Multi-omics integration reveals convergent extracellular matrix remodelling and lipid metabolic reprogramming as...

Adipogenesis from mouse embryonic stem cells (mESCs) offers a tractable model for dissecting early adipocyte commitment, yet the mechanisms coordinating this transition across multiple biological layers remain incompletely understood. Here we present the first simultaneous five-layer multi-omics characterization of mESC-derived adipocyte differentiation, integrating transcriptomics, proteomics, secretomics, lipidomics, and metabolomics from matched adipogenic (Pos) and non-differentiating (Neg) cell populations at day 30. Applying Multi-Omics Factor Analysis (MOFA+), we identified a dominant shared latent axis that perfectly segregated Pos from Neg cells across all five views. Layer-specific functional enrichment converged on two principal biological axes: ECM remodeling - encompassing collagens, laminins, thrombospondins, and lysyl oxidases - and lipid metabolic reprogramming, with phospholipid and glycerolipid metabolic processes dominating the lipidomics/metabolomics layer. Ensemble Machine Learning Feature Ranking (EMFR) identified a secreted factor (Scpep1; importance score 0.90) as the top-ranked discriminatory feature. Network analysis revealed indirect ECM-lipid connectivity mediated by four bridging nodes (Plod1, Thbs2, Plg, Pmp22) through a hub subnetwork of phospholipid-metabolizing enzymes. Targeted qPCR validation of six candidate regulators (Itga5, Igfbp6, Pik3cg, Lpl, Acer3, Sirt1) confirmed RNA-seq concordance. These findings establish convergent ECM remodeling and lipid metabolic reprogramming as central axes of adipocyte identity acquisition from mESCs.