Exploring the Role of Bifenthrin in Recurrent Implantation Failure and Pregnancy Loss Through Network Toxicology and Molecular Docking

Abstract

Bifenthrin (BF) is a widely used pyrethroid pesticide recognized as an endocrine-disrupting chemical (EDC). Previous studies have confirmed that chronic exposure to BF is associated with various health risks. However, its potential association with recurrent implantation failure (RIF) and recurrent pregnancy loss (RPL) remains unclear. In this study, the potential targets of BF were identified using several databases, including the Comparative Toxicogenomics Database (CTD), TargetNet, GeneCards, SwissTargetPrediction, and STITCH. Differentially expressed genes (DEGs) associated with RIF were obtained from bulk RNA-seq datasets in the GEO database. Candidate targets were identified by intersecting the predicted BF-related targets with the RIF-associated DEGs, followed by functional enrichment analysis using the DAVID and g:Profiler platforms. Subsequently, hub genes were identified based on the STRING database and Cytoscape. A diagnostic model was then constructed based on these hub genes in the RIF cohort and validated in an independent recurrent pregnancy loss (RPL) cohort. Additionally, we performed single-cell type distribution analysis and immune infiltration profiling based on single-cell RNA-seq and bulk RNA-seq data, respectively. Molecular docking analysis using AutoDock Vina was conducted to evaluate the binding affinity between BF and the four hub proteins, as well as several hormone-related receptors. Functional enrichment results indicated that the candidate genes were mainly involved in apoptotic and oxidative stress-related pathways. Ultimately, four hub genes-BCL2, HMOX1, CYCS, and PTGS2-were identified. The diagnostic model based on these genes exhibited good predictive performance in the RIF cohort and was successfully validated in the RPL cohort. Single-cell transcriptomic analysis revealed a significant increase in the proportion of myeloid cells in RPL patients, while immune infiltration analysis showed a consistent downregulation of M2 macrophages in both RIF and RPL. Moreover, molecular docking analysis revealed that BF exhibited high binding affinity to all four hub proteins and demonstrated strong binding potential with multiple hormone receptors, particularly pregnane X receptor (PXR), estrogen receptor α (ESRα), and thyroid hormone receptors (TR). In conclusion, the association of BF with four hub genes and multiple hormone receptors suggests a potential link to immune and endocrine dysregulation observed in RIF and RPL. However, in vivo and in vitro experimental evidence is currently lacking, and further studies are needed to elucidate the mechanisms by which BF may contribute to RIF and RPL.

Keywords: bifenthrin; endocrine-disrupting chemical; molecular docking; recurrent implantation failure; recurrent pregnancy loss.

Figure 1

Flowchart of the study design.

Figure 2

Analysis of bifenthrin-associated genes and DEGs in the RIF cohort. (A) Volcano plot showing the DEGs in the RIF cohort (GSE111974) identified by “limma” analysis. (B) Venn diagram showing the overlap between bifenthrin-targeted genes and RIF-associated DEGs. (C) g:Profiler enrichment analysis of the 18 overlapping genes. (D) GO enrichment analysis of the 18 overlapping genes. (E) KEGG pathway enrichment analysis of the 18 overlapping genes. (F) Spearman correlation heatmap of the 18 target genes. Abbreviations: DEG, differentially expressed gene; RIF, recurrent implantation failure; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes.

Figure 3

Identification of hub genes. (A) PPI network ranked by Degree value. (B) Genes screened through eight network metrics. (C) UpSet plot displaying overlapping genes identified by eight network metrics in CytoHubba. Abbreviations: MNC, Maximum Neighborhood Component; MCC, Maximal Clique Centrality; EPC, Edge Percolated Component.

Figure 4

Logistic Regression Prediction Model of the RIF cohort. (A) Forest plot of univariate logistic regression analysis for four hub genes in the RIF cohort. (B) Nomogram of the prediction model.

Figure 5

Evaluation of the predictive model. (A) Calibration curve in the RIF cohort. (B) Decision curve analysis (DCA) in the RIF cohort. (C) ROC curve in the RIF cohort. (D) ROC curves of the four hub genes in the RIF cohort. (E) DCA curve in the RPL cohort. (F) ROC curve in the RPL cohort. (G) ROC curves of the four hub genes in the RPL cohort. (H) Scatter plot of AUC values from 1000 models constructed by randomly selecting four genes in the RIF training set and applying to the RPL test set. The red dot represents the model based on PTGS2, BCL2, HMOX1, and CYCS.

Figure 6

scRNA-seq analysis of RIF and RPL. (A) UMAP visualization of cellular heterogeneity in the RIF cohort. (B) Compositional bar plot depicting cell type proportions in the RIF cohort. (C) UMAP visualization of cellular heterogeneity in the RPL cohort. (D) Compositional bar plot depicting cell type proportions in the RPL cohort. (E) Z-score heatmap showing the expression patterns of four hub genes across different cell types in the RIF cohort. (F) Z-score heatmap showing the expression patterns of four hub genes across different cell types in the RPL cohort.

Figure 7

Analysis of immune cell infiltration in the RIF and RPL cohort. (A) Boxplots of immune infiltration analysis for immune cells in the RIF cohort. (B) Boxplots of immune infiltration analysis for immune cells in the RPL cohort. (C) Heatmap of the Spearman correlation between PTGS2, HMOX1, CYCS, BCL2, and immune cells in the RIF cohort. (D) Heatmap of the Spearman correlation between PTGS2, HMOX1, CYCS, BCL2, and immune cells in the RPL cohort. * p < 0.05, ** p < 0.01.