Ferroptosis, an iron-dependent form of cell death driven by overwhelming lipid peroxidation, represents a vulnerability in cancers, and therapeutic strategies to further potentiate ferroptosis hold great potential for melanoma treatment.
To systematically identify drugs that sensitize ferroptosis, we initially calculated ferroptosis score (FPS) using our published algorithm model[
1], and conducted spearman correlation analysis between FPS and the cell sensitivities to various anti-cancer drugs across 859 cancer cell lines. Remarkably, drugs targeting epigenetic regulators were significantly associated with high FPS, especially bromodomain and extra-terminal domain (BET) inhibitors (I-BET151, JQ1, and GSK525762A;
Fig. 1a). Further investigations revealed a strong synergy in melanoma cells when BET inhibitors (JQ1, NHWD-870, OTX015, and I-BET151) were combined with the ferroptosis inducer RSL3, an inhibitor of glutathione peroxidase 4 (GPX4), with combination index values less than 1 and fewer colony numbers (Additional file 1: Fig. S1a-j). Consistently, BET inhibitors also sensitize melanoma cells to genetic inhibition of
GPX4 (Additional file 1: Fig. S1k-s). Notably, the cytotoxicity of the co-treatment of BET inhibitors and RSL3 could be completely abrogated by the ferroptosis inhibitor ferrostatin-1 and the iron chelator deferoxamine, but not by inhibitors of apoptosis (Z-VAD-FMK), necroptosis [necrostatin-1s (Nec-1s)], or autophagy [chloroquine (CQ)] in melanoma cells (
Fig. 1b; Additional file 1: Fig. S2a). The combined treatment triggered prominent ferroptosis-related characteristics, including more lipid peroxidation (
Fig. 1c; Additional file 1: Fig. S2b-c), and shrunken mitochondria with increased membrane density (
Fig. 1d). Moreover, the BET inhibitors-enhanced cell death in
GPX4-deficient melanoma cells was restored by ferroptosis inhibitors (Additional file 1: Fig. S2d-g). These results suggest that BET inhibitors potentiate
GPX4 inhibition-induced ferroptosis in melanoma.
Drug target analysis revealed that
BRD4, but not
BRD2/3, was negatively correlated with FPS in melanoma cohorts (Additional file 1: Fig. S3a). Genetic inhibition of
BRD4, but not
BRD2/3, enhanced RSL3-induced ferroptosis (
Fig. 1e; Additional file 1: Fig. S3b-h), and
BRD4 overexpression resisted ferroptosis in melanoma cells (Additional file 1: Fig. S3i-m), suggesting that BET inhibitors potentiate RSL3-induced ferroptosis by targeting BRD4. RNA-seq analysis demonstrated that
AKR1C2 was most dramatically downregulated by pharmacological and genetic inhibition of
BRD4 among 7 ferroptosis suppressors (
Fig. 1f; Additional file 1: Fig. S4a-b), which is consistent with other melanoma datasets after I-BET151 treatment (Additional file 1: Fig. S4c). BET inhibitors could significantly decrease AKR1C2 expression at both mRNA and protein levels (
Fig. 1g-h; Additional file 1: Fig. S4d-e). Likewise, genetic inhibition or overexpression of BRD4 suppressed or upregulated the expression of AKR1C2 in melanoma cells, respectively (Additional file 1: Fig. S4f-n). Notably, BET inhibitors could not affect the expression and transcription activity of NRF2, the known transcription factor of AKR1C2[
2], and GPX4 expression (Additional file 1: Fig. S4o-p), ruling out the possibility that BET inhibitors sensitize melanoma ferroptosis through GPX4 or NRF2 repression. ChIP-seq data showed a prominent BRD4 binding peak in the
AKR1C2 gene promoter, while the amplitude of the binding peak was diminished upon NHWD-870 treatment, which is consistent with Zhang
et al.[
3]. ChIP-seq data analysis (
Fig. 1i) suggesting that BRD4 transcriptionally regulates
AKR1C2 expression. We previously reported that BET inhibitors suppress STAT3 signaling through the BRD4/IL-6 axis[
4]. Inhibiting STAT3 activity by shRNA or the inhibitor stattic significantly suppressed AKR1C2 expression (Additional file 1: Fig. S5a-d). ChIP-seq analysis also demonstrated that STAT3 binds to the
AKR1C2 promoter (Additional file 1: Fig. S5e), suggesting that BET inhibitors directly inhibit
AKR1C2 expression by BRD4, or indirectly by the BRD4/IL-6/STAT3 axis. AKR1C2 was reported to inhibit ferroptosis by degrading lipid peroxides[
5], which was validated in our study. Pharmaceutical and genetic inhibition of AKR1C2 sensitized melanoma cells to ferroptosis (Additional file 1: Fig. S6a-h). On the contrary,
AKR1C2 overexpression conferred resistance to RSL3-induced ferroptosis in melanoma cells (Additional file 1: Fig. S6i-l). Strikingly, BET inhibitors failed to further potentiate RSL3-induced ferroptosis in the presence of AKR1C2 inhibitors or
AKR1C2 genetic silencing (
Fig. 1j; Additional file 1: Fig. S6m-s). Overexpression of
AKR1C2 partially rescued cytotoxicity caused by the co-treatment of BET inhibitors and RSL3 (
Fig. 1k; Additional file 1: Fig. S6t-u). These findings suggest that BET inhibitors potentiate RSL3-induced ferroptosis at least partially through AKR1C2 inhibition.
In vivo,
GPX4 knockout exhibited minimal impact on melanoma progression, and NHWD-870 treatment alone moderately inhibited melanoma growth (
Fig. 1l; Additional file 1: Fig. S7a-c). However, combining
GPX4 knockout with NHWD-870 significantly impaired melanoma growth and reduced tumor weight (
Fig. 1l; Additional file 1: Fig. S7a-d), with markedly increased staining of 4-HNE, an end product of lipid peroxidation (
Fig. 1m; Additional file 1: Fig. S7e). Cancer immunotherapy has been regarded as an important ferroptosisassociated pathological model
in vivo[
6]. We found that BET inhibitor administration potentiated the efficacy of anti-PD-1 antibody (
Fig. 1n; Additional file 1: Fig. S7f-i), leading to an elevated proportion of IFN-γ
+CD8
+ and GZMB
+CD8
+ T cells (
Fig. 1o-p; Additional file 1: Fig. S7j), despite no increase in total tumor-infiltrating CD8
+ T cells (Additional file 1: Fig. S7k). These results suggested that BET inhibitors sensitize melanoma to GPX4 inhibition-induced ferroptosis and immunotherapy
in vivo. We further observed that
BRD4 expression was negatively associated with FPS in three melanoma single-cell RNA-seq datasets (Additional file 1: Fig. S8a-d), as well as in melanoma regions from spatial transcriptome data (
Fig. 1q-r). Notably,
BRD4 is significantly upregulated, while ferroptosis level is downregulated in immunotherapy-resistant malignant melanoma subpopulations (Additional file 1: Fig. S8e). Melanoma patients with high-ferroptosis plus low-
BRD4/AKR1C2 predict a better prognosis (
Fig. 1s-t; Additional file 1: Fig. S8f) and improved response to immunotherapy (
Fig. 1u; Additional file 1: Fig. S8g-h). These findings suggest that BRD4/AKR1C2 is associated with reduced ferroptosis level and poor efficacy of immunotherapy from multiomics characterization.
Overall, our data illustrate that BET inhibitors potentiate GPX4 inhibition-induced ferroptosis through the dual downregulation of AKR1C2, and provide the rationale for combining BET inhibitors with GPX4 inhibitors or immunotherapy for melanoma treatment (Additional file 1: Fig. S9).
PDF
Full
55
31
关闭全屏
No. 86 Xueyuan South Road, Haidian District, Beijing
010-62199257
qkjq@cast.org.cn
