Data Supplementary Materials Data Availability Statement Data are available upon reasonable request.