Identifying subpopulations in multicellular systems by quantitative chemical imaging using label-free hyperspectral CARS microscopy

Abstract

Quantitative hyperspectral coherent Raman scattering microscopy merges imaging with spectroscopy and utilises quantitative data analysis algorithms to extract physically meaningful chemical components, spectrally and spatially-resolved, with sub-cellular resolution. This label-free non-invasive method has the potential to significantly advance our understanding of the complexity of living multicellular systems. Here, we have applied an in-house developed hyperspectral coherent anti-Stokes Raman scattering (CARS) microscope, combined with a quantitative data analysis pipeline, to imaging living mouse liver organoids as well as fixed mouse brain tissue sections xenografted with glioblastoma cells. We show that the method is capable of discriminating different cellular sub-populations, on the basis of their chemical content which is obtained from an unsupervised analysis, i.e. without prior knowledge. Specifically, in the organoids, we identify sub-populations of cells at different phases in the cell cycle, while in the brain tissue, we distinguish normal tissue from cancer cells, and, notably, tumours derived from transplanted cancer stem cells versus non-stem glioblastoma cells. The ability of the method to identify different sub-populations was validated by correlative fluorescence microscopy using fluorescent protein markers. These examples expand the application portfolio of quantitative chemical imaging by hyperspectral CARS microscopy to multicellular systems of significant biomedical relevance, pointing the way to new opportunities in non-invasive disease diagnostics.