Mechanistic evaluation of pachymic acid in breast cancer via network pharmacology and in vitro validation Authors: Yibing Zhou, Zhuotao Yang, Yuchen Duan, Yijie Cheng, Wenqian Xu, Bo Wang, Jingyuan Xu, Da Qian, Chaoshen Wu Published: 2026 ABSTRACT BackgroundBreast cancer(BC) is a major cause of cancer-related mortality in women worldwide, with current therapies often limited by toxicity and drug resistance. Pachymic acid (PA), a natural compound, demonstrates antitumor potential with relatively low systemic toxicity. However, its mechanism in triple-negative breast cancer (TNBC) remains unclear.MethodsThis study aims to deeply explore the anti-tumor effect of PA on TNBC. The intersection targets of PA and BC were screened through multiple databases, and the key genes were determined through TCGA survival analysis. The potential mechanism was simulated and studied through methods such as network pharmacology, molecular docking and molecular dynamics. Cell proliferation ability was evaluated by CCK-8 assay and colony formation assay. Cell migration and invasion abilities were detected by scratch assay and Transwell assay respectively. Gene expression levels were analyzed by qPCR and Western Blot, and the degree of lipid peroxidation was evaluated by malondialdehyde (MDA) assay.ResultsThrough multi-database screening and TCGA survival analysis, PTGS2, NOS2 and ABCB1 were identified as key targets. The TNBC cell line MDA-MB-231 was used as the experimental model. PA significantly inhibited the proliferation, migration and invasion of MDA-MB-231 cells. Mechanistic studies demonstrated that PA concurrently upregulates PTGS2 and NOS2 expression while downregulating ABCB1. Notably, the upregulation of PTGS2 was significantly correlated with increased lipid peroxidation, suggesting its crucial role in ferroptosis. The coordinated upregulation of NOS2 and downregulation of ABCB1 further enhanced cellular sensitivity to ferroptosis.ConclusionThis study is the first to reveal that PA induces ferroptosis in TNBC cells by regulating the PTGS2/NOS2/ABCB1 signaling axis, providing new potential targets for TNBC treatment. However, this study has limitations, as validation has been conducted only at the cellular level and not yet confirmed in animal models or clinical samples. Future research should validate this mechanism in more TNBC models and across different molecular subtypes of breast cancer to promote the clinical application of PA.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12906-026-05420-4. Introduction Globally, breast cancer (BC) is the most commonly diagnosed cancer and the major cause of cancer-associated deaths among women. Approximately one million new cases are reported annually, representing nearly 18% of all female cancers, with incidence rates rising significantly with age [1, 2]. Significant disparities exist in BC survival rates globally, with a 5-year survival rate of around 80% in developed countries and < 40% in developing regions [3]. In China, cancer has emerged as the primary cause of mortality, as reported by the National Cancer Center [4]. Globally, BC is a prevalent cancer, and its treatment options vary based on molecular subtypes. They are generally divided into two main categories: locoregional control (surgery and radiotherapy) and systemic therapies, including chemotherapy, endocrine therapy for hormone receptor-positive disease, and HER2-targeted treatment [5–7]. While current treatments achieve high cure rates in early-stage non-metastatic cases, advanced-stage disease with distant metastasis remains mostly incurable [2, 8].The emergence of drug resistance in breast cancer treatment is the main cause of treatment failure. Overcoming drug resistance has become the core driving force of new drug development [9]. Recent research emphasizes the potential of Traditional Chinese Medicine (TCM) in cancer treatment, especially in suppressing tumor growth, preventing recurrence and spread, reversing drug resistance, and enhancing immune responses. TCM has also shown effectiveness in reducing treatment-related heart toxicity. Its therapeutic benefits, based on interactions across multiple targets and pathways, support a comprehensive and integrative approach to cancer care [10, 11]. Poria cocos, a traditional medicinal fungus widely used in Oriental medicine, has been extensively researched for its bioactive properties [12]. Pachymic acid (PA), a lanostane-type triterpenoid and one of the main active compounds in Poria cocos, exerts its pharmacological effects by inhibiting signaling pathways such as NF-κB and MAPK, emonstrating anti-inflammatory, antioxidant, and antitumor properties. Studies have shown that PA can inhibit tumor cell migration and invasion in various cancers [13–15]. The toxicity of PA is still under exploration. In animal experiments, no obvious toxic effects were observed in the liver, kidney and other important organs of mice [16]. Network pharmacology is an interdisciplinary field combining systems biology with network informatics, and it has become increasingly prominent in drug discovery recently [17]. In systems biology, complex disorders such as cancer do not originate from single gene mutations but from disturbances in biological networks caused by multiple genetic changes. This field examines these network interactions to evaluate the comprehensive effects of multi-component therapies. It identifies molecular targets for bioactive compounds, enhances treatment effectiveness, and decreases adverse drug reactions [18]. Molecular docking is a computational modeling method that simulates interactions between small molecules and target proteins at the atomic level. This approach predicts ligand-receptor binding conformations and estimates their binding affinities, providing a cost-effective and precise tool for rational drug design and mechanistic studies [19]. This study employed an integrated approach of network pharmacology and experimental validation to investigate the active components of PA and their molecular mechanisms in breast cancer treatment. Potential PA targets were identified using the PubChem and SwissTargetPrediction databases. Breast cancer-associated targets were retrieved from the GeneCards and OMIM databases. The intersection of these target sets yielded PA-breast cancer common targets, which were used to construct and visualize a protein-protein interaction (PPI) network. Key hub genes (PTGS2, NOS2, ABCB1) were selected from the PPI network and TCGA database for survival prognosis analysis. Subsequent in vitro experiments were conducted to validate the network pharmacology-predicted mechanisms underlying PA’s effects on breast cancer. Our findings indicate that PA modulates the expression of key proteins PTGS2, NOS2, and ABCB1, effectively inhibiting breast cancer cell proliferation, migration, and invasion. Crucially, PA demonstrates low cytotoxicity against normal breast epithelial cells. Its anti-tumor efficacy exhibits a positive correlation with both concentration and treatment duration, highlighting its potential as a therapeutic candidate for breast cancer. Materials and methods Screening of the main components and targets of pachymic acid Identify genes that overlap between PA potential drug targets and breast cancer-associated targets. The chemical structure of pachymic acid was obtained from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/), a comprehensive chemical information repository maintained by the National Center for Biotechnology Information (NCBI) under the United States National Institutes of Health (NIH). The predicted biological targets of pachymic acid were identified using the SwissTargetPrediction database (http://www.swisstargetprediction.ch/), an online tool designed to predict potential protein targets of small molecules based on chemical similarity and biological activity data. Potential drug targets of pachymic acid were predicted based on its structural formula. Moreover, disease-associated targets related to BC were obtained from the Online Mendelian Inheritance in Man (OMIM) database and the GeneCards database. A Venn diagram was constructed to integrate the predicted targets of PA with the pathological targets of BC, thus identifying potential mechanisms through which PA may exert therapeutic effects against BC. Functional analysis of GO and KEGG Gene targets of PA relevant to BC, identified in Sect. "Screening of the main components and targets of pachymic acid", were uploaded to the DAVID 6.8 online platform for GO and KEGG pathway enrichment analyses. To address multiple comparisons, adjustments were applied using the false discovery rate (FDR) method. GO enrichment analysis was conducted across three main categories: biological processes (BP), cellular components (CC), and molecular functions (MF). The top 10 enriched GO terms and the top 20 KEGG pathways, ranked by p-value, were visualized with bar plots and bubble charts. All visualizations were created using the Bioinformatics online analysis tool (http://www.bioinformatics.com.cn/). Intersection target acquisition and PPI network construction Drug-related and disease-associated targets were uploaded into the Venny 2.1 online tool to plot a Venn diagram, identifying overlapping targets as potential therapeutic targets of PA in BC. These intersecting targets were submitted to the STRING database (https://string-db.org/) [20] to construct a PPI network. The resulting interaction data were further examined and illustrated via Cytoscape software (v3.9.1) to isolate core functional targets through network topology analysis. Component-target-disease network map Some key signaling cascades and their target genes, identified through KEGG enrichment analysis, were integrated into Cytoscape 3.9.1 to develop a visual network illustrating the interactions among PA components, target genes, and breast cancer-related pathways. Data collection Raw mRNA expression data for BCr were retrieved from the TCGA database (https://portal.gdc.cancer.gov/), a comprehensive repository of genomic and transcriptomic cancer data. A total of 1,231 samples were obtained, comprising 113 normal tissue samples and 1,118 tumor tissue samples. A volcano plot was constructed using these data. Survival analysis of 33 overlapping genes based on their median expression levels revealed statistically significant differences in NOS2, PTGS2, and ABCB1 between high and low expression groups. Furthermore, the study compared these three genes in normal tissues versus breast cancer tissues. Immune infiltration analysis The CIBERSORT algorithm was used to evaluate immune cell populations in the tumor microenvironment (TME). Based on support vector regression, this method deconvolves the gene expression matrix to estimate the proportions of 22 different immune cell types, identified by 547 gene expression biomarkers. These cell types include various subsets of T cells, B cells, plasma cells, and myeloid cells. In this analysis, CIBERSORT was used to examine transcriptomic data, allowing for the determination of immune cell proportions and examining the correlation between immune cell infiltration (ICI) and transcription patterns. miRNA analysis miRNAs are short, non-coding RNA molecules, mainly responsible for controlling gene expression by promoting mRNA degradation or suppressing translation. To explore the regulatory role of miRNAs in the expression of key genes, miRNA-gene interactions were identified using the miRcode database (http://www.mircode.org/index.php). The resulting miRNA-target gene interaction network was then developed and detected via Cytoscape software. Identify miRNAs associated with the regulation of NOS2, PTGS2, and ABCB1. Molecular docking The structure of proteins (3D) encoded by key genes was obtained from the RCSB Protein Data Bank (https://www.rcsb.org/) and the AlphaFold database. The chemical structure of PA was retrieved from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/). Molecular docking simulations were performed using PyMOL and AutoDock software to evaluate the binding affinities and interaction conformations between PA and target proteins. Chemicals and reagents Pachymic acid (PA; CAS No. 29070-92-6, Cat. No. HY-N0371, purity > 99%) was sourced as a DMSO-soluble compound and stored at − 80 °C. An equivalent volume of DMSO was added to the control group (0 µM PA) medium to eliminate potential experimental errors between the control and treatment groups. Primary antibodies against NOS2 (Cat. No. 22226-1-AP, 1:1000), PTGS2 (Cat. No. 12375-1-AP, 1:2000), ABCB1 (Cat. No. 22336-1-AP, 1:1000), and GAPDH (Cat. No. 60004-1-Ig, 1:1000) were procured from Proteintech (China). The secondary antibody, Goat Anti-Rabbit IgG H&L conjugated to horseradish peroxidase (HRP; Cat. No. A0208, 1:1000), was obtained from Beyotime (China). Cell culture Human BC cell line (MDA-MB-231) and the human normal breast epithelial cell line (MCF-10 A) were used for in vitro experiments. Both cell lines were grown in 10% FBS-enriched DMEM and kept at 37 °C with 5% CO2 humidified incubator. Cell viability assay In order to evaluate the toxicity of PA on MDA-MB-231 cells and screen the effective concentration of subsequent experiments, CCK-8 cell viability was detected. Cell viability was measured using the Cell Counting Kit-8 (CCK-8) assay. Approximately 2 × 103 cells were plated per well in 96-well plates and allowed to grow at 37 °C with 5% CO₂ for 48 h. Cells were exposed to PA (0, 10, 20, 30, 40, 50, and 60 µM) and kept for 24, 48, and 72 h of incubation. After treatment, the medium was removed, and 100 µL of fresh medium with 10% CCK-8 solution was added to all wells. Plates were kept for 2 h at 37 °C, and absorbance (OD) was detected at 450 nm using a microplate reader. All experiments were repeated at least three times, with three samples per group. Dose-response curves were generated via GraphPad Prism 8.0 (San Diego, USA), and IC50 values were calculated via nonlinear regression analysis. Clonogenic assay To investigate whether PA weakens the long-term self-renewal and proliferative potential of MDA-MB-231 cells, we performed a clonogenic assay. MDA-MB-231 cells in the exponential (log) phase were seeded (500 cells/well) into 6-well plates. After cell attachment, media containing different concentrations of PA were added, with untreated wells serving as negative controls. After a 9-day incubation period, culture media were discarded, and cells were washed with PBS. Fixation was performed using 1 ml of 4% paraformaldehyde (PFA) for 40 min at 25 °C. Cells were then stained with 0.1% crystal violet for 50 min, rinsed with PBS, and air-dried. Colonies were imaged and quantified, and mean values were calculated. The experiment was conducted in triplicate. Scratch assay To evaluate the effect of PA on MDA-MB-231 cell migration, we performed scratch test and observed the wound closure of cells at 0 and 24 h. MDA-MB-231 cells in the exponential (log) phase were plated (5000 cells/well) into 6-well plates. When they reached approximately 95% confluence, linear scratches were induced via a sterile pipette tip (10 µL). After discarding the medium and washing with PBS to remove cellular debris, cells were cultured in DMEM under standard conditions. Cell migration into the scratch area was evaluated by capturing images at 0 and 24 h with a light microscope at 100x magnification. Multiple fields were analyzed to evaluate wound closure (scratch gap). All experiments were repeated at least three times, with three samples per group. Cell invasion assay In order to more precisely explore whether PA can inhibit the key invasion step of MDA-MB-231 cells through the basement membrane, we carried out a matrigel based cell invasion test. Cell invasion capability was evaluated using Transwell chambers pre-coated with Matrigel and placed in a 24-well plate. The upper (top layer) chamber was filled with 200 µL of FBS-free medium and kept at 37 °C for 2 h. MDA-MB-231 cells in log phase were starved for 12 h in FBS-free medium. Cell suspensions (1 × 105 cells/ml) with different concentrations of PA were added to the top layer. The lower chamber (bottom layer) contained 600 µL of 20% FBS-enriched medium as a chemoattractant. Each experimental condition was tested in triplicate. After 24 h, supernatants were discarded, and cells were washed with PBS. Fixation was performed using 4% paraformaldehyde (PFA) for 30 min, followed by crystal violet (0.1%) staining for 50 min. When dried, five randomly selected microscopic fields per well were imaged. All experiments were repeated at least three times, with three samples per group. All migrated cells were counted, and the invasion rate was measured using the formula:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm{Cell}\:\mathrm{invasion}\:\mathrm{rate}\left(\%\right)=\left(\mathrm{Total}\:\mathrm{invasive}\:\mathrm{cells}\:\mathrm{in}\:\mathrm{the}\:\mathrm{treated}\:\mathrm{group}\:/\:\mathrm{Total}\:\mathrm{invasive}\:\mathrm{cells}\:\mathrm{in}\:\mathrm{the}\:\mathrm{control}\:\mathrm{group}\:\times\:100\%\right).\:\text{The assay was repeated in triplicate for consistency}.$$\end{document} Western blot Total content of protein was isolated from MDA-MB-231 cells treated with different concentrations of PA. The concentrations of isolated protein were quantified, and samples were run on SDS-PAGE, then transferred to polyvinylidene difluoride (PVDF) membranes (blots). These blots were blocked with 5% non-fat milk and incubated at 4 °C for 24 h with primary antibodies against NOS2 (1:1000), PTGS2 (1:2000), ABCB1 (1:1000), and GAPDH (1:1000). After washing, blots were incubated with HRP-linked secondary antibodies. Protein bands were detected using ECL, and band intensities were measured using grayscale analysis with ImageJ software to determine relative protein expression levels. All experiments were repeated at least three times, with three samples per group. Gene expression analysis via qRT-PCR MDA-MB-231 cells were treated with PA at 0, 20, and 40 µmol/L concentrations for 48 h. Total RNA was isolated, and cDNA was synthesized using a Vazyme kit as per the manufacturer’s protocol. The qRT-PCR was performed under the following conditions: initial denaturation at 95 °C for 30 s, followed by 40 amplification cycles of 95 °C for 15 s and 60 °C for 30 s. GAPDH served as the internal reference control, and gene expression was quantified using the 2⁻ΔΔCt method. All experiments were repeated at least three times, with three samples per group. All primer sequences (Forward: A and Reverse: R) used were as follows:ABCB1:F: CAGCCATACAGCAAATCCTTGR: CAAATGTGATCTGGATGTCAACPTGS2:F: CAGCCATACAGCAAATCCTTGR: CAAATGTGATCTGGATGTCAACNOS2:F: GGAGGAGCTGATGGAGTAGTAGCGGR: CTACCTACCTGGGGAACACCTGGGGAPDH:F: AGAAAAACCTGCCAAATATGATGACR: TGGGTGTCGCTGTTGAAGTC ABCB1:F: CAGCCATACAGCAAATCCTTGR: CAAATGTGATCTGGATGTCAAC ABCB1:F: CAGCCATACAGCAAATCCTTGR: CAAATGTGATCTGGATGTCAAC F: CAGCCATACAGCAAATCCTTG F: CAGCCATACAGCAAATCCTTG R: CAAATGTGATCTGGATGTCAAC R: CAAATGTGATCTGGATGTCAAC PTGS2:F: CAGCCATACAGCAAATCCTTGR: CAAATGTGATCTGGATGTCAAC PTGS2:F: CAGCCATACAGCAAATCCTTGR: CAAATGTGATCTGGATGTCAAC F: CAGCCATACAGCAAATCCTTG F: CAGCCATACAGCAAATCCTTG R: CAAATGTGATCTGGATGTCAAC R: CAAATGTGATCTGGATGTCAAC NOS2:F: GGAGGAGCTGATGGAGTAGTAGCGGR: CTACCTACCTGGGGAACACCTGGG NOS2:F: GGAGGAGCTGATGGAGTAGTAGCGGR: CTACCTACCTGGGGAACACCTGGG F: GGAGGAGCTGATGGAGTAGTAGCGG F: GGAGGAGCTGATGGAGTAGTAGCGG R: CTACCTACCTGGGGAACACCTGGG R: CTACCTACCTGGGGAACACCTGGG GAPDH:F: AGAAAAACCTGCCAAATATGATGACR: TGGGTGTCGCTGTTGAAGTC GAPDH:F: AGAAAAACCTGCCAAATATGATGACR: TGGGTGTCGCTGTTGAAGTC F: AGAAAAACCTGCCAAATATGATGAC F: AGAAAAACCTGCCAAATATGATGAC R: TGGGTGTCGCTGTTGAAGTC R: TGGGTGTCGCTGTTGAAGTC Statistical analysis Data was statistically examined via R software (v4.2.2). Statistical variances between groups were evaluated using one-way ANOVA, followed by Tukey’s post hoc test for pairwise comparisons. Results with a p-value < 0.05 were interpreted as statistically significant. Result Bioinformatics characteristics of intersecting genes Potential pharmacological targets of PA were identified using PubChem and SwissTargetPrediction, while BC-related genes were retrieved from GeneCards and OMIM. After removing duplicates, a comparative analysis identified 33 common targets (Fig. 1A). GO classification of these targets revealed 281 enriched terms, including 181 in biological processes (BP), 33 in cellular components (CC), and 67 in molecular functions (MF). The enriched BPs mainly involved responses to stress and metabolic regulation of organic compounds. The CC terms, “response to organic substances” and “cellular stress response,” suggested a significant role within intracellular membrane-associated organelles. MF categories, especially “membrane-bound organelle” and “intracellular organelle,” emphasized the redox activity and catalytic functions of the involved genes (Fig. 1B-E). KEGG-based pathway enrichment identified a total of 78 signaling cascades, with the top 20 detected for downstream analysis. Among the enriched pathways, commonly included “Drug metabolism-cytochrome P450” and “Retinol metabolism,” suggesting that the genes are involved in metabolic processes related to xenobiotic clearance and vitamin A processing. Moreover, pathways such as “IL-17 signaling” and “TNF signaling” showed significant gene enrichment, indicating that immune and inflammatory responses are key factors in regulating the expression of these genes (Fig. 1F). The enriched genes were systematically categorized into BP, CC, and MF to illustrate their functions, with BP-related genes being the most prevalent, highlighting their key role in coordinating complex physiological responses (Fig. 1G). These gene products are involved in oxidative stress responses, metabolic processes, and immune signaling. Their activities within cellular compartments include catalysis and redox mechanisms. Fig. 1Functional enrichment, pathway network and hierarchical regulation of PPI of 33 core targets. A Venn Diagram. B-E GO and KEGG enrichment. F KEGG enriched pathways for differential genes. G Differential gene classification. H Protein interaction network Functional enrichment, pathway network and hierarchical regulation of PPI of 33 core targets. A Venn Diagram. B-E GO and KEGG enrichment. F KEGG enriched pathways for differential genes. G Differential gene classification. H Protein interaction network A PPI network consisting of 33 overlapping targets was developed using data from the STRING database, and a graphical visualization was created with Cytoscape. The left side of the visualization displays the overall connectivity among proteins encoded by differentially expressed genes (DEGs), while the right side highlights a central cluster (indicated by an orange ring) representing a key regulatory subnetwork. This PPI structure reflects a hierarchical regulatory system, where hub proteins act as primary control points, often involved in transcriptional regulation or key signaling events, while surrounding proteins help carry out downstream cellular functions. The increasing density of interactions indicates a complex network of protein relationships, emphasizing the formation of coordinated functional units that regulate key biological processes such as immune response and metabolic control (Fig. 1H). Immune microenvironment, prognosis and therapeutic targets in breast cancer A volcano plot was generated to visualize variations in gene expression between tumor and non-tumor tissues. Genes upregulated in tumor samples are shown in red, while those with reduced expression are depicted in green. Significant dysregulation observed in tumor tissues provided a basic dataset for identifying candidate genes of interest (Fig. 2A). To further identify key genes influencing breast cancer, we performed survival analysis based on the median expression values of the 33 overlapping genes obtained from the previous step. The results demonstrated that NOS2, PTGS2, and ABCB1 showed statistically significant differences (P < 0.05) between high and low expression groups, indicating their strong prognostic value in breast cancer. Therefore, these three genes were selected as key targets for our subsequent investigation (Fig. 2B-D). Comparative profiling of ICI between tumor and normal tissues revealed significant differences in the abundance of various immune populations, such as neutrophils and macrophages. The distinct distribution patterns (red: tumor; blue: normal) highlighted substantial remodeling of the immune cell composition within the TME (Fig. 2E-G). The immune microenvironment is primarily composed of immune cells, extracellular matrix, various growth factors, inflammatory factors, and unique physical-chemical properties, which collectively significantly influence disease diagnosis and clinical treatment sensitivity. Through analysis of the relationship between key genes and immune infiltration in the breast cancer dataset, we further investigated the potential molecular mechanisms by which these key genes influence breast cancer progression, and demonstrated both the immune cell content and correlations among different immune cells (Fig. 2H-I). The results demonstrated significant differences in 14 immune cell types between the two groups, including naive B cells, eosinophils, and M0 macrophages (Fig. 2J). Furthermore, all key genes demonstrated significant correlations with immune cells. Specifically, NOS2 showed significant positive correlations with M0 macrophages, monocytes, and activated dendritic cells, while exhibiting significant negative correlations with resting CD4 memory T cells and resting mast cells (Fig. 2K). PTGS2 demonstrated significant positive correlations with activated dendritic cells, activated mast cells, and T follicular helper cells, but significant negative correlations with regulatory T cells (Tregs) and resting mast cells (Fig. 2L). ABCB1 displayed significant positive correlations with naive B cells, resting CD4 memory T cells, and plasma cells, whereas it showed significant negative correlations with M0 macrophages, regulatory T cells (Tregs), and M1 macrophages (Fig. 2M). Fig. 2Multi-dimensional analysis of immune microenvironment remodeling, survival prognosis and therapeutic targets in breast cancer. A Volcano map of differential gene screening. B-D ABCB1, NOS2 and PTGS2 related survival analysis. E-G Differences in immune cell infiltration of ABCB1, NOS2 and PTGS2. H Immune cell heat map. I Immune cell correlation heat map. J. Distribution of immune cell subsets. K-M Association of ABCB1, NOS2 and PTGS2 with immune cells Multi-dimensional analysis of immune microenvironment remodeling, survival prognosis and therapeutic targets in breast cancer. A Volcano map of differential gene screening. B-D ABCB1, NOS2 and PTGS2 related survival analysis. E-G Differences in immune cell infiltration of ABCB1, NOS2 and PTGS2. H Immune cell heat map. I Immune cell correlation heat map. J. Distribution of immune cell subsets. K-M Association of ABCB1, NOS2 and PTGS2 with immune cells Key gene-mediated immune molecular network and TME regulation We analyzed the correlations between the three key genes and various immune factors, including immunosuppressive factors, immunostimulatory factors, chemokines, and their receptors. These analyses revealed strong correlations between the key genes and immune cell infiltration levels, suggesting their crucial roles in the immune microenvironment. A correlation analysis showed a positive network between the expressions of ABCB1, NOS2, and PTGS2 and most chemokines (red dots)(Fig. 3A). Immunosuppressive markers, such as PDCD1LG2 and LAG3, also had strong positive associations with the same genes(Fig. 3B). Immune stimulators such as CD27 and ICOS show positive correlations with ABCB1, NOS2, and PTGS2 (red dots), while exhibiting negative correlations with specific stimulatory ligands like TNFSF14 (blue dot)(Fig. 3C). Most major histocompatibility complex (MHC) components, especially class II molecules, showed positive correlations with key genes (red dots)(Fig. 3D). Moreover, receptors like CXCR3 and CCR7 were also positively correlated with these genes(Fig. 3E). Fig. 3Key gene mediated immune molecular network and tumor microenvironment regulation. A Chemokine association diagram. B Immunoassay inhibitory factor association diagram. C Immune stimulus factor association diagram. D MHC molecular association diagram. E Immune receptor association diagram Key gene mediated immune molecular network and tumor microenvironment regulation. A Chemokine association diagram. B Immunoassay inhibitory factor association diagram. C Immune stimulus factor association diagram. D MHC molecular association diagram. E Immune receptor association diagram Tumor lncRNA regulatory network and drug targeting potential Using the GeneCards database (https://www.genecards.org/), we obtained regulatory genes associated with ferroptosis. By comparing lncRNA expression profiles between non-tumor (blue) and tumor (yellow) tissues, we analyzed intergroup differences in gene expression and identified 14 genes with significant differences, including AIFM2, GPX4, and SLC7A11 (Fig. 4A). Subsequently, we performed correlation analysis between the key genes (ABCB1, NOS2, PTGS2) and ferroptosis-related regulatory genes. The results demonstrated significant correlations between the expression levels of key genes and regulatory genes, with PTGS2 showing a significant positive correlation with ACSL4(cor = 0.487)and ABCB1 exhibiting a significant negative correlation with SLC3A2༈cor = − 0.294༉(Fig. 4B). Furthermore, we conducted reverse prediction of the three key genes using the miRcode database, identifying 70 miRNAs that formed 119 mRNA-miRNA relationship pairs. These interactions were visualized using Cytoscape. Notably, PTGS2 demonstrated a significant number of miRNA interactions, suggesting its potential role as a hub regulator within the network (Fig. 4C). We performed molecular docking on the key genes. The selected protein and compound pairs for the key genes were as follows: NOS2: 3e7g-Pachymic acid; PTGS2: 5f19-Pachymic acid; ABCB1: P08183-Pachymic acid. Molecular docking results showed that the binding energy for NOS2: 3e7g-Pachymic acid was − 5.76 kcal/mol; for PTGS2: 5f19-Pachymic acid was − 4.32 kcal/mol; and for ABCB1: P08183-Pachymic acid was − 2.23 kcal/mol (Fig. 4D-F). Fig. 4Tumor lncRNA remodeling, co-expression network, ceRNA hub and drug targeting potential. A Genetic expression differences. B Gene co-expression and differential analysis. C CeRNA network analysis. D-F Molecular docking Tumor lncRNA remodeling, co-expression network, ceRNA hub and drug targeting potential. A Genetic expression differences. B Gene co-expression and differential analysis. C CeRNA network analysis. D-F Molecular docking Inhibitory role of PA on MDA-MB-231 cellular growth To evaluate the potential cytotoxicity of PA (Fig. 5A) on normal breast epithelial cells, MCF-10 A cells were treated with PA for 24, 48, and 72 h. Cell viability via the CCK-8 assay depicted that PA concentrations above 40 µM remarkably affected the viability of normal breast cells (Fig. 5B). In comparison, MDA-MB-231 BC cells showed a significant reduction in viability after PA exposure, with the inhibitory effect depending on both concentration and duration (Fig. 5C). Based on these observations, the 0, 20, and 40 µM concentrations of PA at the 48 h time point were selected for further experiments. A dose-dependent decrease in colony formation was observed, indicating that PA effectively suppressed the clonogenic potential of MDA-MB-231 cells, with stronger effects at higher concentrations (Fig. 5D). In the wound healing assay, all groups initially showed similar scratch widths at 0 h. However, after 48 h, the scratch area in the NC group was fully closed, while the PA-treated groups (20 and 40 µM) displayed incomplete wound closure. However, the 40 µM group showed minimal healing, demonstrating that PA substantially impaired the migratory capacity of BC cells in a dose-dependent manner (Fig. 5E). Transwell invasion assays, stained with crystal violet, showed that the NC group had a high density of invasive cells, whereas the number of cells that migrated through the membrane was considerably lower in the 20 and 40 µM PA-treated groups (Fig. 5F). Western blot also confirmed that protein levels of PTGS2 and NOS2 increased with higher PA concentrations, while ABCB1 expression gradually decreased (Fig. 5G). Quantitative RT-PCR results supported these findings, showing upregulated mRNA expression of PTGS2, NOS2, and ABCB1 with increasing PA concentrations (Fig. 5H-J). Next, we investigated PA-induced ferroptosis in MDA-MB-231 cells. Given that glutathione (GSH) depletion and malondialdehyde (MDA) accumulation are also key features of ferroptosis [21], changes in MDA expression levels and GSH/GSSG ratios are of significant research value. As shown in Fig. 5K, compared with the control group, the relative MDA level in the pachymic acid-treated groups increased significantly in a concentration-dependent manner, while the relative GSH/GSSG ratio decreased markedly, with all differences being statistically significant. These results collectively demonstrate that pachymic acid effectively induces ferroptosis in MDA-MB-231 cells in a concentration-dependent manner, further supporting the mechanistic hypothesis that pachymic acid exerts its anti‑triple‑negative breast cancer effects through oxidative stress and lipid peroxidation pathways (Fig. 5K). Fig. 5PA inhibited the proliferation of MDA-MB-231 cells. A PA chemical formula. B Normal breast cells cck-8. C Breast cancer cell cck8. D Cell clone. E Expression of PTG2, NOS2 and ABCB1 proteins. F Cell streaks. G Transwell test. H-J Expression of PTG2, NOS2 and ABCB1 mRNA. K Expression of MDA and GSH/GSSG PA inhibited the proliferation of MDA-MB-231 cells. A PA chemical formula. B Normal breast cells cck-8. C Breast cancer cell cck8. D Cell clone. E Expression of PTG2, NOS2 and ABCB1 proteins. F Cell streaks. G Transwell test. H-J Expression of PTG2, NOS2 and ABCB1 mRNA. K Expression of MDA and GSH/GSSG Discussion Poria cocos, widely recognized as both a medicinal and edible substance, serves as a key ingredient in many TCM prescriptions and patented formulas. Due to its healing effects, it has a range of pharmacological effects, including promoting urination, reducing fluid retention, enhancing qi, strengthening the spleen, eliminating dampness, and supporting restful sleep. Pachymic acid (PA), a unique triterpenoid compound with a four-ring structure derived from Poria cocos, has shown strong anti-tumor effects by inducing apoptosis in various cancer cell types and inhibiting tumor growth, invasion, and metastasis via multi-target mechanisms. Extensive research has highlighted the various biological functions of PA. For example, it inhibits the propagation and invasion of human ovarian cancer cells by modulating the cell cycle and disrupting the E-cadherin/β-catenin axis [22]; it blocks gastric cancer development by targeting PDGFRB to induce Ferroptosis [23]; and it activates the AMPK pathway in cervical cancer cells by triggering endoplasmic reticulum stress responses [24]. In this study, PA was found to affect the expression of key proteins, PTGS2, NOS2, and ABCB1, and substantially reduce the growth, migration, and invasiveness of BC cells, while revealing minimal toxicity toward normal healthy breast cells. These anti-cancer effects were dose- and time-dependent, indicating that PA has strong potential as a treatment option for BC. PTGS2 is a well-known inflammatory gene involved in tumor development, playing a crucial role in promoting cancer cell propagation, metastasis, and recurrence [25]. Its expression consistently increases during Ferroptosis in different cell types, mainly due to lipid peroxidation-induced stress, which activates redox-sensitive signaling pathways like the Nrf2/ARE axis. Integrating with our previous findings that PA upregulates PTGS2, these results strengthen the evidence supporting PTGS2 as a functional component in PA-induced ferroptosis. Although the precise molecular interaction between PTGS2 and lipid peroxidation requires further elucidation, our data suggest that PTGS2 not only serves as a correlative marker but may represent a potential key node in the ferroptosis signaling network activated by PA. Therefore, PTGS2 mRNA and protein levels are commonly used as markers to monitor ferroptosis in experimental settings [26]. In this study, PA was found to considerably enhance PTGS2 expression at both the transcriptional and protein levels. Considering the association between PTGS2 upregulation and ferroptotic responses, the findings indicate that PA may induce ferroptosis in BC cells through a PTGS2-dependent pathway. This suggested mechanism may underlie PA’s anti-cancer effects and offers a promising therapeutic direction for further research. NOS2 is the gene that encodes the inducible nitric oxide (NO) synthetase that produces nitric oxide [27]. Evidence from mouse models shows the anti-tumor potential of NOS2 in colorectal cancer, and ectopic expression of NOS2 in mouse melanoma cells decreased tumor growth and metastasis [28]. The observed upregulation of NOS2 presents an apparently paradoxical yet potentially crucial finding: while NOS2 is frequently associated with tumor progression, it can also mediate antitumor effects within specific contexts. We detected concurrent NOS2 upregulation with PA-induced cancer suppression. Based on this discovery, we propose an innovative hypothesis: PA may strategically “weaponize” NOS2, converting it into an intracellular instrument that activates a self-limiting stress program triggered by elevated nitric oxide (NO) levels. Sustained high-level NO production can ultimately induce cellular apoptosis through mechanisms involving DNA damage and disruption of energy metabolism. NO is vital in anti-tumor immunity by exerting direct cytotoxic effects on cancer cells, promoting the secretion of pro-inflammatory mediators and chemokines to attract immune effector cells, suppressing tumor-induced angiogenesis, and promoting the presentation of neoantigens to T cells [29]. These mechanisms collectively contribute to the eventual induction of tumor cell apoptosis. Chemoresistance remains a major challenge in effectively treating BC. Its underlying mechanisms are multifactorial, involving intricate interactions between receptor tyrosine kinases and downstream signaling cascades, disturbances in cell cycle and apoptosis regulation, and modulation by tumor-infiltrating immune cells [30]. Among the main factors, ABCB1, a gene that encodes a multidrug resistance (MDR) efflux pump, has been associated with resistance patterns in BC. High levels of ABCB1 are connected to an increased risk of disease relapse and are a hallmark of treatment failure, as tumors with high ABCB1 often show broad resistance to chemotherapy. In triple-negative breast cancer (TNBC), ABCB1 also functions as a marker of metastatic potential, where its overexpression enhances cellular invasion and poorer clinical outcomes [31]. In this study, PA was found to reduce ABCB1 levels in BC cells, suggesting that PA might combat MDR by targeting a key efflux transporter, thus restoring chemosensitivity. These results provide mechanistic support for considering PA as a chemosensitizing agent in therapy-resistant BC. Elevated expression of PTGS2 and NOS2 has been shown to produce a synergistic antitumor effect, mainly through the complex interaction between inflammatory signaling and oxidative stress pathways [32]. Higher PTGS2 levels promote the accumulation of prostaglandin E₂ (PGE₂), which activates EP3 receptors, resulting in the transcriptional upregulation of NOS2 and enhanced production of nitric oxide (NO) [33, 34]. At high levels, NO enhances the tumor-suppressing roles of PTGS2 by S-nitrosylating its key cysteine residues, increasing its enzymatic activity, and stimulating additional PGE2 synthesis. This synergistic mechanism supports the induction of apoptosis in malignant cells [35]. Both PGE₂, derived from PTGS2, and NO, produced by NOS2, suppress the stimulation of the NF-κB pathway, which is often associated with tumor progression [36, 37]. In comparison, the expression of ABCB1, a crucial mediator of multidrug resistance, is mainly controlled by inflammatory signals, primarily through the NF-κB pathway. Proinflammatory stimuli like TNF-α, IL-1β, or chemotherapeutic-induced cellular stress activate NF-κB, which binds to the promoter region of ABCB1, leading to its overexpression and thus increasing resistance to anticancer drugs in solid tumors [38]. Metabolites of PTGS2 and NOS2, specifically PGE₂ and NO, can oppose ABCB1-mediated resistance through various mechanisms, including blocking upstream signaling pathways that activate ABCB1, decreasing its functional capacity, and reducing its induction by the microenvironment [39–41]. The tumor-inhibitory effects from increased PTGS2 and NOS2 expression are due to the biological activity of their metabolites at high concentrations, which mediate cytotoxic effects, immune modulation, and remodeling of the TME. These mechanisms are guided by two key factors: dose-dependency and microenvironment adaptation. This observation suggests that adjusting the levels of PTGS2 and NOS2, rather than completely blocking them, could provide a new therapeutic approach for certain cancers. However, the anti-tumor effects of this strategy depend on specific conditions, requiring careful regulation based on the tumor’s molecular profile and microenvironment. Therefore, clinical implementation would require precisely calibrated methods based on the tumor’s molecular and microenvironmental characteristics. This study reports for the first time that PA significantly upregulates PTGS2 and NOS2 expression while downregulating ABCB1 in triple-negative breast cancer, providing new directions for exploring the molecular mechanisms underlying PA-mediated suppression of TNBC progression. It should be noted that our hypothesis proposing PTGS2/NOS2/ABCB1 as key downstream effectors of PA is primarily based on gene expression and phenotypic correlation data. Although these correlations are statistically significant and biologically informative, establishing definitive causal relationships will require further validation through gain- and loss-of-function experiments. Our selection of the MDA-MB-231 cell model was motivated by the need to thoroughly investigate PA’s mechanism of action in this TNBC subtype, which remains particularly challenging due to limited treatment options and poor prognosis. However, given the substantial heterogeneity of breast cancer and the distinct signaling pathways driving malignant progression across different molecular subtypes, the PA-PTGS2/NOS2/ABCB1 signaling axis identified in our TNBC model may not be directly applicable to other breast cancer subtypes. Based on these considerations, we will prioritize using siRNA/shRNA to specifically knock down the aforementioned genes or employ specific inhibitors to observe whether they can reverse PA effects as our primary future research task. Simultaneously, we will validate the universality of this mechanism across various breast cancer models including Luminal subtypes. These investigations are crucial for confirming the causal relationships in this signaling axis, identifying potential beneficiary patient populations, and advancing translational medical applications. Conclusion PA demonstrates significant antitumor efficacy in breast cancer MDA-MB-231 cells, manifested through the inhibition of cell proliferation, migration, and invasion capabilities. The underlying mechanism involves a unique molecular regulatory network characterized by concurrent upregulation of PTGS2 and NOS2, coupled with downregulation of ABCB1. We have proposed and validated an innovative “dual-pathway activation” hypothesis: PA simultaneously promotes lipid peroxidation through PTGS2 upregulation and induces high-level NO production via NOS2 induction, thereby synergistically triggering ferroptosis. This distinctive mechanism employs multiple pathways including direct cytotoxicity, metabolic reprogramming, and oxidative stress to collectively induce tumor cell death. Furthermore, the observed downregulation of ABCB1 suggests PA’s potential to enhance tumor cell sensitivity to chemotherapeutic agents, providing a theoretical foundation for combination therapy strategies. It should be emphasized that the conclusions of this study are based on cell models of a specific TNBC subtype. The highly heterogeneous nature of breast cancer implies that the universality of this mechanism across other subtypes still requires systematic validation. Thus, future research should validate this pathway in multiple breast cancer models including Luminal subtypes and confirm the causal relationship of PTGS2/NOS2/ABCB1 in this process through genetic approaches, which will establish a solid foundation for developing precise therapeutic strategies targeting TNBC. Supplementary Information Supplementary Material 1. Supplementary Material 1.