Metronomic Capecitabine Triggers Ferroptosis in Hepatocellular Carcinoma Cells Through Inhibition of TYMS

ABSTRACT

Objective

Capecitabine (CAP) is an orally administered prodrug of fluorouracil, predominantly utilized in the treatment of solid tumors. Triggering tumor ferroptosis is an important mechanism for the treatment of hepatocellular carcinoma (HCC). However, the potential of CAP to induce ferroptosis in HCC, along with the underlying mechanisms, remains unknown.

Methods

In this study, a subcutaneous HCC model was constructed using the syngeneic Hepa1‐6 cell line in C57BL/6 mice, followed by treatment with metronomic CAP (mCAP). The anti‐tumor effects of mCAP were evaluated by monitoring tumor volume, performing pathological staining, and evaluating tumor oxidative stress levels. In vitro, various thymidylate synthase (TYMS) inhibitors were used to treat both mouse and human HCC cell lines. Furthermore, TYMS‐overexpressing plasmids were transfected into mouse and human HCC cell lines to directly investigate their impact on intracellular oxidative stress. Intracellular oxidative stress and ferroptosis‐related markers were detected using flow cytometry, transmission electron microscopy, and Western blotting.

Results

5‐FU, an active metabolite of CAP, as well as TYMS‐specific inhibitors, suppressed the proliferation of Hepa1‐6 and HepG2 cells. These treatments promoted p67^phox expression, activated nicotinamide adenine dinucleotide phosphate hydrogen oxidase (NOX), induced reactive oxygen species production, and increased ferrous ion accumulation. Electron microscopy revealed mitochondrial alterations characteristic of ferroptosis. Raltitrexed, another TYMS inhibitor, also induced ferroptosis in hepatocellular carcinoma cells. In vivo, the anti‐tumor effect of mCAP was antagonized by co‐treatment with an NOX inhibitor.

Conclusion

CAP targets TYMS to induce ferroptosis by activating NOX, thereby inhibiting HCC progression.

In the syngeneic tumor mouse model, capecitabine inhibits TYMS by metabolizing to 5‐fluorouracil in vivo, which leads to the activation of intracellular NOX. The massive release of ROS leads to the accumulation of ferrous ions in mitochondria, and the imbalance of intracellular oxidative stress and the accumulation of ferrous ions cause ferroptosis of liver cancer cells.

Abbreviations

5‐FU

5‐fluorouracil

CAP

Capecitabine

dTMP

Deoxythymidine monophosphate

dUMP

Deoxy‐uridine monophosphate

GPX4

Glutathione Peroxidase 4

GSH

Glutathione

HCC

Hepatocellular carcinoma

MDA

Malondialdehyde

mCAP

Metronomic capecitabine

NOX

Nicotinamide adenine dinucleotide phosphate hydrogen oxidase

RAL

Raltitrexed

ROS

Reactive oxygen species

TYMS

Thymidylate synthase

1. Introduction

Liver cancer is a highly aggressive malignancy and ranks as the third leading cause of cancer‐related death worldwide [1]. Hepatocellular carcinoma (HCC) accounts for 75%–85% of primary liver cancer cases, and the 5‐year survival rate for focal HCC is 32.6%, whereas it drops to only 10.8% and 2.4% for regional and metastatic HCC, respectively [2]. Treatment options for HCC include surgical resection, radiofrequency ablation, immunotherapy, radiotherapy, and chemotherapy; however, tumor recurrence remains a significant challenge. Therefore, investigating the mechanisms underlying drug treatment for HCC is of paramount importance.

Metronomic chemotherapy, characterized by the continuous administration of low‐dose agents, exerts both anti‐tumor and anti‐angiogenic effects. This strategy affords enhanced therapeutic efficacy and fewer side effects compared with conventional dosing [3]. Capecitabine (CAP) is a classical anti‐metabolite drug, and relevant clinical studies have confirmed that metronomic CAP (mCAP) effectively inhibits the progression of HCC [4, 5, 6]. Thymidylate Synthase (TYMS) catalyzes the conversion of deoxy‐uridine monophosphate (dUMP) to deoxythymidine monophosphate (dTMP) [7]. CAP selectively inhibits TYMS, thereby blocking the conversion of dUMP to dTMP [8]. This inhibition suppresses tumor cell proliferation, thereby mediating the observed anti‐tumor effect.

Ferroptosis is a form of programmed cell death characterized by the accumulation of intracellular iron ions and lipid peroxidation. The accumulation of intracellular reactive oxygen species (ROS) is a key factor in inducing ferroptosis [9]. Triggering ferroptosis has emerged as a strategy to inhibit HCC progression [10, 11]. Several clinically relevant ferroptosis inducers, such as sorafenib, sulfasalazine, statins, and artemisinin, have been used to induce ferroptosis in HCC and further inhibit tumor growth [12]. Previous studies have shown that CAP modulates CD4⁺T cell ferroptosis [13, 14]. However, it remains unclear whether CAP can induce ferroptosis in HCC, and the relationship between TYMS and ferroptosis is poorly understood. Nicotinamide adenine dinucleotide phosphate hydrogen oxidase (NOX) is a major source of intracellular ROS. During tumor development, alterations in cellular metabolism and the intracellular microenvironment activate NOX, leading to elevated ROS levels and cellular oxidative stress [15]. In the liver, elevated NOX expression correlates with ROS production and drives the progression of HCC [16]. The NOX complex comprises multiple subunits, among which the upregulation of p67^phox directly correlates with increased NOX activity [17]. Inhibiting TYMS is a proven anti‐tumor strategy that blocks DNA synthesis and induces cell death [18]. TYMS inhibitors have been shown to induce oxidative stress and subsequent cell death in colon cancer cells [19]. This effect is mediated by NOX‐dependent elevation of intracellular ROS.

This study demonstrates the anti‐tumor efficacy of mCAP in a syngeneic tumor mouse model of HCC. Additionally, we verify that TYMS inhibition triggers NOX‐induced intracellular ROS accumulation and ferroptosis in hepatocellular carcinoma cells in vitro. These findings provide a solid theoretical basis for its application in HCC treatment.

2. Methods

2.1. Cell Culture

C57BL/6 mice‐derived Hepa 1–6 cell line, and HepG2 human HCC cell line were purchased from Procell Life Science & Technology (Wuhan, China). Hepa1‐6 cells were cultured in DMEM complete medium containing sodium pyruvate (Procell, Wuhan, China), while HepG2 cells were cultured in MEM complete medium containing NEAA (Procell, Wuhan, China), both supplemented with 10% fetal bovine serum and 1% penicillin–streptomycin solution (Procell, Wuhan, China). The cells were cultured at 37°C with 5% carbon dioxide. All cell lines were authenticated by short tandem repeat (STR) profiling and verified as mycoplasma‐free before experimentation.

2.2. Establishment and Treatment of Syngeneic Tumor Mouse Model

Male C57BL/6 mice aging 6–8 weeks and weighing 20–23 g used in this study were purchased from Beijing Huafukang Biological Science. The animal experiments were conducted in SPF‐grade animal facilities. The animals were provided with ad libitum access to clean food and water and were maintained on a 12‐h light/dark cycle. The subcutaneous area of C57BL/6 mice was injected with 2 × 10⁶ Hepa1‐6 cells resuspended in 100 μL PBS [20]. Tumor volume was monitored by measuring its length (L) and width (W) with a vernier caliper. The tumor volume was calculated using the formula (L × W²) × 0.5. After the tumor volume reached 80–100 mm³, the animals were divided randomly into three groups (n = 6 mice per group): The untreated group, which only received tumor volume measurement without any special treatments; the mCAP group, which received 100 mg/kg/d of CAP (MCE, Shanghai, China) by gavage; and the mCAP+VAS group, which received the mCAP combined with VAS2870 group was administered 100 mg/kg/d CAP by gavage and VAS2870 (MCE, Shanghai, China) 10 mg/kg/d by intraperitoneal injection. Tumor volume was measured at 3‐day intervals. Mice were euthanized upon reaching a tumor volume of 1,000 mm³. Subcutaneous tumor images were collected, and tumor samples were obtained for subsequent experiments. In this study, we used the previously published mouse dose of 100 mg/kg/d as the metronomic dose [21]. The animal experiments followed the ARRIVE guidelines and were approved by the Ethics Committee of the Institute of Radiation Medicine, Chinese Academy of Medical Sciences (Approval NO. IRM/2‐IACACC‐2312‐001).

2.3. Hematoxylin–Eosin Staining

The tumor tissues were fixed in 10% paraformaldehyde for 48 h, dehydrated, and embedded in paraffin. Subsequently, 4 μm sections of the tumor tissues were cut for hematoxylin–eosin (H&E) staining as described (Solarbio, Beijing, China).

2.4. Immunohistochemical Staining

Immunohistochemistry was utilized to detect Glutathione Peroxidase 4 (GPX4) expression in tumor tissues [22]. After dewaxing and rehydrating tissue sections and performing thermal repair with sodium citrate antigen retrieval solution (Elabscience, Hubei, China), tumor sections were incubated with the primary anti‐GPX4 (ab125066; Abcam, Cambridge, MA, USA) at 4°C overnight. The enhanced HRP‐conjugated goat anti‐mouse/rabbit IgG polymer was used at room temperature for 20 min, followed by color development with diaminobenzidine (DAB) chromogen and counterstaining with hematoxylin. The DP70 digital camera was used to acquire digital images with an Olympus microscope (BX‐61, New York, USA). Ten high‐power fields (×200) were randomly selected, and the positive areas were histologically analyzed using Image J software.

2.5. Detection of Oxidative Stress Levels

The Glutathione (GSH) and malondialdehyde (MDA) contents in tumor tissues were measured using GSH and MDA kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, China) in accordance with the manufacturer's instructions. Cells were harvested 48 h after drug treatment, and intracellular GSH and MDA contents were measured using GSH and MDA kits (Nanjing Jiancheng Bioengineering Institute, Nanjing, China) in accordance with the manufacturer's instructions.

2.6. Determination of Ferrous Ion Content in Tissues

According to the instructions, the ferrous ion content in tumor tissue was determined using the Ferrous Iron Colorimetric Assay Kit (Elabscience, Hubei, China).

2.7. Cell Viability Assay

Cell Counting Kit‐8 (Glpbio, Montclair, USA) was used to detect hepatocellular carcinoma cells' viability. The cells in logarithmic phase were seeded a density of 6000 cells per well in 96‐well plates and treated with different drugs (5‐fluorouracil (5‐FU), MCE, Shanghai, China; Raltitrexed (RAL), MCE, Shanghai, China) and inhibitors (Diphenyleneiodonium chloride (DPI), MCE, Shanghai, China; VAS2870, MCE, Shanghai, China) after 48 h, the medium was discarded and the complete medium containing 10% CCK‐8 solution was added to each well. The plate was then incubated at 37°C for 1 h, and the OD value at 450 nm was recorded to calculate cell viability.

2.8. Flow Cytometry

After 48 h of treatment with 5‐FU and RAL, cells were collected and loaded with the DCFH‐DA probe. The cells were then incubated in a cell incubator at 37°C for 20 min, with the probes being reversed every 5 min to ensure full loading. Following this, the cells were washed three times with serum‐free cell medium, and intracellular ROS levels were measured using flow cytometry. To detect the ferrous ion level in tumor cells, the FerroOrange probe was utilized. The cells were loaded with the probe and subsequently incubated in a cell incubator at 37°C for 20 min. After incubation, the cells were detected using flow cytometry.

2.9. Detection of NOX

Cells were harvested 48 h after drug treatment, and intracellular NOX levels were measured using the NADP+/NADPH colorimetric assay kit (Elabscience, Wuhan, China) according to the manufacturer's instructions. The NOX levels in tumor tissues were measured using the NADP+/NADPH colorimetric assay kit (Elabscience, Wuhan, China) according to the manufacturer's instructions.

2.10. Transmission Electron Microscopy

After 48 h of IC₅₀ drug treatment, the medium was discarded, and precooled 2.5% glutaraldehyde was fixed at 4°C for 1 h. The cells were then scraped and transferred to a centrifuge tube, where they were collected after centrifugation. The samples were fixed with osmium tetroxide and uranyl acetate, dehydrated with a graded series of ethanol and acetone, and embedded in epoxy resin before being cut into ultrathin sections. Transmission electron microscopy (HT7800; Hitachi, Tokyo, Japan) was used to observe organelle morphology.

2.11. Plasmids and Cell Transfection

Empty vector, pcDNA3.1‐TYMS, and pcDNA3.1‐Tyms were ordered from GenePharma (Shanghai, China) for overexpression assays. HepG2 and Hepa1‐6 cells were transiently transfected with Lipofectamine 3000 following the manufacturer's instructions.

2.12. Statistical Analysis

Statistical analysis was performed with SPSS 26.0 and GraphPad 9.4.1. Data were expressed as mean ± standard deviation (SD). A t‐test was used for comparisons between two groups, and one‐way analysis of variance (ANOVA) was used for comparisons between three or more groups. A p‐value of less than 0.05 was considered statistically significant for the test results.

3. Results

3.1. mCAP Inhibits Tumor Growth in Syngeneic Tumor Mouse Model

Given the clinical applicability and favorable safety profile of mCAP in HCC patients, we established a syngeneic HCC‐bearing mouse model to determine the mCAP‐mediated effects and underlying mechanisms (Figure 1A). As shown in Figure 1B,C, the tumor sizes in the mCAP‐treated group were remarkably smaller than those in the vehicle‐treated group (P < 0.0001), indicating that mCAP significantly suppresses HCC progression in vivo. In addition, histological H&E staining revealed that mCAP treatment led to substantial tissue necrosis, disrupted tumor architecture, and augmented cell death compared with the vehicle‐treated group (Figure 1D). The ferrous ion content in tumor tissues was measured using a colorimetric assay. The results showed that the ferrous ion content in the tumor tissues of the mCAP group was higher than that in the untreated group (p < 0.001) (Figure 1E). Collectively, these results indicate that mCAP treatment effectively inhibits HCC progression in vivo, a process that involves ferroptosis.

FIGURE 1.

Metronomic capecitabine inhibits syngeneic tumor growth in mice. (A) The subcutaneous administration of the mouse HCC cell line Hepa1‐6 was performed on C57BL/6 mice. (B) When the syngeneic tumor size reached 80–100 mm³, the mice were randomly divided into 2 groups (n = 5 per group). The metronomic capecitabine group was gavaged with capecitabine (100 mg/kg/d) the next day, and the control group was left untreated. The tumor growth curve was monitored. (C) The treatment will end when the tumor size reaches 1,000 mm³, mice were sacrificed, and tumors were excised to measure tumor volume. (D) Hematoxylin and eosin staining of syngeneic tumor sections. (E) The level of ferrous ions in tumor tissue has been detected. Untreated group: Without any additional specific treatment; mCAP group: Metronomic capecitabine group, mice gavaged with 100 mg/kg. ns, p > 0.05, ***p < 0.001, ****p < 0.0001.

3.2. 5‐FU Induces Ferroptosis in HCC Cell Lines in Vitro

As 5‐FU is the ultimate active metabolite of CAP in vivo, we used 5‐FU to treat HCC cell lines in vitro. Thus, we investigated the impact of 5‐FU on cell viability, ROS, and Fe²⁺ levels in Hepa1‐6 and HepG2 cells. Firstly, the IC₅₀ values of 5‐FU were determined in the mouse Hepa1‐6 HCC cell line and the human HepG2 cell line. The IC₅₀ values for Hepa1‐6 and HepG2 cells were 5.53 μM and 11.22 μM, respectively (Figure 2A). ROS and iron ions are both indicators and inducers of ferroptosis. After 48 h of treatment, flow cytometry analysis observed increased levels of ROS and Fe²⁺ in both cell lines (Figure 2B,C). Transmission electron microscopy revealed that after 5‐FU treatment, mitochondria decreased in size, membrane density increased, and cristae decreased or disappeared (Figure 2D). Consistent with these findings, 5‐FU treatment significantly altered key ferroptosis biomarkers. In Hepa1‐6 cells, cellular GSH level significantly decreased from 4.583 ± 0.1401 (mean ± SD) to 3.49 ± 0.1323 compared with the control (p < 0.001) (Figure 2E). In HepG2 cells, GSH levels also decreased significantly with 5‐FU treatment compared with the control (p < 0.001) (Figure 2E). MDA, a characteristic product of lipid peroxidation, is tightly linked with ferroptosis. Accordingly, 5‐FU treatment markedly increased the intracellular MDA level in HCC (Figure 2E). Furthermore, the protein level of ACSL4 was significantly upregulated by 5‐FU treatment (Figure 2F). Collectively, these results confirm that 5‐FU induces ferroptosis in HCC cells.

FIGURE 2.

The induction of intracellular ROS production through NOX activation by 5‐fluorouracil ultimately leads to ferroptosis in hepatocellular carcinoma cells. (A) Hepa1‐6 and HepG2 cells were cultured with varying concentrations of 5‐fluorouracil for 48 h. Cell viability was measured using the CCK8 method. (B) ROS levels were measured using DCFH‐DA after treating Hepa1‐6 and HepG2 cells with IC₅₀ doses of 5‐fluorouracil for 48 h. (C) Intracellular Fe²⁺ production was measured after treating Hepa1‐6 and HepG2 cells with an IC₅₀ dose of 5‐fluorouracil for 48 h. (D) Transmission electron microscopy images of mitochondria from Hepa1‐6 and HepG2 cells after 5‐fluorouracil treatment revealed disrupted mitochondrial membranes, as evidenced by increased membrane density, reduced or absent cristae (yellow arrows), and altered morphology (Scale bar, 5.0 um and 2.5 um). (E) GSH and MDA levels were measured after treating Hepa1‐6 and HepG2 cells with an IC50 dose of 5‐fluorouracil for 48 h. (F) ACSL4 protein expression in Hepa1‐6 and HepG2 cells after treatment with 5‐FU or RAL at their respective IC₅₀ for 48 h was determined by Western blotting. CON group: Without any additional specific treatment; 5‐FU group: Hepatocellular carcinoma cells were treated with 5‐fluorouracil (Hepa1‐6: 5 μM; HepG2: 10 μM). ns, p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001.

3.3. TYMS‐Specific Inhibitor RAL Induces Excessive Oxidative Stress and Triggers Ferroptosis in HCC

To investigate whether TYMS inhibition triggers ferroptosis in vitro, we used the TYMS‐specific inhibitor Raltitrexed (RAL), considering that TYMS is the primary target of CAP and 5‐FU in vivo. The IC₅₀ values of RAL were determined using the CCK‐8 assay after 48 h of treatment. The IC₅₀ values for Hepa1‐6 and HepG2 cells were calculated to be 10.42 μM and 10.91 μM, respectively (Figure 3A). Subsequently, we evaluated the levels of ROS and Fe²⁺ in hepatocellular carcinoma cells treated with their respective IC₅₀ dose of RAL. Our findings revealed that RAL treatment led to elevated levels of ROS and Fe²⁺ in both Hepa1‐6 and HepG2 cells (Figure 3B,C), suggesting that TYMS inhibitors impede tumor cell growth by modulating ferroptosis. Moreover, the distinctive features of ferroptosis were observed in hepatoma cells treated with RAL by transmission electron microscopy (Figure 3D). Specifically, RAL‐treated cells displayed reduced mitochondrial shrinkage, increased mitochondrial membrane density, and reduced or absent cristae compared with the control group. Furthermore, RAL significantly increased intracellular MDA levels (Figure 3E) and ACSL4 protein expression (Figure 3F) in HCC cells. Similarly, RAL treatment reduced the GSH level in Hepa1‐6 and HepG2 cells (Figure 3E). These results further demonstrated that the TYMS‐specific inhibitor RAL induces ferroptosis in HCC.

FIGURE 3.

The TYMS inhibitor induces intracellular production of reactive oxygen species and triggers ferroptosis in hepatocellular carcinoma cells. (A) Hepa1‐6 and HepG2 cells were cultured with varying concentrations of raltitrexed for 48 h. Cell viability was measured using the CCK8 method. (B) ROS levels were measured using DCFH‐DA after treating Hepa1‐6 and HepG2 with IC₅₀ doses of raltitrexed for 48 h. (C) Intracellular Fe²⁺ production was measured after treating Hepa1‐6 and HepG2 with an IC₅₀ dose of raltitrexed for 48 h. (D) Transmission electron microscopy images of mitochondria from Hepa1‐6 and HepG2 cells after raltitrexed treatment revealed disrupted mitochondrial membranes, as evidenced by increased membrane density, reduced or absent cristae (yellow arrows), and altered morphology (Scale bar, 5.0 um and 2.5 um). (E) GSH and MDA levels were measured after treating Hepa1‐6 and HepG2 cells with an IC50 dose of raltitrexed for 48 h. CON group: Without any additional specific treatment; RAL group: Hepatocellular carcinoma cells were treated with raltitrexed (Hepa1‐6: 10 μM; HepG2: 10 μM). ns, p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001.

3.4. Inhibition of TYMS Activates NOX to Trigger Ferroptosis in HCC

Previous studies have demonstrated that NOX activation induces excessive oxidative stress in tumors. However, the involvement of TYMS in HCC ferroptosis via NOX activation remains uncertain. Additionally, we initially observed that 5‐FU treatment upregulated NOX activity (Figure 4A), implying a potential role for NOX in 5‐FU‐induced ferroptosis in liver tumor cells. To determine the role of NOX in 5‐FU and RAL‐mediated effects in HCCs, we first determined NOX activity in HCC cells treated with TYMS inhibitors. The results showed that both 5‐FU and RAL significantly increased NOX activity in mouse and human HCC cell lines (Figure 4B), indicating that TYMS inhibition upregulates NOX activity. To explore the role of NOX in TYMS inhibitor‐induced ferroptosis, we combined NOX inhibitors with 5‐FU or RAL. The NOX inhibitors DPI and VAS2870 were used in combination with TYMS inhibitors to assess cell viability. The results showed that VAS2870 in combination with 5‐FU increased cell viability compared to 5‐FU alone, while DPI decreased cell viability compared to 5‐FU alone due to its cytotoxicity, but increased cell viability compared to DPI alone. Similarly, when cells were treated with RAL, the results were consistent (Figure 4C). Furthermore, the TYMS inhibitor‐induced accumulation of intracellular ROS and Fe²⁺ was attenuated by NOX inhibition (Figure 4D,E). Collectively, these findings indicate that TYMS inhibitors induce ferroptosis in HCC cells by activating NOX.

FIGURE 4.

Role of NOX in TYMS inhibitor‐induced ferroptosis in hepatocellular carcinoma cells. (A) Intracellular NOX activity was measured after treating Hepa1‐6 and HepG2 cells with an IC₅₀ dose of 5‐fluorouracil for 48 h. (B) The NOX activity of cells treated with various TYMS inhibitors and NOX inhibitors for 48 h was measured. (C) The CCK‐8 assay detected changes in cell viability from treatment with TYMS inhibitors of 5‐fluorouracil and raltitrexed (Hepa1‐6: 5 μM 5‐fluorouracil or 10 μM raltitrexed; HepG2: 10 μM 5‐fluorouracil or 10 μM raltitrexed), alone or with investigation into the effects of different NOX inhibitors (20 μM VAS2870 or 10 μM Diphenyleneiodonium chloride) after 48 h. (D) The levels of ROS were measured using DCFH‐DA through flow cytometric analysis after treatment. (E) The levels of ferrous iron were detected using the FerroOrange probe through flow cytometric analysis after treatment. ns, p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

3.5. Overexpression of TYMS Inhibits NOX to Suppress Ferroptosis in HCC

In the early stage of this study, we confirmed that TYMS inhibition by 5‐FU or RAL activates NOX, leading to ROS production and ferroptosis. We then tried to find direct evidence linking TYMS inhibition to NOX activation in HCC. The p67^phox subunit is a critical regulator of the NOX complex, and the upregulation of p67^phox directly enhances NOX enzymatic activity [17]. To further investigate whether TYMS inhibition regulates NOX activation via p67phox, we treated Hepa1‐6 and HepG2 cells with the TYMS‐specific inhibitors 5‐FU and RAL and assessed the effects on p67^phox expression. The results revealed that 5‐FU and RAL directly increased the expression level of p67^phox in HCC cells (Figure 5A). To further verify the functional relationship between TYMS and NOX activation in HCC ferroptosis, we constructed TYMS overexpressing plasmids, and the level of TYMS protein was markedly increased compared with the vector control group (Figure 5B). We examined the p67^phox protein levels of Hepa1‐6 and HepG2 cells and found that the p67^phox levels of HCC cells were significantly reduced after overexpression of TYMS (Figure 5C). Additionally, we observed a marked suppression of NOX activity following TYMS overexpression in Hepa1‐6 and HepG2 cells (Figure 5D). These findings demonstrated that TYMS overexpression directly suppresses NOX activation. Further, we also assessed the effect of TYMS overexpression on intracellular oxidative stress in HCC. The levels of GSH were significantly increased, while the levels of MDA were significantly reduced in HCC cells following overexpression of TYMS (Figure 5D). We further examined the protein expression of ACSL4, a key marker of ferroptosis, and found notable suppression following TYMS overexpression (Figure 5E). These results demonstrate that NOX‐mediated ferroptosis in HCC cells was significantly inhibited after TYMS overexpression. In summary, our findings indicate that TYMS modulates HCC ferroptosis by directly regulating NOX activity.

FIGURE 5.

Overexpression of TYMS inhibited NOX to suppress ferroptosis in HCC. (A) p67^phox protein expression in Hepa1‐6 and HepG2 cells after treatment with 5‐FU or RAL at their respective IC₅₀ for 48 h was determined by Western blotting. (B) TYMS protein expression in Hepa1‐6 and HepG2 cells after transfection with TYMS overexpressing plasmids was determined by Western blotting. (C) p67^phox protein expression in Hepa1‐6 and HepG2 cells after transfection with TYMS overexpressing plasmids was determined by Western blotting. (D) NOX activity, GSH, and MDA levels in Hepa1‐6 and HepG2 cells were measured after transfection with TYMS overexpressing plasmids. (E) ACSL4 protein expression in Hepa1‐6 and HepG2 cells after transfection with TYMS overexpressing plasmids was determined by Western blotting. CON group: Without any additional specific treatment; 5‐FU group: Hepatocellular carcinoma cells were treated with 5‐fluorouracil (Hepa1‐6: 5 μM; HepG2: 10 μM); RAL group: Hepatocellular carcinoma cells were treated with raltitrexed (Hepa1‐6: 10 μM; HepG2: 10 μM); Vector group: Hepa1‐6 and HepG2 cells after transfection with vector; pcDNA‐TYMS group: HepG2 cells after transfection with TYMS overexpressing plasmid; pcDNA‐TYMS group: Hepa1‐6 cells after transfection with TYMS overexpressing plasmid. ns, p > 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

3.6. NOX Inhibitor Antagonizes the Anti‐Tumor Effect of mCAP

To investigate whether CAP inhibits liver tumor growth through NOX activation and subsequent ferroptosis, we administered mCAP (100 mg/kg/d) orally alone or in combination with the specific NOX inhibitor VAS2870 (10 mg/kg/d) via intraperitoneal injection. The impact on tumor progression was assessed in vivo using a syngeneic mouse model. The results demonstrated that the anti‐cancer efficacy of mCAP was significantly counteracted by co‐treatment with the NOX inhibitor VAS2870 (Figure 6A). Compared with the untreated and combined treatment groups, mCAP significantly delayed tumor growth and reduced tumor nodule size (Figure 6B). H&E staining indicated that mCAP treatment caused extensive tumor cell death, loss of cell morphology, and nuclear rupture (Figure 6C). When combined with a NOX inhibitor, the anti‐tumor effect was reversed. The expression levels of GSH and MDA in tumor tissues confirmed that mCAP activates NOX in vivo to induce ferroptosis (Figure 6D). In this study, we supplemented the detection of NOX activity within the tumors from in vivo samples, which showed that mCAP could increase NOX activity in the tumors, while VAS2870 blocked the effects (Figure 6D). Additionally, immunohistochemical staining for GPX4 also demonstrated that mCAP induces ferroptosis of tumor cells in vivo. Ferroptosis was inhibited after combined treatment with an NOX inhibitor (Figure 6E). In summary, the results observed in the syngeneic tumor mouse model are consistent with the in vitro findings, suggesting that mCAP induces ferroptosis in HCC by activating NOX to inhibit tumor growth. Conversely, the effect of mCAP is antagonized by NOX inhibitors.

FIGURE 6.

NOX inhibitor antagonized mCAP‐induced ferroptosis and inhibited anti‐tumor effect in the syngeneic tumor mouse model. (A) The tumor growth curve was monitored. (B) Representative images of a mouse syngeneic model and statistical analysis of tumor volume after subcutaneous injection are presented. (C) Hematoxylin and eosin staining of tumor sections. (D) NOX activity, GSH, and MDA levels in the tumor. (E) Immunohistochemical staining for glutathione peroxidase 4 and corresponding statistics of positive areas in tumor tissue sections; mCAP+VAS group: Metronomic capecitabine and NOX inhibitor VAS2870 group, mice gavaged with capecitabine 100 mg/kg and intraperitoneal injection of VAS2870 10 mg/kg. mCAP group: Metronomic capecitabine group (mice gavaged with 100 mg/kg). Untreated group: Without any additional specific treatment. ns, p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

4. Discussion

In our study, we found that mCAP effectively inhibits tumor growth in mice, as evidenced by observations of tumor size and pathological staining in a syngeneic tumor mouse model (Figure 1B–D). Additionally, 5‐FU activates NOX by inhibiting TYMS, thereby inducing excessive ROS production and triggering ferroptosis in tumor cells. During the progression of HCC, cancer cells evade the harmful effects of excessive iron accumulation through their antioxidant defense mechanisms. Recent studies have shown a strong correlation between ferroptosis and HCC [2, 23, 24], indicating that HCC cells resistant to traditional treatment or with a high metastatic tendency may be more vulnerable to ferroptosis. Ferroptosis is a form of regulated cell death; targeted regulation of ferroptosis may offer a novel treatment strategy for refractory and resistant tumors in traditional drug therapy [10]. First, a variety of cancer‐related signaling pathways, such as Ras, TP53, and NFE2L2, modulate iron metabolism and thereby impact tumor cell death [12]. Second, tumor cells are more susceptible to ferroptosis due to their unique metabolism and high intracellular ROS load [25], providing a therapeutic window for ferroptosis‐targeted therapy. Additionally, HCC has shown potential sensitivity to compounds capable of inducing ferroptosis [26]. Recent studies in HCC have positioned ferroptosis as a therapeutically actionable vulnerability closely associated with tumor progression, treatment resistance, and immunotherapy response [27]. In this context, inducing ferroptosis is increasingly regarded not simply as a nonspecific oxidative death process, but as a mechanism‐driven approach with direct translational relevance for HCC. Therefore, inducing ferroptosis in tumor cells represents a promising strategy to improve therapeutic efficacy and potentially augment anti‐tumor immunity.

TYMS is highly expressed in malignant tumors, reflecting elevated DNA biosynthesis relative to non‐tumor tissues [28]. Overexpression of TYMS further promotes tumor growth, invasion, and metastasis [29]. Moreover, TYMS critically mediates drug resistance in cancer cells [30]. Therefore, inhibiting TYMS induces cancer cell death and suppresses tumor growth, making it a common target for anti‐tumor drugs. TYMS catalyzes the biochemical conversion of dUMP to dTMP, which is essential for the de novo synthesis of the DNA precursor thymine [31]. Additionally, studies have shown that biological complexes involved in the de novo synthesis of pyrimidine regulate tumor cell proliferation and ferroptosis [32]. CAP, an oral prodrug of 5‐FU, is metabolized to dUMP, which binds to and inhibits TYMS, thereby disrupting dTMP production [33]. This study also utilized RAL, a quinazoline folate analogue that acts on the folate binding site of TYMS in place of the folate cofactor mTHF, as another specific inhibitor of TYMS. Blocking the process of folate activation specifically inhibits the catalysis of TYMS, leading to DNA fragmentation and cell death. In addition, independent HCC studies have revealed that TYMS downregulation alone is sufficient to suppress malignant phenotypes, further supporting TYMS as a biologically relevant vulnerability in HCC [34]. Importantly, recent HCC‐focused studies suggest that TYMS status may predict fluoropyrimidine response and prognosis, whereas substantial intertumoral variability in capecitabine‐metabolizing enzymes could influence CAP efficacy in HCC [35]. In this study, ferroptosis was observed in tumor cells treated with TYMS inhibitors. Specifically, treatment with the TYMS inhibitors 5‐FU and RAL led to a significant accumulation of ferrous ions in both Hepa1‐6 and HepG2 cells (Figures 2C and 3C). In HCC cells treated with RAL and 5‐FU, we observed the distinctive features of ferroptosis; specifically, mitochondria decreased in size, membrane density increased, and cristae decreased or disappeared (Figures 2D and 3D). This profile stands in sharp contrast to the relatively intact mitochondria in apoptosis and the characteristically swollen mitochondria in necrosis. Furthermore, both 5‐FU and RAL significantly modulated the expression levels of key ferroptosis markers, including MDA, GSH, and ACSL4 (Figures 2E,F and 3E). Taken together, the comprehensive evaluation of the above evidence confirms that both 5‐FU and RAL induce ferroptosis in HCC cells.

NOX is involved in the synthesis of intracellular ROS and plays a key role in the ferroptosis process. Furthermore, NOX activation promotes ferroptosis and reverses lenvatinib resistance in HCC cells [36]. A recent study has shown that flavoprotein ferroptosis suppressor protein (FSP1) inhibits cellular ferroptosis by oxidizing NOX [37]. In addition, NOX activity may serve as a biomarker for predicting cellular sensitivity to ferroptosis [38]. Recent studies in HCC have further established that oxidative stress and NOX‐related ferroptosis networks modulate drug sensitivity and resistance. In parallel, NOX‐dependent oxidative ferroptosis circuits have been implicated in reversing lenvatinib resistance in HCC, further underscoring the therapeutic relevance of redox‐regulated ferroptosis in this malignancy [36]. Consequently, NOX‐driven ferroptosis emerges as a promising therapeutic strategy for HCC. In this study, we found that TYMS inhibitors increased intracellular NOX activity in tumor cells (Figure 4B), confirming that TYMS inhibitors induce cell ferroptosis by activating NOX, and that CAP exerts anti‐tumor effects by targeting ferroptosis in tumor cells. Additionally, ROS production is a crucial factor in the process of ferroptosis, and there is a positive correlation between NOX activation and ROS production. This correlation was confirmed by detecting intracellular NOX activity and ROS expression levels, which revealed that enhanced NOX activity resulted in a significant increase in intracellular ROS production. The enhancement of NOX activity induced by TYMS inhibitor was significantly inhibited by co‐treatment with the NOX inhibitors DPI and VAS2870 (Figure 4D). DPI is a widely used NOX inhibitor in hepatocellular carcinoma research [39]. In this study, we found that DPI suppresses the characteristic features of ferroptosis induced by 5‐FU or RAL. However, DPI is now recognized by some investigators as a pan‐flavoprotein inhibitor that may also interfere with other ROS‐related or flavin‐dependent targets, including mitochondrial electron transport, nitric oxide synthases, xanthine oxidase, and calcium‐handling pathways [40]. Owing to its potential off‐target effects, we also employed the more selective NOX inhibitor VAS2870 to confirm the effects of NOX during TYMS‐mediated induction of ferroptosis and to make our results more reliable. VAS2870, a chemically distinct NOX inhibitor, provided orthogonal pharmacological support. The results of this study showed that VAS2870, as a second, chemically distinct NOX inhibitor, led to concordant attenuation of 5‐FU or RAL‐induced ROS accumulation, Fe²⁺ elevation, and ferroptotic phenotypes. The NOX inhibitor effectively reduced ROS release and intracellular Fe²⁺ levels (Figure 4E), demonstrating that inhibiting NOX reduces ROS production and further prevents ferroptosis. These results demonstrated that TYMS inhibition by 5‐FU or RAL activates NOX. Subsequently, we sought to identify the direct link between TYMS inhibition and NOX activation. The p67^phox regulatory subunit directly activates the NOX complex, thereby inducing ROS production [41]. To identify if TYMS inhibition could regulate p67^phox expression to activate NOX, we did use TYMS specific inhibitors (5‐FU and RAL) to treat Hepa1‐6 and HepG2 cell lines in vitro, and detected whether p67^phox expression, NOX activity, as well as NOX‐mediated ferroptosis are regulated. The results showed that both 5‐FU and RAL could increase the expression of p67^phox protein (Figure 5A) and NOX activity (Figure 4A). Furthermore, we also did TYMS overexpression in both HCC cell lines (Figure 5B), and detected if p67^phox expression and NOX activity are downregulated, which showed that when overexpressing TYMS in HCC cell lines, p67^phox expression (Figure 5C) and NOX activity (Figure 5D) are significantly decreased, while cell ferroptosis marker MDA level is significantly inhibited (Figure 5D), as well as ACSL4 (Figure 5E). Taken together, these results indicate that TYMS modulates HCC ferroptosis by directly regulating NOX activity. These findings are also consistent with recent HCC studies showing that ferroptosis‐regulatory redox networks influence sensitivity to systemic therapies, including sorafenib and lenvatinib. Accordingly, our data suggest that the TYMS–NOX–ROS axis may represent a previously underappreciated mechanism through which fluoropyrimidine‐based treatment exploits ferroptotic vulnerability in HCC cells [42].

Accumulating evidence indicates that oxidative stress contributes to HCC progression. Specifically, excessive ROS production disrupts the balance of oxidation–reduction reactions in cancer cells, leading to depletion of GSH and downregulation of GPX4 expression. This process exacerbates the excessive accumulation of lipid oxides, leading to lipid peroxidation, metabolic dysfunction, and increased susceptibility of tumor cells to ferroptosis [43]. GSH is utilized to eliminate intracellularly overproduced ROS to sustain the balance of intracellular oxidation–reduction reactions. Therefore, a decrease in GSH levels renders tumor cells more prone to oxidative stress. Simultaneously, NOX plays a crucial role in modulating ROS generation. In vitro, TYMS inhibitors trigger a series of reactions through NOX activation, leading to ferroptosis in tumor cells (Figure 4B–E). To further validate this, we co‐administered the specific NOX inhibitor VAS2870 and assessed the effect of NOX inhibition on ferroptosis in cancer cells. The results revealed that mCAP inhibited HCC tumor growth and increased NOX activity, while VAS2870 blocked the effects (Figure 6A,B,D). H&E and IHC results further showed a reduction in the necrotic area and GPX4 positive area in the mCAP combined with VAS2870 group compared to the mCAP alone (Figure 6C,E). The role of GPX4 in preventing ferroptosis is crucial, as its inactivation results in the accumulation of lipid peroxidation, thereby triggering ferroptosis [44]. The dysregulation of ROS levels leads to depletion of GSH, accumulation of MDA, and a decrease in GPX4 expression. Consistent with previous studies, our findings demonstrate that NOX plays an important role in promoting tumor ferroptosis (Figure 6D).

The tumor microenvironment (TME) is a complex structure with various functions that play a crucial role in tumor development. Within the TME, tumor cells interact with immune cells to regulate cancer progression. Upon external stimulation, tumor cells undergo a transformation from non‐immunogenic to immunogenic, resulting in cell death. This process, which mediates an anti‐tumor immune response, is referred to as immunogenic cell death (ICD) [45]. ICD has emerged as a novel approach in the field of immunotherapy, aiming to activate the immune system by exposing many antigens. Certain chemotherapeutic drugs have been discovered to induce ICD and enhance the immune system's ability to eliminate tumor cells, thereby facilitating targeted tumor destruction. Notably, 5‐FU has been reported to induce ICD [46]. In contrast, ferroptosis effectively generates immunogenicity and subsequently inhibits tumor growth. In future studies, we aim to investigate whether CAP promotes the advancement of anti‐tumor therapy by eliciting ICD. Emerging evidence in HCC indicates that ferroptosis can extend beyond direct tumor killing to remodel the immune microenvironment and enhance responses to immune checkpoint blockade [47]. However, ferroptosis‐induced inflammation may also exert context‐dependent protumor effects, suggesting that the timing and extent of ferroptosis induction require careful control in HCC [48]. Recent HCC studies have already shown that nanoplatforms can induce ICD and enhance immunotherapy efficacy [49]. Moreover, emerging classes of nano‐probes have been shown to significantly enhance NOX activity via the photothermal effect, inducing ROS generation and explosive ferroptosis with a synergistic inhibition of tumor growth in vitro and in vivo [50]. Integrating conventional drugs with nanomedicine represents a promising direction for HCC therapy. Adjuvant chemotherapy is a crucial component of the treatment regimen for HCC patients. Elucidating drug mechanisms is essential given chemotherapy‐induced immunosuppression and rising chemoresistance in HCC. Our previous results showed that mCAP induces T‐cell apoptosis and exerts an immunosuppressive effect in a mouse gavage model [21]. In addition, we further confirmed that mCAP suppresses T‐cell responses after liver transplantation in rats [13]. Meanwhile, we have previously demonstrated that CAP alleviates acute rejection in mouse heart allografts by targeting TYMS [51]. This study confirms that CAP exerts anti‐tumor effects by inducing ferroptosis in tumor cells. Recent studies in HCC indicate that metronomic capecitabine retains measurable antitumor activity in advanced HCC, supporting further biomarker‐driven investigation of capecitabine‐based combination strategies [52]. Furthermore, a recent study found that combining 5‐FU with certain TYMS inhibitors synergistically inhibits HCC progression, expanding the versatility of CAP in clinical applications [53]. Therefore, evaluating the synergistic effects of mCAP and TYMS inhibitors is essential for developing more effective therapeutic strategies.

In conclusion, our findings suggest that CAP inhibits TYMS, leading to NOX activation, increased ROS production, and ultimately ferroptosis in HCC cells. It is important to note that these conclusions derive from experiments conducted on tumor‐bearing mice and in vitro cell lines. Metronomic chemotherapy also offers a potential solution to 5‐FU resistance. Future work will employ HCC xenograft models and clinical specimens to delineate the anti‐tumor efficacy of CAP. We aim to clarify the mechanism of CAP as a chemotherapy drug for HCC patients and optimize therapeutic regimens.

5. Conclusion

This study demonstrates that mCAP effectively inhibits tumor growth in a syngeneic tumor mouse model. Mechanistically, CAP is metabolized to 5‐FU, which inhibits TYMS and thereby activates NOX‐mediated elevation of ROS, triggering ferroptosis in HCC.

Author Contributions

Yuan Chang: conceptualization, data curation, formal analysis, methodology, project administration, visualization, writing – original draft, writing – review and editing. Jiaowen Yang: conceptualization, data curation, formal analysis, methodology, project administration, visualization, writing – original draft, writing – review and editing. Ruining Yang: conceptualization, data curation, formal analysis, methodology, project administration, visualization, writing – original draft, writing – review and editing. Dejun Kong: data curation, formal analysis, visualization. Jinliang Duan: investigation, validation. Zhan Feng: formal analysis, data curation, conceptualization, writing – original draft. Hao Wang: formal analysis, project administration, visualization. Qian Sun: validation, visualization, formal analysis. Jiashu Ren: data curation, formal analysis, visualization. Zhenglu Wang: conceptualization, formal analysis, project administration, visualization, validation. Shaofeng Chen: data curation, project administration, visualization. Sei Yoshida: validation, project administration. Hong Zheng: conceptualization, data curation, formal analysis, project administration, validation, visualization, writing – original draft, writing – review and editing. Lei Cao: conceptualization, data curation, project administration.

Funding

This work was supported by the Key Project of the Natural Science Foundation of Tianjin (21JCZRJC00160), the General Project of the Natural Science Foundation of Tianjin (22JCYBJC01230), the Tianjin Postgraduate Research Innovation Project (2022SKY217), the Incubation Fund of Tianjin First Central Hospital (No. 2025FYMS03), and the Tianjin Key Medical Discipline Construction Project (No. TJYXZDXK‐3‐006A).

Ethics Statement

All the animal experimental procedures were performed in compliance with the guidelines of ARRIVE and had obtained the approval from the ethics committee of the Institute of Radiation Medicine, Chinese Academy of Medical Sciences (Approval NO. IRM/2‐IACACC‐2312‐001).

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.