Introduction

Nasopharyngeal carcinoma (NPC) is a malignancy originating from the epithelial cells of the nasopharyngeal mucosa, with a notable geographical distribution, primarily prevalent in southern China and Southeast Asia1. The etiology of NPC involves Epstein-Barr virus (EBV) infection, genetic predisposition, and environmental factors such as the consumption of foods containing volatile nitrosamines2. Currently, radiotherapy-based multimodal therapy remains the primary treatment for NPC, significantly improving survival rates. However, approximately 10% of patients with NPC develop resistance to radiotherapy, resulting in tumor recurrence and treatment failure, presenting a significant clinical challenge in NPC management3,4. Therefore, understanding the mechanisms of radiotherapy resistance and identifying effective radiosensitizers to overcome this resistance are crucial for improving the prognosis of patients with NPC.

Natural compounds and their analogs play a key role in drug discovery, particularly in oncology5. Epigallocatechin gallate (EGCG), a polyphenolic compound derived from green tea, demonstrates a wide range of pharmacological properties, including antitumor, antioxidant, anti-inflammatory, cardiovascular, and neuroprotective effects6. Recent studies indicate that EGCG inhibits NPC cell proliferation and enhances radiosensitivity, suggesting its potential as a radiosensitizer [7,8]. However, the precise molecular targets and regulatory mechanisms underlying EGCG’s effect on radiotherapy resistance in NPC remain unclear and require further investigation.

This study aims to explore the potential of EGCG in overcoming radiotherapy resistance in NPC by identifying its molecular targets and mechanisms of action through RNA sequencing, network pharmacology, molecular docking, and molecular dynamics simulations. These findings will provide a theoretical basis for advancing the biomedical application of EGCG. The research workflow is depicted in Fig. 1.

Fig. 1
figure 1

Flow diagram of the study.

Full size image

Results

Identification and enrichment analysis of EGCG targets in NPC related to radiotherapy

To identify potential targets of EGCG, data were extracted from GeneCards, TCMSP, TargetNet, and SwissTargetPrediction databases, resulting in 588, 140, 114, and 100 targets, respectively. After removing duplicates, a total of 734 unique EGCG targets were compiled (Supplementary Table S1). Differential expression analysis of the GSE48501 dataset revealed 205 DEGs associated with radiotherapy (Fig. 2A). Intersection analysis subsequently identified 21 potential therapeutic targets of EGCG in NPC related to radiotherapy (Fig. 2B). Co-expression correlation analysis of these 21 targets revealed predominantly positive correlations (Fig. 2C). The PPI network illustrated extensive protein interactions among the candidate targets (Fig. 2D). GO enrichment analysis revealed that the enriched biological processes (BP) primarily involved responses to xenobiotic stimulus, chemical stress, hypoxia, decreased oxygen levels, and steroid hormones. Key cellular components (CC) included RNA polymerase II transcription regulator complexes, focal adhesions, cell-substrate junctions, cyclin-dependent kinase holoenzyme complexes, and cornified envelopes. Molecular functions (MF) were enriched in GTP binding, guanyl nucleotide binding, guanyl ribonucleotide binding, DNA-binding transcription factor binding, and GTPase activity (Fig. 2E). KEGG pathway analysis indicated that the candidate targets were primarily enriched in pathways related to Salmonella infection, pathogenic Escherichia coli infection, lipid and atherosclerosis, Yersinia infection, osteoclast differentiation, gastric cancer, non-alcoholic fatty liver disease, pancreatic cancer, pertussis, and Kaposi sarcoma-associated herpesvirus infection (Fig. 2F).

Fig. 2
figure 2

Identification and enrichment analysis of EGCG targets in NPC related to radiotherapy. (A) DEGs between radiotherapy-sensitive and radiotherapy-resistant NPC tissues. (B) A Venn diagram shows 21 shared EGCG targets obtained through intersection. (C) Correlation analysis of 21 targets. (D) PPI network of 21 targets. (E) GO enrichment analysis. (F) KEGG pathway enrichment analysis.

Full size image

Construction and validation of the prognostic model

To identify potential prognostic targets of EGCG in relation to radiotherapy in NPC, Cox regression and LASSO regression analyses were performed on the 21 candidate targets from the GSE102349 dataset. Univariate Cox regression identified two genes (ABCC1, CASP1) significantly associated with prognosis (p < 0.05) (Fig. 3A). Following LASSO regression and multivariate Cox regression, these two genes were retained to construct the prognostic signature for radiotherapy in NPC associated with EGCG (Fig. 3B). The risk score was calculated based on their Cox coefficients as follows: Risk score = (0.908221728762324 × expression level of ABCC1) + (−0.669911227543324 × expression level of CASP1).

Patients with NPC were categorized into high- and low-risk groups based on the median risk score. Heatmap analysis demonstrated that the 2 genes were differentially expressed in both high- and low-risk groups (Fig. 3C). The risk score distribution and survival status plots showed higher mortality in the high-risk group (Fig. 3D), indicating an unfavorable prognosis for these patients. Kaplan–Meier curve analysis revealed significant differences in PFS between the two groups. The low-risk group demonstrated a better PFS than the high-risk group (p = 0.023577, Fig. 3E). Time-dependent ROC curve analysis indicated that the area under the curve (AUC) values were 0.82, 0.76, and 0.752 for the predicted survival rates at 1, 2, and 3 years, respectively, thus demonstrating robust prognostic power (Fig. 3F).

Fig. 3
figure 3

Identification and validation of the prognostic signature. (A) Univariate Cox regression analysis identifying 2 prognostic targets. (B) LASSO analysis. (C) Heatmap of the signature genes correlated with risk-score. (D) Risk score distribution and survival status of NPC samples based on the risk score. (E) Kaplan-Meier curves of survival between patients in the high- and low-risk group. (F) Time-dependent ROC curves of 1-, 2-, and 3-years of NPC patients.

Full size image

Immune activity and tumor microenvironment

The CIBERSORT algorithm was employed to analyze the relationship between the two-gene prognostic risk model and immune cell infiltration in NPC using the GSE102349 dataset. Significant differences in immune cell subtypes were observed between the high- and low-risk groups. Specifically, B cells naive, B cells memory, and T cells gamma delta were more abundant in the low-risk group, while Plasma cells, T cells follicular helper, resting Dendritic cells, and activated Dendritic cells showed increased infiltration in the high-risk group (Fig. 4A). Spearman correlation analysis revealed a positive correlation between the risk score and the infiltration of Plasma cells, activated Mast cells, and resting Dendritic cells, whereas a negative correlation was found with T cells gamma delta, B cells naive, and B cells memory (Fig. 4B). Further analysis using the ssGSEA algorithm showed that the low-risk group exhibited higher infiltration levels of various immune cells, including activated B cells, activated CD4 + and CD8 + T cells, activated dendritic cells, immature B cells, eosinophils, macrophages, neutrophils, and regulatory T cells (Fig. 4C). These differences in immune cell infiltration suggest alterations in immune function, prompting further comparison of immune function scores. Significant increases in immune function scores related to Antigen-Presenting Cell Co-Inhibition (APC-co-inhibition), C-C Chemokine Receptor (CCR), checkpoint regulation, cytolytic activity, human leukocyte antigen (HLA) expression, inflammation promotion, major histocompatibility complex class I (MHC-class I), parainflammation, T cell co-inhibition and co-stimulation, and Type I and II Interferon (IFN) responses were observed in the low-risk group (Fig. 4D). Additionally, stromal, immune, and ESTIMATE scores were significantly higher in the low-risk group compared to the high-risk group (Fig. 4E). These results indicate considerable differences in the immune microenvironment between the two groups, suggesting that patients with lower risk scores may experience a more active and robust immune response, which is linked to a better prognosis. Analysis of immune checkpoint gene expression between the two groups revealed higher expression levels of CTLA4, BTLA, and PDCD1 in the low-risk group (Fig. 4F). These results imply that the two-gene prognostic risk model could serve as a valuable tool for assessing NPC treatment responses and guiding clinical decision-making.

Fig. 4
figure 4

Identification of the immune landscape between the two risk groups. (A) The differences of immune cells infiltration between two risk groups in boxplots. (B) The correlations between immune cells infiltration and risk scores. (C-D) Box plot of the expression levels of 23 immune cell types 13 immune functions between two risk groups by ssGSEA. (E) Comparisons of the stromal, immune and ESTIMATE score in the two risk groups. (F) The expression levels of immune checkpoints between two risk groups.

Full size image

Correlation analysis between core genes and known radiation resistance-associated genes

Spearman correlation analysis also demonstrated that ABCC1 was positively correlated with known radioresistant genes (XRCC5, FASN, TRAF4, ZBTB33, PRKDC, HILPDA, KIF15, TRAF2, ATF5, CPT1A) in NPC, while CASP1 was negatively correlated with these genes (Fig. 5A). These results further support the reliability of these findings.

Assessing egcg’s impact on ABCC1 and CASP1 expression in NPC cell lines

To validate the effects of EGCG on the expression of ABCC1 and CASP1 in NPC, EGCG treatment experiments were conducted in NPC cell lines. Following EGCG treatment, a significant reduction in the expression of ABCC1 and a significant increase in the expression of CASP1 were observed (p < 0.05, Fig. 5B). These results suggest that ABCC1 and CASP1 may serve as potential targets for EGCG in the treatment of radioresistant NPC.

GeneMANIA-based functional association network analysis of 2 hub targets and enrichment analysis

Further, GMFA network analysis was performed on these two targets, resulting in a GMFA-ED comprising 42 genes. GO and KEGG enrichment analyses of the GMFA-ED provided a comprehensive overview of the BP, CC, MF, and KEGG pathways associated with EGCG’s potential targets in radioresistant NPC (Fig. 5C). GO enrichment analysis identified key BPs, including regulation of the inflammatory response, positive regulation of the inflammatory response, and interleukin-1 production. The CCs primarily involved the canonical inflammasome complex, nuclear envelope, and apical plasma membrane. MFs were enriched in peptidase regulator activity, peptidase activator activity, and endopeptidase regulator activity. KEGG pathway analysis post-GMFA indicated enrichment in the NOD-like receptor signaling pathway, cytosolic DNA-sensing pathway, legionellosis, ABC transporters, and Yersinia infection. These results offer valuable insights into the complex molecular mechanisms underlying EGCG’s potential antimetastatic effects in radioresistant NPC.

Fig. 5
figure 5

Validation of core targets and functional association network analysis. (A) Correlation between core targets and radiation resistance-associated genes. (B) Validation of the regulatory effect of EGCG on the expression of core targets in NPC cell lines. (C) GeneMANIA-based functional association network analysis of 2 hub targets and enrichment analysis.

Full size image

Molecular Docking and dynamics of EGCG with ABCC1 and CASP1

The docking results showed that EGCG had low binding energy values with CASP1(−7.7 kcal/mol) and ABCC1(−7.6 kcal/mol), indicating strong affinities. In the CASP1-EGCG complex, ILE155, MET156, LYS158, HIS404, and PHE401 form van der Waals interactions with EGCG, GLN142 forms carbon-hydrogen bonds, and ASP157 forms conventional hydrogen bonds. The binding free energy of −81.545 kJ/mol indicates a strong affinity (Fig. 6A). In the ABCC1-EGCG complex, LEU856, ALA863, PRO833, SER830, and TYR852 form van der Waals interactions with EGCG, MET829 forms conventional hydrogen bonds, and ARG867 exhibits π-cation and π-sulfur interactions. The binding free energy of −33.081 kJ/mol also suggests a high affinity (Fig. 6B). These results provide structural insights into the molecular interactions between EGCG and its core target proteins. The RMSD is an indicator used to measure the stability of molecular structures in molecular dynamics simulations. Small RMSD fluctuations in ligands and proteins indicate a stable system with minimal displacement or conformational changes. The equilibrium of the CASP1-EGCG and ABCC1-EGCG complexes was assessed using RMSD. The CASP1-EGCG complex reached equilibrium after 50 ns, with fluctuations around 5 Å, while the ABCC1-EGCG complex reached equilibrium after 20 ns, with fluctuations around 1.5 Å (Fig. 6C). This indicates that EGCG maintains high stability when bound to both target proteins.The Rg is used to evaluate the tightness of the architecture. The smaller values indicate a tighter protein. The Rg of the CASP1-EGCG and ABCC1-EGCG complexes remained relatively small throughout the simulation, indicating that these complexes maintained relatively compact and stable structures (Fig. 6D). SASA is the area of the protein surface accessible by the solvent, which measures the area of the protein surface in contact with the solvent. The results showed that the SASA of the CASP1-EGCG and ABCC1-EGCG complexes was relatively stable (Fig. 6E). The number of hydrogen bonds in the complexes was also analyzed. For CASP1-EGCG, the hydrogen bond count ranged from 0 to 7, with approximately 3 hydrogen bonds in most cases. For ABCC1-EGCG, the number ranged from 0 to 7, with approximately 4 hydrogen bonds in most cases. These results suggest strong hydrogen bonding interactions between EGCG and both target proteins (Fig. 6F). The RMSF reflects the fluctuation of the site structure of amino acid residues during the simulation. The smaller the value is, the more stable the residue at this position is. The results showed that the RMSF values of both complexes were relatively low (mostly below 3 Å), indicating low flexibility and high stability (Fig. 6G, H).

Fig. 6
figure 6

Molecular docking and molecular dynamics simulation studies. (A) Molecular docking of EGCG with CASP1 (B) Molecular docking of EGCG with ABCC1 (C) RMSD values of the two complexes. (D) Rg values of the two complexes. (E) SASA values of the two complexes. (F) Number of hydrogen bonds in the two complexes. (G-H) RMSF values of the two complexes.

Full size image

ABCC1 in single-cell RNA-sequencing data

The UMAP analysis displayed five distinct cell types, each color-coded, including B cells, cancer-associated fibroblasts (CAFs), epithelial cells, myeloid cells, and T cells (Fig. 7A). The expression and distribution of ABCC1 were visualized, revealing high expression levels in tumor cells (Fig. 7B). ABCC1 was found to be enriched in B cells, epithelial cells, myeloid cells, and T cells (Fig. 7C-D). Moreover, CASP1 was also found to be enriched in B cells, epithelial cells, myeloid cells, and T cells (Supplementary Figure S1).

Fig. 7
figure 7

Single-cell RNA-sequencing analysis. (A) Umap plot showing the clustering of different cell types. (B) Expression of ABCC1 in tumor tissues and normal tissues. (C-D) The expression distribution of ABCC1 across various cell types within tumor tissues.

Full size image

Materials and methods

Data acquisition

Drug target identification was conducted using the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP), SwissTargetPrediction, TargetNet, and GeneCards databases, resulting in 734 potential EGCG-related targets after deduplication. Protein names were converted to gene symbols using the UniProt database. Expression matrices and clinical data from the GSE48501, GSE150430, and GSE102349 datasets were downloaded from the GEO database. The GSE48501 dataset contains mRNA expression profiles from two radioresistant NPC CNE2-IR cell samples and two radiosensitive NPC CNE2 cell samples. CNE2-IR cells were derived from the poorly differentiated NPC cell line CNE2 by treating the cells with four rounds of sublethal doses of radiation. The GSE150430 dataset includes 15 primary NPC tumor samples and one normal sample. Additionally, the GSE102349 dataset includes complete progression-free survival (PFS) data for 88 out of 113 patients, which was utilized for further analysis.

Identification and enrichment analysis of radiotherapy-related differential genes

The “limma” package in R was employed to identify differentially expressed genes (DEGs) between the radioresistant and radiosensitive groups in the GSE48501 dataset. Genes with a threshold of |log2FoldChange| > 1 and an adjusted p-value < 0.05 were considered significant radiotherapy-related DEGs. Subsequently, intersection analysis was performed between EGCG-related targets and radiotherapy-related DEGs, resulting in a set of shared genes. These shared genes were identified as potential targets through which EGCG modulates the radiotherapy response in NPC. The STRING database. was used to construct protein-protein interaction (PPI) networks for potential targets. The networks were then visualized using Cytoscape software (version 3.6.1). Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) pathway enrichment analyses were conducted on the identified potential targets using the “cluster profile” R package to investigate the underlying molecular mechanisms, with statistical significance set at p.adjust < 0.05.

Construction and validation of the prognostic model

We developed a prognostic model by using a three-step approach. First, genes significantly related to prognosis were screened out through univariate analysis in the GSE102349 dataset. Second, least absolute shrinkage and selection operator (LASSO) Cox regression was conducted using the “glmnet” R package to prevent overfitting. Finally, multivariate Cox regression analysis was conducted to establish an optimized risk score model. The risk score was calculated using the following formula:

Risk score = expression of gene a × coefficient a + expression of gene b × coefficient b+ …… + expression of gene n × coefficient n7.

Subsequently, the samples were divided into low- and high-risk groups based on the risk score (median cut-off value). To evaluate the grouping accuracy, the PFS between the two risk subgroups was compared using Kaplan–Meier analysis. Time-dependent receiver operating characteristic (ROC) curves were generated using the “survivalROC” R package.

Immune analysis

Differences in immune cell infiltration between the two risk groups were analyzed and compared using single-sample Gene Set Enrichment Analysis (ssGSEA) and CIBERSORT. Spearman correlation analysis was used to explore the relationships between the risk model and immune status. The ESTIMATE algorithm was used to evaluate the immune score, stromal score, and ESTIMATE score. Additionally, the expression levels of immune checkpoint genes were compared between the two risk groups.

Correlation analysis between core genes and known radiotherapy resistance-associated genes

A comprehensive literature search on PubMed focused on radiotherapy resistance in NPC, identifying several genes associated with this resistance, including XRCC5, FASN, TRAF4, ZBTB33, PRKDC, HILPDA, KIF15, TRAF2, ATF5, and CPT1A8,9,10,11,12,13,14,15,16,17. Spearman correlation analysis was then conducted to examine the relationships between core genes and these identified radiotherapy resistance-related genes.

Transcriptome sequencing of NPC cell line HK1 based on EGCG treatment

The HK1 cell line, a widely utilized in vitro model for NPC research, was established in 1990 by a research team at the University of Hong Kong. Derived from the poorly differentiated squamous carcinoma tissue of a nasopharyngeal tumor in a patient from South China, this cell line closely mirrors the biological characteristics of clinical tumors. As such, it is highly suitable for investigations into the mechanisms underlying tumor development, drug sensitivity, and the regulation of metastasis. The method for processing NPC cell lines was as follows: EGCG (Sigma-Aldrich, St. Louis, USA) was dissolved in dimethyl sulfoxide (DMSO) to prepare a stock solution. The NPC cell line HK1 was seeded in 100 mm culture dishes at a density of 1.0 × 10⁵ cells/well. Once the cells reached the logarithmic growth phase, three independent experiments were set up. In the experimental group, EGCG at a concentration of 10 µg/mL (IC50) was used, while in the control group, DMSO at a volume ratio of 0.1% was applied. After 24 h of treatment, RNA was extracted for subsequent transcriptome sequencing18.

GeneMANIA-based functional association network analysis of core targets

To systematically identify genes associated with the core targets of EGCG, a novel approach within the GeneMANIA framework19 was employed. The GeneMANIA functional association (GMFA) method integrates co-expression, genetic interaction, and physical interaction data to identify genes pertinent to disease processes. The newly identified genes were then combined with the initial two core targets to form the GMFA-based Expanded Database (GMFA-ED). The GMFA-ED dataset underwent extensive GO and KEGG pathway enrichment analyses, highlighting the top five KEGG pathways and the top ten terms in each GO category (BP, CC, and MF).

In-depth analysis of EGCG interaction with core targets by molecular docking and molecular dynamics simulation studies

To investigate EGCG’s binding interactions with core target proteins, molecular docking was performed. The crystal structures of core target proteins from the RCSB PDB were processed in PyMOL (removal of water and ligands) and imported into AutoDock Tools v1.5.7 for hydrogenation, charge optimization, and non-polar hydrogen optimization. AutoDock Vina was used for docking via command-line with specified grid and GA parameters20. Results were visualized using PyMOL and Discovery Studio 2019, revealing the binding affinities and mechanisms of EGCG with key proteins. Additionally, 100 ns molecular dynamics (MD) simulations were conducted on the complex using Gromacs 2022 software. The Charmm 36 force field21 was applied for the protein, and Gaff2 was chosen for the ligand. The TIP3P water model was used for solvating the protein-ligand system, with a water box and periodic boundary of 1.2 nm22. Electrostatic interactions were managed using Particle Mesh Ewald (PME) and Verlet al.gorithms. The system underwent 100,000 equilibration steps under isothermal-isochoric (NVT) and isothermal-isobaric (NPT) conditions at 300 K and 1 bar, with a coupling constant of 0.1 ps and a 100 ps duration. Subsequently, a 100 ns molecular dynamics production run was conducted, with trajectories being saved every 2 fs20. Trajectory analysis was performed using Gromacs tools to calculate: root mean square deviation (RMSD), root mean square fluctuation (RMSF), number of hydrogen bonds, radius of gyration (Rg), solvent-accessible surface area (SASA), and potential of mean force (PMF) derived from the free energy landscape.

Single-cell RNA-sequencing analysis

The dataset GSE150430 was analyzed with the “Seurat” R package. Principal component analysis (PCA) assessed the significance of principal components across tissues or cells, and the datasets were visualized using uniform manifold approximation and projection (UMAP). The distribution of ABCC1 and CASP1 expression within cell clusters was also explored.

Statistical analysis

All statistical analyses were performed using R software (version 4.4.1) and GraphPad Prism (version 5.01). The t-test was applied to validate EGCG’s regulatory effect on core target expression in NPC cell lines. A p-value of < 0.05 was considered statistically significant for all tests (ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

Discussion

The combination of natural compound supplements with radiotherapy has demonstrated significant potential in enhancing therapeutic efficacy, offering new insights and strategies for the comprehensive treatment of NPC. EGCG has been shown to increase the radiosensitivity of radioresistant cells, positioning it as a promising radiosensitizer capable of substantially improving therapeutic outcomes in NPC treatment8,23. However, elucidating the mechanisms underlying EGCG’s role in NPC radiotherapy is essential for its clinical application. To address this gap, the present study employed a multidisciplinary approach integrating RNA sequencing, network pharmacology, and molecular docking techniques to explore the mechanisms driving EGCG’s effects in NPC radiotherapy and to identify key therapeutic targets.

This investigation identified 21 potential targets associated with EGCG’s effects in NPC radiotherapy. Functional enrichment analysis of these targets provided insights into how EGCG modulates radiotherapy resistance in NPC. GO analysis revealed significant enrichment in biological processes such as response to xenobiotic stimulus, cellular response to chemical stress, response to hypoxia, and response to decreased oxygen levels. Hypoxia is a critical factor contributing to distant metastasis, radiotherapy resistance, and poor prognosis in various tumors, including NPC24. The effectiveness of radiotherapy hinges on tumor cells’ ability to manage radiation-induced oxidative stress and DNA damage25. Oxygen availability enhances tumor sensitivity to radiotherapy, while radiosensitizers offer additional potential to improve radiotherapy outcomes26. Research has shown that the hypoxic microenvironment promotes resistance to chemotherapy and radiotherapy, fostering malignant tumor phenotypes such as enhanced DNA damage repair, increased invasiveness and metastatic potential, and decreased survival rates. These processes are regulated transcriptionally by HIF-1α. EGCG has been shown to directly destabilize HIF-1α27,28, suggesting that EGCG can counteract radiotherapy resistance in NPC cells and improve patient survival by modulating hypoxia-related mechanisms. KEGG analysis revealed that the candidate targets are predominantly enriched in pathways related to Salmonella infection, pathogenic Escherichia coli infection, lipid metabolism and atherosclerosis, Yersinia infection, osteoclast differentiation, gastric cancer, non-alcoholic fatty liver disease, pancreatic cancer, pertussis, and Kaposi sarcoma-associated herpesvirus infection. The invasion of host cells by Salmonella and other intracellular pathogens triggers macroautophagy (hereafter referred to as autophagy), a metabolic process that transports cytoplasmic components to lysosomes for degradation29. Autophagy is a critical defense mechanism for clearing intracellular pathogens and plays an essential role in regulating immune responses30. Additionally, autophagy is central to tumorigenesis and cancer progression, enabling tumor cells to withstand tissue damage caused by radiotherapy and chemotherapy, which contributes to acquired resistance to these treatments31. EGCG has been shown to significantly inhibit Salmonella, pathogenic Escherichia coli, and Yersinia32, suggesting that EGCG may enhance the radiosensitivity of tumor cells by inhibiting pathogen infections and reducing autophagy. Furthermore, macrophage autophagy serves as an anti-atherogenic mechanism, maintaining lipid homeostasis by promoting the catabolism of cytosolic lipid droplets33. Abnormal lipid metabolism is pivotal in tumor development and radiotherapy resistance, a phenomenon confirmed in NPC13. EGCG has been shown to ameliorate abnormal lipid metabolism and fatty liver, as well as to mitigate atherosclerosis34,35. These findings suggest that EGCG may combat tumor radiotherapy resistance through multiple pathways, providing strong support for our study.

In this study, two core targets (ABCC1, CASP1) were identified as being associated with tumor prognosis, aligning with prior research. ATP-binding cassette subfamily C member 1 (ABCC1) is a critical component of the ABC transporter superfamily36. As a multispecific efflux transporter, ABCC1 is localized on the plasma membrane, where it facilitates the extrusion of chemotherapeutic drugs, preventing the accumulation of toxic drug levels within the cell and conferring resistance to chemotherapeutic agents in vitro37. Beyond its drug efflux activity, ABCC1 also transports glutathione, inflammatory mediators (such as leukotrienes and prostaglandins), and bioactive lipids (including sphingosine-1-phosphate and lysophosphatidylinositol), potentially playing a key role in cancer initiation, progression, invasion, and radiotherapy resistance, independent of its drug efflux function38. Further studies have demonstrated that increased ABCC1 expression significantly enhances radiotherapy resistance in breast cancer cells39. In summary, high expression of ABCC1 is closely linked to poor prognosis in cancers, including NPC40. These findings corroborate our results and underscore the pivotal role of ABCC1 in tumor resistance and prognosis. Additionally, scRNA-seq analysis revealed significant differences in ABCC1 distribution across various cell types. Based on these observations, it is hypothesized that alterations in cell type interactions and the distinct expression patterns of ABCC1 may contribute to radiotherapy resistance in NPC. CASP1, a key member of the CASP protein family, is involved in regulating a variety of cellular processes, including apoptosis, inflammatory responses, and necrosis. Specifically, CASP1 modulates immune responses and disease pathogenesis by activating the inflammasome41. Moreover, pyroptosis induced by CASP1 acts as a natural immune defense against intracellular bacterial infections42. In various cancers, such as lung adenocarcinoma, and breast cancer, CASP1 expression is closely associated with prognosis, with higher levels typically correlating with better outcomes43,44. Silencing CASP1 in non-small cell lung cancer cells promotes tumor growth and invasion41. Therefore, CASP1 plays a significant role in tumor initiation, progression, and prognosis. These findings strongly support our research, further validating its scientific merit and reliability.

This study investigated the potential interactions between two core genes and EGCG using molecular docking. The results revealed that EGCG exhibits a strong binding affinity with both ABCC1 and CASP1. The molecular dynamics simulation results strongly supported the stability of the docking poses and the predicted binding interactions of EGCG with CASP1 and ABCC1. The RMSD analysis showed that both complexes reached equilibrium, with the ABCC1-EGCG complex exhibiting particularly low fluctuations (around 1.5 Å), indicating high stability. Additionally, the relatively small and stable Rg values further confirmed the compactness and stability of the complexes. The stable SASA and consistent hydrogen bond counts (3–4 bonds on average) highlighted the strong and persistent interactions between EGCG and the target proteins. Moreover, the low RMSF values (mostly below 3 Å) indicated minimal flexibility and high stability of the binding sites. Overall, these results validated the docking poses and binding interactions, demonstrating that EGCG maintained stable and strong interactions with both CASP1 and ABCC1. This suggests that its biological effects may be mediated through interactions with these targets. This finding is corroborated by research from Jokar MH and colleagues, who demonstrated that EGCG significantly reduces the expression of ABCC1 in leukemia HL60 cells, enhancing drug resistance45. Collectively, these observations support the hypothesis of a functional interaction between EGCG and these core genes, with potential biological implications. Future studies should further elucidate the mechanisms underlying these interactions in tumor cells and explore their potential in addressing radiotherapy resistance in NPC.

This study has several limitations. First, the evaluation of the non-target effects of EGCG is limited, so the potential roles of other non-target proteins cannot be completely ruled out, and future research will focus on identifying more specific targets of EGCG. Second, molecular docking and dynamics simulations are mainly based on static structure predictions, which may not fully capture the dynamic characteristics and potential allosteric effects of the interactions in a biological environment. Third, although EGCG’s regulatory effects on ABCC1 and CASP1 expression have been confirmed, its direct contribution to changes in radiotherapy resistance remains to be verified. Future research will explore these mechanisms through more in-depth in vitro and in vivo experiments to provide more precise strategies for NPC radiotherapy. Finally, the single-cell dataset used has a limited number of samples, which may affect the generalizability and robustness of the results. Future studies should validate the findings using larger and more diverse datasets to improve the accuracy and reliability of the conclusions.

Conclusion

By screening core genes involved in EGCG-mediated improvement of radiotherapy resistance in NPC, constructing a risk prediction model, and integrating multidimensional approaches such as molecular docking, molecular dynamics simulations, single-cell sequencing, and experimental validation, this research highlights the critical roles of these genes in NPC radiotherapy and their potential clinical value. Future work will aim to explore the therapeutic potential of these core genes as EGCG targets in NPC radiotherapy, offering new avenues for the development of personalized precision medicine strategies.