The contrasting regulatory effects of valproic acid on ferroptosis and disulfidptosis in hepatocellular carcinoma

Abstract

Background: Valproic acid (VPA), a branched short-chain fatty acid, is extensively utilized as both an antiepileptic medication and a mood stabilizer. However, the complete pharmacological functions of VPA on programmed cell death are still not fully understood. In this study, we investigated the role of VPA in modulating ferroptosis and disulfidptosis, which are emerging forms of programmed cell death triggered by lipid peroxidation and disulfide stress respectively. Methods: Herein, the network pharmacology analysis, genome-wide mRNA transcription assay and metabolomics analysis were performed to predict the major pharmacological action and potential targets of VPA. To confirm the hypothesis, pharmacological targeting model and gene knockdown model was created in our work. The pharmacological action of VPA on ferroptosis and disulfidptosis was evaluated respectively. Results: Our findings primarily indicated that the potential targets of VPA were linked to hepatocarcinogenesis and programmed cell death. Additionally, omics data suggested that VPA could significantly influence iron transport and glucose homeostasis. Notably, VPA heightened the susceptibility of hepatocellular carcinoma (HCC) cells to ferroptosis by increasing the labile iron pool, facilitating the accumulation of free iron through enhanced cellular ferritinophagy and reduced ferritin expression. Furthermore, VPA promoted the transcription of glucose-6-phosphate dehydrogenase (G6PD) and impacted glutathione (GSH) metabolism. The activation of the NRF2-G6PD pathway induced by VPA further augmented the production of NADPH and GSH, which subsequently inhibited the formation of disulfide bonds among various cytoskeletal proteins, as well as disulfidptosis in HCC cells. Conclusion: Overall, our results highlight the significant role of VPA in differentially regulating ferroptosis and disulfidptosis in HCC cells, thereby offering a precise avenue for addressing drug-resistant HCC in clinical practice.

Keywords: G6PD; disulfidptosis; ferroptosis; labile iron pool; valproic acid.

Figure 1

VPA treatment sensitizes HCC to ferroptosis. Both MHCC97-H and HepG2 cells were subjected to VPA treatment (1 mM and 2 mM) for a duration of 24 h, then exposed to Erastin (Era, 5 μM), Sorafenib (Sor, 1 μM), RSL3 (2 μM)) and Cisplatin (Cis, 10 μM) to induce cell death. Both crystal violet staining (A) and CCK-8 assay (B) were utilized to evaluate cell viability in different groups. In addition, the level of MDA was measured to access the lipid peroxidation (C), which was further confirmed via BODIPY staining (D). Additionally, the impact of VPA on different ferroptosis regulators was examined using western blot (E) and the level of free Fe²⁺ was measured using FerroOrange staining (F-G, Scale bar = 50 μm). Data are expressed as mean ± SD. The P value less than 0.05 was considered statistically significant and the value of Cohen's d over 0.8 represents a large effect size. *: P < 0.05 and Cohen's d > 0.8 compared to Ctrl group.

Figure 2

VPA influences various genes essential for the regulation of oxidative stress and ferroptosis. In this study, the HCC cell line MHCC97-H was subjected to VPA treatment (2 mM) for a duration of 24 h. Subsequently, the cell samples were collected for RNA sequencing analysis. Initially, a volcano plot was created to assess the impact of VPA treatment on gene transcription (A). Following this, differentially expressed genes were utilized for cluster analysis via Metascape. The network illustrating the top 20 enriched terms for these differentially expressed genes was color-coded by cluster ID, with the same terms further distinguished by p-value (B). The elements related to metal ion transport and oxidative stress were compiled in Figure 2C, while the key genes were presented in a heatmap (D). Furthermore, both upregulated genes (E) and downregulated genes (F) depicted in the heatmap were analyzed using the STING database, and the interactions among all aforementioned genes were illustrated in a separate Protein-Protein Interaction Network (G).

Figure 3

Ferritinophagy holds a crucial position underlying the action of VPA concerning ferroptosis. Both MHCC97-H and HepG2 cells were subjected to VPA treatment (1 mM and 2 mM) for a duration of 24 h, then harvested for western blot assay. The autophagy markers, p62 and LC3, were measured in our work (A-D). In addition, the effect of VPA treatment in different time points was also analyzed (E-F). The impact of VPA treatment on autophagic flux was evaluated using live cell immunofluorescence assay, and the red puncta (white arrow) represented autolysosome (G, Scale bar = 10 μm). Moreover, MHCC97-H cells were treated with VPA plus iron chelators (Deferasirox (DFS) and Deferoxamine mesylate (DFOM)), then exposed to ferroptosis inducers. Cell death was analyzed in each group via PI staining (H, Scale bar = 10 μm). The levels of lipid peroxidation across various groups were accessed via MDA measurement (I) and BODIPY staining (J-K). Finally, the level of free Fe²⁺ was determined using FerroOrange staining (L-M). Data are expressed as mean ± SD. The P value less than 0.05 was considered statistically significant and the value of Cohen's d over 0.8 represents a large effect size. *: P < 0.05 and Cohen's d > 0.8 compared to Ctrl group or compared between different groups.

Figure 4

The therapeutic effectiveness of the combination of Erastin and VPA in vivo against HCC. Wild type HCC cells were injected into SCID mice, and the mice were further treated with Erastin (Era), VPA and DFS. Herein, tumor volume (A) and tumor weight (B, Scale bar = 1 cm) were measured respectively. In addition, the levels of free iron (C), MDA (D) and expression of PTGS2 (E) were determined in tumor tissues from each group, and the transcription of FTL, FTH1 and NQO1 in tumor tissues were evaluated using RT-qPCR in our work (F-H). Data are expressed as mean ± SD. The P value less than 0.05 was considered statistically significant and the value of Cohen's d over 0.8 represents a large effect size. *: P < 0.05 and Cohen's d > 0.8 compared between different groups.

Figure 5

VPA modulates multiple metabolites related with programmed cell death. In this study, the HCC cell line MHCC97-H underwent treatment with VPA (2 mM) for a duration of 24 h. Subsequently, the cell samples were collected for metabolomics analysis. The primary metabolic alterations were assessed using both Positive Mode and Negative Mode (A). The analysis of secondary metabolites highlighted both upregulated and downregulated metabolites, as illustrated in Figure 5B. Furthermore, the relationship between mass-to-charge ratio and P value was examined in Figure 5C, while the interactions among differential metabolites and key metabolic pathways were depicted in a network plot (D). Lastly, Figure 5E presented the significant metabolic pathways in HCC cells that were affected by VPA.

Figure 6

VPA suppresses the sensitivity of HCC cells to disulfidptosis. Herein, MHCC97-H, Hep3B, HepG2 and BEL-7404 cells were initially subjected to treatment with VPA (2 mM) for a duration of 24 h, followed by glucose deprivation for 8 to 12 h to induce disulfidptosis. The subsequent changes in cell morphology and cell death were assessed across each experimental group (A, Scale bar = 20 μm). A heatmap was generated to represent the expression of genes associated with glucose and energy metabolism as derived from RNA-seq data (B). Furthermore, the formation of disulfide bonds in cytoskeletal proteins, specifically FLNA and DREBRIN, induced by glucose starvation was analyzed through non-reducing western blotting (C). The expression level of Group III was considered as “1”. Additionally, phalloidin staining was utilized to examine the presence of actin filaments (F-actin) in the various groups (D, Scale bar = 10 μm). The transcription levels of several critical enzymes involved in glycolysis and the pentose phosphate pathway (PPP) were further assessed using qPCR (E-F). Data are expressed as mean ± SD. The P value less than 0.05 was considered statistically significant and the value of Cohen's d over 0.8 represents a large effect size. *: P < 0.05 and Cohen's d > 0.8 compared to Ctrl group or compared between different groups.

Figure 7

VPA treatment promotes the activation of G6PD transcription in HCC cells. Both MHCC97-H and HepG2 cells were subjected to VPA treatment for a duration of 24 h, then harvested for qPCR and western blot assay. The mRNA and protein levels of G6PD were measured in our work (A-B). In addition, HCC cells were treated with SFN (1 and 2 μM) for 16 h, and the protein level of NRF2 was determined in each group (C). The transcription of G6PD and another NRF2 target gene, NQO1, was accessed using qPCR (D). Moreover, the function of potential AREs in human G6PD promotor region was analyzed via luciferase assay. MHCC97-H and HepG2 cells transfected with ARE-firefly luciferase or TK-renilla luciferase vectors for 24 h were left untreated or treated with SFN (2 μM) for 16 h. The cells were harvested for dual luciferase assay (E). Finally, The influence of VPA treatment on NRF2 expression was further determined using immunoblot (F), and the endogenous interaction between NRF2 and ARE1 in human G6PD gene was validated through ChIP-qPCR (G). Data are expressed as mean ± SD. The P value less than 0.05 was considered statistically significant and the value of Cohen's d over 0.8 represents a large effect size. *: P < 0.05 and Cohen's d > 0.8 compared to Ctrl group or compared between different groups.

Figure 8

G6PD plays a pivotal role in the action of VPA on disulfidptosis. G6PD knockdown cells were generated through siRNA transfection, and the effectiveness of the knockdown model was assessed via western blot analysis. HCC cells (MHCC97-H) were exposed to varying concentrations of VPA for 24 h, and the protein levels of G6PD were measured to verify the knockdown effect (A), and si-2# was chosen for following assays. Furthermore, both wild type and G6PD knockdown HCC cells underwent treatment with VPA for 24 h, followed by glucose deprivation (10-12 h) to trigger disulfidptosis. The morphological changes and cell death were subsequently analyzed in each group (B, Scale bar = 20 μm). Additionally, the formation of disulfide bonds in cytoskeletal proteins (FLNA and DREBRIN) induced by glucose starvation was assessed using non-reducing western blot (C). The expression level of Group III was considered as “1”. Phalloidin staining was then utilized to examine the actin filament (F-actin) in the various groups (D, Scale bar = 10 μm). Given the significance of G6PD in the pentose phosphate pathway and NADPH production (E), the activity of G6PD (F) and NADP⁺-NADPH metabolism (G) were measured herein. Moreover, the activity of GR (H) and GSH-GSSH metabolism (I) were also investigated in this study. Data are expressed as mean ± SD. The P value less than 0.05 was considered statistically significant and the value of Cohen's d over 0.8 represents a large effect size. *: P < 0.05 and Cohen's d > 0.8 compared to Ctrl group or compared between different groups.