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

Area of Science:

• Biotechnology
• Cell Biology
• Spectroscopy

Background:

• Immortalized cell lines are vital for in vitro research but can develop subpopulations, complicating data interpretation.
• Heterogeneity in cell lines arises from mutations or asynchronous cell cycles, impacting study reproducibility.
• A method to classify and separate these subpopulations is needed to enhance understanding of cellular responses.

Purpose of the Study:

• To develop and validate a method for classifying and separating cell subpopulations based on biochemical differences.
• To utilize spectroscopic signals from magnetically sorted cells to identify unique subpopulations.
• To improve the reliability and interpretation of in vitro studies using heterogeneous cell lines.

Main Methods:

• SW620 colon cancer cells were incubated with RGDfC-functionalized nanoparticles for magnetic sorting.
• Cells were sorted based on α5β1 integrin binding activity.
• Simultaneous surface-enhanced Raman scattering (SERS) and optical-photothermal infrared spectroscopy (O-PTIR) were used for imaging and analysis.

Main Results:

• Pearson correlations in SERS visualized the cyclic RGDfC-integrin α5β1 interaction.
• Partial least-squares discriminant analysis of O-PTIR spectra successfully classified sorted cell subpopulations.
• The study demonstrated the ability to sort cells based on biochemical changes using an integrin-activity assay.

Conclusions:

• Simultaneous SERS and O-PTIR spectroscopy combined with magnetic cell sorting can effectively differentiate and separate cell subpopulations.
• This integrin-activity-based assay provides a novel approach to address cellular heterogeneity in in vitro studies.
• The developed method enhances the understanding of biochemical variations within a single cell line, leading to more reliable research outcomes.