Integrating Mendelian randomization with single-cell and spatial transcriptomics to investigate gut microbiota-associated genetic signals and molecular characteristics in prostate cancer.

Background: Prostate cancer (PRAD) represents one of the most prevalent malignancies among men globally. Accumulating evidence implicates gut microbiota dysbiosis in the initiation and progression of prostate cancer; however, the causal relationship and underlying molecular mechanisms remain incompletely elucidated. This study integrated Mendelian randomization (MR), bioinformatics, single-cell RNA sequencing, and spatial transcriptomics to systematically investigate the genetic associations between gut microbiota and prostate cancer risk and characterize the expression patterns of microbiota-associated candidate genes within the tumor microenvironment. Methods: Two-sample MR analyses were performed using summary statistics from the OpenGWAS database to evaluate causal associations between gut microbiota and PRAD risk, with sensitivity analyses conducted to ensure robustness. Significant SNPs were functionally annotated and mapped to candidate genes using g:SNPense, and differentially expressed associated genes (DEAGs) were identified through integration with GEO datasets. Subsequent analyses encompassed immune infiltration profiling, machine learning model construction, and nomogram validation. Single-cell RNA sequencing and spatial transcriptomics were employed to delineate cell type-specific expression and spatial distribution patterns of signature genes. Candidate gene expression alterations in prostate cancer cells were experimentally validated by quantitative real-time PCR (qRT-PCR). Results: MR analyses identified significant positive causal associations between PRAD risk and Brevibacillaceae, Brevibacillales, CAG-110, and Pararhizobium, whereas Lawsonibacter sp000492175 exhibited a significant inverse association. A total of 21 differentially expressed genes were identified, with RASSF3, CROCC, UBE4B, SMOC2, and ACACA emerging as key signature genes. Immune infiltration analyses revealed a significant positive correlation between FRMD3 and regulatory T cells (Tregs), and a negative correlation between ACACA and neutrophil infiltration. The random forest classifier demonstrated optimal predictive performance, achieving an area under the curve (AUC) of 0.853 in the validation cohort. Single-cell and spatial transcriptomic analyses revealed cell-type-associated expression patterns of ACACA, SMOC2, and UBE4B across different cellular populations. These findings suggest potential associations between signature genes and specific cellular contexts, which require further experimental validation. qRT-PCR validated differential expression patterns of these candidate genes in prostate cancer cells. Conclusion: This study provides genetic causal inference evidence supporting a close association between gut microbiota and prostate cancer risk, and identifies multiple key signature genes implicated in immune microenvironment regulation and tumor progression. Integrated single-cell and spatial transcriptomic analyses suggest that these genes exhibit distinct expression patterns among different cellular populations and may provide clues for understanding their potential roles in PRAD progression. These findings provide genetic evidence supporting the association between gut microbiota and prostate cancer risk and highlight potential microbiota-associated molecular signatures. Further functional studies are required to clarify whether these genes directly contribute to microbiota-related prostate cancer progression.