The gut microbiome influences drug metabolism and therapeutic efficacy. Still, the lack of a general label-free approach for monitoring bacterial or host metabolic contribution hampers deeper insights. Here, a 2D nuclear magnetic resonance (NMR) approach is introduced that enables real-time monitoring of the metabolism of Levodopa (L-dopa), an anti-Parkinson drug, in both live bacteria and bacteria-host (Caenorhabditis elegans) symbiotic systems. The quantitative method reveals that discrete Enterococcus faecalis substrains produce different amounts of dopamine in live hosts, even though they are a single species and all have the Tyrosine decarboxylase (TyrDC) gene involved in L-dopa metabolism. The differential bacterial metabolic activity correlates with differing Parkinson's molecular pathology concerning alpha-synuclein aggregation as well as behavioral phenotypes. The gene's existence or expression is not an indicator of metabolic activity is also shown, underscoring the significance of quantitative metabolic estimation in vivo. This simple approach is widely adaptable to any chemical drug to elucidate pharmacomicrobiomic relationships and may help rapidly screen bacterial metabolic effects in drug development.