The Gene Expression Omnibus database (https://www.ncbi.nlm.nih.gov/gds) was used to download GSE136247 dataset (23), using R package ‘FactoMineR (https://cran.r-project.org/web/packages/FactoMineR/index.html)’, ‘factoextra (https://cran.r-project.org/web/packages/factoextra/index.html)’. An advanced multivariate statistical method, known as Principal Component Analysis (PCA), was utilized to analyze the data compilations derived from the two disparate groups of sample cohorts. Subsequently, the ‘limma (https://www.bioconductor.org/packages/release/bioc/html/limma.html)’ package was utilized to delve into the disparities in gene expression between these two cohorts. A total of 527 differentially expressed genes were obtained according to P<0.05 and |logFC|>1. A graphical representation of the gene volcano plot was crafted utilizing the R library ‘ggplot2,’ whereas the graphical depiction of the Top 20 differentially expressed genes volcano plot was created with the assistance of the R package ‘pheatmap’.

For 527 genes from the GSE136247 dataset, GO and KEGG functional enrichment was performed using the R package ‘clusterProfiler (https://www.bioconductor.org/packages/release/bioc/html/clusterProfiler.html)’, and the top-ranking results were displayed.

The STRING repository (accessed at http://cn.string-db.org/) was employed to build a network of protein interactions utilizing the aforementioned 527 overlapping genes.

The GEPIA2 (http://gepia2.cancer-pku.cn/#analysis) database was accessed, which provides cancer gene expression data based on the TCGA database. The name of the TOP2A gene was input and HCC [TCGA-liver hepatocellular carcinoma (LIHC)] was selected as the research object. In the expression analysis section, GEPIA2 was used to draw the expression comparison chart of TOP2A in liver cancer tissues and normal liver tissues, and box plots were used to demonstrate the expression differences between different groups. In order to assess the impact of TOP2A expression on the survival outcomes of liver cancer sufferers, the GEPIA2 database survival analysis feature was utilized. Patients with liver cancer were categorized into two groups based on the level of TOP2A expression: One with elevated expression and the other with reduced expression. Through the online website Kaplan-Meier Plotter (https://kmplot.com/analysis/), ‘Liver cancer’ was selected, the gene name ‘TOP2A’ was input, the follow-up time endpoint was set to sixty months in ‘Follow up threshold’, and the survival curve was drawn. The log-rank method was employed to compare the survival discrepancies between these two subsets, aiming to establish a link between TOP2A expression levels and the patients' prognostic outlook.

Initially, the three-dimensional structure of TOP2A (with PDB identifier: 1ZXM) was sourced from the Protein Data Bank (PDB; http://www.rcsb.org/). The PHN file in SDF format was retrieved from the PubChem database (accessible at http://pubchem.ncbi.nlm.nih.gov). Subsequently, Open Babel (http://openbabel.org/index.html) was utilized to transform this file into the PDB format. Following this, the protein underwent dehydration and hydrogenation processes, after which its charge was computed. Using Autodock tools version 1.5.7 (https://autodocksuite.scripps.edu/adt/), the protein was then formatted into pdbqt.

The 3D structural data of TOP2A, identified by PDB code 1ZXM, was acquired from the PDB repository. The Small Molecule Data File (SDF) for PHN was fetched from the PubChem database, and then converted into PDB format via Open Babel. The protein structure was subsequently stripped of water molecules and hydrogenated, with its electrical charge assessed. The conversion to pdbqt format was facilitated by Autodock tools version 1.5.7. Ligand processing involved hydrogenation and the calculation of torsional energies, culminating in its transformation into pdbqt format as well. The boundaries for the docking simulation were set, and the molecular docking simulation was carried out using Autodock vina. Visual inspection and three-dimensional analysis were conducted with pymol 2.1.0 (Schrödinger Corporation), yielding a comprehensive 3D graphical representation.

Human HCC cell lines JHH7, HEP3B217, SNU878, JHH2 and normal liver cell line THLE-2 were purchased from the Qingqi (Shanghai) Biotechnology Development Co., Ltd. The frozen cells were thawed after retrieval from liquid nitrogen storage and revitalized. These cells were subsequently cultivated in a DMEM solution (cat. no. 11995; Beijing Solarbio Science & Technology Co., Ltd.), which consisted of a blend of 10% newborn calf serum (cat. no. S9020; Beijing Solarbio Science & Technology Co., Ltd.) and 1% penicillin-streptomycin mixture. PHN (cat. no. F798884) was purchased from Shanghai Maclin Biochemical Technology Co., LTD.

The T25 cell culture vial was removed from the cell incubator and the vial contained the exponential growth stage of the liver cancer cell line. The old medium was discarded, 2 ml sterile PBS was added, the T25 cell culture bottle was gently shaken, the excess PBS was discarded, and then 1 ml trypsin digestion solution (cat. no. T1300; Beijing Solarbio Science & Technology Co., Ltd.) was added and incubated at 37°C for 2 min. Digestion was terminated by adding 2 ml DMEM medium. The cells were counted by Thermo Fisher Scientific, Inc. (Countess 3) and inoculated into six wells with 1×10⁵ cells per well and placed in a cell incubator at 37°C (Countess 3; Thermo Fisher Scientific, Inc.) overnight. The old medium was discarded, add the premixed solution of Lip2000 (cat. no. 11668030; Thermo Fisher Scientific, Inc.) and SiTOP2A interference sequence (1,500 ng per well) or overexpression plasmid (2,500 ng per well) were added and cells were cultured in serum-deprived DMEM solution for a period of 5 h. Subsequently, fresh DMEM was introduced, and the culture was maintained for an additional 48 h to proceed with the subsequent investigation. The specific sequences for the small interfering RNA (siRNA) and plasmid are presented in Table SI.

siRNA was used to knockdown TOP2A (si-TOP2A), and plasmids were used for overexpression of TOP2A (OE-TOP2A). In JHH7 cells, they were divided into three groups: Si-negative control (NC), PHN + Si-NC, and PHN + si-TOP2A. In the PHN + si-TOP2A group, JHH7 was transfected with siRNA and incubated for 48 h. The Si-NC and PHN + Si-NC groups were transfected with the control sequence and incubated for 48 h. After that, PHN was added to the PHN + Si-NC and PHN + si-TOP2A groups and incubated for another 48 h. Cells were extracted for subsequent experiments. In JHH2 cells, they were divided into three groups: Vector, Vector + PHN, and OE-TOP2A + PHN. In the Vector + PHN and OE-TOP2A + PHN groups, PHN was added to JHH2 cells and incubated for 48 h. After that, the OE-TOP2A + PHN group was transfected with the overexpression TOP2A plasmid and continued to be incubated for 48 h. The Vector and Vector + PHN groups were transfected with the control vector and continued to be incubated for 48 h. Cells were extracted for subsequent experiments.

Cell transfection is performed in the same way as aforementioned. The successfully transfected cells were removed from the incubator, cell digestion and centrifugation were consistent with previous methods, and cell precipitation was collected. The cultures were seeded into 96-microwell trays at a seeding density of 3,000 cells per well and subsequently exposed to varying dosages of PHN for periods of 0, 24, 48 and 72 h. Following this treatment, 10 µl of the CCK-8 solution (cat. no. C0038; Beyotime Institute of Biotechnology) was introduced into each individual well, after which the incubation at 37°C was prolonged for an additional 4 h. Finally, the 96-well plates were removed from the cell incubator and placed on the SpectraMax Mini multi-function enzyme label to detect OD (450 nm) value (Molecular Devices, LLC). The subsequent data were statistically analyzed by Graph Prism 9.5.0 software (Dotmatics).

Exponential growth cells were received, cell digestion was carried out, add 1×10⁵ cells per well into six-well plates and incubated overnight in a cell incubator. Transfection EDU solution (10 µmol/l; cat. no. C0071S; Beyotime Institute of Biotechnology) was added to the transfected cells and incubated in a cell incubator for 2 h. The old medium was removed, and 1 ml of cell fixing solution (cat. no. P0099; Beyotime Institute of Biotechnology) was added to each well for incubation at room temperature (20°C) for 30 min, and then 1 ml of 0.3% Triton X-100 (cat. no. P0096; Beyotime Institute of Biotechnology) was added to each well and incubated at room temperature (20°C) for 10 min. The permeable solution was removed and 1 ml Hoechst 33342 solution (cat. no. C1022; Beyotime Institute of Biotechnology) was added to each well and incubated for 10 min at room temperature (20°C) without light. The Hoechst 33342 solution was removed, cleaned with PBS for 3 times, observed with inverted fluorescence microscope (model AMF7000; Thermo Fisher Scientific, Inc.). Images were captured, and then statistical analysis was performed using Image J 1.52a (National Institutes of Health).

Transfection of HCC was performed by adding a suitable volume of TRIzol reagent (cat. no. 15596026; Beijing Solarbio Science & Technology Co., Ltd.) was utilized for the isolation of total RNA, adhering to the provided protocol. Subsequently, the RNA concentration was assessed using the NanoDrop One spectrophotometer. Following this, complementary DNA was synthesized through reverse transcription, following the manufacturer's protocol (cat. no. D7168M; Beyotime Institute of Biotechnology). Thereafter, the BeyoFast^™ SYBR Green RT-qPCR Mix (cat. no. D7262; Beyotime Institute of Biotechnology) was incorporated into the reaction mixture. The template DNA was fully denatured by pre-heating at 95°C for 3 min on the PCR apparatus, and then entered the amplification cycle. In each cycle, the template was denatured by holding at 95°C for 30 sec, and then the temperature was lowered to 60°C for 30 sec, so that the primer and the template were fully annealed. The mixture was held at 72°C for 1 min (primer extended on the template, DNA synthesized, a cycle complete, then repeat 40 cycles) and finally set 4°C. The 7900HT fluorescence quantitative PCR instrument (cat. no. 4351405; Thermo Fisher Scientific, Inc.) was set up according to the instructions. Relative gene expression data were analyzed by real-time quantitative PCR and the 2^−ΔΔCq method (24). The subsequent data were statistically analyzed by Graph Prism 9.5.0 software. Details of the primer sequences are described in Table SI.

HCC cell transfection was conducted as aforementioned; cells in each group were expanded to >2×10⁶ to extract cell precipitation for use. The HCC cell specimens underwent processing in compliance with the iron concentration testing kit (cat. no. MAK025; MilliporeSigma). The intracellular Fe²⁺ concentration was assayed, and the optical density (OD) at 593 nm was obtained using the SpectraMax Mini multi-purpose enzyme analyzer. Additionally, the cells affected by HCC were subjected to the MDA measurement protocol as per the kit instructions (cat. no. S0131S; Beyotime Institute of Biotechnology). The OD readings at 532 nm were recorded with the aid of the SpectraMax Mini multifunctional enzyme analyzer. The subsequent data were statistically analyzed by Graph Prism 9.5.0 software.

After treating each group of cells by the aforementioned method, the cells were fixed with 2.5% glutaraldehyde and 1% osmium tetroxide for 2 h respectively at 4°C in sequence, and then dehydrated in a series of acetone gradients. Subsequently, the morphological changes of mitochondria in the tissues were observed by transmission electron microscopy (JEM-1400FLASH; JEOL, Ltd.), and the images were captured.

The successfully transfected JHH7 and JHH2 cells were inoculated in six-well plates at 1×10⁵ cells/well, and incubated overnight. Then, the culture medium was removed and the cells were gently washed twice with PBS. Next, 2 ml of ROS fluorescent working solution (cat. no. S0033M; Beyotime Institute of Biotechnology) were added, and the mixture was incubated for 30 min away from light. After the incubation, the cells were washed again with PBS to remove the ROS staining solution. Subsequently, DAPI staining solution (1 µg/ml) was used to stain the cell nuclei, and cells were incubated away from light for 5 min. After the staining was completed, the cells were washed with PBS. Finally, images were observed and captured under a fluorescence microscope, and the relative expression level of intracellular ROS was analyzed using ImageJ 1.5.2a software (National Institutes of Health).

The treated cells were selected, and these cell samples were co-incubated with RIPA lysis buffer (cat. no. R0010; Beijing Solarbio Science & Technology Co., Ltd.) at 4°C for 30 min. Subsequently, protein extracts were isolated from HCC cells. The protein levels were quantified using a BCA assay kit. Based on the protein concentration readings, the volume of the sample was adjusted correspondingly. An equal volume of the protein (30 µg) mixture was pipetted into the wells of the gel matrix (containing 30% acrylamide), and the electrical potential was fixed at 120 volts for the purpose of conducting gel electrophoresis. Following a 75 min run, the gel was transferred onto a PVDF membrane, and the electrical setting was then increased to 200 volts. After a 30 min transfer, the PVDF membrane was immersed in a solution of 5% non-fat milk powder and allowed to sit for a duration of 2 h at 25°C. TBST (containing 0.1% Tween 20) was cleaned three times after incubation, TBST was cleaned three times after the primary antibody was incubated overnight at 4°C, and chemiluminescent solution was added for 10 sec after the secondary antibody was incubated for 2 h at 25°C. Images were stored in the chemiluminescence apparatus (SH-Cute523; Hangzhou Shenhua Technology Co., Ltd.) for subsequent statistical analysis. WB antibody information is included in Table SII. Densitometric analysis was carried out using ImageJ 1.5.2a software (National Institutes of Health).

All statistical evaluations were conducted with the SPSS version 22.0 program (IBM Corp.). Quantitative results are presented as the average value ± standard deviation. For comparisons between the two distinct groups, an independent samples t-test was applied. To assess variations among several groups, a one-factor ANOVA followed by Tukey's multiple comparison test was utilized. Comparisons across various time intervals were executed with the Bonferroni correction for post hoc analysis. P<0.05 was considered to indicate a statistically significant difference.

PCA was first performed on GSE136247 data to obtain a total of 527 differentially expressed genes, of which 93 were upregulated and 434 downregulated (Fig. 1A). Subsequently, differential genes were selected to draw volcano maps (Fig. 1B) and cluster heat maps (Fig. 1C). It was found that compared with normal liver tissues, AFP, PEG10, SULT1C2, GPC3 and acyl-CoA synthetase family member 4 (ACSL4) were most significantly upregulated in liver cancer tissues, while MT1G, MT1G, C9, CYP2C8 and CTP1A2 were most significantly downregulated. These significant differential genes can be used as potential therapeutic targets for liver cancer. KEGG functional enrichment analysis showed that differential genes were mainly enriched in complement and coagulation cascades, drug metabolism and metabolism of xenobiotics by cytochrome P450 (Fig. 1D and E). Subsequently, differential gene PPI network interaction was conducted (Fig. 1F).

The differential genes of liver cancer tissue and normal liver tissue obtained as aforementioned were intersected with PHN drug targets to draw Wayne's diagram. It was found that only one intersection gene was TOP2A (Fig. 2A). Later, the levels of TOP2A expression in TCGA liver cancer dataset (TCGA-LIHC) were investigated and a significant increase was observed in TOP2A levels within liver cancer samples compared with the normal tissue controls (Fig. 2B). This observation aligns with the current analysis of TOP2A expression patterns. Moreover, there was a significant association between TOP2A expression and liver cancer outcomes, with patients exhibiting high TOP2A expression demonstrating poorer prognostic outcomes compared with those with low TOP2A expression levels (Fig. 2C). The 2D and 3D molecular structure formula of PHN is demonstrated in Fig. 2D. In order to further confirm the existence of binding sites between PHN and TOP2A, molecular docking was simulated in silico, and the docking results showed that there were binding sites between PHN and TOP2A, along with a strong binding effect (Fig. 2E).

Through our online database (https://sites.broadinstitute.org/ccle/), the different TOP2A expression was screened in HCC cell line (Fig. 3A). The pair of HCC cell lines exhibiting the most prominent contrast in TOP2A expression levels was chosen, along with healthy liver cells, to investigate TOP2A expression via WB and RT-qPCR analysis. The findings indicated that TOP2A was most prominently expressed in the JHH7 cellular line while the expression was lowest in JHH2 cells (Fig. 3B). This further proves the accuracy of the database prediction information. Therefore, JHH7 cells with high TOP2A expression and JHH2 with low TOP2A expression were selected for subsequent experiments.

To verify the effects of PHN on the proliferation of JHH7 and JHH2 HCC cells, the effects of different concentrations (6.25, 12.5, 25, 50, 100 and 200 µmol/l) of PHN on HCC cell proliferation were examined after 24 h of treatment by CCK-8 assay. The results of CCK-8 revealed that PHN had no detailed inhibitory effect on the viability of HCC at concentrations of 6.25 µmol and 12.5 µmol/l. PHN had a strong inhibitory effect on HCC from the concentration interval of 25 to 200 µmol/l, accompanied by a concentration dependence. The IC₅₀ of PHN-treated JHH7 cells for 24 h was 106.3 µmol/l, and that of JHH2 cells was 123.4 µmol/l. A PHN concentration of 100 µmol/l (<IC₅₀) was selected for the subsequent experiment (Fig. 4A). To ascertain the impact of PHN on the protein expression of TOP2A, the cell line JHH7 was chosen, known for its high TOP2A expression, and it was subjected to varying concentrations of PHN for a duration of 48 h. Subsequent analysis revealed that the protein expression of TOP2A was suppressed by PHN in a manner directly proportional to the concentration administered (Fig. 4B).

To further verify the specific role of TOP2A in hepatoma cell lines, TOP2A expression was either knocked down or upregulated. Subsequently, enhanced expression was observed in the JHH7 and JHH2 clones, respectively. Immunoblotting and RT-qPCR analyses revealed that each of the three si-TOP2A constructs significantly suppressed the mRNA and protein synthesis of TOP2A (Fig. 2C). Notably, si-TOP2A1 exhibited the most profound suppression of both mRNA and protein synthesis for TOP2A. Conversely, an upsurge in the mRNA and protein indicators of TOP2A was evident within the OE-TOP2A experimental cohort (Fig. 4D). The experimental results demonstrated that TOP2A-downregulated cell models in JHH7 cells and TOP2A-upregulated cell models in JHH2 cells were successfully constructed.

CCK-8 and EDU experiments confirmed that PHN group could significantly inhibit the proliferation of JHH7 liver cancer cells compared with the control group, and PHN + si-TOP2A group could inhibit the proliferation of JHH7 liver cancer cells, but there was no statistical significance compared with PHN group. PHN also inhibited the proliferation of JHH2 HCC cells. After PHN treatment, the overexpression of TOP2A plasmid could partially restore the vitality of JHH2 cells. The aforementioned results confirmed that PHN inhibited the proliferation of JHH7 and JHH2 HCC cells by targeting TOP2A (Fig. 5A and B).

To determine the ability of PHN to trigger ferroptosis in hepatoma cells, the changes in mitochondrial morphology and structure were observed by electron microscopy. The electron microscopy structure showed that mitochondria in the control group were oval or rod-shaped, with a double-membrane structure. The outer membrane was smooth, and the inner membrane folded inward to form cristae. However, after treatment with PHN, the outer membrane of mitochondria ruptured, the cristae decreased, and the mitochondrial morphology was disrupted. Nevertheless, overexpression of TO2AP could reverse this phenomenon (Fig. 6A). Attention was also paid to the changes in the intracellular ROS level. The results revealed that PHN treatment could significantly enhance the fluorescence intensity of intracellular ROS, while overexpression of TO2AP could restore the intracellular ROS expression level. In addition, the fluctuations in intracellular Fe²⁺ concentration and MDA content were monitored. The results indicated that PHN significantly increased the intracellular Fe²⁺ concentration and MDA level in hepatoma cells. Furthermore, when PHN was introduced after initially inhibiting TOP2A, it further increased the levels of Fe²⁺ and MDA. Conversely, subsequent overexpression of TOP2A after PHN restored the concentrations of Fe²⁺ and MDA in hepatoma cells to the original levels (Fig. 6B-F). This suggests that PHI can disrupt the mitochondrial structure of hepatoma cells, trigger oxidative stress, increase the expression levels of MDA and Fe²⁺ intracellularly, and promote ferroptosis in hepatoma cells. The mechanism is likely to be achieved by inhibiting the expression level of TOP2A protein.

To delve deeper into the underlying mechanism by which PHN initiates ferroptosis, the activation of the JAK2/STAT3 signaling axis and the concentration of proteins linked to ferroptotic processes were examined utilizing WB. The findings indicated that PHN significantly inhibited the activation (phosphorylation) of both JAK2 and STAT3, while it had no substantial impact on the overall protein quantities of JAK2 and STAT3. Additionally, PHN was found to decrease the synthesis of glutathione peroxidase 4 (GPX4) and factor inhibiting hypoxia-inducible factor 1 alpha (FIH1), whereas it concurrently enhanced the synthesis of proteins pertinent to ferroptosis, such as COX2, ACSL4 and NOX1. After TOP2A was downregulated, JHH7 cells were treated with PHN, which enhanced the inhibition of the JAK2/STAT3 signaling pathway. After PHN treatment, overexpression of TOP2A could resist the regulatory effects of PHN on JHH2 cell pathway proteins and ferroptosis-related proteins (Fig. 7A and B). These results suggest that PHN can induce ferroptosis in HCC by inhibiting the JAK2/STAT3 signaling pathway and regulating the expression level of ferroptosis-related proteins.