Label-free imaging and machine learning reveal cyclin-dependent kinase 6 (CDK6)-regulated metabolic phenotypes and...

Cyclin-dependent kinase 6 (CDK6) regulates adipogenesis and beige fat formation, but its role in preadipocyte metabolism is unclear. The mouse embryonic fibroblast cell line 3 T3-L1 is a well-established preadipocyte model with high adipogenic potential under adipogenic stimulation. Here, we employed label-free, non-invasive Raman imaging to investigate the metabolic profile of 3 T3-L1 cells following CDK6 knockout (KO) at the single-cell level. To improve the specificity of spectral interpretation, isolated cellular mitochondria and mouse skin tissue were used as reference standards for mitochondrial and collagen Raman fingerprints, respectively. True component analysis (TCA) was applied to Raman hyperspectral datasets to decompose complex spectral signals into distinct biochemical components with corresponding spatial distributions. Compared to wild-type (WT) cells, CDK6 ablation increased mitochondrial content and the unsaturated-to-saturated fatty acid ratio, but reduced collagen and overall lipid content. These shifts suggest that CDK6 deletion promotes mitochondrial metabolism and suppresses extracellular matrix synthesis, thereby conferring beneficial metabolic changes that support differentiation potential toward beige adipocytes. Four Raman spectral biomarkers (I_809/937, I_754/2918, I_937/1585, I_1244/1444) enabled effective cell-type discrimination. For hyperspectral data processing, we employed a machine learning-based pipeline comprising preprocessing (baseline correction and normalization), dimensionality reduction via principal component analysis (PCA), and classification using multiple algorithms. Among these, the linear discriminant analysis (LDA) model achieved the highest classification accuracy of 98.3% in discriminating WT and KO cells. This study demonstrates that Raman spectroscopy not only enables label-free analysis of subtle metabolic phenotypes at the single-cell level but also provides a novel platform for predicting adipocyte differentiation fate.