Transcriptomic signatures of inter-tissue communication in metabolic adaptation to high-fat high-sugar diet and metformin across 20 CC-RIX genetic backgrounds.

Abstract

Type 2 diabetes is characterized by systemic metabolic dysfunction and heterogeneous therapeutic responses, limiting the effectiveness of standard treatments such as metformin. Here, we integrate multi-tissue transcriptomic profiling, network analysis, and machine learning to investigate the molecular basis of metabolic dysregulation and variability in treatment response within a genetically diverse mouse model. Transcriptomic data were obtained from the Gene Expression Omnibus (GEO; GSE237750), comprising RNA-seq datasets from skeletal muscle (GSE237747, n = 703), adipose tissue (GSE237737, n = 705), and liver (GSE237743, n = 705). Samples were derived from Collaborative Cross Recombinant Inbred Intercross (CC-RIX) mice representing 20 genetic backgrounds and included both male and female animals subjected to a high-fat high-sugar (HFHS) diet with metformin treatment. We show that HFHS exposure induces coordinated metabolic reprogramming across skeletal muscle, liver, and adipose tissue, characterized by mitochondrial dysfunction, altered lipid metabolism, and chronic inflammation. Network analysis reveals extensive inter-organ signaling interactions, supporting a systems-level model of metabolic disease and positioning skeletal muscle as a central node of dysregulation. Mechanistically, HFHS exposure activates FOXO-dependent proteolytic and autophagic pathways, linking insulin resistance to muscle wasting. Metformin treatment partially reverses these transcriptional alterations, restoring aspects of metabolic function in a subset of individuals. Using baseline transcriptomic profiles, we identified gene expression patterns associated with metformin response. However, predictive performance was modest, with area under the curve (AUC) values close to random classification and permutation testing indicating no significant separation from null expectations. Despite this, feature importance analysis consistently highlighted pathways related to mitochondrial function, lipid metabolism, and immune signaling, suggesting that coordinated biological processes, rather than individual genes, underlie variability in therapeutic response. External validation in an independent cohort demonstrated preservation of directional expression changes, supporting the robustness of pathway-level signals despite differences in cohort composition and experimental platforms. To further explore potential mechanisms of drug action, we performed molecular docking analysis, which predicted that metformin can occupy a defined binding pocket within the target protein. A potential hydrogen bond interaction suggests binding stability, providing structural context for a plausible molecular mechanism linking drug action to observed transcriptional responses. Together, these findings highlight the importance of systems-level regulation in metabolic disease and underscore the complexity of predicting therapeutic response using transcriptomic data alone. While predictive accuracy remains limited, the integration of multi-tissue transcriptomics, network analysis, and structural modeling provides a framework for understanding metabolic dysfunction and supports the development of more refined, mechanism-driven approaches.