UNDERSTANDING CELL DIFFERENTIATION AND MIGRATION WITH MULTIVARIATE CELL SHAPE QUANTIFICATION
MetadataShow full item record
This thesis focuses on developing multivariate quantification methods of cell shape to facilitate understanding of physiological processes such as cell differentiation and migration. Cell shape reflects complex intracellular and extracellular factors affecting cell function. However, analyses associating cell shape and cell function need to account for challenges of multivariate interpretation, single-cell heterogeneity and reproducibility. Specifically, Human Bone Marrow Stromal Cells (hBMSCs) population in nanofiber scaffolds can develop osteogenic differentiation without chemical cues. I developed a method based on Support Vector Machine (SVM) to train classifiers as boundaries in the shape metric spaces to identify the day 1 cell shape phenotype of hBMSCs population in nanofiber scaffolds. To reduce the effect of single-cell heterogeneity in the population phenotyping, the “supercell” method was introduced to generate average measurements of small groups of cells for SVM training. To overcome the multivariate complexity in biological interpretation, a feature selection process was implemented to select the most significant cell shape metrics. The predictive potential of the achieved classifiers was validated by subsampling. It was found that in nanofiber scaffolds, hBMSCs were narrower with more elongated and dendritic shape and rougher cell boundary. Further, I found that increase in nanofiber density enhanced hBMSCs osteogenic differentiation potential. The pre-trained classifiers successfully predict the modulation of nanofiber density on hBMSCs fates and single-cell shape. While much can be learned from cell shapes alone, it is important to note that shapes can change with time, especially for migrating cells. The second part of my thesis focuses on analysis of shape dynamics. Quantification for cell shape dynamics at the subcellular level was developed to understand the coordination of the subcellular myosin localization and the cell boundary dynamics in neutrophil migration in vivo. The correlation of myosin localization and positive cell boundary curvature was identified as a unique in vivo neutrophil migration phenotype. Correlations of myosin localization and local cell boundary dynamics in vivo were found to be affected by cell motility and polarization. This analysis framework shown here can also be used to study the link between other subcellular features and neutrophil migration and shape dynamics.