Machine Learning Classification of Integrin-Expression-Based Magnetic Sorted SW 620 Cells by Simultaneous O-PTIR and SERS

Immortalized cell lines are commonly used for in vitro studies such as drug efficacy, toxicology, and life cycle due to their cost effectiveness and accessibility; however, subpopulations within a cell line can arise from random mutations or asynchronous cell cycles which may lead to results that make interpretation difficult. A method that could classify these differences and separate unique subpopulations would increase our understanding of heterogeneous cellular responses. In the present work, we explore spectroscopic signals associated with subpopulations of cells magnetically sorted on the basis of α₅β₁ integrin binding to cyclic-RGDfC which mimics fibronectin in the extracellular matrix. SW620 colon cancer cells were incubated with cyclic-RGDfC functionalized gold-coated, iron core nanoparticles and magnetically sorted. The subpopulations from the sort were imaged (<i>N</i> = 10 positive and <i>N</i> = 10 negative, number of cells) via simultaneous surface-enhanced Raman scattering (SERS) and optical-photothermal infrared spectroscopy (O-PTIR). Pearson correlations of the standard peptide-protein interaction in the SERS channel allowed for visualization of the cyclic RGDfC-integrin α₅β₁ interaction. Partial least-squares discriminant analysis of the O-PTIR spectra collected from cell maps successfully classified the positively or negatively sorted cells. These results demonstrate that biochemical changes within a single cell line can be sorted via an integrin-activity-based assay using simultaneous SERS and O-PTIR.