A microbiota–substrate–metabolite–target network suggests AKT1-associated links between gut microbiota-derived metabolites and osteoporosis

OBJECTIVE: To build a computational, hypothesis-generating framework that links gut microbiota-related metabolites with osteoporosis (OP) while retaining upstream microbial source and substrate information, and to prioritize candidate host targets, pathways, and modules for subsequent testing. METHODS: Metabolites and human intestinal host targets were obtained from gutMGene and structures from PubChem; candidate targets were predicted with SEA and SwissTargetPrediction, and osteoporosis targets from CTD, GeneCards, and OMIM. Shared targets were analyzed by protein-protein interaction networks, GO and KEGG enrichment, a layered microbiota-substrate-metabolite-target network, ADMET profiling, and molecular docking. The candidate hubs were further tested for differential expression in three independent human Gene Expression Omnibus (GEO) datasets. RESULTS: Forty-eight shared targets were identified, with TP53, AKT1, IL6, TNF, and IL1B among the most connected candidate hubs and no single target dominant across complementary centrality metrics. The layered network comprised 121 nodes and 150 edges and yielded 479 directed candidate paths-a graph-theoretical enumeration of layered connectivity rather than independently validated chains-highlighting tryptophan-indole and flavonoid modules. Independently, AKT1 and TP53 were upregulated in osteoporotic bone-marrow mesenchymal stem cells (log2FC 1.59 and 1.55; FDR 0.0099 and 0.037), whereas hub expression was weak or heterogeneous in circulating immune cells, and docking offered only qualitative structural plausibility. CONCLUSION: The source-preserving M-S-M-T framework prioritizes candidate multi-target links between gut microbiota-derived metabolites and OP-related inflammatory and bone-remodeling pathways, with AKT1 as one, rather than the unique, candidate signaling hub; independent transcriptomic data provide convergent but cell-type-specific support, and functional validation remains required.