Integrating network toxicology, machine learning, gut microbiome analysis, and structural validation to reveal the molecular mechanism linking PFOA and PFOS exposure to age-related macular degeneration

Integrating network toxicology, machine learning, gut microbiome analysis, and structural validation to reveal the molecular mechanism linking PFOA and PFOS exposure to age-related macular degeneration | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Integrating network toxicology, machine learning, gut microbiome analysis, and structural validation to reveal the molecular mechanism linking PFOA and PFOS exposure to age-related macular degeneration Zhenyu Guo, Yujun Peng, Yiwei Lin, Jiaxi Liu, Yongjie Qin, Hongyang Zhang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8563077/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Purpose Perfluorooctanoic acid (PFOA) and perfluorooctane sulfonic acid (PFOS) are two ubiquitous persistent organic pollutants. Currently, little is known about the potential mechanisms of PFOA and PFOS in ophthalmic diseases. This study systematically investigates the effects of PFOA and PFOS exposure on age-related macular degeneration (AMD), aiming to identify potential targets and elucidate associated molecular mechanisms. Methods Differential expression analysis across multiple transcriptomic datasets was performed to identify AMD-associated genes. Network toxicology, machine learning, Mendelian randomization, gut microbiota profiling, molecular docking, and MD simulations were subsequently integrated to characterize the interactions between PFOA/PFOS and their protein targets. Results We first performed toxicity prediction to identify 97 potential targets linking these toxins to AMD. Subsequently, integrating bioinformatics with 128 machine learning algorithms further identified four key core genes: SLC1A4, ADAM17, EGFR, and MAPK10. Potential regulatory mechanisms were mapped using TF-miRNA-mRNA networks and GeneMANIA. Mendelian randomization (MR) validated causal relationships between core genes SLC1A4 and ADAM17 and AMD. Molecular docking and kinetic simulations confirmed stable interactions between PFOA/PFOS and their targets. Our findings reveal the pivotal roles of inflammatory responses, oxidative stress, and amino acid metabolism disorders in the pathogenic mechanisms of these toxins. Analysis of the gutMGene gut microbiome database identified specific bacteria potentially acting through metabolites to influence human genes enriched in inflammatory signaling pathways, thereby promoting the toxic effects of PFOA/PFOS on AMD. This established a “toxin-gut bacteria-metabolite-gene” relationship network. Conclusion This study elucidates the underlying molecular mechanisms and signaling pathways linking PFOA/PFOS exposure to AMD, identifying SCL1A4 and ADAM17 as key targets for PFOA/PFOS-induced AMD. These findings not only provide new insights into the role of environmental pollutants in AMD pathogenesis but also offer an analytical framework for elucidating the effects of other environmental toxins on ocular diseases. PFOA and PFOS Age-related macular degeneration Bioinformatics Network toxicology machine learning Gut microbiota interactions Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Introduction Per-and polyfluoroalkyl substances (PFAS) represent a class of highly fluorinated aliphatic compounds characterized by exceptional physicochemical stability. Owing to their unique surface activity and resistance to heat and degradation, PFAS are widely incorporated into industrial and consumer products such as textiles, surfactants, furniture coatings, food packaging, and firefighting foams[ 1 ]. These compounds readily infiltrate global ecosystems through wastewater discharge, atmospheric transport, and bioaccumulation. Among them, perfluorooctanoic acid (PFOA) and perfluorooctanesulfonic acid (PFOS) have drawn particular attention because of their extensive production and demonstrated environmental persistence and toxicity[ 2 ]. They are persistent organic pollutants listed in the Stockholm Convention, which have become serious environmental pollutants due to their extreme persistence and transboundary mobility[ 3 ]. Age-related macular degeneration (AMD) causes irreversible vision loss or blindness[ 4 ] and is the leading cause of blindness in older worldwide[ 5 ]. Its pathogenesis results from complex interactions between genetics and environmental risk factors. With aging, increased production of reactive oxygen species coupled with impaired clearance functions leads to cumulative oxidative damage in the retina. Over time, the accumulation of oxidative stress and alterations in antioxidant responses result in impaired function of retinal pigment epithelial cells, diminishing their capacity to support and maintain photoreceptor function[ 6 ]. Concurrently, environmental factors associated with AMD (e.g., smoking) exacerbate oxidative stress and compromise blood supply[ 7 ]. Therefore, reducing environmental risk factors may help decrease the prevalence and progression of AMD, offering new perspectives and insights for preventing and alleviating its economic burden[ 8 ]. Current research findings on environmental toxins and AMD remain limited and isolated. It has been noted that human visual impairment is closely associated with perfluoroalkyl and polyfluoroalkyl substances[ 9 ]. Studies indicate that perfluorooctanoic acid (PFOA) exposure can induce inflammatory responses in retinal pigment epithelial cells[ 10 ]. Furthermore, epidemiological research has identified an association between PFOA/PFOS exposure and AMD. Chen et al[ 11 ].examined high-detection-level serum perfluoroalkyl substances (PFAS) including PFOA and PFOS in 1,605 participants recruited during the 2005–2008 NHANES. They identified an association between PFOS exposure and AMD risk. However, the association between PFOA/PFOS and age-related macular degeneration remains confined to the level of epidemiological correlation. The complex biological environment hinders the elucidation of precise pathogenic mechanisms, making targeted research into the mechanisms by which toxin exposure influences AMD critically important. To bridge this gap, this study employs a systematic network toxicology approach integrating toxicological targets of pollutants with AMD-associated disease genes: Refining the PFOA/PFOS-driven AMD pathogenesis network through multi-omics data integration; Identifying key hub molecules via machine learning-guided topological and functional enrichment analysis; Validating causal relationships between core genes and AMD using MR; and validated the binding effects between toxins and targets through molecular docking and molecular dynamics simulations. This study aims to provide scientific evidence for the prevention and treatment of ophthalmic diseases involving PFOA/PFOS. It proposes a “Environmental toxin-Eye” relationship network, offering novel insights into the link between environmental pollutants and ophthalmic diseases, and establishing a more comprehensive, reliable, and in-depth scientific foundation for the interdisciplinary comprehensive assessment of environmental pollutants. Methods Toxicity prediction and target identification The molecular structures and SMILES strings of PFOA and PFOS were obtained from the PubChem database ( https://pubchem.ncbi.nlm.nih ). The ADMETLab 3.0 platform was then used to predict their physicochemical and toxicological properties. Potential molecular targets were identified through the ChEMBL, SwissTarget Prediction, and PharmMapper databases, restricting the search to Homo sapiens to ensure biological relevance. The UniProt database ( https://www.uniprot.org/ ) was used for cross-validation and standardization of target information. We integrated, deduplicated, and merged the targets for PFOA and PFOS to ensure their completeness, ultimately compiling a comprehensive toxicant database. Acquisition of disease-associated genes The Gene Expression Omnibus (GEO) contains high-throughput gene expression data[ 12 ] submitted by research institutions worldwide. We curated three age-related macular edema (ARME) transcriptome datasets (GSE99248, GSE1719, and GSE29801) from the GEO database ( https://www.ncbi.nlm.nih.gov/geo/ ). GSE99248 (including 15 control samples and 16 AMD samples) and GSE1719 (including 18 control samples and 18 AMD samples) serve as the discovery cohort, while GSE29801 (including 150 control samples and 142 AMD samples from macular RPE-choroid or extramacular retina tissue) functions as the validation cohort. To mitigate batch effects, a multi-stage normalization process was implemented, including substitute variable analysis (SVA) and gene expression standardization (GES). Corrected principal component analysis (PCA) revealed improved clustering of samples across batches in the reduced-dimension space, confirming successful data harmonization. To comprehensively identify AMD disease targets for subsequent analysis, we systematically collected AMD-related targets by integrating GeneCards, OMIM disease databases, and GEO database resources. Protein-Protein Interaction (PPI) network and core target identification Intersection analysis between toxin-related and AMD-related targets was performed using the Venny tool. The overlapping targets were uploaded to the STRING database ( https://cn.string-db.org/ ) to construct a PPI network, which was visualized in Cytoscape 3.10.3. Key hub genes and subnetworks were identified using the CytoHubba plugin Functional enrichment analysis Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were conducted using the DAVID platform ( https://davidbioinformatics.nih.gov/ ) and the clusterProfiler package in R. Biological process (BP), cellular component (CC), and molecular function (MF) categories were analyzed, with significance thresholds set at P < 0.05. Machine learning for key target recognition To systematically identify biomarkers for toxins and diseases, we constructed an integrated machine learning prediction framework combining multiple algorithms. Based on training set expression profiles, we employed multiple classical machine learning algorithms—Lasso regression (glmnet, alpha = 1), Ridge regression (glmnet, alpha = 0), Elastic Net (glmnet, alpha between 0–1), Random Forest (randomForestSRC, ntree = 1000, nodesize = 5), GBM (gbm, n.trees = 10000, interaction.depth = 3, shrinkage = 0.001, cv.folds = 10), Support Vector Machine (e1071), Naive Bayes (e1071), Partial Least Squares Generalized Linear Model (plsRglm), BART, etc.—to develop 128 predictive models. Hyperparameter optimization was performed using 10-fold cross-validation, with data partitioned into training and internal validation subsets via stratified sampling. Optimal single-model predictions were subsequently integrated using a stacked ensemble learning strategy. Finally, gene expression patterns were visualized using the pheatmap package. Functional prediction of toxin-disease genes and construction of miRNA-TF-mRNA regulatory networks To systematically investigate the functional roles of toxin-disease-associated genes in disease pathogenesis, comprehensive functional and pathway prediction analyses were conducted using GeneMANIA( http://genemania.org/ ). Simultaneously, by integrating three prediction databases (Starbase, miRTarBase, and TargetScan), we identified miRNAs targeting toxin- and disease-associated genes, from which 23 miRNAs with the strongest disease association were selected. Furthermore, using the NetworkAnalyst database ( https://www.networkanalyst.ca/ ), we identified the most strongly associated transcription factors (TFs) among toxin- and disease-associated genes. Finally, based on the above data, a visual miRNA-TF-mRNA regulatory network was constructed using Cytoscape 3.10.3, elucidating the potential molecular mechanisms linking toxins and diseases. Mendelian randomization analysis of core genes and validation using the HPA database To further infer causality, the TwoSampleMR R package (v0.5.6; https://mrcieu.github.io/TwoSampleMR/ ) performed a two-sample mendelian randomization analysis on pooled GWAS SNP data to evaluate causal relationships between key genes and AMD. For the exposure variable (SLC1A4 expression), we utilized the eQTL dataset (OpenGWAS ID: eqtl-a-ENSG00000115902, Vosa U, 2018), which comprises 31,684 European individuals and 18,504 SNPs; For the exposure variable (ADAM17 expression), we utilized the eQTL dataset (OpenGWAS ID: eqtl-a-ENSG00000151694, Vosa U, 2018), which includes 31,644 European individuals and 19,103 SNPs. For the outcome variable (Age-related macular degeneration), the GWAS dataset (OpenGWAS ID: finn-b-H7_AMD, 2021) was used, comprising 3,763 cases and 205,359 controls (total sample size = 209,122) of European ancestry, covering 16,380,424 SNPs. Instrument variables were selected using stringent criteria: P < 5e-8, r2 < 0.001, and kb < 10,000. Results were validated using inverse variance weighting (IVW) combined with sensitivity, heterogeneity, and pleiotropy tests. All analyses were performed in R 4.5.0 using TwoSampleMR v0.5.6 and MRPRESSO v1.0. Additionally, by integrating the HPA database ( https://www.proteinatlas.org/ ) and employing single-cell distribution analysis, we focused on observing gene expression in vision-related cell types such as Müller glia cells, photoreceptor cells, and bipolar cells. UMap visualization was employed to analyze gene distribution patterns across different cell populations. Fluorescence microscopy images were collected and organized for subcellular localization validation, enabling observation of protein localization within specific cellular compartments such as the nucleus, cytoplasm, and endoplasmic reticulum. Molecular docking and molecular dynamics (MD) simulations The two-dimensional (2D) molecular structure data for PFOA and PFOS were obtained from the PubChem database. The PDB format files for core genes were retrieved from the RCSB database ( https://www.rcsb.org/ ). AutoDock Vina software was employed for molecular docking, with results visualized using pymol software. Binding energy was utilized to evaluate interactions between ligands and receptors: A binding energy below 0 kcal/mol indicates spontaneous binding, while values below − 5 kcal/mol suggest stable binding. A 100 ns molecular dynamics simulation was performed under standard temperature and pressure conditions using GROMACS 2025.2 software, employing the CHARMM36 force field and TIP3P water model for solvation. Gut microbiota analysis Given that the gut microbiota also plays an indispensable role in ophthalmic diseases, and considering that PFOA/PFOS are readily absorbed and poorly metabolized in the gut[ 13 ], we utilized the gut microbial gene database (gutMGene; http://www.gutmgene.org/ ) to retrieve all host-gene-based associations[ 14 ]. We focused on the “human” species filter to identify gut microbiota sharing common target genes with the toxins. Metabolites associated with these microbiota were retrieved through the search module and rigorously filtered based on host species = “human,” association source = “literature association,” and PubChem CID availability. We then employed the Similarity ensemble approach (SEA, https://sea.bkslab.org/ ) and SwissTargetPrediction ( https://swisstargetprediction.ch/ ) to gather potential targets for these metabolites. GO and KEGG analyses were performed on metabolite-associated genes to elucidate potential metabolic pathways. Finally, network diagrams linking toxins, target genes, gut microbiota, metabolites, and host genes were constructed using STRING and Cytoscape. Results Toxicity assessment and target identification of PFOA and PFOS The ADMEtLab website was used to assess the toxicity profiles of PFOA and PFOS. The results predicted extensive eye irritation and eye corrosion toxicity for PFOA and PFOS, indicating a risk of ocular damage (Supplemental Digital Content Fig. 1A-B). Based on these findings, we employed network toxicology to analyze the role of PFOA/PFOS in promoting AMD progression, aiming to further understand whether these toxins increase the risk of developing AMD. By integrating data from SwissTargetPrediction, ChEMBL, and pharmmapper databases, we compiled a total of 439 potential targets for PFOA and PFOS, thereby gaining insights into the possible molecular pathways through which these toxins may induce pathological changes. Identification of AMD-associated genes We analyzed transcriptome data from the GEO database (GSE99248, GSE1719) to compare gene expression profiles between AMD patients and healthy controls. To minimize batch effects, we merged the GSE99248 and GSE1719 datasets and performed comprehensive normalization of the gene expression matrix. Principal Component Analysis (PCA) revealed improved data distribution after normalization, with the normalized dataset exhibiting clearer clustering patterns (Fig. 2A-B). Subsequent differential expression analysis identified 47 genes significantly altered in AMD. Expression changes for these genes were visualized using volcano plots and heatmaps (Fig. 2C-D). Subsequently, WGCNA analysis was performed. Based on this parameter, a Topology Overlap Matrix (TOM) was constructed and hierarchical clustering was conducted to identify co-expression modules. Nine independent gene modules were analyzed (Fig. 2E). Module-trait association analysis revealed that specific modules were significantly associated with AMD (P < 0.05) (Fig. 2F). To comprehensively capture potential genes involved in PFOA/PFOS-induced AMD and ensure subsequent analysis reliability, we also screened AMD-related target genes from GeneCards and OMIM disease databases. By integrating differentially expressed genes with WGCNA module genes and removing duplicates, we identified 1,687 disease-associated candidate genes. Finally, by integrating PFOA/PFOS toxicant targets, we identified 97 overlapping genes potentially involved in toxicant-mediated AMD (Fig. 2G). Protein–Protein Interaction (PPI) network and functional enrichment analysis The 97 overlapping targets were uploaded to the STRING database to construct a protein–protein interaction (PPI) network, which was visualized using Cytoscape 3.10.3 (Fig. 3A). Functional enrichment analysis with DAVID identified 2,158 significant GO terms—comprising 1,929 biological processes (BP), 56 cellular components (CC), and 173 molecular functions (MF). BP terms were mainly enriched in oxidative stress, reactive oxygen species response, and chemical stress response (Fig. 3B–C). CC enrichment involved vesicle lumen, secretory granule lumen, and cytoplasmic vesicle lumen, while MF terms included phosphatase binding, MAP kinase activity, and lipoprotein particle binding (Fig. 3B–C). KEGG analysis highlighted enrichment in lipid metabolism, atherosclerosis, and endocrine resistance pathways (Fig. 3D). These findings collectively suggest that PFOA and PFOS may participate in the onset and progression of AMD by regulating these key biological pathways. Identification of core genes using machine learning and prediction of core gene functions and construction of miRNA-TF-mRNA regulatory networks Through comprehensive machine learning analysis of candidate targets, we constructed 128 predictive models to identify core genes affected by PFOA/PFOS in AMD. We selected Stepglm[both] + RF as the ensemble model due to its outstanding performance with high accuracy in both training and validation phases (Fig. 4A). Calibration curves validated the predictive accuracy of the machine learning model on the training dataset (Fig. 4B), identifying four key genes: SLC1A4, ADAM17, EGFR, and MAPK10. ROC curve analysis confirmed the diagnostic potential of these core genes (AUC > 0.75, Fig. 4C). The diagnostic performance of these feature genes was separately evaluated on the training and validation sets using ROC curves: the AUC value for the test set was 0.954, and the AUC value for the validation set was 0.787 (Figs. 4D-E). Subsequently, nomograms were constructed to score each gene's expression, precisely quantifying AMD risk. SLC1A4 and ADAM17 emerged as clear risk factors for the disease (Fig. 4F). Finally, we mapped the selected genes to their chromosomal locations (Fig. 4G). Functional prediction of the four core genes via GeneMANIA revealed their involvement in the ERBB signaling pathway and amino acid phosphorylation regulation (Fig. 4H). Potential miRNAs and transcription factors (TFs) targeting these genes were identified using StarBase, miRTarBase, TargetScan, and NetworkAnalyst databases. The final regulatory network visualized in Cytoscape 3.10.3 encompassed 47 miRNAs, 13 TFs, and 4 core genes, revealing intricate regulatory interactions among these molecules (Fig. 4I). Mendelian randomization and HPA validation of hub genes Next, we performed Mendelian randomization (MR) analysis using the four core genes as exposure variables and AMD as the outcome variable to investigate the causal relationship between the core genes and AMD. Results indicated (Fig. 5A) that SLC1A4 (OR = 1.4096) and ADAM17 (OR = 1.3649) were risk factors for AMD, suggesting that increased expression of these genes correlates with a higher probability of AMD occurrence (Figs. 5B-C). Both genes had P-values < 0.05, indicating highly significant causal relationships with AMD and confirming their statistical significance. Validation via the HPA database revealed their widespread distribution in photoreceptor cells through single-cell localization analysis (Supplemental Digital Content Fig. 2A). Subcellular localization analysis indicated SLC1A4 predominantly resides in the centrosome, while ADAM17 is localized in the cytosol (Supplemental Digital Content Fig. 2B). Molecular docking and Molecular dynamic simulations To further investigate the binding efficiency of the toxins PFOA/PFOS with MR-validated target proteins, molecular docking analyses were performed for SLC1A4 and ADAM17 proteins with both toxins. It can be observed that SLC1A4 exhibits high binding energies with both toxins. This is attributed to the presence of ASN-402 and ASN-402, MET-221, SER-170 residues in the PFOA-SLC1A4 and PFOS-SLC1A4 complexes, respectively, indicating strong intermolecular hydrogen bonding interactions between the toxins and the core protein (Fig. 6A). To further validate binding stability, molecular dynamics simulations were performed on the PFOA-SLC1A4 and PFOS-SLC1A4 complexes (Fig. 6B). Models constructed using Gromacs software revealed that the root mean square deviation (RMSD) values for both PFOA-SLC1A4 and PFOS-SLC1A4 complexes exhibited minimal fluctuations. Equilibrium was reached after approximately 10 and 30 nanoseconds, respectively, with sustained relative stability, indicating a tightly bound structure. Solvent-accessible surface area (SASA), an indicator for evaluating protein surface area, showed stable SASA values for both PFOA-SLC1A4 and PFOS-SLC1A4 complexes. Following ligand binding to the receptor, the SASA of both complexes exhibited no significant changes over time, suggesting minimal structural impact from ligand binding. RMSF curves reflect the degree of amino acid residue fluctuations during molecular dynamics simulations. Both PFOA-SLC1A4 and PFOS-SLC1A4 complexes exhibited low RMSF values, indicating minimal fluctuations and higher structural stability. Furthermore, the number of hydrogen bonds reflects the strength of protein-ligand interactions. The PFOA-SLC1A4 and PFOS-SLC1A4 complexes also demonstrated stable hydrogen bond density and strength throughout the simulation. In summary, the PFOA-SLC1A4 and PFOS-SLC1A4 complex systems exhibit stable binding, with the complexes demonstrating robust hydrogen bonding interactions Involvement of gut microbiota and metabolites in PFOA/PFOS-induced AMD We investigated the role of the gut microbiota in PFOA/PFOS-induced AMD through microbiome-host gene association studies. By integrating the gutMGene gut microbiota database, we consolidated the interrelationships among gut bacteria, metabolites, and genes. After collating and deduplicating data from SEA and Swisstarget databases, we identified 570 human target genes associated with these gut microbes. Intersecting these with PFOA/PFOS target sites revealed 29 potential core genes, enabling construction of a network diagram linking toxins and gut microbiota (Fig. 7A). We identified four key bacterial strains sharing host genes with PFOA/PFOS (Fig. 7B): Enterococcus faecalis、Streptococcus salivarius、Turicibacter、 Lacticaseibacillus paracasei. These strains co-expressed with host genes IL-1B and PPAR, respectively. Subsequent analysis revealed these strains produce 29 unique metabolites ( Enterococcus faecalis : 9; Streptococcus salivarius : 1; Turicibacter : 15; Lacticaseibacillus paracasei : 4). Concurrent GO and KEGG enrichment analyses were conducted. BP analysis indicated associations with biological processes including response to lipopolysaccharide and response to molecule of bacterial origin (Fig. 7C-D). CC analysis highlighted their influence on membrane rafts and membrane microdomains (Fig. 7C-D). MF analysis indicated their role in DNA and RNA transcription factor binding (Fig. 7C-D). KEGG pathway analysis revealed associations with inflammation-related pathways including IL-17 signaling and TNF signaling (Fig. 7E). Collectively, these multi-omics correlations suggest that gut microbiota-derived metabolites may modulate host gene expression and signaling networks, thereby participating in the PFOA/PFOS-mediated AMD pathogenesis. Discussion This study proposes an “Environmental toxin-Eye” relationship axis and develops a rapid and comprehensive method for assessing the toxicological properties of environmental toxins. systematically revealing that PFOA and PFOS may drive AMD pathophysiology through directly binding to core targets SLC1A4 and ADAM17, thereby disrupting amino acid metabolism, triggering inflammatory responses, and inducing oxidative stress. Although growing evidence suggests an epidemiological association between PFOA/PFOS and AMD[ 15 ], the specific molecular mechanisms and signaling pathways linking PFOA/PFOS to AMD remain poorly understood. This research gap warrants particular attention. By integrating bioinformatics, network toxicology, machine learning, Mendelian randomization, gut microbiota and metabolite analysis, molecular docking, and molecular dynamics simulations, we successfully uncovered the underlying molecular mechanisms. This approach established a comprehensive, mutually reinforcing chain of evidence, significantly enhancing the credibility of our findings. It is worth emphasizing that we employ multiple complementary approaches rather than relying on a single method. This ensures that our findings are not solely derived from bioinformatics predictions but are cross-validated through analyses at different levels, thereby enhancing the reliability and interpretability of our identification of key targets and pathways. This is crucial for comprehensively understanding the extensive and long-term effects of PFOA/PFOS on AMD and developing strategies to mitigate associated risks, underscoring the necessity for in-depth investigation into the potential ocular health impacts of PFOA/PFOS and other environmental toxins. Unlike traditional, time-consuming, and costly animal studies, computational toxicology enables efficient and rapid screening of toxins and their potential health risks. The use of molecular docking and molecular dynamics simulation techniques further enhances the accuracy of toxicological prediction and assessment. These methods serve as valuable tools for understanding the complex interactions between environmental pollutants and diseases[ 16 ]. Song et al. systematically revealed how endocrine-disrupting chemicals lead to the onset and progression of diabetic microvascular disease through network toxicology analysis[ 17 ]. Guo et al. comprehensively elucidated the role of ATBC in inducing oral squamous cell carcinoma (OSCC) and its potential molecular mechanisms using network toxicology alongside molecular docking and molecular dynamics simulations, providing crucial support for ATBC toxicology mechanism research[ 18 ].These studies collectively provide a robust methodological foundation for our research. We first identified significant ocular toxicity in PFOA and PFOS through predictions from toxicology prediction websites. By integrating multi-omics datasets with PPI network interactions and diverse machine learning and bioinformatics methods, we identified candidate pathogenic genes linking PFOA/PFOS to AMD. This established a potential molecular network connecting PFOA/PFOS exposure to AMD pathogenesis. Using 128 machine learning algorithms, we identified four core genes associated with both the toxins and AMD: SLC1A4, ADAM17, EGFR, and MAPK10. Subsequently, Mendelian randomization demonstrated significant causal relationships between SLC1A4 and ADAM17 with AMD. Furthermore, the odds ratios (OR) for SLC1A4 and ADAM17 were 1.4096 and 1.3649, respectively, indicating they are important risk factors for AMD. Molecular docking and molecular dynamics simulations provided biological relevance for the aforementioned mechanisms: PFOA/PFOS exhibited strong binding energies and stable interaction patterns with SLC1A4, PFOA/PFOS possess the potential to directly impact amino acid transporter function; their interaction with ADAM17 further suggests that these toxins may amplify inflammatory signaling by affecting enzymatic cleavage or receptor release[ 19 ]. GO and KEGG functional enrichment analyses consistently indicate that oxidative stress and lipid-related pathways are highly enriched in the interacting gene sets, aligning with known AMD pathogenesis mechanisms[ 20 , 21 ]. SLC1A4, as a Na⁺-dependent neutral amino acid transporter, plays a crucial role in maintaining amino acid homeostasis in the retina and RPE [ 22 , 23 ]. Its functional abnormalities can lead to amino acid metabolism disorders. Numerous studies have demonstrated links between amino acid metabolism and retinal pathology, with glutamine, glutamate, and alanine metabolism playing key roles in AMD[ 24 ]. Research indicates that both amino acid biosynthesis and metabolic pathways are closely associated with AMD progression[ 25 ]. Research indicates that SLC36A4, another amino acid transporter, regulates the amino acid pool within RPE lysosomes[ 26 ]. These findings align with our investigation into how SLC1A4 influences AMD by modulating amino acid metabolism. ADAM17 is a crucial protease that plays a significant role in inflammation, angiogenesis, cell adhesion, and migration[ 27 , 28 ].Studies indicate that ADAM17 is a widely expressed protein in the P7 retina[ 29 ]and is crucial for ocular morphogenesis[ 30 ]. Research has discovered that the ADAM17/TNF pathway protects Drosophila retinal cells from age- and activity-related degeneration[ 31 ], offering a new perspective for elucidating the molecular mechanisms by which PFOA/PFOS induce AMD. These findings align with our investigation that the toxins amplify inflammatory signaling by activating ADAM17, leading to oxidative stress. PFOA/PFOS may promote AMD development through a synergistic dual mechanism: simultaneously disrupting amino acid transport (impairing RPE metabolic homeostasis) and amplifying ADAM17-mediated inflammatory signaling. This dual action collectively causes RPE functional imbalance and structural damage. This finding holds significant implications for elucidating the regulatory role of these toxins in AMD and its molecular mechanisms, potentially offering novel insights for AMD prevention strategies. PFOA and PFOS, due to their high persistence, bioaccumulation, and toxicity, can enter the human body and accumulate over time, making them key environmental pollutants[ 32 ]. The high persistence and bioaccumulation of PFOA and PFOS, evidenced by their half-lives exceeding 41 years in groundwater and strong bioaccumulation potential, establish sustained exposure pathways through drinking water and aquatic food chains[ 33 ]. This aligns temporally and spatially with the age-related occurrence of AMD and its increasing global incidence[ 34 ]. Furthermore, through dietary intake and the persistent presence of toxins, coupled with their continuous accumulation within the food chain, humans may ultimately experience direct and prolonged exposure to these environmental toxins. We hypothesize that this toxin may be absorbed into the eye via the intestinal route. Given the slow blood flow in the choroid of the eye, this facilitates prolonged toxin accumulation. Concurrently, the photoreceptor cells and retinal pigment epithelium (RPE) in the retina operate in a highly metabolic state, making them susceptible to toxicity. This further drives the pathophysiological progression of age-related macular degeneration (AMD). The gut microbiota is an indispensable component for maintaining human health. Abnormalities in the gut microbiome have been confirmed to be closely associated with ocular diseases in both humans and animals. Dysbiosis of the gut microbiota can trigger various ocular diseases through the“ Gut-Eye Axis ”[ 35 ]. By integrating the relationship between environmental toxins and the microbiome, we constructed a “Toxin-Gut Microbiota-Metabolite-Gene” relationship network. This revealed that PFOA/PFOS may drive the progression of AMD by disrupting gut microbiota homeostasis, promoting inflammatory responses, and reshaping the dynamic balance between probiotics and pathogens. Additionally, probiotic therapy represents a potentially effective approach for treating AMD. Supplementation with probiotics helps suppress pathogenic microorganisms, reduce inflammatory responses, and enhance the integrity of the intestinal barrier[ 36 ]. We found that the key gene IL-1B, associated with toxins and gut microbiota, is closely linked to Turicibacter and Lactobacillus paracasei within the gut microbiome. We hypothesize that toxins disrupt normal beneficial gut bacteria, thereby accelerating disease progression. Our gut microbiome association analysis expands the research perspective on toxin-host interactions. The high persistence and bioaccumulation properties of PFOA/PFOS in the environment enable them to exert long-term effects on host metabolism and immune homeostasis through oral exposure or accumulation via the food chain. This provides a basis for future exploration and prevention strategies targeting gut microbiome intervention for AMD. We progressively constructed and cross-validated the mechanism by which PFOA/PFOS may promote AMD: by disrupting amino acid metabolism (SLC1A4-related), amplifying inflammatory signaling (ADAM17-related), and leveraging the mediating role of gut microbiota metabolism, ultimately leading to retinal functional imbalance and structural damage. The novelty of this study lies in our first-ever clarification of the mechanism linking environmental toxins to AMD, raising awareness about the connection between environmental toxins and ophthalmic diseases. Although our network-based toxicology approach provides valuable insights into the mechanisms linking environmental toxins to AMD, we acknowledge certain limitations in this study. First, while the sample size meets the requirements for exploratory analysis, it may still limit the generalizability of the findings. Additionally, the predictive results obtained require further experimental validation to confirm their biological significance and accuracy. Therefore, future research should consider expanding sample sizes and enhancing population diversity through large-scale, long-term epidemiological studies to track and analyze the dynamic relationship between toxin exposure and AMD incidence, with particular focus on populations with high PFOA/PFOS exposure. Concurrently, in vivo and in vitro experiments should validate key molecular mechanisms to enhance the reliability and generalizability of research findings. This study's in-depth analysis of the toxic mechanisms of PFOA and PFOS in AMD also holds significant clinical implications: (1) Highlighting the impact of environmental toxins on ophthalmic diseases; (2) Developing preventive strategies for AMD induced by environmental toxins; (3) Identifying high-risk populations for AMD through monitoring of PFOA and PFOS exposures. Conclusion This study reveals a potential molecular mechanism linking PFOA/PFOS exposure to AMD, offering a novel perspective on the integration of environmental health science and ophthalmology. Through a comprehensive approach spanning molecular, genomic, proteomic, and gut microbiome dimensions, it provides hypotheses and insights for AMD prevention and treatment. These findings will drive further investigation into environmental factors in the pathogenesis of AMD and promote the development of effective prevention and treatment strategies. Declarations Author Contributions Z.Y.G and Y.J.P designed and supervised the overall study, performed data interpretation, and drafted the manuscript. Z.Y.G was responsible for bioinformatics analysis, data processing, and contributed to manuscript preparation. Y.J.P, Y.W.L and J.X.L participated in data collection, sample processing, and curation of clinical information. Y.J.Q and H.Y.Z conceived the study, provided critical revisions, and acted as the corresponding author. All authors read and approved the final manuscript. Ethics and Consent to Participate declarations: not applicable. Funding This study was supported by the National Natural Science Foundation of China (grant No. 82171036 & 82471083). Data Availability The corresponding authors of this paper can provide primary data that supports its conclusions upon reasonable request. Ethics declarations Ethical approval and consent to participate Not applicable. Consent for publication The publication has been approved by all co-authors. Competing interests The authors declare no competing interests. References Papadopoulou E, Stratakis N, Basagana X, Brantsaeter AL, Casas M, Fossati S, Grazuleviciene R, Smastuen Haug L, Heude B, Maitre L, et al. Prenatal and postnatal exposure to PFAS and cardiometabolic factors and inflammation status in children from six European cohorts. Environ Int. 2021;157:106853. Zango ZU, Khoo KS, Garba A, Kadir HA, Usman F, Zango MU, Da Oh W, Lim JW. A review on superior advanced oxidation and photocatalytic degradation techniques for perfluorooctanoic acid (PFOA) elimination from wastewater. Environ Res. 2023;221:115326. Shi W, Zhang Z, Li M, Dong H, Li J. Reproductive toxicity of PFOA, PFOS and their substitutes: A review based on epidemiological and toxicological evidence. Environ Res. 2024;250:118485. Guymer RH, Campbell TG. Age-related macular degeneration. Lancet. 2023;401:1459–72. Jonas JB, Cheung CMG, Panda-Jonas S. Updates on the Epidemiology of Age-Related Macular Degeneration. Asia Pac J Ophthalmol (Phila). 2017;6:493–7. Kaidonis G, Lamy R, Wu J, Yang D, Psaras C, Doan T, Stewart JM. Aqueous Fluid Transcriptome Profiling Differentiates Between Non-Neovascular and Neovascular AMD. Invest Ophthalmol Vis Sci. 2023;64:26. Fleckenstein M, Schmitz-Valckenberg S, Chakravarthy U. Age-Related Macular Degeneration: A Review. JAMA. 2024;331:147–57. Wang D, Chen Y, Li J, Wu E, Tang T, Singla RK, Shen B, Zhang M. Natural products for the treatment of age-related macular degeneration. Phytomedicine. 2024;130:155522. Yuan M, Zhang T, Sun Y, Yi S, Han G, Wei P, Chen X, Wang Y, Zhu L. Microglia-Mediated Pathological Retinal Angiogenesis Leading to Visual Impairment in Mice Exposed to Perfluorooctanoic Acid. Environ Sci Technol. 2025;59:15680–91. Tien PT, Lin HJ, Tsai YY, Lim YP, Chen CS, Chang CY, Lin CJ, Chen JJ, Wu SM, Huang YJ, Wan L. Perfluorooctanoic acid in indoor particulate matter triggers oxidative stress and inflammation in corneal and retinal cells. Sci Rep. 2020;10:15702. Chen X, Li J, Xu N, Li X, Li J, Guo Q, Zhang J, Miao H, Huang L. Serum lipids mediate the association of per- and polyfluoroalkyl substances exposure and age-related macular degeneration. PLoS ONE. 2025;20:e0317678. Clough E, Barrett T, Wilhite SE, Ledoux P, Evangelista C, Kim IF, Tomashevsky M, Marshall KA, Phillippy KH, Sherman PM, et al. NCBI GEO: archive for gene expression and epigenomics data sets: 23-year update. Nucleic Acids Res. 2024;52:D138–44. Zhang W, Tian Y, Chen B, Xu S, Wu L. PFOA/PFOS Facilitated Intestinal Fatty Acid Absorption by Activating the PPARalpha Pathway: Insights from Organoids Model. Environ Health (Wash). 2024;2:85–94. Qi C, He G, Qian K, Guan S, Li Z, Liang S, Liu J, Ke X, Zhang S, Lu M, et al. gutMGene v2.0: an updated comprehensive database for target genes of gut microbes and microbial metabolites. Nucleic Acids Res. 2025;53:D783–8. Yuan M, Yi S, Wang X, Han G, Wei P, Lv Z, Gui B, Chen X, Wang Y, Zhu L. Promoted Translocation of Perfluorooctanoic Acid across the Blood-Retinal Barrier due to its Inhibition of Tight Junction Assembly by Antagonizing LPAR1. Environ Sci Technol. 2025;59:4807–19. Feng C, Yan W, Mei Z, Luo X. Exploring the toxicological impact of bisphenol a exposure on psoriasis through network toxicology, machine learning, and multi-dimensional bioinformatics analysis. J Environ Manage. 2025;385:125708. Song S, Huang L, Zhou X, Yu J. Exploring the toxicological network in diabetic microvascular disease. Int J Surg. 2025;111:3895–907. Guo Y, Liu Y, Chen Y, Du S, Zheng Y, Wang L. The mechanisms of environmental pollutant acetyl tributyl citrate induced oral squamous cell carcinoma using network toxicology, molecular docking and molecular dynamics simulation. Int J Surg. 2025;111:7873–85. Saad MI, Jenkins BJ. The protease ADAM17 at the crossroads of disease: revisiting its significance in inflammation, cancer, and beyond. FEBS J. 2024;291:10–24. Fleckenstein M, Keenan TDL, Guymer RH, Chakravarthy U, Schmitz-Valckenberg S, Klaver CC, Wong WT, Chew EY. Age-related macular degeneration. Nat Rev Dis Primers. 2021;7:31. Tan LX, Germer CJ, La Cunza N, Lakkaraju A. Complement activation, lipid metabolism, and mitochondrial injury: Converging pathways in age-related macular degeneration. Redox Biol. 2020;37:101781. Elazar D, Alvarez N, Drobeck S, Gunn TM. SLC1A4 and Serine Homeostasis: Implications for Neurodevelopmental and Neurodegenerative Disorders. Int J Mol Sci 2025, 26. Hansman DS, Du J, Casson RJ, Peet DJ. Eye on the horizon: The metabolic landscape of the RPE in aging and disease. Prog Retin Eye Res. 2025;104:101306. Xia M, Zhang F. Amino Acids Metabolism in Retinopathy: From Clinical and Basic Research Perspective. Metabolites 2022, 12. Hou XW, Wang Y, Pan CW. Metabolomics in Age-Related Macular Degeneration: A Systematic Review. Invest Ophthalmol Vis Sci. 2020;61:13. Shang P, Valapala M, Grebe R, Hose S, Ghosh S, Bhutto IA, Handa JT, Lutty GA, Lu L, Wan J, et al. The amino acid transporter SLC36A4 regulates the amino acid pool in retinal pigmented epithelial cells and mediates the mechanistic target of rapamycin, complex 1 signaling. Aging Cell. 2017;16:349–59. Caolo V, Swennen G, Chalaris A, Wagenaar A, Verbruggen S, Rose-John S, Molin DG, Vooijs M, Post MJ. ADAM10 and ADAM17 have opposite roles during sprouting angiogenesis. Angiogenesis. 2015;18:13–22. Wang X, Wang T, Kaneko S, Kriukov E, Lam E, Szczepan M, Chen J, Gregg A, Wang X, Fernandez-Gonzalez A, et al. Photoreceptors inhibit pathological retinal angiogenesis through transcriptional regulation of Adam17 via c-Fos. Angiogenesis. 2024;27:379–95. Toonen JA, Ronchetti A, Sidjanin DJ. A Disintegrin and Metalloproteinase10 (ADAM10) Regulates NOTCH Signaling during Early Retinal Development. PLoS ONE. 2016;11:e0156184. Sel S, Kalinski T, Enssen I, Kaiser M, Nass N, Trau S, Wollensak G, Brauer L, Jager K, Paulsen F. Expression analysis of ADAM17 during mouse eye development. Ann Anat. 2012;194:334–8. Muliyil S, Levet C, Dusterhoft S, Dulloo I, Cowley SA, Freeman M. ADAM17-triggered TNF signalling protects the ageing Drosophila retina from lipid droplet-mediated degeneration. EMBO J. 2020;39:e104415. Wee SY, Aris AZ. Environmental impacts, exposure pathways, and health effects of PFOA and PFOS. Ecotoxicol Environ Saf. 2023;267:115663. Cheng X, Wei Y, Zhang Z, Wang F, He J, Wang R, Xu Y, Keerman M, Zhang S, Zhang Y, et al. Plasma PFOA and PFOS Levels, DNA Methylation, and Blood Lipid Levels: A Pilot Study. Environ Sci Technol. 2022;56:17039–51. Collaborators GBDGA. Global burden of vision impairment due to age-related macular degeneration, 1990–2021, with forecasts to 2050: a systematic analysis for the Global Burden of Disease Study 2021. Lancet Glob Health. 2025;13:e1175–90. Floyd JL, Grant MB. The Gut-Eye Axis: Lessons Learned from Murine Models. Ophthalmol Ther. 2020;9:499–513. Jin Q, Wang S, Yao Y, Jiang Q, Li K. The gut-eye axis: from brain neurodegenerative diseases to age-related macular degeneration. Neural Regen Res. 2025;20:2741–57. Additional Declarations No competing interests reported. Supplementary Files Supplement.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8563077","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":581386633,"identity":"3a2c8d86-1b73-4c2a-88c9-5ecfac607e07","order_by":0,"name":"Zhenyu Guo","email":"","orcid":"","institution":"Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zhenyu","middleName":"","lastName":"Guo","suffix":""},{"id":581386635,"identity":"2cb23e1a-a67c-451d-92e1-91ce1ad56de7","order_by":1,"name":"Yujun Peng","email":"","orcid":"","institution":"Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yujun","middleName":"","lastName":"Peng","suffix":""},{"id":581386637,"identity":"89ee6bd2-9ef7-4c5e-9ff2-0ef97d51709f","order_by":2,"name":"Yiwei Lin","email":"","orcid":"","institution":"Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yiwei","middleName":"","lastName":"Lin","suffix":""},{"id":581386639,"identity":"ea4dfc17-69c6-4372-a349-c44c31f1205f","order_by":3,"name":"Jiaxi Liu","email":"","orcid":"","institution":"Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jiaxi","middleName":"","lastName":"Liu","suffix":""},{"id":581386641,"identity":"aebe5812-0ad8-41b2-99c6-ec5334d72eaf","order_by":4,"name":"Yongjie Qin","email":"","orcid":"","institution":"Guangdong Provincial People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"Yongjie","middleName":"","lastName":"Qin","suffix":""},{"id":581386645,"identity":"efb8346f-c5c0-4a73-93bc-6617745026c7","order_by":5,"name":"Hongyang Zhang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA1ElEQVRIie3QMQrCMBSA4ScFuzypY0qlNxAiBUtB6lUSBLs6OlaEuHgAxXuIY0SwSw9Qt2Z3qLugdXRKRsF8c/6XvABY1g+i4ErJny/03NVKNWYJ8vqROwN/e1lHxCyBaLTPnQndZaKPJklM2DzonbpIfSWAQBoOc02S7Pg16JWIccBFvYBZNJa6h1XtLdglmBz4hhKQ/GiQjNuEIr2dBUHDpF1fMKRVxzQp77xuSon+lrefTE12KTIp2VJOPbdQqlmmoTaBPvuaoDv+4WmHWpZl/b03cfRGJYLxZBoAAAAASUVORK5CYII=","orcid":"","institution":"Southern Medical University","correspondingAuthor":true,"prefix":"","firstName":"Hongyang","middleName":"","lastName":"Zhang","suffix":""}],"badges":[],"createdAt":"2026-01-09 16:53:30","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8563077/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8563077/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":101440967,"identity":"4b2d1759-1cd5-45b7-b33d-00d0a5a0ade8","added_by":"auto","created_at":"2026-01-29 16:57:45","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":31371242,"visible":true,"origin":"","legend":"\u003cp\u003eWorkflow of data processing and analysis in this study\u003c/p\u003e","description":"","filename":"Fig.1.png","url":"https://assets-eu.researchsquare.com/files/rs-8563077/v1/4c99cd2b4861dde698d2b17d.png"},{"id":101440970,"identity":"630ed081-5583-4b2f-8692-60fd88fe96ac","added_by":"auto","created_at":"2026-01-29 16:57:45","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":4966597,"visible":true,"origin":"","legend":"\u003cp\u003eIdentification of PFOA/PFOS and AMD-related target genes: (A) PCA scatter plot shows distinct separation between GSE99248 and GSE1719 datasets before batch correction, indicating batch effects. (B) PCA scatter plot after batch correction shows the integration of GSE99248 and GSE1719 datasets, indicating reduced batch effects. (C) Volcano plot shows DEGs based on logFC and significance. Red dots are upregulated, blue dots are downregulated, and gray dots are nonsignificant. (D) Heatmap shows DEG expression patterns across samples. Red indicates upregulation, blue indicates downregulation. (E) Gene dendrogram from WGCNA shows hierarchical clustering based on co-expression. Module colors in the lower panel represent different gene modules. (F) Module-trait relationships heatmap shows correlations between WGCNA-identified modules and sample traits (Normal vs. Disease). Values in boxes indicate correlation coefficients and P-values. (G) Intersection of predicted PFOA/PFOS targets and AMD-associated genes\u003c/p\u003e","description":"","filename":"Fig.2.png","url":"https://assets-eu.researchsquare.com/files/rs-8563077/v1/8e3be4eac570de0ce691cbf4.png"},{"id":101440968,"identity":"4fc58eeb-46ac-4697-8106-00ec98fcbd46","added_by":"auto","created_at":"2026-01-29 16:57:45","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":19807696,"visible":true,"origin":"","legend":"\u003cp\u003ePPI, GO, and KEGG enrichment analysis: (A). PPI Network. GO Analysis: (B). Enrichment circle visualization. (C) Bar chart analyses. (D). KEGG pathways.\u003c/p\u003e","description":"","filename":"Fig.3.png","url":"https://assets-eu.researchsquare.com/files/rs-8563077/v1/7abd7e86334cace99501d9be.png"},{"id":101440965,"identity":"71ff3439-5118-4912-b62a-97b29ee19ab9","added_by":"auto","created_at":"2026-01-29 16:57:44","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":14942444,"visible":true,"origin":"","legend":"\u003cp\u003eIdentification of core genes in PFOA/PFOS-induced AMD and prediction of core gene functions and construction of miRNA-TF-mRNA regulatory networks: (A) Model performance comparison: Heatmap displays AUC values for various models across cohorts. Left column = models, right column = AUC (higher = better). Colors indicate cohort sources (B) Calibration curve assessing the prediction accuracy of the nomogram for the training set; higher convergence of the three curves indicates better alignment between predicted and actual probabilities. (C) ROC Curves: ROC curves for key genes (SLC1A4, ADAM17, EGFR, MAPK10). AUC indicates predictive performance. (D) ROC curves evaluate the diagnostic performance of the feature genes for the training set. (E) ROC curves evaluate the diagnostic performance of the feature genes for the validation set. (F) Nomogram predicting the risk in the training set based on PFOA/PFOS-AMD-Related Genes. (G) Circle graphs showed the chromosomal locations of the PFOA/PFOS-AMD-Related Genes. (H) Functional and pathway prediction analyses using GeneMANIA for PFOA/PFOS-AMD-Related Genes. Surrounding nodes represent predicted genes, with larger node sizes indicating stronger correlation. Colors represent functional enrichment types. (I) Diagram of the miRNA-TF-mRNA regulatory network, where red circles represent genes, green diamonds represent predicted transcription factors (TFs), and orange V-shapes represent miRNAs.\u003c/p\u003e","description":"","filename":"Fig.4.png","url":"https://assets-eu.researchsquare.com/files/rs-8563077/v1/1093d449f621ee48126cf1c0.png"},{"id":101440966,"identity":"ecdc63f2-1ea5-4249-9065-fe6dd2380ac9","added_by":"auto","created_at":"2026-01-29 16:57:44","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":15490426,"visible":true,"origin":"","legend":"\u003cp\u003eMendelian randomization: (A) Validate the causal relationship between key target protein and AMD (id, beta, se, OR, or_lci95, or_uci95 and P - values were for the respective MR analysis); (B-C) Results of mendelian randomization analysis. (Different colored lines in the figure represent five different analysis methods （MR Egger, Weighted median, Inverse variance weighted, Simple mode, Weighted mode）\u003c/p\u003e","description":"","filename":"Fig.5.png","url":"https://assets-eu.researchsquare.com/files/rs-8563077/v1/3a6eaa7549f1e951ba2d19d0.png"},{"id":101440969,"identity":"b256239e-81db-4053-8a09-db8b5bb821fa","added_by":"auto","created_at":"2026-01-29 16:57:45","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":41673799,"visible":true,"origin":"","legend":"\u003cp\u003ePrediction of direct interaction between PFOA/PFOS and SLC1A4/ADAM17 protein: (A) Molecular docking visualization of toxins and core targets; (B) Molecular dynamics simulation of toxins targeting core sites.\u003c/p\u003e","description":"","filename":"Fig.6.png","url":"https://assets-eu.researchsquare.com/files/rs-8563077/v1/468187a8982d2bec0960a9ed.png"},{"id":101440971,"identity":"34567f2d-9080-4ab5-8458-27d7bd26fa7d","added_by":"auto","created_at":"2026-01-29 16:57:45","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":30498583,"visible":true,"origin":"","legend":"\u003cp\u003eGut microbiota-host gene association analysis reveals potential mechanisms of PFOA/PFOS-induced AMD: (A) Common targets of toxins and microorganisms. (B) Protein-protein interaction network constructed using the SEA database. Top 5 enriched terms for (C) Biological processes and molecular functions, and (D) Biological processes and cellular components. (E) KEGG pathways.\u003c/p\u003e","description":"","filename":"Fig.7.png","url":"https://assets-eu.researchsquare.com/files/rs-8563077/v1/0eadee984df22e19fb20c660.png"},{"id":109296246,"identity":"874cae7f-e0e8-45ef-96b7-0b1506b19c97","added_by":"auto","created_at":"2026-05-15 08:46:21","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":128688357,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8563077/v1/68019fc3-78fd-4235-9f22-4595aee6616e.pdf"},{"id":101440964,"identity":"dfc07c2c-f523-480f-ac10-42c7fc9d2a32","added_by":"auto","created_at":"2026-01-29 16:57:44","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":752688,"visible":true,"origin":"","legend":"","description":"","filename":"Supplement.docx","url":"https://assets-eu.researchsquare.com/files/rs-8563077/v1/22b4b75649ba7c10507de86b.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Integrating network toxicology, machine learning, gut microbiome analysis, and structural validation to reveal the molecular mechanism linking PFOA and PFOS exposure to age-related macular degeneration","fulltext":[{"header":"Introduction","content":"\u003cp\u003ePer-and polyfluoroalkyl substances (PFAS) represent a class of highly fluorinated aliphatic compounds characterized by exceptional physicochemical stability. Owing to their unique surface activity and resistance to heat and degradation, PFAS are widely incorporated into industrial and consumer products such as textiles, surfactants, furniture coatings, food packaging, and firefighting foams[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. These compounds readily infiltrate global ecosystems through wastewater discharge, atmospheric transport, and bioaccumulation. Among them, perfluorooctanoic acid (PFOA) and perfluorooctanesulfonic acid (PFOS) have drawn particular attention because of their extensive production and demonstrated environmental persistence and toxicity[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. They are persistent organic pollutants listed in the Stockholm Convention, which have become serious environmental pollutants due to their extreme persistence and transboundary mobility[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAge-related macular degeneration (AMD) causes irreversible vision loss or blindness[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e] and is the leading cause of blindness in older worldwide[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Its pathogenesis results from complex interactions between genetics and environmental risk factors. With aging, increased production of reactive oxygen species coupled with impaired clearance functions leads to cumulative oxidative damage in the retina. Over time, the accumulation of oxidative stress and alterations in antioxidant responses result in impaired function of retinal pigment epithelial cells, diminishing their capacity to support and maintain photoreceptor function[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Concurrently, environmental factors associated with AMD (e.g., smoking) exacerbate oxidative stress and compromise blood supply[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Therefore, reducing environmental risk factors may help decrease the prevalence and progression of AMD, offering new perspectives and insights for preventing and alleviating its economic burden[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eCurrent research findings on environmental toxins and AMD remain limited and isolated. It has been noted that human visual impairment is closely associated with perfluoroalkyl and polyfluoroalkyl substances[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Studies indicate that perfluorooctanoic acid (PFOA) exposure can induce inflammatory responses in retinal pigment epithelial cells[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Furthermore, epidemiological research has identified an association between PFOA/PFOS exposure and AMD. Chen et al[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e].examined high-detection-level serum perfluoroalkyl substances (PFAS) including PFOA and PFOS in 1,605 participants recruited during the 2005\u0026ndash;2008 NHANES. They identified an association between PFOS exposure and AMD risk.\u003c/p\u003e \u003cp\u003eHowever, the association between PFOA/PFOS and age-related macular degeneration remains confined to the level of epidemiological correlation. The complex biological environment hinders the elucidation of precise pathogenic mechanisms, making targeted research into the mechanisms by which toxin exposure influences AMD critically important. To bridge this gap, this study employs a systematic network toxicology approach integrating toxicological targets of pollutants with AMD-associated disease genes: Refining the PFOA/PFOS-driven AMD pathogenesis network through multi-omics data integration; Identifying key hub molecules via machine learning-guided topological and functional enrichment analysis; Validating causal relationships between core genes and AMD using MR; and validated the binding effects between toxins and targets through molecular docking and molecular dynamics simulations. This study aims to provide scientific evidence for the prevention and treatment of ophthalmic diseases involving PFOA/PFOS. It proposes a \u0026ldquo;Environmental toxin-Eye\u0026rdquo; relationship network, offering novel insights into the link between environmental pollutants and ophthalmic diseases, and establishing a more comprehensive, reliable, and in-depth scientific foundation for the interdisciplinary comprehensive assessment of environmental pollutants.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eToxicity prediction and target identification\u003c/h2\u003e \u003cp\u003eThe molecular structures and SMILES strings of PFOA and PFOS were obtained from the PubChem database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubchem.ncbi.nlm.nih\u003c/span\u003e\u003cspan address=\"https://pubchem.ncbi.nlm.nih\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). The ADMETLab 3.0 platform was then used to predict their physicochemical and toxicological properties. Potential molecular targets were identified through the ChEMBL, SwissTarget Prediction, and PharmMapper databases, restricting the search to Homo sapiens to ensure biological relevance. The UniProt database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.uniprot.org/\u003c/span\u003e\u003cspan address=\"https://www.uniprot.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was used for cross-validation and standardization of target information. We integrated, deduplicated, and merged the targets for PFOA and PFOS to ensure their completeness, ultimately compiling a comprehensive toxicant database.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eAcquisition of disease-associated genes\u003c/h3\u003e\n\u003cp\u003eThe Gene Expression Omnibus (GEO) contains high-throughput gene expression data[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e] submitted by research institutions worldwide. We curated three age-related macular edema (ARME) transcriptome datasets (GSE99248, GSE1719, and GSE29801) from the GEO database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.ncbi.nlm.nih.gov/geo/\u003c/span\u003e\u003cspan address=\"https://www.ncbi.nlm.nih.gov/geo/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). GSE99248 (including 15 control samples and 16 AMD samples) and GSE1719 (including 18 control samples and 18 AMD samples) serve as the discovery cohort, while GSE29801 (including 150 control samples and 142 AMD samples from macular RPE-choroid or extramacular retina tissue) functions as the validation cohort.\u003c/p\u003e \u003cp\u003eTo mitigate batch effects, a multi-stage normalization process was implemented, including substitute variable analysis (SVA) and gene expression standardization (GES). Corrected principal component analysis (PCA) revealed improved clustering of samples across batches in the reduced-dimension space, confirming successful data harmonization. To comprehensively identify AMD disease targets for subsequent analysis, we systematically collected AMD-related targets by integrating GeneCards, OMIM disease databases, and GEO database resources.\u003c/p\u003e\n\u003ch3\u003eProtein-Protein Interaction (PPI) network and core target identification\u003c/h3\u003e\n\u003cp\u003eIntersection analysis between toxin-related and AMD-related targets was performed using the Venny tool. The overlapping targets were uploaded to the STRING database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://cn.string-db.org/\u003c/span\u003e\u003cspan address=\"https://cn.string-db.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) to construct a PPI network, which was visualized in Cytoscape 3.10.3. Key hub genes and subnetworks were identified using the CytoHubba plugin\u003c/p\u003e\n\u003ch3\u003eFunctional enrichment analysis\u003c/h3\u003e\n\u003cp\u003eGene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were conducted using the DAVID platform (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://davidbioinformatics.nih.gov/\u003c/span\u003e\u003cspan address=\"https://davidbioinformatics.nih.gov/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and the clusterProfiler package in R. Biological process (BP), cellular component (CC), and molecular function (MF) categories were analyzed, with significance thresholds set at P\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/p\u003e\n\u003ch3\u003eMachine learning for key target recognition\u003c/h3\u003e\n\u003cp\u003eTo systematically identify biomarkers for toxins and diseases, we constructed an integrated machine learning prediction framework combining multiple algorithms. Based on training set expression profiles, we employed multiple classical machine learning algorithms\u0026mdash;Lasso regression (glmnet, alpha\u0026thinsp;=\u0026thinsp;1), Ridge regression (glmnet, alpha\u0026thinsp;=\u0026thinsp;0), Elastic Net (glmnet, alpha between 0\u0026ndash;1), Random Forest (randomForestSRC, ntree\u0026thinsp;=\u0026thinsp;1000, nodesize\u0026thinsp;=\u0026thinsp;5), GBM (gbm, n.trees\u0026thinsp;=\u0026thinsp;10000, interaction.depth\u0026thinsp;=\u0026thinsp;3, shrinkage\u0026thinsp;=\u0026thinsp;0.001, cv.folds\u0026thinsp;=\u0026thinsp;10), Support Vector Machine (e1071), Naive Bayes (e1071), Partial Least Squares Generalized Linear Model (plsRglm), BART, etc.\u0026mdash;to develop 128 predictive models. Hyperparameter optimization was performed using 10-fold cross-validation, with data partitioned into training and internal validation subsets via stratified sampling. Optimal single-model predictions were subsequently integrated using a stacked ensemble learning strategy. Finally, gene expression patterns were visualized using the pheatmap package.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eFunctional prediction of toxin-disease genes and construction of miRNA-TF-mRNA regulatory networks\u003c/h2\u003e \u003cp\u003eTo systematically investigate the functional roles of toxin-disease-associated genes in disease pathogenesis, comprehensive functional and pathway prediction analyses were conducted using GeneMANIA(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://genemania.org/\u003c/span\u003e\u003cspan address=\"http://genemania.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Simultaneously, by integrating three prediction databases (Starbase, miRTarBase, and TargetScan), we identified miRNAs targeting toxin- and disease-associated genes, from which 23 miRNAs with the strongest disease association were selected. Furthermore, using the NetworkAnalyst database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.networkanalyst.ca/\u003c/span\u003e\u003cspan address=\"https://www.networkanalyst.ca/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), we identified the most strongly associated transcription factors (TFs) among toxin- and disease-associated genes. Finally, based on the above data, a visual miRNA-TF-mRNA regulatory network was constructed using Cytoscape 3.10.3, elucidating the potential molecular mechanisms linking toxins and diseases.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eMendelian randomization analysis of core genes and validation using the HPA database\u003c/h3\u003e\n\u003cp\u003eTo further infer causality, the TwoSampleMR R package (v0.5.6; \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://mrcieu.github.io/TwoSampleMR/\u003c/span\u003e\u003cspan address=\"https://mrcieu.github.io/TwoSampleMR/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) performed a two-sample mendelian randomization analysis on pooled GWAS SNP data to evaluate causal relationships between key genes and AMD. For the exposure variable (SLC1A4 expression), we utilized the eQTL dataset (OpenGWAS ID: eqtl-a-ENSG00000115902, Vosa U, 2018), which comprises 31,684 European individuals and 18,504 SNPs; For the exposure variable (ADAM17 expression), we utilized the eQTL dataset (OpenGWAS ID: eqtl-a-ENSG00000151694, Vosa U, 2018), which includes 31,644 European individuals and 19,103 SNPs. For the outcome variable (Age-related macular degeneration), the GWAS dataset (OpenGWAS ID: finn-b-H7_AMD, 2021) was used, comprising 3,763 cases and 205,359 controls (total sample size\u0026thinsp;=\u0026thinsp;209,122) of European ancestry, covering 16,380,424 SNPs. Instrument variables were selected using stringent criteria: P\u0026thinsp;\u0026lt;\u0026thinsp;5e-8, r2\u0026thinsp;\u0026lt;\u0026thinsp;0.001, and kb\u0026thinsp;\u0026lt;\u0026thinsp;10,000. Results were validated using inverse variance weighting (IVW) combined with sensitivity, heterogeneity, and pleiotropy tests. All analyses were performed in R 4.5.0 using TwoSampleMR v0.5.6 and MRPRESSO v1.0.\u003c/p\u003e \u003cp\u003eAdditionally, by integrating the HPA database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.proteinatlas.org/\u003c/span\u003e\u003cspan address=\"https://www.proteinatlas.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and employing single-cell distribution analysis, we focused on observing gene expression in vision-related cell types such as M\u0026uuml;ller glia cells, photoreceptor cells, and bipolar cells. UMap visualization was employed to analyze gene distribution patterns across different cell populations. Fluorescence microscopy images were collected and organized for subcellular localization validation, enabling observation of protein localization within specific cellular compartments such as the nucleus, cytoplasm, and endoplasmic reticulum.\u003c/p\u003e\n\u003ch3\u003eMolecular docking and molecular dynamics (MD) simulations\u003c/h3\u003e\n\u003cp\u003eThe two-dimensional (2D) molecular structure data for PFOA and PFOS were obtained from the PubChem database. The PDB format files for core genes were retrieved from the RCSB database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.rcsb.org/\u003c/span\u003e\u003cspan address=\"https://www.rcsb.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). AutoDock Vina software was employed for molecular docking, with results visualized using pymol software. Binding energy was utilized to evaluate interactions between ligands and receptors: A binding energy below 0 kcal/mol indicates spontaneous binding, while values below \u0026minus;\u0026thinsp;5 kcal/mol suggest stable binding. A 100 ns molecular dynamics simulation was performed under standard temperature and pressure conditions using GROMACS 2025.2 software, employing the CHARMM36 force field and TIP3P water model for solvation.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eGut microbiota analysis\u003c/h2\u003e \u003cp\u003eGiven that the gut microbiota also plays an indispensable role in ophthalmic diseases, and considering that PFOA/PFOS are readily absorbed and poorly metabolized in the gut[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e], we utilized the gut microbial gene database (gutMGene; \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.gutmgene.org/\u003c/span\u003e\u003cspan address=\"http://www.gutmgene.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) to retrieve all host-gene-based associations[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. We focused on the \u0026ldquo;human\u0026rdquo; species filter to identify gut microbiota sharing common target genes with the toxins. Metabolites associated with these microbiota were retrieved through the search module and rigorously filtered based on host species = \u0026ldquo;human,\u0026rdquo; association source = \u0026ldquo;literature association,\u0026rdquo; and PubChem CID availability. We then employed the Similarity ensemble approach (SEA, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://sea.bkslab.org/\u003c/span\u003e\u003cspan address=\"https://sea.bkslab.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and SwissTargetPrediction (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://swisstargetprediction.ch/\u003c/span\u003e\u003cspan address=\"https://swisstargetprediction.ch/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) to gather potential targets for these metabolites. GO and KEGG analyses were performed on metabolite-associated genes to elucidate potential metabolic pathways. Finally, network diagrams linking toxins, target genes, gut microbiota, metabolites, and host genes were constructed using STRING and Cytoscape.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eToxicity assessment and target identification of PFOA and PFOS\u003c/h2\u003e \u003cp\u003eThe ADMEtLab website was used to assess the toxicity profiles of PFOA and PFOS. The results predicted extensive eye irritation and eye corrosion toxicity for PFOA and PFOS, indicating a risk of ocular damage (Supplemental Digital Content Fig.\u0026nbsp;1A-B). Based on these findings, we employed network toxicology to analyze the role of PFOA/PFOS in promoting AMD progression, aiming to further understand whether these toxins increase the risk of developing AMD. By integrating data from SwissTargetPrediction, ChEMBL, and pharmmapper databases, we compiled a total of 439 potential targets for PFOA and PFOS, thereby gaining insights into the possible molecular pathways through which these toxins may induce pathological changes.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eIdentification of AMD-associated genes\u003c/h2\u003e \u003cp\u003eWe analyzed transcriptome data from the GEO database (GSE99248, GSE1719) to compare gene expression profiles between AMD patients and healthy controls. To minimize batch effects, we merged the GSE99248 and GSE1719 datasets and performed comprehensive normalization of the gene expression matrix. Principal Component Analysis (PCA) revealed improved data distribution after normalization, with the normalized dataset exhibiting clearer clustering patterns (Fig.\u0026nbsp;2A-B). Subsequent differential expression analysis identified 47 genes significantly altered in AMD. Expression changes for these genes were visualized using volcano plots and heatmaps (Fig.\u0026nbsp;2C-D). Subsequently, WGCNA analysis was performed. Based on this parameter, a Topology Overlap Matrix (TOM) was constructed and hierarchical clustering was conducted to identify co-expression modules. Nine independent gene modules were analyzed (Fig.\u0026nbsp;2E). Module-trait association analysis revealed that specific modules were significantly associated with AMD (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Fig.\u0026nbsp;2F). To comprehensively capture potential genes involved in PFOA/PFOS-induced AMD and ensure subsequent analysis reliability, we also screened AMD-related target genes from GeneCards and OMIM disease databases. By integrating differentially expressed genes with WGCNA module genes and removing duplicates, we identified 1,687 disease-associated candidate genes. Finally, by integrating PFOA/PFOS toxicant targets, we identified 97 overlapping genes potentially involved in toxicant-mediated AMD (Fig.\u0026nbsp;2G).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eProtein\u0026ndash;Protein Interaction (PPI) network and functional enrichment analysis\u003c/h2\u003e \u003cp\u003eThe 97 overlapping targets were uploaded to the STRING database to construct a protein\u0026ndash;protein interaction (PPI) network, which was visualized using Cytoscape 3.10.3 (Fig.\u0026nbsp;3A). Functional enrichment analysis with DAVID identified 2,158 significant GO terms\u0026mdash;comprising 1,929 biological processes (BP), 56 cellular components (CC), and 173 molecular functions (MF). BP terms were mainly enriched in oxidative stress, reactive oxygen species response, and chemical stress response (Fig.\u0026nbsp;3B\u0026ndash;C). CC enrichment involved vesicle lumen, secretory granule lumen, and cytoplasmic vesicle lumen, while MF terms included phosphatase binding, MAP kinase activity, and lipoprotein particle binding (Fig.\u0026nbsp;3B\u0026ndash;C). KEGG analysis highlighted enrichment in lipid metabolism, atherosclerosis, and endocrine resistance pathways (Fig.\u0026nbsp;3D). These findings collectively suggest that PFOA and PFOS may participate in the onset and progression of AMD by regulating these key biological pathways.\u003c/p\u003e \u003cp\u003e \u003cb\u003eIdentification of core genes using machine learning and prediction of core gene functions and construction of miRNA-TF-mRNA regulatory networks\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThrough comprehensive machine learning analysis of candidate targets, we constructed 128 predictive models to identify core genes affected by PFOA/PFOS in AMD. We selected Stepglm[both]\u0026thinsp;+\u0026thinsp;RF as the ensemble model due to its outstanding performance with high accuracy in both training and validation phases (Fig.\u0026nbsp;4A). Calibration curves validated the predictive accuracy of the machine learning model on the training dataset (Fig.\u0026nbsp;4B), identifying four key genes: SLC1A4, ADAM17, EGFR, and MAPK10. ROC curve analysis confirmed the diagnostic potential of these core genes (AUC\u0026thinsp;\u0026gt;\u0026thinsp;0.75, Fig.\u0026nbsp;4C). The diagnostic performance of these feature genes was separately evaluated on the training and validation sets using ROC curves: the AUC value for the test set was 0.954, and the AUC value for the validation set was 0.787 (Figs.\u0026nbsp;4D-E). Subsequently, nomograms were constructed to score each gene's expression, precisely quantifying AMD risk. SLC1A4 and ADAM17 emerged as clear risk factors for the disease (Fig.\u0026nbsp;4F). Finally, we mapped the selected genes to their chromosomal locations (Fig.\u0026nbsp;4G). Functional prediction of the four core genes via GeneMANIA revealed their involvement in the ERBB signaling pathway and amino acid phosphorylation regulation (Fig.\u0026nbsp;4H). Potential miRNAs and transcription factors (TFs) targeting these genes were identified using StarBase, miRTarBase, TargetScan, and NetworkAnalyst databases. The final regulatory network visualized in Cytoscape 3.10.3 encompassed 47 miRNAs, 13 TFs, and 4 core genes, revealing intricate regulatory interactions among these molecules (Fig.\u0026nbsp;4I).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eMendelian randomization and HPA validation of hub genes\u003c/h2\u003e \u003cp\u003eNext, we performed Mendelian randomization (MR) analysis using the four core genes as exposure variables and AMD as the outcome variable to investigate the causal relationship between the core genes and AMD. Results indicated (Fig.\u0026nbsp;5A) that SLC1A4 (OR\u0026thinsp;=\u0026thinsp;1.4096) and ADAM17 (OR\u0026thinsp;=\u0026thinsp;1.3649) were risk factors for AMD, suggesting that increased expression of these genes correlates with a higher probability of AMD occurrence (Figs.\u0026nbsp;5B-C). Both genes had P-values\u0026thinsp;\u0026lt;\u0026thinsp;0.05, indicating highly significant causal relationships with AMD and confirming their statistical significance.\u003c/p\u003e \u003cp\u003eValidation via the HPA database revealed their widespread distribution in photoreceptor cells through single-cell localization analysis (Supplemental Digital Content Fig.\u0026nbsp;2A). Subcellular localization analysis indicated SLC1A4 predominantly resides in the centrosome, while ADAM17 is localized in the cytosol (Supplemental Digital Content Fig.\u0026nbsp;2B).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eMolecular docking and Molecular dynamic simulations\u003c/h2\u003e \u003cp\u003eTo further investigate the binding efficiency of the toxins PFOA/PFOS with MR-validated target proteins, molecular docking analyses were performed for SLC1A4 and ADAM17 proteins with both toxins. It can be observed that SLC1A4 exhibits high binding energies with both toxins. This is attributed to the presence of ASN-402 and ASN-402, MET-221, SER-170 residues in the PFOA-SLC1A4 and PFOS-SLC1A4 complexes, respectively, indicating strong intermolecular hydrogen bonding interactions between the toxins and the core protein (Fig.\u0026nbsp;6A). To further validate binding stability, molecular dynamics simulations were performed on the PFOA-SLC1A4 and PFOS-SLC1A4 complexes (Fig.\u0026nbsp;6B). Models constructed using Gromacs software revealed that the root mean square deviation (RMSD) values for both PFOA-SLC1A4 and PFOS-SLC1A4 complexes exhibited minimal fluctuations. Equilibrium was reached after approximately 10 and 30 nanoseconds, respectively, with sustained relative stability, indicating a tightly bound structure. Solvent-accessible surface area (SASA), an indicator for evaluating protein surface area, showed stable SASA values for both PFOA-SLC1A4 and PFOS-SLC1A4 complexes. Following ligand binding to the receptor, the SASA of both complexes exhibited no significant changes over time, suggesting minimal structural impact from ligand binding. RMSF curves reflect the degree of amino acid residue fluctuations during molecular dynamics simulations. Both PFOA-SLC1A4 and PFOS-SLC1A4 complexes exhibited low RMSF values, indicating minimal fluctuations and higher structural stability. Furthermore, the number of hydrogen bonds reflects the strength of protein-ligand interactions. The PFOA-SLC1A4 and PFOS-SLC1A4 complexes also demonstrated stable hydrogen bond density and strength throughout the simulation. In summary, the PFOA-SLC1A4 and PFOS-SLC1A4 complex systems exhibit stable binding, with the complexes demonstrating robust hydrogen bonding interactions\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eInvolvement of gut microbiota and metabolites in PFOA/PFOS-induced AMD\u003c/h2\u003e \u003cp\u003eWe investigated the role of the gut microbiota in PFOA/PFOS-induced AMD through microbiome-host gene association studies. By integrating the gutMGene gut microbiota database, we consolidated the interrelationships among gut bacteria, metabolites, and genes. After collating and deduplicating data from SEA and Swisstarget databases, we identified 570 human target genes associated with these gut microbes. Intersecting these with PFOA/PFOS target sites revealed 29 potential core genes, enabling construction of a network diagram linking toxins and gut microbiota (Fig.\u0026nbsp;7A). We identified four key bacterial strains sharing host genes with PFOA/PFOS (Fig.\u0026nbsp;7B): \u003cem\u003eEnterococcus faecalis、Streptococcus salivarius、Turicibacter、 Lacticaseibacillus paracasei.\u003c/em\u003e These strains co-expressed with host genes IL-1B and PPAR, respectively. Subsequent analysis revealed these strains produce 29 unique metabolites (\u003cem\u003eEnterococcus faecalis\u003c/em\u003e: 9; \u003cem\u003eStreptococcus salivarius\u003c/em\u003e: 1; \u003cem\u003eTuricibacter\u003c/em\u003e: 15; \u003cem\u003eLacticaseibacillus paracasei\u003c/em\u003e: 4). Concurrent GO and KEGG enrichment analyses were conducted. BP analysis indicated associations with biological processes including response to lipopolysaccharide and response to molecule of bacterial origin (Fig.\u0026nbsp;7C-D). CC analysis highlighted their influence on membrane rafts and membrane microdomains (Fig.\u0026nbsp;7C-D). MF analysis indicated their role in DNA and RNA transcription factor binding (Fig.\u0026nbsp;7C-D). KEGG pathway analysis revealed associations with inflammation-related pathways including IL-17 signaling and TNF signaling (Fig.\u0026nbsp;7E). Collectively, these multi-omics correlations suggest that gut microbiota-derived metabolites may modulate host gene expression and signaling networks, thereby participating in the PFOA/PFOS-mediated AMD pathogenesis.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study proposes an \u0026ldquo;Environmental toxin-Eye\u0026rdquo; relationship axis and develops a rapid and comprehensive method for assessing the toxicological properties of environmental toxins. systematically revealing that PFOA and PFOS may drive AMD pathophysiology through directly binding to core targets SLC1A4 and ADAM17, thereby disrupting amino acid metabolism, triggering inflammatory responses, and inducing oxidative stress.\u003c/p\u003e \u003cp\u003eAlthough growing evidence suggests an epidemiological association between PFOA/PFOS and AMD[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e], the specific molecular mechanisms and signaling pathways linking PFOA/PFOS to AMD remain poorly understood. This research gap warrants particular attention. By integrating bioinformatics, network toxicology, machine learning, Mendelian randomization, gut microbiota and metabolite analysis, molecular docking, and molecular dynamics simulations, we successfully uncovered the underlying molecular mechanisms. This approach established a comprehensive, mutually reinforcing chain of evidence, significantly enhancing the credibility of our findings. It is worth emphasizing that we employ multiple complementary approaches rather than relying on a single method. This ensures that our findings are not solely derived from bioinformatics predictions but are cross-validated through analyses at different levels, thereby enhancing the reliability and interpretability of our identification of key targets and pathways. This is crucial for comprehensively understanding the extensive and long-term effects of PFOA/PFOS on AMD and developing strategies to mitigate associated risks, underscoring the necessity for in-depth investigation into the potential ocular health impacts of PFOA/PFOS and other environmental toxins.\u003c/p\u003e \u003cp\u003eUnlike traditional, time-consuming, and costly animal studies, computational toxicology enables efficient and rapid screening of toxins and their potential health risks. The use of molecular docking and molecular dynamics simulation techniques further enhances the accuracy of toxicological prediction and assessment. These methods serve as valuable tools for understanding the complex interactions between environmental pollutants and diseases[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Song et al. systematically revealed how endocrine-disrupting chemicals lead to the onset and progression of diabetic microvascular disease through network toxicology analysis[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Guo et al. comprehensively elucidated the role of ATBC in inducing oral squamous cell carcinoma (OSCC) and its potential molecular mechanisms using network toxicology alongside molecular docking and molecular dynamics simulations, providing crucial support for ATBC toxicology mechanism research[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e].These studies collectively provide a robust methodological foundation for our research. We first identified significant ocular toxicity in PFOA and PFOS through predictions from toxicology prediction websites. By integrating multi-omics datasets with PPI network interactions and diverse machine learning and bioinformatics methods, we identified candidate pathogenic genes linking PFOA/PFOS to AMD. This established a potential molecular network connecting PFOA/PFOS exposure to AMD pathogenesis. Using 128 machine learning algorithms, we identified four core genes associated with both the toxins and AMD: SLC1A4, ADAM17, EGFR, and MAPK10. Subsequently, Mendelian randomization demonstrated significant causal relationships between SLC1A4 and ADAM17 with AMD. Furthermore, the odds ratios (OR) for SLC1A4 and ADAM17 were 1.4096 and 1.3649, respectively, indicating they are important risk factors for AMD. Molecular docking and molecular dynamics simulations provided biological relevance for the aforementioned mechanisms: PFOA/PFOS exhibited strong binding energies and stable interaction patterns with SLC1A4, PFOA/PFOS possess the potential to directly impact amino acid transporter function; their interaction with ADAM17 further suggests that these toxins may amplify inflammatory signaling by affecting enzymatic cleavage or receptor release[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. GO and KEGG functional enrichment analyses consistently indicate that oxidative stress and lipid-related pathways are highly enriched in the interacting gene sets, aligning with known AMD pathogenesis mechanisms[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. SLC1A4, as a Na⁺-dependent neutral amino acid transporter, plays a crucial role in maintaining amino acid homeostasis in the retina and RPE [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. Its functional abnormalities can lead to amino acid metabolism disorders. Numerous studies have demonstrated links between amino acid metabolism and retinal pathology, with glutamine, glutamate, and alanine metabolism playing key roles in AMD[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Research indicates that both amino acid biosynthesis and metabolic pathways are closely associated with AMD progression[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. Research indicates that SLC36A4, another amino acid transporter, regulates the amino acid pool within RPE lysosomes[\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. These findings align with our investigation into how SLC1A4 influences AMD by modulating amino acid metabolism. ADAM17 is a crucial protease that plays a significant role in inflammation, angiogenesis, cell adhesion, and migration[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e].Studies indicate that ADAM17 is a widely expressed protein in the P7 retina[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]and is crucial for ocular morphogenesis[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. Research has discovered that the ADAM17/TNF pathway protects Drosophila retinal cells from age- and activity-related degeneration[\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e], offering a new perspective for elucidating the molecular mechanisms by which PFOA/PFOS induce AMD. These findings align with our investigation that the toxins amplify inflammatory signaling by activating ADAM17, leading to oxidative stress. PFOA/PFOS may promote AMD development through a synergistic dual mechanism: simultaneously disrupting amino acid transport (impairing RPE metabolic homeostasis) and amplifying ADAM17-mediated inflammatory signaling. This dual action collectively causes RPE functional imbalance and structural damage. This finding holds significant implications for elucidating the regulatory role of these toxins in AMD and its molecular mechanisms, potentially offering novel insights for AMD prevention strategies.\u003c/p\u003e \u003cp\u003ePFOA and PFOS, due to their high persistence, bioaccumulation, and toxicity, can enter the human body and accumulate over time, making them key environmental pollutants[\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. The high persistence and bioaccumulation of PFOA and PFOS, evidenced by their half-lives exceeding 41 years in groundwater and strong bioaccumulation potential, establish sustained exposure pathways through drinking water and aquatic food chains[\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. This aligns temporally and spatially with the age-related occurrence of AMD and its increasing global incidence[\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. Furthermore, through dietary intake and the persistent presence of toxins, coupled with their continuous accumulation within the food chain, humans may ultimately experience direct and prolonged exposure to these environmental toxins. We hypothesize that this toxin may be absorbed into the eye via the intestinal route. Given the slow blood flow in the choroid of the eye, this facilitates prolonged toxin accumulation. Concurrently, the photoreceptor cells and retinal pigment epithelium (RPE) in the retina operate in a highly metabolic state, making them susceptible to toxicity. This further drives the pathophysiological progression of age-related macular degeneration (AMD). The gut microbiota is an indispensable component for maintaining human health. Abnormalities in the gut microbiome have been confirmed to be closely associated with ocular diseases in both humans and animals. Dysbiosis of the gut microbiota can trigger various ocular diseases through the\u0026ldquo; Gut-Eye Axis \u0026rdquo;[\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. By integrating the relationship between environmental toxins and the microbiome, we constructed a \u0026ldquo;Toxin-Gut Microbiota-Metabolite-Gene\u0026rdquo; relationship network. This revealed that PFOA/PFOS may drive the progression of AMD by disrupting gut microbiota homeostasis, promoting inflammatory responses, and reshaping the dynamic balance between probiotics and pathogens. Additionally, probiotic therapy represents a potentially effective approach for treating AMD. Supplementation with probiotics helps suppress pathogenic microorganisms, reduce inflammatory responses, and enhance the integrity of the intestinal barrier[\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. We found that the key gene IL-1B, associated with toxins and gut microbiota, is closely linked to \u003cem\u003eTuricibacter\u003c/em\u003e and \u003cem\u003eLactobacillus paracasei\u003c/em\u003e within the gut microbiome. We hypothesize that toxins disrupt normal beneficial gut bacteria, thereby accelerating disease progression. Our gut microbiome association analysis expands the research perspective on toxin-host interactions. The high persistence and bioaccumulation properties of PFOA/PFOS in the environment enable them to exert long-term effects on host metabolism and immune homeostasis through oral exposure or accumulation via the food chain. This provides a basis for future exploration and prevention strategies targeting gut microbiome intervention for AMD.\u003c/p\u003e \u003cp\u003eWe progressively constructed and cross-validated the mechanism by which PFOA/PFOS may promote AMD: by disrupting amino acid metabolism (SLC1A4-related), amplifying inflammatory signaling (ADAM17-related), and leveraging the mediating role of gut microbiota metabolism, ultimately leading to retinal functional imbalance and structural damage. The novelty of this study lies in our first-ever clarification of the mechanism linking environmental toxins to AMD, raising awareness about the connection between environmental toxins and ophthalmic diseases. Although our network-based toxicology approach provides valuable insights into the mechanisms linking environmental toxins to AMD, we acknowledge certain limitations in this study. First, while the sample size meets the requirements for exploratory analysis, it may still limit the generalizability of the findings. Additionally, the predictive results obtained require further experimental validation to confirm their biological significance and accuracy. Therefore, future research should consider expanding sample sizes and enhancing population diversity through large-scale, long-term epidemiological studies to track and analyze the dynamic relationship between toxin exposure and AMD incidence, with particular focus on populations with high PFOA/PFOS exposure. Concurrently, in vivo and in vitro experiments should validate key molecular mechanisms to enhance the reliability and generalizability of research findings. This study's in-depth analysis of the toxic mechanisms of PFOA and PFOS in AMD also holds significant clinical implications: (1) Highlighting the impact of environmental toxins on ophthalmic diseases; (2) Developing preventive strategies for AMD induced by environmental toxins; (3) Identifying high-risk populations for AMD through monitoring of PFOA and PFOS exposures.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis study reveals a potential molecular mechanism linking PFOA/PFOS exposure to AMD, offering a novel perspective on the integration of environmental health science and ophthalmology. Through a comprehensive approach spanning molecular, genomic, proteomic, and gut microbiome dimensions, it provides hypotheses and insights for AMD prevention and treatment. These findings will drive further investigation into environmental factors in the pathogenesis of AMD and promote the development of effective prevention and treatment strategies.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eZ.Y.G and Y.J.P designed and supervised the overall study, performed data interpretation, and drafted the manuscript. Z.Y.G was responsible for bioinformatics analysis, data processing, and contributed to manuscript preparation. Y.J.P, Y.W.L and J.X.L participated in data collection, sample processing, and curation of clinical information. Y.J.Q and H.Y.Z conceived the study, provided critical revisions, and acted as the corresponding author. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics and Consent to Participate declarations:\u0026nbsp;\u003c/strong\u003enot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was supported by the National Natural Science Foundation of China (grant No. 82171036 \u0026amp; 82471083).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe corresponding authors of this paper can provide primary data that supports its conclusions upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics declarations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthical approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe publication has been approved by all co-authors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003ePapadopoulou E, Stratakis N, Basagana X, Brantsaeter AL, Casas M, Fossati S, Grazuleviciene R, Smastuen Haug L, Heude B, Maitre L, et al. Prenatal and postnatal exposure to PFAS and cardiometabolic factors and inflammation status in children from six European cohorts. Environ Int. 2021;157:106853.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZango ZU, Khoo KS, Garba A, Kadir HA, Usman F, Zango MU, Da Oh W, Lim JW. A review on superior advanced oxidation and photocatalytic degradation techniques for perfluorooctanoic acid (PFOA) elimination from wastewater. Environ Res. 2023;221:115326.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShi W, Zhang Z, Li M, Dong H, Li J. Reproductive toxicity of PFOA, PFOS and their substitutes: A review based on epidemiological and toxicological evidence. Environ Res. 2024;250:118485.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuymer RH, Campbell TG. Age-related macular degeneration. Lancet. 2023;401:1459\u0026ndash;72.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJonas JB, Cheung CMG, Panda-Jonas S. Updates on the Epidemiology of Age-Related Macular Degeneration. Asia Pac J Ophthalmol (Phila). 2017;6:493\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaidonis G, Lamy R, Wu J, Yang D, Psaras C, Doan T, Stewart JM. Aqueous Fluid Transcriptome Profiling Differentiates Between Non-Neovascular and Neovascular AMD. Invest Ophthalmol Vis Sci. 2023;64:26.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFleckenstein M, Schmitz-Valckenberg S, Chakravarthy U. Age-Related Macular Degeneration: A Review. JAMA. 2024;331:147\u0026ndash;57.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang D, Chen Y, Li J, Wu E, Tang T, Singla RK, Shen B, Zhang M. Natural products for the treatment of age-related macular degeneration. Phytomedicine. 2024;130:155522.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYuan M, Zhang T, Sun Y, Yi S, Han G, Wei P, Chen X, Wang Y, Zhu L. Microglia-Mediated Pathological Retinal Angiogenesis Leading to Visual Impairment in Mice Exposed to Perfluorooctanoic Acid. Environ Sci Technol. 2025;59:15680\u0026ndash;91.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTien PT, Lin HJ, Tsai YY, Lim YP, Chen CS, Chang CY, Lin CJ, Chen JJ, Wu SM, Huang YJ, Wan L. Perfluorooctanoic acid in indoor particulate matter triggers oxidative stress and inflammation in corneal and retinal cells. Sci Rep. 2020;10:15702.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen X, Li J, Xu N, Li X, Li J, Guo Q, Zhang J, Miao H, Huang L. Serum lipids mediate the association of per- and polyfluoroalkyl substances exposure and age-related macular degeneration. PLoS ONE. 2025;20:e0317678.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eClough E, Barrett T, Wilhite SE, Ledoux P, Evangelista C, Kim IF, Tomashevsky M, Marshall KA, Phillippy KH, Sherman PM, et al. NCBI GEO: archive for gene expression and epigenomics data sets: 23-year update. Nucleic Acids Res. 2024;52:D138\u0026ndash;44.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang W, Tian Y, Chen B, Xu S, Wu L. PFOA/PFOS Facilitated Intestinal Fatty Acid Absorption by Activating the PPARalpha Pathway: Insights from Organoids Model. Environ Health (Wash). 2024;2:85\u0026ndash;94.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQi C, He G, Qian K, Guan S, Li Z, Liang S, Liu J, Ke X, Zhang S, Lu M, et al. gutMGene v2.0: an updated comprehensive database for target genes of gut microbes and microbial metabolites. Nucleic Acids Res. 2025;53:D783\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYuan M, Yi S, Wang X, Han G, Wei P, Lv Z, Gui B, Chen X, Wang Y, Zhu L. Promoted Translocation of Perfluorooctanoic Acid across the Blood-Retinal Barrier due to its Inhibition of Tight Junction Assembly by Antagonizing LPAR1. Environ Sci Technol. 2025;59:4807\u0026ndash;19.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFeng C, Yan W, Mei Z, Luo X. Exploring the toxicological impact of bisphenol a exposure on psoriasis through network toxicology, machine learning, and multi-dimensional bioinformatics analysis. J Environ Manage. 2025;385:125708.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSong S, Huang L, Zhou X, Yu J. Exploring the toxicological network in diabetic microvascular disease. Int J Surg. 2025;111:3895\u0026ndash;907.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuo Y, Liu Y, Chen Y, Du S, Zheng Y, Wang L. The mechanisms of environmental pollutant acetyl tributyl citrate induced oral squamous cell carcinoma using network toxicology, molecular docking and molecular dynamics simulation. Int J Surg. 2025;111:7873\u0026ndash;85.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSaad MI, Jenkins BJ. The protease ADAM17 at the crossroads of disease: revisiting its significance in inflammation, cancer, and beyond. FEBS J. 2024;291:10\u0026ndash;24.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFleckenstein M, Keenan TDL, Guymer RH, Chakravarthy U, Schmitz-Valckenberg S, Klaver CC, Wong WT, Chew EY. Age-related macular degeneration. Nat Rev Dis Primers. 2021;7:31.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTan LX, Germer CJ, La Cunza N, Lakkaraju A. Complement activation, lipid metabolism, and mitochondrial injury: Converging pathways in age-related macular degeneration. Redox Biol. 2020;37:101781.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eElazar D, Alvarez N, Drobeck S, Gunn TM. SLC1A4 and Serine Homeostasis: Implications for Neurodevelopmental and Neurodegenerative Disorders. Int J Mol Sci 2025, 26.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHansman DS, Du J, Casson RJ, Peet DJ. Eye on the horizon: The metabolic landscape of the RPE in aging and disease. Prog Retin Eye Res. 2025;104:101306.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXia M, Zhang F. Amino Acids Metabolism in Retinopathy: From Clinical and Basic Research Perspective. Metabolites 2022, 12.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHou XW, Wang Y, Pan CW. Metabolomics in Age-Related Macular Degeneration: A Systematic Review. Invest Ophthalmol Vis Sci. 2020;61:13.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShang P, Valapala M, Grebe R, Hose S, Ghosh S, Bhutto IA, Handa JT, Lutty GA, Lu L, Wan J, et al. The amino acid transporter SLC36A4 regulates the amino acid pool in retinal pigmented epithelial cells and mediates the mechanistic target of rapamycin, complex 1 signaling. Aging Cell. 2017;16:349\u0026ndash;59.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCaolo V, Swennen G, Chalaris A, Wagenaar A, Verbruggen S, Rose-John S, Molin DG, Vooijs M, Post MJ. ADAM10 and ADAM17 have opposite roles during sprouting angiogenesis. Angiogenesis. 2015;18:13\u0026ndash;22.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang X, Wang T, Kaneko S, Kriukov E, Lam E, Szczepan M, Chen J, Gregg A, Wang X, Fernandez-Gonzalez A, et al. Photoreceptors inhibit pathological retinal angiogenesis through transcriptional regulation of Adam17 via c-Fos. Angiogenesis. 2024;27:379\u0026ndash;95.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eToonen JA, Ronchetti A, Sidjanin DJ. A Disintegrin and Metalloproteinase10 (ADAM10) Regulates NOTCH Signaling during Early Retinal Development. PLoS ONE. 2016;11:e0156184.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSel S, Kalinski T, Enssen I, Kaiser M, Nass N, Trau S, Wollensak G, Brauer L, Jager K, Paulsen F. Expression analysis of ADAM17 during mouse eye development. Ann Anat. 2012;194:334\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMuliyil S, Levet C, Dusterhoft S, Dulloo I, Cowley SA, Freeman M. ADAM17-triggered TNF signalling protects the ageing Drosophila retina from lipid droplet-mediated degeneration. EMBO J. 2020;39:e104415.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWee SY, Aris AZ. Environmental impacts, exposure pathways, and health effects of PFOA and PFOS. Ecotoxicol Environ Saf. 2023;267:115663.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCheng X, Wei Y, Zhang Z, Wang F, He J, Wang R, Xu Y, Keerman M, Zhang S, Zhang Y, et al. Plasma PFOA and PFOS Levels, DNA Methylation, and Blood Lipid Levels: A Pilot Study. Environ Sci Technol. 2022;56:17039\u0026ndash;51.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCollaborators GBDGA. Global burden of vision impairment due to age-related macular degeneration, 1990\u0026ndash;2021, with forecasts to 2050: a systematic analysis for the Global Burden of Disease Study 2021. Lancet Glob Health. 2025;13:e1175\u0026ndash;90.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFloyd JL, Grant MB. The Gut-Eye Axis: Lessons Learned from Murine Models. Ophthalmol Ther. 2020;9:499\u0026ndash;513.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJin Q, Wang S, Yao Y, Jiang Q, Li K. The gut-eye axis: from brain neurodegenerative diseases to age-related macular degeneration. Neural Regen Res. 2025;20:2741\u0026ndash;57.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"PFOA and PFOS, Age-related macular degeneration, Bioinformatics, Network toxicology, machine learning, Gut microbiota interactions","lastPublishedDoi":"10.21203/rs.3.rs-8563077/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8563077/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003ePurpose\u003c/h2\u003e \u003cp\u003ePerfluorooctanoic acid (PFOA) and perfluorooctane sulfonic acid (PFOS) are two ubiquitous persistent organic pollutants. Currently, little is known about the potential mechanisms of PFOA and PFOS in ophthalmic diseases. This study systematically investigates the effects of PFOA and PFOS exposure on age-related macular degeneration (AMD), aiming to identify potential targets and elucidate associated molecular mechanisms.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eDifferential expression analysis across multiple transcriptomic datasets was performed to identify AMD-associated genes. Network toxicology, machine learning, Mendelian randomization, gut microbiota profiling, molecular docking, and MD simulations were subsequently integrated to characterize the interactions between PFOA/PFOS and their protein targets.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eWe first performed toxicity prediction to identify 97 potential targets linking these toxins to AMD. Subsequently, integrating bioinformatics with 128 machine learning algorithms further identified four key core genes: SLC1A4, ADAM17, EGFR, and MAPK10. Potential regulatory mechanisms were mapped using TF-miRNA-mRNA networks and GeneMANIA. Mendelian randomization (MR) validated causal relationships between core genes SLC1A4 and ADAM17 and AMD. Molecular docking and kinetic simulations confirmed stable interactions between PFOA/PFOS and their targets. Our findings reveal the pivotal roles of inflammatory responses, oxidative stress, and amino acid metabolism disorders in the pathogenic mechanisms of these toxins. Analysis of the gutMGene gut microbiome database identified specific bacteria potentially acting through metabolites to influence human genes enriched in inflammatory signaling pathways, thereby promoting the toxic effects of PFOA/PFOS on AMD. This established a \u0026ldquo;toxin-gut bacteria-metabolite-gene\u0026rdquo; relationship network.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eThis study elucidates the underlying molecular mechanisms and signaling pathways linking PFOA/PFOS exposure to AMD, identifying SCL1A4 and ADAM17 as key targets for PFOA/PFOS-induced AMD. These findings not only provide new insights into the role of environmental pollutants in AMD pathogenesis but also offer an analytical framework for elucidating the effects of other environmental toxins on ocular diseases.\u003c/p\u003e","manuscriptTitle":"Integrating network toxicology, machine learning, gut microbiome analysis, and structural validation to reveal the molecular mechanism linking PFOA and PFOS exposure to age-related macular degeneration","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-29 16:57:38","doi":"10.21203/rs.3.rs-8563077/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"d6d71cb5-9f12-4689-b412-d2757f3d1b7a","owner":[],"postedDate":"January 29th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-07T09:12:42+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-29 16:57:38","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8563077","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8563077","identity":"rs-8563077","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}