SERS-Cytomics for macrophage phenotyping and metabolic profiling

Abstract

Tumor-associated macrophages, the largest population of immune cells in the tumor microenvironment (TME), dominate the complexity of TME due to their highly phenotypic plasticity, leading to personalized tumor progression, differentiated drug response, and tolerance. However, current methods for the interpretation of cellular heterogeneity involve destructive, cumbersome, and high-cost sample pretreatment procedures. Herein, we present surface-enhanced Raman scattering cytomics (SERS-Cytomics) as a non-destructive and low-cost approach for macrophage phenotyping and metabolic profiling. SERS-Cytomics is capable of providing the whole molecular fingerprints of living cells at a single-cell level. A deep-learning model enables accurate classification of three phenotypes of macrophages exceeding 95%. Further, the Shapley additive explanations analysis is adopted to screen the differential metabolic signals, where two metabolic biomarkers at 1071 (glucose) and 1440 (cholesterol) cm−1 display the maximum weight in the classification of M1 and M2 macrophages, respectively. Correlation explanation of SERS-Cytomics and transcriptomics reveals the involvement of the glycolytic pathway and lipid metabolism in M1 and M2 macrophages, respectively. Therefore, SERS-Cytomics enables an efficient and non-destructive analytical tool for cellular heterogeneity explanation at a single living cell level.