Overview of study design and workflow.

(A) Main data sources: Single-cell and bulk cis-eQTL datasets from European-ancestry individuals were used to define immune-cell-specific gene expression. GWAS summary statistics for prostate cancer were obtained from the PRACTICAL and FinnGen consortia. (B) Primary analyses: Two-sample Mendelian randomization (MR) was used to infer causal effects, followed by weighted correlation analysis across immune cell types and Bayesian colocalization to assess shared genetic variants. (C) Follow-up analyses: Functional enrichment, protein-protein interaction (PPI) network analysis, single-cell RNA sequencing (scRNA-seq) validation, and drug repurposing via STRING and DrugBank were conducted to explore biological relevance and therapeutic potential.

Correlation, colocalization, and protein interaction network in prostate cancer.

(A) Pearson correlation matrix of MR results across immune cell types and bulk-eQTL, showing shared genetic influence. (B) Number of prostate cancer-associated eGenes in each immune cell type with varying levels of colocalization evidence. (C) PPI network of colocalized eGenes prioritized by Bayesian colocalization (PP.H4 > 0.50), with node size representing connectivity.

Pathway and network analyses of prostate cancer-associated eGenes.

(A) KEGG pathway enrichment analysis of colocalized eGenes using Metascape. Dot size indicates count number; color represents pathway category. (B) MCODE clustering of PPI networks highlights tightly connected modules (MCODE1–4). Red-labeled proteins show colocalization with prostate cancer risk loci.

The scRNA-seq profiling of prostate tumor microenvironment.

(A–C) UMAP plots showing sample identity, clustering, and major cell types identified. (D) Dot plot of canonical markers used for cell type annotation. (E–F) Cell composition proportions by cluster and by patient sample. (G) Expression patterns of prioritized bulk prostate cancer-associated eGenes across cell types. (H–K) Feature plots of representative eGenes (MYO6, MRPS34, VAMP8, MMAB) across cell types.

Causal effect estimates of prioritized eGenes across PRACTICAL, FinnGen, and meta-analysis.

Only eGenes remained significant (p < 0.05) in meta-analysis are shown.

Drug repurposing network of prostate cancer-associated targets.

This network diagram illustrates three key relationships: (1) links between proteins encoded by prioritized prostate cancer-associated causal eGenes and their potential targeted drugs; (2) connections between existing prostate cancer drugs and their known protein targets; and (3) protein-protein interactions between proteins encoded by prioritized prostate cancer-associated causal eGenes and proteins targeted by approved prostate cancer drugs.