How does ferrocene correlate with ferroptosis? Multiple approaches to explore ferrocene-appended GPX4 inhibitors as anticancer agents

Abstract

Ferroptosis has emerged as a form of programmed cell death and exhibits remarkable promise for anticancer therapy. However, it is challenging to discover ferroptosis inducers with new chemotypes and high ferroptosis-inducing potency. Herein, we report a new series of ferrocenyl-appended GPX4 inhibitors rationally designed in a "one stone kills two birds" strategy. Ferroptosis selectivity assays, GPX4 inhibitory activity and CETSA experiments validated the inhibition of novel compounds on GPX4. In particular, the ROS-related bioactivity assays highlighted the ROS-inducing ability of 17 at the molecular level and their ferroptosis enhancement at the cellular level. These data confirmed the dual role of ferrocene as both the bioisostere motif maintaining the inhibition capacity of certain molecules with GPX4 and also as the ROS producer to enhance the vulnerability to ferroptosis of cancer cells, thereby attenuating tumor growth in vivo. This proof-of-concept study of ferrocenyl-appended ferroptosis inducers via rational design may not only advance the development of ferroptosis-based anticancer treatment, but also illuminate the multiple roles of the ferrocenyl component, thus opening the way to novel bioorganometallics for potential disease therapies.

Fig. 1. (A) Illustration of the basis of developing ferrocene-based ferroptosis inducers. (B) Typical small molecule ferroptosis inducers.

Fig. 2. (A) Illustration of the ferrocene-based organometallic compound library construction and screening. (B) Structure-based drug design (SBDD) of ferrocene-appended GPX4 inhibitors. (C) Molecular structures of the ferrocene complexes 12, 15 and 17 with thermal ellipsoids shown at 50% probability.

Fig. 3. (A) The percentage of cell viability after treatment of ML162 or 17 (0.4 μM) with Z-VAD-FMK (10 μM), Nec-1 (10 μM), Wortmannin (30 μM), Fer-1 (2 μM), and DFO (30 μM); analysis results represented the mean ± SD. (B) Transwell assays were used to detect the invasion of cancer cells after treatment of 17 (0.5 and 2 μM). (C) CETSA experiments were performed on OS-RC-2 cells treated with DMSO, ML162 and 17 (5 μM) for 4 h. The protein levels were analyzed by western blotting after heating at different temperatures (30–60 °C), and the apparent T_agg values for GPX4 in OS-RC-2 cells in the absence and presence of ML162 and 17 were 44.09 °C to 51.42 °C, and 44.05 to 58.00 °C, respectively. (D) Western blot analysis of GPX4 levels in OS-RC-2 cells following treatment of 17 in 3 and 24 h. *P < 0.05, **P < 0.01, and ***P < 0.001 versus the control groups.

Fig. 4. (A) Monitoring of the generation of ROS of corresponding compounds (100 μM) using fluorescence spectroscopy in combination with ROS-sensitive DCFH-DA (λ_ex = 501 nm and λ_em = 531 nm). (B) OS-RC2 cells were treated with ML162 (1 μM) and 17 (0.5, 1, and 2 μM) for 4 h, and then the content of MDA was determined. (C) Flow cytometry analysis for intracellular LPO by C11-BODIPY of OS-RC-2 cells treated with DMSO, ML162 and 17 (0.5 μM) with and without fer-1 (1.5 μM) respectively. (D) Flow cytometry analysis for intracellular ROS by DCFH-DA of OS-RC-2 cells treated with DMSO, ML162 and 17 (0.5 μM) with and without fer-1 (1.5 μM) respectively. (E) Representative confocal images and analysis of C11-BODIPY staining of OS-RC-2 cells after incubation with ML162 and 17 (0.5 μM) with and without fer-1 (1.5 μM) respectively. (F) Representative confocal images and analysis of DCFH-DA staining of OS-RC-2 cells after incubation with ML162 and 17 (0.5 μM) with and without fer-1 (1.5 μM) respectively. (G) Intracellular Fe²⁺ was visualized with the FerroOrange fluorescent probe. OS-RC-2 cells were treated with ML162 and 17 (0.5 μM) for 4 h, and the content of intracellular Fe²⁺ was determined. *P < 0.05, **P < 0.01, and ***P < 0.001 versus the control groups.

Fig. 5. The binding models of ferrocenyl ligands in the GPX4 crystal structure (PDB code 6HKQ). Compounds 15 (A), 17 (B), 12 (C), and 13 (D) were docked to GPX4 by Watvina with the template-based method; then all the structures were minimized with the xtb program. The original ligand ML162 was shown in green transparent stick mode, and ferrocenyl ligands in cyan stick mode. The hydrogen bonds were represented with a yellow dashed line by VMD software.

Fig. 6. (A) In vivo antitumor activity of ML162 and 17 in a human OS-RC2 renal carcinoma model (BALB/c nude mice, given medicine every three days, n = 6). (B) Picture of excised tumors from the different groups after the treatment; (C) tumor volume curves during the treatment period; (D) tumor weights after treatment; (E) tumor inhibition ratio after the treatment period. (F) Body weight curves of different groups after the treatment; (G) H&E staining of the major organs of mice after treatment with compounds. (H) Anti-GPX4 immunohistochemistry images and analysis from the OS-RC-2 xenograft model. (I) Western blot analysis of GPX4 levels in tumor tissue following treatment of the corresponding compounds. *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001 versus the control groups.

Fig. 7. RNA-seq analysis of samples from OS-RC-2 xenograft model rats treated with 17 (20 mg kg⁻¹) and the control. (A) Volcano plot of all DEGs; red indicates upregulation, whereas blue indicates downregulation. (B) Protein–protein interaction (PPI) networks constructed by Metascape and the protein function were enriched. (C) GO term enrichment analysis of DEGs from the 17 (20 mg kg⁻¹) treatment group. (D) GSEA analysis of DEGs in the JAK-STAT signaling pathway. (E) KEGG enrichment analysis of DEGs from the 17 (20 mg kg⁻¹) treatment group. (F and G) The genes as the driver and suppressor of ferroptosis overlap with DEGs that are up-regulated and down-regulated by using a Venn diagram.

Fig. 8. Proposed mechanisms of ferrocene-appended ferroptosis inducers.