Network toxicology and molecular docking reveal the mechanistic basis of bisphenol A-mediated cholestatic liver injury

Abstract

Background

Bisphenol A (BPA), a widespread environmental endocrine-disrupting chemical, is suspected to contribute to liver injury, yet the underlying mechanisms, particularly for cholestatic liver injury (CLI), remain poorly defined. This study aims to systematically elucidate the molecular pathways by which BPA induces CLI.

Methods

We applied an integrated network toxicology approach. BPA targets were predicted using chemical databases (ChEMBL, SwissTargetPrediction, TargetNet), while disease targets for CLI were sourced from GeneCards and OMIM. Bioinformatics analyses, including Gene Ontology and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment, were conducted on overlapping genes. A protein-protein interaction network was constructed to identify hub genes, followed by molecular docking simulations to validate BPA's binding affinity to these key targets.

Results

We identified 200 common genes linking BPA exposure to CLI. Pathway analysis revealed that BPA perturbs multiple biological processes, including chemical detoxification, energy metabolism, inflammation, and bile secretion. Ten core genes (TP53, TNF, and key CYP450 enzymes) were pinpointed as central players. Molecular docking confirmed that BPA binds strongly to these hub targets, substantiating their mechanistic role.

Conclusion

Our findings provide a comprehensive mechanistic framework explaining how BPA exposure may lead to cholestatic liver injury. The study establishes a novel predictive strategy for evaluating the hepatotoxicity of environmental pollutants.

Graphical Abstract

1. Introduction

Cholestatic liver injury (CLI) result from disrupted bile synthesis or secretion, leading to systemic accumulation of bile components [1]. These disorders encompass a spectrum of clinical manifestations, ranging from mild biochemical abnormalities to fulminant hepatic failure or hepatobiliary neoplasia[2]. The underlying pathophysiology may involve congenital defects, immune-mediated mechanisms, structural abnormalities (including obstructive or vascular lesions), or toxic insults, all of which can impair bile duct integrity or hepatocellular function [3]. Following hepatic injury, mature cholangiocytes and hepatocytes undergo compensatory proliferation, which can trigger pathological remodeling, including periductular fibrosis, progressive biliary fibrosis, and ultimately cirrhosis [4]. Emerging evidence implicates environmental factors in dysregulating immune homeostasis and inducing CLI in genetically predisposed individuals [5]. The interplay between environmental exposures and genetic susceptibility, mediated through chemical-gene associations, is now recognized as a fundamental mechanism driving CLI pathogenesis, offering new perspectives on this multifactorial autoimmune condition [6].

Bisphenol A (BPA) is classified as an endocrine-disrupting chemical (EDC), a category of xenobiotics that alter hormonal homeostasis through multiple mechanisms [7]. The Endocrine Society defines EDCs as exogenous compounds capable of interfering with hormone synthesis, secretion, transport, binding, or elimination [8]. BPA serves as a fundamental industrial chemical, primarily employed in manufacturing polycarbonate plastics, food-can coatings, dental materials, and various plastic products [9]. The widespread use of BPA in food-contact materials and dental applications has resulted in its pervasive environmental dissemination, correlating with significant human exposure levels and raising important toxicological concerns regarding its xenobiotic effects. BPA accumulates in various tissues and induces adverse effects through multiple mechanistic routes [10]. A key aspect of its toxicity lies in two synergistic mechanisms that drive multi-organ damage. First, BPA acts as a direct agonist of the Toll-like receptor 4 (TLR4)/myeloid differentiation factor 2 (MD-2) complex, activating the downstream nuclear factor kappa B (NF-κB) and mitogen-activated protein kinase (MAPK) signaling cascades, which promote the release of pro-inflammatory cytokines and lead to tissue injury and immune dysregulation [11], [12], [13]. Second, BPA disrupts cellular redox homeostasis by directly inhibiting major antioxidant enzymes, including superoxide dismutase (SOD), catalase (CAT), and glutathione peroxidase (GPx), upregulating the pro-oxidant enzyme NADPH oxidase 4 (NOX4), and suppressing the nuclear factor erythroid 2-related factor 2 (Nrf2)-mediated antioxidant response, collectively resulting in oxidative damage and apoptosis [14]. The crosstalk between inflammatory activation and oxidative stress constitutes a central mechanism underlying BPA-induced systemic toxicity [15]. As the primary site for xenobiotic detoxification and metabolism, the liver plays a critical role in BPA processing and is particularly vulnerable to its toxic effects, Consequently, BPA exposure profoundly perturbs hepatic immune and metabolic homeostasis, contributing to a range of pathological outcomes [16]. Epidemiological evidence further supports this association, linking higher BPA body burdens to an increased incidence of metabolic liver diseases, such as non-alcoholic fatty liver disease and abnormal liver enzyme profiles [17]. Although previous studies indicate that BPA-mediated disruption of nuclear receptor signaling may impair hepatic lipid homeostasis, trigger Kupffer cell activation, and exacerbate oxidative stress [16], the precise molecular mechanisms linking BPA exposure to CLI progression remain incompletely characterized. Specifically, it is undetermined whether BPA directly perturbs hepatobiliary signaling networks or indirectly promotes liver damage through systemic endocrine disruption.

To elucidate these questions, the present study was designed with the following major objectives: to identify the core molecular targets and overlapping pathways shared between BPA exposure and CLI pathogenesis using an integrated network toxicology approach; to delineate the key biological processes and signaling networks through which BPA may drive cholestatic liver damage; and to validate the binding interactions between BPA and the identified core protein targets via molecular docking simulations. This integrated strategy aims not only to reveal the pathogenic nexus between BPA and CLI but also to provide molecular insights for developing targeted hepatoprotective strategies.

2. Methods

2.1. Network-based toxicity prediction for BPA

To systematically evaluate BPA-induced hepatotoxicity, we implemented an integrated network toxicology strategy. Structural data and canonical SMILES strings for BPA were first extracted from PubChem [18]. Employing computational toxicology platforms including ProTox 3.0 [19] and ADMETlab 3.0 [20], we comprehensively evaluated the compound's toxicological properties, with particular emphasis on blood-brain barrier penetration, oral bioavailability, and organ-specific toxicities (hepatotoxicity, immunotoxicity, and endocrine disruption potential).

To enhance prediction reliability, we performed cross-platform validation by comparing ProTox 3.0 and ADMETlab 3.0 outputs with established toxicological data for both BPA and structurally related compounds, thereby confirming prediction accuracy through empirical concordance analysis [21]. All computational analyses were conducted using standard human (Homo sapiens) physiological parameters to ensure biological relevance and translational applicability.

2.2. Collection of BPA targets

The molecular targets of BPA (CAS 80–05–7) were delineated through an integrative computational workflow that combined comprehensive database interrogation with structure-based prediction. Initial chemical annotation was performed using PubChem (CID: 6623) (https://pubchem.ncbi.nlm.nih.gov/), which verified the molecular formula C₁₅H₁₆O₂ and provided the canonical SMILES string (CC(C)(C1 =CC C(C C1) O) C2 =CC C(C C2) O) together with high-resolution 2D and 3D structural coordinates.

Using the BPA SMILES string as a query, we systematically identified potential molecular targets through comprehensive screening of three authoritative databases: ChEMBL [22], SwissTargetPrediction [23], and TargetNet (http://targetnet.scbdd.com/). All predictions with probability scores > 0 were initially retained as candidate targets. The resulting target names were subsequently translated into official gene symbols using the bioDBnet conversion tool (https://biodbnetabcc.ncifcrf.gov/db/db2db.php). Following integration of targets from multiple sources, we performed rigorous deduplication and structural validation procedures, ultimately generating a high-confidence BPA target library.

2.3. Identification of CLI-related targets

CLI-associated targets were retrieved from GeneCards and OMIM databases using the keywords “cholestasis,” “cholestatic liver injury,” and “jaundice.” To balance biological relevance with analytical feasibility, only protein-coding genes with a GeneCards relevance score > 1 were retained. Parallel OMIM searches with matching keywords expanded our CLI target pool. After merging and deduplicating database outputs, we generated an optimized target set that improved both the specificity and interpretability of subsequent network biology analyses.

2.4. "BPA-CLI " network plot

Venn diagram analysis implemented via the ggvenn R package revealed consensus targets between BPA exposure and CLI pathogenesis. By processing BPA-target interactions alongside CLI-related gene networks, this approach generated a validated set of overlapping molecular targets, structured interaction files for network construction, and comprehensive node descriptors to facilitate subsequent topological analyses.

2.5. Functional enrichment analysis

Functional enrichment analysis was conducted to characterize the biological significance of identified targets using the clusterProfiler R package (v4.10.1) [24]. Gene Ontology (GO) annotation categorized targets into three ontological domains: biological processes (BP), cellular components (CC), and molecular functions (MF). Concurrently, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis elucidated potential signaling pathways. Statistical significance was determined using a dual threshold of p-value < 0.05 and false discovery rate (FDR) < 0.05. Resultant data were visualized through ggplot2 for comprehensive graphical representation.

2.6. Protein-protein interaction (PPI) network analysis and hub targets identification

The STRING database (https://string-db.org/) was employed to analyze both experimentally validated and computationally predicted protein-protein interactions [25]. To elucidate the functional interplay among BPA-responsive CLI targets, we constructed a high-confidence PPI network with STRING, restricting the search to Homo sapiens. Interactions exhibiting confidence scores < 0.9 were excluded to prioritize experimentally validated associations. The TSV file was subsequently imported into Cytoscape (version 3.9.1), and the Network Analyzer plugin was employed to quantify global topological properties and to extract hub nodes [26]. Node color was mapped to degree centrality, with darker hues corresponding to higher connectivity, thereby highlighting candidate central targets within the network.

2.7. Molecular docking

We performed structure-based docking analyses to predict interaction modes between BPA and hub protein targets. The three-dimensional structures of BPA were obtained from PubChem database (https://pubchem.ncbi.nlm.nih.gov/). Human protein structures were acquired from UniProt database (https://www.uniprot.org/), with corresponding three-dimensional structural data obtained from the RCSB Protein Data Bank (https://www.rcsb.org/). The prepared receptor structures and BPA ligand were then processed through CB-Dock2 for blind molecular docking analysis [27]. CB-Dock2 automatically prepares the submitted structures by sequentially stripping crystallographic waters, discarding non-essential heteroatoms, and refining intermolecular interactions. From the output interface, we extracted the docking results including the binding free energy (ΔG) values for the top five ranked conformations, where more negative free energy values indicate thermodynamically more favorable binding poses.

3. Result

3.1. BPA toxicity network predictions

The ProTox database analysis predicted significant estrogenic activity for BPA, with molecular modeling revealing high-affinity binding to the estrogen receptor (Supplementary material S1). These results indicate that BPA may perturb normal physiological homeostasis and promote hepatotoxicity by disrupting estrogen receptor signaling pathways. ADMETlab 3.0 analysis additionally revealed potential neurotoxicity and cutaneous toxicity associated with BPA exposure (Supplementary Material S2). Comprehensive inspection of the predictive outputs further revealed that BPA may disrupt mitochondrial membrane potential and perturb metabolic homeostasis.

3.2. Target analysis of BPA

We systematically compiled a comprehensive inventory of BPA molecular targets by integrating data from three independent databases: ChEMBL, SwissTargetPrediction, and TargetNet. Initial screening identified 661 potential targets from ChEMBL, 39 from SwissTargetPrediction, and 212 from TargetNet. Following dataset integration and duplicate removal through Venn analysis, we established a final curated set of 820 unique BPA targets.

3.3. CLI-related targets

We systematically identified CLI-associated genes by interrogating two authoritative databases: GeneCards and OMIM. From GeneCards, we extracted 3075 CLI-related protein-coding genes meeting the selection criterion of a relevance score > 1. OMIM contributed an additional 59 candidate genes. Following dataset integration and removal of redundant entries through Venn analysis, we established a final curated set of 2229 unique CLI-associated genes.

3.4. Potential targets of BPA in CLI pathogenesis

Integrative analysis of the 2229 CLI-associated genes and 820 BPA targets identified 200 overlapping candidates through Venn analysis (Fig. 1), suggesting potential molecular mechanisms by which BPA exposure might influence CLI development.

Fig. 1.

Venn diagram of the targets of BPA and CLI.

3.5. Functional enrichment analysis

To elucidate how BPA influences CLI at the molecular level, we conducted GO and KEGG pathway enrichment analyses on 200 shared targets using the R package clusterProfiler. The top five significantly enriched GO-BP terms, ranked by ascending adjusted P-values, were primarily linked to processes including response to xenobiotic stimulus, steroid metabolic process, positive regulation of kinase activity, reactive oxygen species metabolic process, and unsaturated fatty acid metabolic process. The top five enriched GO-CC terms were associated with key cellular components, including membrane rafts, collagen-containing extracellular matrix, complexes of collagen trimers, focal adhesions, and protein complexes involved in cell adhesion. Correspondingly, the most significant GO-MF terms primarily involved molecular functions such as nuclear receptor activity, ligand-activated transcription factor activity, cytokine receptor binding, transmembrane receptor protein kinase activity, and receptor ligand activity. KEGG pathway enrichment analysis revealed 136 significantly enriched pathways (adjusted P < 0.05), with the most prominent being lipid and atherosclerosis, PI3K-Akt signaling pathway, steroid hormone biosynthesis, drug metabolism - cytochrome P450, bile secretion, viral carcinogenesis, ECM-receptor interaction, TNF signaling pathway, hepatocellular carcinoma, and PPAR signaling pathway (Fig. 2).

Fig. 2.

Functional annotation of shared BPA-CLI targets. (A) Top five significantly enriched GO biological processes (adjusted *p* < 0.05), visualized as a bubble plot with color intensity representing −log10(*p*-value) and bubble size proportional to gene number. (B) Top ten enriched KEGG pathways ranked by statistical significance. (C) Integrated network of top five enriched GO functional terms. (D) Integrated network of top five enriched KEGG functional terms.

3.6. PPI network construction and core target identification

A PPI network was generated using the STRING database. The composite network was constructed and visualized using Cytoscape 3.9.1, with strategic visual encoding (color gradients, geometric nodes, and force-directed layouts) to distinguish functional modules (Fig. 3). This systems-level analysis facilitated comprehensive mapping of BPA-CLI molecular interrelationships. Core genes associated with BPA-induced CLI were systematically identified through comprehensive evaluation using Network Analyzer plugin (Fig. 4). Nodes were ranked according to three key centrality metrics: degree centrality (number of direct connections), betweenness centrality (frequency of acting as a bridge on shortest paths), and closeness centrality (proximity to all other nodes in the network). We established specific threshold criteria for hub gene selection: nodes simultaneously ranking in the top 20 % for degree centrality, top 15 % for betweenness centrality, and top 25 % for closeness centrality across the entire network were identified as hub genes. This multi-metric approach ensured the selection of nodes that are not only highly connected but also strategically positioned within the network architecture. Integrative analysis identified TP53, CYP450 enzymes (CYP2B6, CYP3A4, CYP2C9, CYP2C8, CYP2C19, CYP2E1, CYP1A2, CYP1A1), and TNF as pivotal mediators in BPA-induced CLI pathogenesis.

Fig. 3.

Network analysis of molecular interactions in BPA-induced CLI.

Fig. 4.

PPI network in BPA-induced CLI. Network topology showing candidate core targets (highly connected nodes) and their interactions. Node color intensity scales with degree centrality (darker hues indicate greater connectivity), reflecting potential functional importance in the BPA-CLI pathological network.

3.7. Molecular docking analysis of BPA with CLI hub targets

To elucidate the molecular interactions between BPA and key biological targets, molecular docking simulations were performed with ten critical proteins: CYP2B6 (PDB ID: 3IBD), CYP3A4 (PDB ID: 1TQN), CYP2C9 (PDB ID: 1R90), CYP2C8 (PDB ID: 2NNJ), CYP2C19 (PDB ID: 4GQS), CYP2E1 (PDB ID: 3E6I), CYP1A2 (PDB ID: 2HI4), CYP1A1 (PDB ID: 4I8V), TP53 (PDB ID: 1YC5), and TNF (PDB ID: 2AZ5). CB-Dock2 analysis revealed that BPA binds to the target proteins with binding energies ranging from −5.4 to −8.2 kcal/mol (Table 1). In molecular docking studies, binding energy values < 0 typically indicate favorable binding interactions, with values below −5.0 kcal/mol representing strong binding affinity, where more negative values correlate with greater interaction strength. Our results revealed robust binding affinities between BPA and the ten core targets, suggesting spontaneous complex formation and implicating these interactions in BPA-induced CLI pathogenesis. The nine most favorable binding interactions (lowest energy states) between BPA and the target proteins were further visualized (Fig. 5).

Table 1.

Computational docking results for BPA-core target binding affinities (kcal/mol).

Target Name	PDB ID	Binding Energy (kcal/mol)	Key Contact residues
CYP2B6	3IBD	−7.5	ARG98 ILE101 VAL104 VAL113 ILE114 PHE115 ARG125 SER128 VAL129 MET132
CYP3A4	1TQN	−7.8	ARG105 ARG106 PRO107 PHE108 PHE137 SER180 VAL183 ILE184 THR185 THR187
CYP2C9	1R9O	−7.5	PRO37 ILE45 ILE50 LEU54 LEU67 PHE69 LYS72 ILE74 PHE100 PRO101
CYP2C8	2NNJ	−7.0	ARG97 ILE102 SER103 ARG105 ILE106 THR107 ILE113 SER114 ARG200 PHE201
CYP2C19	4GQS	−8.1	VAL38 ILE39 ILE42 ILE45 ILE47 VAL50 SER51 PHE69 GLY70 MET74
CYP2E1	3E6I	−8.2	ARG100 ILE114 ILE115 PHE116 ARG126 LEU133 PHE207 ASP295 LEU296 PHE298
CYP1A2	2HI4	−7.7	ARG108 LEU123 THR124 PHE125 TRP133 ARG137 LEU144 PHE226 VAL227 PHE260
CYP1A1	4I8V	−7.5	ARG106 SER120 MET121 SER122 PHE123 TRP131 ARG135 LEU142 PHE224 ASP313
TP53	1YC5	−7.9	GLY21 ALA22 GLY23 SER25 ILE30 PRO31 ASP32 PHE33 PHE48 MET71
TNF	2AZ5	−5.4	PRO20 GLN21 ALA22 GLU23 GLY24 GLN25 LEU26 LYS65 GLN67 PRO139

Abbreviation: PDB ID, Protein Data Bank Identifier; BPA, bisphenol A; CYP2B6, cytochrome P450 family 2 subfamily B member 6; CYP3A4, cytochrome P450 family 3 subfamily A member 4; CYP2C9, cytochrome P450 family 2 subfamily C member 9; CYP2C8, cytochrome P450 family 2 subfamily C member 8; CYP2C19, cytochrome P450 family 2 subfamily C member 19; CYP2E1, cytochrome P450 family 2 subfamily E member 1; CYP1A2, cytochrome P450 family 1 subfamily A member 2; CYP1A1, cytochrome P450 family 1 subfamily A member 1; TP53, tumor protein p53; TNF, tumor necrosis factor.

Fig. 5.

Molecular docking analysis of BPA with key proteins. (A) CYP1A1_BPA complex (Binding energy: − 7.5 kcal/mol); (B) CYP1A2_BPA complex (-7.7 kcal/mol); (C) CYP2B6_BPA complex (-7.5 kcal/mol); (D) CYP2C8_BPA complex (-7.0 kcal/mol); (E) CYP2C9_BPA complex (-7.5 kcal/mol); (F) CYP2C19_BPA complex (-8.1 kcal/mol); (G)CYP2E1_BPA complex (-8.2 cal/mol); (H)CYP3A4_BPA complex (-7.8 kcal/mol); (I)TP53_BPA complex (-7.9 kcal/mol).

4. Discussion

BPA, a ubiquitous environmental endocrine disruptor, is extensively employed in the production of polycarbonate plastics and epoxy resins, resulting in pervasive human exposure via food-contact materials and a broad spectrum of consumer goods [28]. Accumulating evidence demonstrates that BPA exposure disrupts endocrine function and promotes metabolic dysregulation, including obesity, diabetes mellitus, and cardiovascular disorders [29]. While human populations are routinely exposed to complex environmental pollutant mixtures that may exhibit synergistic or antagonistic interactions, the current investigation specifically examines BPA owing to the paucity of mechanistic evidence regarding its involvement in CLI pathogenesis. Here, we leveraged an integrated strategy that unites network toxicology, molecular docking, and evidence-based literature mining to dissect the molecular mechanisms by which BPA triggers CLI.

Through systematic bioinformatics interrogation, we delineated 200 high-probability molecular targets that potentially bridge BPA exposure with CLI progression, providing novel insights into the disease's environmental etiology. Our comprehensive enrichment analysis revealed several critical biological pathways and molecular mechanisms that may underlie BPA-induced CLI. The GO biological process analysis identified five predominant pathways, with the most significant being response to xenobiotic stimulus, strongly suggesting that BPA exposure triggers characteristic detoxification responses in hepatocytes. The enrichment of steroid metabolic processes and unsaturated fatty acid metabolism is particularly noteworthy, as these pathways are fundamentally implicated in bile acid homeostasis and lipid metabolism - two key processes disrupted in cholestasis [30], [31]. At the cellular component level, the significant enrichment of membrane raft and focal adhesion suggest potential disruption of hepatocyte polarity and bile canalicular integrity, which are critical for proper bile flow [32]. The prominence of extracellular matrix components, including collagen-containing matrices, may reflect early fibrotic changes preceding overt biliary fibrosis [33]. The molecular function analysis revealed striking enrichment of nuclear receptor activity and ligand-activated transcription factor function, implicating BPA's endocrine-disrupting properties in CLI pathogenesis. This is mechanistically consistent with the observed enrichment of cytokine receptor binding, suggesting a potential interplay between endocrine disruption and inflammatory signaling in BPA-induced CLI [34]. The KEGG pathway analysis yielded 136 significantly enriched pathways, with several having direct relevance to CLI. The prominence of bile secretion and steroid hormone biosynthesis pathway provide compelling evidence for BPA's disruption of biliary function. The enrichment of drug metabolism-cytochrome P450 pathways suggests potential metabolic interactions between BPA and pharmaceutical compounds in hepatotoxicity. The significant enrichment of the PPAR signaling pathway is of particular mechanistic relevance to CLI. PPARα activation is a known regulator of bile acid homeostasis, and its suppression—as suggested by our network pharmacology results—is directly linked to the transcriptional downregulation of key hepatobiliary transporters, including the bile salt export pump (BSEP/ABCB11) and multidrug resistance-associated protein 2 (MRP2/ABCC2) [35]. The PI3K-Akt axis is known to post-translationally regulate the activity and membrane localization of canalicular transporters like BSEP [36]. Studies have shown that insulin-induced Akt activation can promote the retrieval of BSEP from the canalicular membrane to intracellular compartments, effectively reducing biliary bile acid secretion capacity [37]. Collectively, these omics-level signatures delineate a multi-hit model in which BPA simultaneously dysregulates xenobiotic sensing, lipid metabolism, oxidative stress, and extracellular-matrix remodeling to initiate and propagate CLI.

Subsequent network analysis revealed ten pivotal molecular targets (TP53, CYP2B6, CYP3A4, CYP2C9, CYP2C8, CYP2C19, CYP2E1, CYP1A2, CYP1A1, and TNF) potentially mediating BPA-triggered CLI. These targets cluster into three functional categories: tumor suppressor signaling (TP53), cytochrome P450-mediated metabolism (CYP2B6, CYP3A4, CYP2C9, CYP2C8, CYP2C19, CYP2E1, CYP1A2, CYP1A1), and inflammatory regulation (TNF), collectively suggesting a multi-factorial pathogenic mechanism for BPA-induced CLI. As a critical tumor suppressor, TP53 may contribute to BPA-induced CLI through multiple mechanisms. Previous studies have demonstrated that BPA exposure can induce TP53-mediated apoptosis in hepatocytes [38]. In cholestasis, sustained TP53 activation may promote excessive cell death, exacerbating liver injury [39]. Additionally, TP53 regulates bile acid homeostasis by modulating transporters like bile salt export pump, suggesting its potential role in bile flow impairment [40]. The prominence of multiple CYP isoforms (CYP2B6, CYP3A4, CYP2C9/8/19, CYP2E1, CYP1A1/2) suggests BPA may disrupt both phase I metabolism and bile acid synthesis. Specifically, BPA-induced overexpression of CYP3A4, the dominant drug-metabolizing enzyme responsible for bile acid hydroxylation [41], may deplete hepatocyte glutathione reserves, thereby exacerbating oxidative stress - a well-established mediator of cholangiocyte injury [42]. Concurrently, the upregulation of ROS-generating enzymes CYP1A1/2 and CYP2E1 likely contributes to the observed lipid peroxidation in BPA-exposed hepatocytes, leading to canalicular membrane damage and subsequent bile flow impairment [43]. Furthermore, BPA-mediated inhibition of the CYP2C subfamily (particularly CYP2C8/9), which metabolizes arachidonic acid to vasoactive eicosanoids, may perturb hepatic microcirculation, potentially exacerbating the ischemic pathology characteristic of CLI [44]. TNF signaling emerged as the central inflammatory mediator in BPA-induced CLI. Mechanistically, BPA activates Kupffer cells to overexpress TNFα, which subsequently exerts multiple hepatotoxic effects [45]. This triad of TNF -mediated effects collectively contributes to the development of CLI by impairing bile secretion, promoting biliary epithelial dysfunction, and initiating fibrotic remodeling [46].

The molecular docking simulations revealed several critical findings regarding BPA's potential to disrupt CLI through direct interactions with key molecular targets. The robust binding affinities observed (range: −5.4 to −8.2 kcal/mol) strongly suggest that BPA can spontaneously form stable complexes with multiple proteins implicated in CLI pathogenesis. These computational results provide mechanistic support for epidemiological evidence linking BPA exposure to hepatobiliary dysfunction. The particularly strong binding affinities observed with cytochrome P450 isoforms (CYP2B6, CYP3A4, CYP2C9/8/19, CYP2E1, CYP1A1/2) align with previous reports of BPA's metabolic disruption [47]. Notably, the high-affinity interaction with CYP2E1 (ΔG = −8.2 kcal/mol) provides structural evidence for BPA's potential to exacerbate oxidative stress in hepatocytes, consistent with observed lipid peroxidation in BPA-exposed models [48]. The significant binding to TP53 (ΔG = −7.9 kcal/mol) offers structural insights into BPA's potential to stabilize this tumor suppressor, which may explain reported downregulation of bile acid transporters in exposed hepatocytes [49]. The moderate but physiologically relevant interaction with TNF (ΔG = −5.4 kcal/mol) suggests BPA might modulate inflammatory signaling, supporting previous observations of Kupffer cell activation following exposure [50]. The molecular docking results collectively suggest that BPA may induce CLI through multiple interconnected mechanisms involving direct modulation of bile acid-metabolizing enzymes, disruption of redox homeostasis via oxidative stress-inducing enzymes, interference with cellular stress response pathways, and potential alteration of inflammatory signaling cascades. These computational findings are particularly significant as all observed binding energy values exceeded the −5.0 kcal/mol threshold considered biologically relevant for molecular interactions [51], strongly supporting the physiological plausibility of BPA's direct engagement with these critical hepatobiliary targets. The predicted interactions with CYP450 isoforms, TP53, and TNF provide a structural basis for understanding how BPA exposure could simultaneously disrupt metabolic, oxidative, and inflammatory homeostasis in CLI. These results offer novel mechanistic insights into the molecular initiating events that may underlie BPA's hepatotoxicity, bridging the gap between epidemiological observations of BPA-associated CLI and specific molecular interactions at the protein level.

5. Limitations of the study

While this study provides a systematic framework and mechanistic insights into BPA-induced CLI, several limitations should be acknowledged. The primary constraint stems from its reliance on computational predictions, which lack experimental validation. Although molecular docking suggested strong binding affinities between BPA and the identified hub targets, this method does not fully capture protein dynamics or physiological conditions. Future work would benefit from molecular dynamics simulations to assess complex stability and from experimental validation of key targets (e.g., TP53, TNF) and pathways (e.g., PI3K-Akt signaling) in relevant in vitro or in vivo models.

Furthermore, the translational relevance of our findings must be interpreted with caution. Our model did not account for real-world human exposure scenarios, including the potential synergistic or antagonistic effects of co-exposure to complex environmental EDC mixtures. Critical exposure parameters—such as the developmental timing of exposure, cumulative dose effects, and potential non-monotonic dose responses—remain unaddressed. Consequently, future longitudinal studies are needed to establish the exposure-response relationship and determine whether environmentally achievable tissue concentrations of BPA sufficiently engage the predicted targets in vivo.

Finally, our analysis focused specifically on BPA-related mechanisms and did not comprehensively examine other CLI etiologies, such as genetic predisposition, viral hepatitis, or autoimmune disorders. Elucidating the distinct and shared pathways across these varied causes of cholestasis will be essential for a holistic understanding of CLI pathogenesis.

6. Conclusion

In conclusion, this study utilized an integrated computational strategy combining network toxicology with molecular docking to systematically investigate the role of BPA in CLI. We identified 200 shared molecular targets and elucidated their functional involvement in critical processes such as xenobiotic sensing, metabolic reprogramming, PI3K-Akt signaling, redox homeostasis, and TNF-mediated inflammation. Through network topology analysis, ten hub genes—TP53, TNF, and several cytochrome P450 enzymes (CYP2B6, CYP3A4, CYP2C9, CYP2C8, CYP2C19, CYP2E1, CYP1A2, CYP1A1)—were prioritized, and molecular docking confirmed strong binding affinities between BPA and these key targets, supporting their mechanistic relevance in BPA-induced hepatotoxicity. Despite its computational nature, this work establishes a foundational framework for understanding BPA-driven CLI and highlights the value of integrative bioinformatics in toxicological profiling. The findings not only reveal potential pathogenic pathways but also nominate candidate biomarkers for subsequent experimental validation and environmental risk assessment.

CRediT authorship contribution statement

Yinpei Tang: Writing – original draft, Validation, Methodology, Formal analysis. Zhengwang Zhu: Supervision, Software. Pingsheng Zhu: Supervision, Project administration, Conceptualization. Linjing She: Writing – review & editing, Data curation. Yaping Sun: Writing – review & editing, Data curation.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Handling Editor: Lawrence Lash

Appendix A. Supplementary material

Supplementary material

Supplementary material

Data availability

Data will be made available on request.