Fig. 1

BET inhibitor-mediated downregulation of AKR1C2 sensitizes melanoma to ferroptosis induced by GPX4 inhibition. a Signaling pathways targeted by drugs that are sensitive (blue) or resistant (red) to the ferroptosis score (FPS) from CTRP database. Drug names are listed on the x-axis and the signaling pathway targeted by the drug on the y-axis. The upper bar plot denotes the Spearman correlation between FPS and drug sensitivity. r < 0 defines drug sensitivity, r > 0 defines drug resistance; the bar plot on the right shows the number of drugs targeting each signaling pathway; The size of the point indicates the significance of the correlation. b A375 melanoma cells were pretreated with JQ1 (1 μmol/L) or NHWD-870 (10 nmol/L) for 24 h, and then cotreated with RSL3 (2.5 μmol/L). DMSO, necrostatin-1s (Nec-1s, 10 μmol/L), chloroquine (CQ, 10 μmol/L), Z-VAD-FMK (10 μmol/L), ferrostatin-1 (Fer-1, 4 μmol/L), or deferoxamine (DFO, 100 μmol/L) were added in combination groups for 10 h, and cell viability was assessed. c Lipid peroxidation production in A375 cells was measured by flow cytometry using BODIPY-C11. Cells were first treated with 1 μmol/L JQ1 or 10 nmol/L NHWD-870 for 24 h, alone or in combination with 2 μmol/L RSL3, 2 μmol/L RSL3 plus 4 μmol/L Fer-1, 2 μmol/L RSL3 plus 1 mmol/L N-acetyl-cysteine (NAC) for another 6 h as indicated. d Transmission electron microscopy of A375 cells pretreated with JQ1 (1 μmol/L) for 24 h and then co-treated with RSL3 (2.5 μmol/L) in combination group for 6 h. Red arrow indicates morphological change of mitochondria. Scale bar = 4 μm (upper), 1 μm (bottom). e Dose–response curves of RSL3-induced death in control (shCon) and BRD4 knockdown (shBRD4) A375 cells. f Heatmap representation of changes in gene expression in JQ1-treated/siBRD4 versus control A375 cells (P < 0.05). Each horizontal line represents one gene, ordered by gene expression. g The regulatory effect of BET inhibitors on AKR1C2 expression after 48 h treatment based on Western blotting. DMSO (0.02%); NHWD-870: 10 nmol/L; JQ1: 2 μmol/L; OTX015: 2 μmol/L; I-BET151: 2 μmol/L. h Quantification of AKR1C2 expression in A375 cells treated with the same BET inhibitors as g by immunofluorescence images. Scale bar = 20 μm. i ChIP-seq analysis of BRD4 binding peak in the AKR1C2 promoter in DMSO-treated [2] and NHWD-870-treated melanoma cells (upper) or in DMSO-treated [3] and JQ1-treated cells (lower). Red boxes indicate 2 kb regions around the transcription start site. The arrow indicates the transcription direction. j Relative viability of 5 μmol/L RSL3-treated shCon and shAKR1C2 A375 cells at the presence of DMSO or 10 nmol/L NHWD-870. k Relative viability of 5 μmol/L RSL3-treated control and AKR1C2 overexpression A375 cells in the presence of DMSO or 10 nmol/L NHWD-870. l Tumor growth in control, GPX4 knockout (sgGPX4), NHWD-870, and combination groups. m Representative immunohistochemistry images of 4-HNE and GPX4 in the four groups. Scale bar = 50 μm. n Tumor growth in the isotype IgG2α, NHWD-870, anti-PD-1 antibody, and combination groups. The percentage of cells expressing IFN-γ (o) and GZMB (p) in tumor-infiltrating CD8⁺ T cells by flow cytometry analysis. q Spatial enhanced-resolution clustering performed by the BayesSpace algorithm at the left panel identified 4 clusters corresponding to the original histopathological annotations. Spatial heatmap shows the level of BRD4 and FPS expression among 4 clusters at the right panel. r Scatter plot shows the spearman correlation between the expression level of BRD4 and FPS. Kaplan–Meier curves compare overall survival (TCGA-SKCM) (s) and progression-free survival of immune checkpoint inhibitors (ICIs) cohorts (PRJEB23709) (t) between the high-FPS + low BRD4 (blue) and low-FPS + high BRD4 (yellow) groups. u The proportion of patients with different responses to immunotherapy in the TCGA-SKCM cohort with TIDE-predicted ICB response and the melanoma ICIs cohort (PRJEB23709). R response, including responder, complete response and partial response, NR non-response, including non-responder, stable disease and progressive disease. Quantification data are presented as mean ± SD, and compared using one-way ANOVA in b-c, o-p, two-way ANOVA in e, j-l, n, log-rank test in s-t, and Fisher’s exact test/chi-square test in u. AKR1C2 aldo–keto reductase 1C2, BET bromodomain and extra-terminal domain, BRD4 bromodomain-containing protein 4, CQ chloroquine, CTRP Cancer Therapeutics Response Portal, DFO deferoxamine, EV empty vector, FDR false discovery rate, Fer-1 ferrostatin-1, FPS ferroptosis score, GPX4 glutathione peroxidase 4, GZMB granzyme B, 4-HNE 4-hydroxynonenal, ICIs immune checkpoint inhibitors, IFN-γ interferon-γ, NAC N-acetyl-cysteine, Nec-1s necrostatin-1s, NR non-response, PD-1 programmed cell death protein-1, R response, SKCM skin cutaneous melanoma, TCGA the cancer genome atlas, ns non-significant, *P < 0.05, **P < 0.01.