Mounting evidence has associated gut microbial composition and systemic metabolic profiles with Parkinson's disease (PD), yet causal pathways and molecular mechanisms remain unclear. We employed a systematic two-stage analytical framework integrating multiple Mendelian randomization (MR) approaches, transcriptome and epigenome-wide regulatory analyses, and colocalization analysis to genetically dissect the microbiota-host regulatory network implicated in PD susceptibility. Initially, two-sample MR was performed using gut microbiota data from the MiBioGen consortium, followed by an independent replication MR analysis using mbQTL data from the Dutch Microbiome Project (DMP). The two-step MR approach was employed to identify potential mediation of blood metabolites. Summary data-based MR (SMR) methods were also used to pinpoint key blood metabolic genes and regulatory regions associated with PD risk, with the combination of PD GWAS data, blood expression quantitative trait loci (eQTLs), and DNA methylation quantitative trait loci (mQTLs). Colocalization analyses were performed to explore interactions between host metabolic gene expression and the gut microbiota using blood expression quantitative trait loci (eQTLs) and fecal microbial quantitative trait loci (mbQTLs). Our multi-stage analyses identified five microbial taxa with nominal positive associations with Parkinson's disease (PD), including class Clostridia, order Bacillales, and genera Clostridium sensu stricto 1, Dorea, and Lachnospiraceae UCG001. By contrast, three additional microbial genera (Butyricimonas, Defluviitaleaceae UCG011 and Marvinbryantia) exhibited nominal protective correlations with PD. Cross-cohort analyses further indicated a shared correlation pattern along the Clostridia-Clostridiales-Clostridiaceae/Lachnospiraceae taxonomics potentially linking to PD risk across both discovery and validation cohorts. Mediation analyses indicated that betaine may mediate the Lachnospiraceae UCG001-PD associations, while androstenediol monosulfate may serve as a mediator linking Clostridium sensu stricto 1 to PD. SMR analysis highlights genes like SH2B1, JUP, as key host metabolic genes with potential causal relevance to PD, with colocalization indicating genetic overlaps supporting their potential involvement in gene-microbiota interactions. These findings deepen our understanding of gut-brain axis dysregulation in PD and may provide novel insights for future mechanistic validation and targeted preventive or therapeutic investigations.