Integrated Metabolomics and Amplicon Sequencing Reveal Microbial Succession and Metabolite Dynamics During Sojae Semen Praeparatum Fermentation

Sojae Semen Praeparatum (SSP) is a traditional fermented soybean product with both dietary and medicinal uses in China. However, the dynamic metabolic changes and microbial community succession during SSP fermentation remain insufficiently characterized. In this study, untargeted metabolomics based on UHPLC-HRMS (Ultra-high-performance liquid chromatography-high-resolution mass spectrometry) was integrated with 16S rRNA gene and ITS amplicon sequencing to investigate metabolite remodeling, bacterial and fungal community succession, and their potential associations during SSP fermentation. Clear stage-dependent separation of metabolite profiles was observed, and 22,343 metabolic features were detected, of which 4281 were putatively annotated. Differential metabolites were mainly assigned to amino acids, dipeptides, lipids, carbohydrates, and isoflavones. Sucrose and other carbohydrates were progressively consumed during fermentation, while N-acetyl amino acids, short dipeptides, proline betaine and a range of lipid metabolites built up towards the later stages. The microbiota did not drift at random: amplicon sequencing resolved a clear successional sequence in both kingdoms. Bacteria converged on genera within the Enterobacteriaceae—mainly Enterobacter, Kosakonia and Citrobacter—whereas the fungi started out dominated by Rhizopus and ended in a co-dominant configuration of Aspergillus, Mucor and several minor taxa. When the two data sets were linked, the dominant taxa were associated with metabolites assigned to nitrogen, lipid and phenolic pathways. Procrustes analysis suggested a potential association between the fungal community and the metabolome, whereas the bacterial association was marginal; however, the difference in coupling strength between the two was not directly tested. These results provide preliminary evidence and a foundation for quality control and process optimization, but require validation across independent batches.