| A strategy in regenerative medicine involves the restitution of functional tissues using biomaterials pre-populated with transplanted cells such as human mesenchymal stem cells (hMSC). However, current design and optimization of extracellular environments to controllably promote tissue-specific regeneration are guided by empiricism, and there is a lack of structure-activity relations underlying cell-biomaterial interactions. This dissertation is focused on using high resolution confocal/multiphoton fluorescence imaging for developing quantifiable descriptors of cell-biomaterial interactions under complex microenvironments, including three-dimensional scaffolds, textural gradients of polymer substrates, and soluble biochemical factors that stimulate differentiation or cancerous transformation.;In the first project (Chapter 2), we demonstrated the feasibility of using multiphoton imaging to quantitatively characterize microstructure of 3D biomaterial scaffolds and pseudo-3D cell morphology. This approach was further expanded, in the second project (Chapter 3), to a multidimensional space of cellular and subcellular features (termed cell descriptors) derived from: morphology, reporter protein expression, localization and spatial organization of protein reporters. Using spatially graded polymer blend substrates of both continuous roughness and discrete chemical compositions, we combined high throughput screening with high content analysis to identify both "global" and "high-content" structure-property relationships between cell adhesion and biomaterial properties such as polymer chemistry and topography.;In the next project (Chapter 4), we developed a novel molecular screening tool based on the high content descriptors of a nuclear reporter, nuclear mitotic apparatus (NuMA). Using high content imaging, data dimension reduction and machine learning techniques we mapped the nuclear features to different stem cell phenomena, specifically, stem cell lineage commitment to osteogenic versus adipogenic lineages. We reported that NuMA based nuclear descriptors captured the early lineage commitment of hMSC vs self-renewal. Moreover, a combined cytoskeletal and nuclear based "composite" profiling was demonstrated to be a robust tool to parse out not only lineage commitment vs self-renewal but within different lineages (e.g. osteogenic vs adipogenic). In the final project (Chapter 5), nuclear feature based modeling was used to discern early subcellular changes during oncogenic transformation. The utility of this approach was demonstrated by parsing a library of synthetic polymer substrates based on their differential potential to modulate carcinogen-induced transformation of stem cells. |
