L-02 hepatocytes from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China) were cultured in Dulbecco’s Modified Eagle Medium (DMEM) (Gibco, Grand Island, NY, United States) with 10% fetal bovine serum (FBS; Gibco) and antibiotics (100 U/mL of penicillin and 100 μg/mL of streptomycin) in a 5% CO₂ incubator at 37 °C. Cell lines were confirmed mycoplasma-free before experiments.

CGA was obtained from Alfabiotech (Chengdu, China). Antibodies against PARP (ab191217), Caspase-9 (ab32539), Cyclin D1 (ab40754), CDK6 (ab124821), p21 (ab109520), Cyclin A2 (ab32386), Cyclin E1 (ab33911), ASNS (ab40850), GPX4 (ab125066), Ferritin (ab75973), TfR1 (ab214039), IREB2 (ab232994), RRM2 (ab57653), c-Myc (ab32072), NDUFV2 (ab183715), NDUFS3 (ab177471), NDUFS8 (ab170936), Grim19 (ab110240), SDHA (ab137040), SDHB (ab178423), UQCRB (ab190360), UQCRC2 (ab203832), UQCRFS1 (ab191079), COX IV (ab202554), ATP5A (ab176569), and TOMM20 (ab186735) were obtained from Abcam (Cambridge, MA). Antibodies against GAPDH (2118S), p-Erk1/2 (4370T), and SLC7A11 (32683S) were bought from Cell Signalling Technology (Danvers, MA). β-actin antibody (AF0003), MTT kit, and CCK-8 kit were purchased from Beyotime Biotechnology (Shanghai, China). MitoSOX red and Carboxy-H₂DCFDA were obtained from Thermo Fisher Scientific (Waltham, MA). SB203581, SP600125, and PD0325901 were purchased from Selleck Chemicals (Houston, TX, United States).

The RTCA operational procedure was described as previously mentioned (Lan et al., 2018). Briefly, the effect of different CGA concentrations (0, 200, 400, and 800 μM) on L-02 hepatocyte growth was assessed using the RTCA system. Cells were seeded at 2 × 10³ per well in a 10-well plate with a microelectronic biosensor chip, similar in size to a 96-well plate. After overnight incubation at 37 °C, the medium was replaced with DMEM containing various CGA concentrations. Cell growth was monitored continuously for 100 h using the RTCA system. The cell index, indicating changes in electrical impedance as a measure of cell migration, was calculated with RTCA Software Package 2.0. Each sample was tested in triplicate across three experiments.

LDH cytotoxicity assay was performed as previously described (Ma et al., 2022). Briefly, L-02 cells were seeded at 2 × 10³ cells per well in 96-well plates. After 24 h at 37 °C, the medium was replaced with fresh DMEM containing 1% FBS and varying CGA concentrations. The wells were categorized into cell-free, untreated control, untreated control for cell lysis (serving as a control for maximum enzyme activity), and CGA-treated groups. The cells incubated at 37 °C for 48 h, followed by the addition of 100 μL of the same medium and another 47-h incubation. LDH release reagent was then added, and after 96 h total incubation, plates were centrifuged at 400 g for 5 min. Finally, 120 μL of supernatant from each well was transferred to a new plate for analysis using an LDH Cytotoxicity Assay Kit (Beyotime Biotechnology, Shanghai, China). Absorbance at 490 nm was measured for each well using a microplate reader (Thermo Fisher Scientific, Waltham, MA), with 600 nm as the reference wavelength. The data were then plotted into bar charts using GraphPad Prism 5.

Cytotoxicity assays and LFQ-MS were performed according to established protocols (Pang et al., 2022). Raw mass spectrometry files were analyzed using Protein Discoverer (v1.4.0.288) and Mascot (v2.3.2).

LDH and alkaline phosphatase (ALP) levels were measured using the rate method on a Siemens ADVIA2400 Automatic Biochemical Analyzer. All procedures were performed in strict adherence to the manufacturer’s instructions for the Siemens original reagents. The data were then plotted into bar charts using GraphPad Prism 5.

The Seahorse XF96 Extracellular Flux Analyzer (Seahorse Bioscience, North Billerica, MA, United States) was utilized to measure real-time integrated cellular oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) as previously outlined (Lan et al., 2018; Wang L. et al., 2017). Data is presented using Agilent’s Wave Desktop software (Agilent Technologies Inc., SC, CA, United States).

WB analysis, RT-qPCR, and ROS analysis were performed as previously described (Pang et al., 2022). Primers for RT-qPCR were synthesized by BGI (Shenzhen, China) and are listed in Supplementary Table S1. Data processing and presentation were conducted using BioRadCFXManager 3.0 (Bio-Rad, Hercules, CA) and flow cytometry software from ACEA Biosciences, United States.

The enzyme activity of three respiratory-chain complexes was quantified in the mitochondria of L-02 hepatocytes treated with CGA and in untreated cells, as previously described (Birch-Machin and Turnbull, 2001; Bao et al., 2018; Liu et al., 2018). Data is acquired and displayed using the Hitachi U3900 UV-vis spectrometer and GraphPad Prism 5.

To detect GSH and GSSG (glutathione disulfide), two 200 μL sample solutions were prepared. L-02 hepatocytes, with or without 600 μM CGA treatment for 96 h, were digested, centrifuged at 4 °C and 1,000 rpm for 4 min, and washed with PBS (phosphate buffered saline). The cell concentration was adjusted to 1 × 10⁷ cells/ml, centrifuged again, and lysed with 80 μL of 10 mM HCl. After two freeze-thaw cycles, 20 μL of 5% 5-sulfosalicylic acid (SSA) was added, followed by centrifugation at 8,000×g for 10 min. The supernatant was transferred, diluted with ddH₂O, and SSA concentration adjusted to 0.5%. GSH and GSSG levels were measured using the GSSG/GSH Quantification Kit from Dojindo Laboratories, Japan, following the manufacturer’s instructions. The data were then plotted into bar charts using GraphPad Prism 5.

L-02 cells were cultured in 10 cm dishes at an initial density of 4 × 10⁵ cells per dish and treated with either 0 or 600 μM CGA. After 96 h, the cells were harvested using pancreatic enzymes, washed twice with PBS, and lysed with 1% TritonX-100 (300 μL for 0 μM CGA and 150 μL for 600 μM CGA) on ice for 10 min. The lysates were centrifuged at 12,000 g and 4 °C for 10 min. A 2 μL sample was used for protein concentration analysis with a BCA assay kit (Thermo Fisher Scientific, Waltham, MA). The remaining sample was resuspended in 5% nitric acid (5 mL) and digested under high pressure. The iron content in the digested sample was then measured using ICP-MS (Agilent) according to the manufacturer’s guidelines and performed as previously described (Ren et al., 2021). The data were then plotted into bar charts using GraphPad Prism 5.

Fifty Kunming (KM) mice (17–20 g, equal male and female) were obtained from Hunan Slack Jingda Experimental Animal Co., LTD. After a week of acclimatization, they were divided into five groups of 10 mice each, maintaining equal gender distribution. Following a 16 h fast, each group received a single injection of CGA at different doses (normal saline, 125 mg/kg, 250 mg/kg, 375 mg/kg, and 500 mg/kg). The 7-day mortality rate and daily mouse weights were tracked, and the LD₅₀ of CGA injection was determined as previously described (Schlede et al., 1992).

Blood was drawn from the eyeballs, incubated at 37 °C for 2 h, and then refrigerated at 4 °C overnight. The next day, the serum was separated by cryocentrifugation at 2000 rotations/min for 10 min and transferred to a sterile centrifuge tube. The serum was diluted with SA buffer (about 300 μL) and analyzed for biochemical indices using the Hitachi 7,060 automatic analyzer, measuring AST, ALT, ALP, T-Bil-V2, UREA, and CREA-S. The data were then plotted into bar charts using GraphPad Prism 5.

Data that follow a normal distribution are presented as the mean ± standard deviation (SD). Comparisons between two groups were performed using an independent Student’s t-test. For comparisons involving three or more groups with a single variable, a one-way ANOVA was employed. When comparing three or more groups with two variables, a two-way ANOVA was utilized. For measurement data not conforming to a normal distribution, comparisons between two groups were conducted using the Mann-Whitney test. For comparisons involving three or more groups with a single variable, the Kruskal-Wallis test was applied. All analyses were conducted using GraphPad Prism 5 software. Statistical significance was determined at a p-value of less than 0.05.

Figures 2A–C show how RTCA, flow cytometry, and CCK-8 (Cell Counting Kit-8) were used together to analyze the growth and dose-response curves of hepatocytes L-02 treated with different concentrations of CGA. The results suggest that concentrations above 400 μM inhibit hepatocyte proliferation after 96 h, with 800 μM showing toxicity. Concentrations between 100 and 200 μM have a slight stimulatory effect on proliferation, but this effect is not statistically significant. The IC50 of CGA for hepatocyte activity was determined to be 613.1 µM (Figure 2C). For subsequent mechanistic experiments, we used 600 µM as a working approximation of the IC50 concentration. LDH activity in the L-02 cell medium was measured after incubation with different concentrations of CGA and FBS for 96 h using a LDH Cytotoxicity Assay Kit and a biochemical analyzer. Lower CGA concentrations (50–300 μM) decreased LDH activity in L-02 cells (p < 0.05), suggesting a positive effect. However, higher concentrations (600 μM) led to increased LDH activity (p < 0.05), indicating hepatocyte toxicity (Figures 2D,E).

In Figure 2F, hepatocyte OCR decreased with oligomycin treatment and increased with FCCP (a mitochondrial oxidative phosphorylation uncoupler) treatment. Hepatocytes treated with 200 and 400 µM of CGA showed a significant increase in OCR compared to controls, with a concentration-dependent effect observed between 200 and 400 µM. Figure 2G and Supplementary Figure S1A-C show that hepatocytes had increased basal, maximal, spare respiration, and intracellular ATP production after a 4 h incubation with 200 μM and 400 µM CGA (p < 0.05). However, 800 µM CGA did not have a significant effect on these mitochondrial energy metabolism indicators in hepatocytes. Hepatocytes treated with 800 µM CGA exhibited decreased OCR compared to the control group. ECAR values for hepatocytes treated with 200–800 µM CGA for 4 h showed no change in glycolysis (Figure 2H). Incubation with 200 µM CGA for 15 days did not significantly impact the clonogenic ability of L-02 cells; however, treatment with 300 µM CGA almost completely inhibited L-02 cell clonogenesis (p < 0.001) (Figures 2I,J).

Figure 3A shows proteomic screening results: 117 differential proteins identified after 48 h and 164 after 96 h of incubation with 600 µM CGA. Venn diagram analysis identified a total of 18 overlapping differential proteins between the two groups, including 10 upregulated and 5 downregulated proteins. The bioinformatics analysis suggests that dysregulation of multiple signaling pathways, such as glutathione metabolism, beta-alanine metabolism, fatty acid degradation, lysine degradation, nucleotide excision repair, valine, leucine, and isoleucine degradation, and ferroptosis, may contribute to hepatocyte toxicity at the IC50 of CGA (Figures 3B,C). Proteomic analysis incidentally revealed alterations in the expression of alkaline phosphatase isozymes (ALPG and ALPP) (Figures 3D,E), and a corresponding increase in ALP activity was detected in the culture supernatant (Figure 3F).

As illustrated in Figure 4A, electron microscopy showed no major changes in the mitochondrial morphology of L-02 cells treated with 600 μM CGA; however, there were alterations in cell size and nucleolar morphology, which indicate early apoptosis. Proteomic analysis indicated a significant upregulation of CTSD (cathepsin D), CAPN1 (calpain 1), and CYCS (cytochrome c) proteins following treatment with 600 µM CGA (p < 0.05) (Figure 4B). Conversely, the same concentration of CGA resulted in a significant downregulation of EIF2S1 (eukaryotic translation initiation factor 2 subunit) protein (p < 0.05) (Figure 4B). Flow cytometry analysis showed that an IC50 of CGA induced apoptosis in hepatocytes, resulting in a 7.2% increase in apoptotic cells compared to the control group (p < 0.001) (Figures 4C,D). Western blot analysis revealed that CGA enhanced the cleavage of PARP and caspase-9 precursors in a dose-dependent manner (Figure 4E). Additionally, flow cytometry analysis indicated that the L-02 cell cycle was arrested in the S phase due to CGA treatment (p < 0.001) (Figures 4F,G). The findings from the Western blot analysis demonstrated that treatment with 600 µM CGA resulted in a significant decrease in the protein expression of Cyclin D1 and CDK6, which are involved in the G1/S phase transition (Figure 4H).

KEGG analysis showed a link between CGA and pathways involving glutathione metabolism, fatty acid degradation, and ferroptosis (Figures 3B,C). Initially, through the application of proteomics, Western blotting, and qRT-PCR, it was verified that treatment with CGA at a concentration of 600 µM resulted in an upregulation of transferrin receptor 1 (TfR1) expression and Chac1 (glutathione-specific gamma-glutamylcyclotransferase 1) expression, as well as a downregulation of glutathione peroxidase 4 (GPX4) expression. All of these changes are associated with ferroptosis (Figure 5F; Supplementary Figure S2A,B). Furthermore, ICP-MS analysis indicated that the IC50 of CGA led to an increase in the total iron content within hepatocytes, although this increase was not statistically significant (Figure 5A). In contrast, the use of the FerroOrange probe demonstrated that the IC50 of CGA reduced the levels of free ferrous ions in hepatocytes (Figure 5B). Additionally, Western blot analysis revealed no significant change in the expression of SLC7A11 (cystine/glutamate transporter), a key protein in the ferroptosis pathway (Figure 5F).

Further investigation found that CGA increased the mRNA and/or protein levels of key enzymes involved in glutathione metabolism (p < 0.05), including RRM2 (ribonucleoside-diphosphate reductase subunit M2), GSR (glutathione reductase), GCLM (glutamate--cysteine ligase regulatory subunit), TM4SF1 (transmembrane 4 L6 family member 1), PGD (6-phosphogluconate dehydrogenase), G6PD (glucose-6-phosphate 1-dehydrogenase), and GSTP1 (glutathione S-transferase P), potentially enhancing glutathione biosynthesis (Figure 5C). The GSSG/GSH Quantification Kit assay showed that CGA increased GSH biosynthesis in hepatocytes after 96 h (p < 0.01), but did not affect the GSH/GSSG ratio (Figures 5D,E). Nevertheless, after 6 days of incubation, CGA significantly decreased the intracellular GSH/GSSG ratio (p < 0.001) (Figures 5D,E).

The live cell counting experiment demonstrated that co-incubation of the ferroptosis inhibitor Ferrostatin-1 (1 μM) with the IC50 concentration of CGA significantly augmented the inhibitory effect of CGA on hepatocyte proliferation compared to the control group. Conversely, co-incubation with the ferroptosis inducer Erastin (1 μM) markedly attenuated the inhibitory impact of the IC50 concentration of CGA on hepatocyte growth (Supplementary Figure S2C). Western blot analysis showed up-regulation of IREB2 (Iron-responsive element-binding protein 2) and down-regulation of Ferritin and c-Myc in hepatocytes treated with CGA at its IC50 concentration (Figure 5F), indicating iron deficiency. Subsequent proteomics and RT-qPCR analyses indicated that the IC50 of CGA may upregulate ALDH1B1 (mitochondrial aldehyde dehydrogenase X) and ACADVL (very long-chain specific acyl-CoA dehydrogenase) expression (p < 0.05) (Supplementary Figure S2A,B), which prevents lipid peroxidation and enhancing fatty acid degradation in hepatocytes.

The IC50 of CGA can reduce oxidative stress in hepatocytes by affecting the redox system; however, it may also deplete ferrous ions and impact various biological processes. The results shown in Figure 6 indicate a significant decrease in the protein expression levels of the iron-sulfur (Fe-S) cluster subunits (NDUFV2, NDUFS3, NDUFS8, SDHB, and UQCRFS1) in mitochondrial complexes I-III (Figure 6A), as well as in the protein expression of mitochondrial cis-aconitase ACO2 (Figure 6B), which contains an iron-sulfur cluster. This decrease leads to reduced enzyme activity in mitochondrial complexes I-III (Figure 6C). Other subunits in mitochondrial complex I-V that do not contain iron-sulfur clusters were not affected (Figure 6A). Despite the decrease in enzyme activity of mitochondrial complexes I-III, the IC50 of CGA did not significantly increase intracellular total ROS and mitochondrial ROS levels (refer to Supplementary Figure S3A,B).

To directly test the hypothesis that CGA toxicity stems from functional iron deficiency, we performed rescue experiments with ferric citrate. First, in cell viability assays, co-incubation with ferric citrate (75 or 150 µM) significantly alleviated the inhibitory effect of CGA (IC50) on hepatocyte activity (Supplementary Figure S3C). Crucially, control groups treated with the same concentrations of ferric citrate alone showed no significant effect on cell viability, confirming that the rescue was not due to non-specific growth promotion by iron. Second, we investigated the molecular basis of this rescue. Western blot analysis revealed that co-treatment with 100 µM ferric citrate effectively reversed the CGA-induced downregulation of key proteins involved in the cell cycle (CDK6, Cyclin D1, c-Myc), mitochondrial electron transport (NDUFS3, SDHB), and iron storage (Ferritin), while also modulating redox balance markers (GPX4) (Figure 6D). Importantly, the control group treated with 100 µM ferric citrate alone provided critical mechanistic insight: it did not reproduce the CGA-induced pattern of protein suppression. Instead, it led to an expected upregulation of Ferritin and had no significant effect on the levels of CDK6, Cyclin D1, c-Myc, NDUFS3, or SDHB (Figure 6D). This stark contrast demonstrates that ferric citrate specifically corrects the molecular perturbations caused by CGA rather than exerting independent, generalized effects. Additionally, to probe the specificity of the cell death pathway, co-incubation with the necrosis inhibitor Nec-1s (10 µM) partially improved cell viability, whereas the NF-κB inhibitor CAPE (1 µM) had no effect (Supplementary Figure S3D). Collectively, these findings provide multi-layered evidence that depletion of the labile iron pool is a central event in CGA-induced hepatotoxicity, and that its consequences—ranging from cell cycle arrest and mitochondrial dysfunction to altered protein expression—can be specifically counteracted by iron repletion.

As previously mentioned, extended exposure of hepatocytes to the IC50 of CGA still disrupts cellular redox homeostasis. Consequently, this study investigated how the antioxidant NAC can protect hepatocytes from toxicity caused by excessive CGA. In vitro experiments showed that co-treatment with NAC (2 mM) and CGA (600 µM) prevented the non-enzymatic discoloration reaction of CGA and effectively mitigated the inhibitory effects of CGA on hepatocyte proliferation (Figures 7A–C). Flow cytometry analysis indicated that NAC prevented ROS accumulation in hepatocyte mitochondria caused by the IC50 of CGA (p > 0.05) (Figure 7D). RT-qPCR and WB experiments demonstrated that, in addition to ferritin, NAC also helped mitigate the effects of CGA on cellular processes in L-02 cells (Figure 7E; Supplementary Figure S3F).

Previous studies have shown that deferoxamine (DFO) inhibits the growth of esophageal squamous cell carcinoma (ESCC) cells by activating the ERK pathway in a ROS-dependent manner (Lan et al., 2018). This research found that CGA concentrations above 300 µM increased Erk1/2 phosphorylation (Supplementary Figure S3E), and treatment with 800 µM CGA led to a significant increase in ROS accumulation in hepatocytes (Figures 7F,G). NAC reduced mitochondrial ROS levels caused by 800 µM CGA and improved L-02 cell activity (Figures 7G,H). PD0325901 (a MEK kinase inhibitor) was also effective in mitigating the effects of CGA (800 µM) (p < 0.05), while SB203581 (a p38 MAPK inhibitor) and SP600125 (a broad-spectrum JNK inhibitor) were not (Figure 7H). Consequently, these experimental outcomes are consistent with a model in which the ROS-MAPK/Erk1/2 pathway may play a crucial role in mediating the hepatocyte toxicity induced by excessive CGA.

In Table 1, it is shown that all KM mice survived for 7 days after receiving 0–125 mg/kg CGA injections; however, mortality rates increased with higher doses, reaching 100% at 500 mg/kg. Symptoms such as convulsions and difficulty breathing were observed in mice receiving higher doses. Nevertheless, mice injected with different doses of CGA did not show significant changes in body weight or serum biochemical indices related to liver and kidney function compared to the control group over a 7-day period (Figures 8A–D). Based on the data, the LD50 of CGA is 382.28 mg/kg, indicating a safe dosage range for a single intravenous injection of 0–125 mg/kg. Considering an estimated average blood volume of approximately 7%–8% of body weight in mice (Diehl et al., 2001), the LD50 dose of 382.28 mg/kg would result in an initial circulating concentration estimated to be in the range of 5–10 mM. This concentration is approximately an order of magnitude higher than the in vitro IC50 of 613 µM for hepatocytes, taking into account the rapid distribution and metabolism occurring in vivo.