An integrated analysis of fibroblast morphology and migration on bioengineered substrata aided by machine vision and learning techniques.

摘要

In vivo, mechanical based signaling cues are nearly ubiquitous within organisms, organ systems, tissues and individual cells. Through the process known as mechanotransduction, various mechanical based stimuli, such as: stretch, shear and compression, are converted to biochemical responses which regulate biological phenomena within the model organs or systems. One specific type of mechanical stimuli involves a bidirectional feedback loop in which individual cells probe and respond to the material properties of extracellular matrix (ECM), specifically the elastic modulus of the substrata.;Both in vivo and in vitro, this signaling pathway is known to have a prominent role in regulating tissue morphogenesis and homeostasis, influencing cellular differentiation, regulating gene transcription and protein translation and determining cellular migration and morphology. The latter two phenomena, cellular migration and morphology, have been quantified in vivo using thin film substrata with tunable material properties composed of cross-linked polyacrylamide (pAAM). However, although both phenomena are the result of directed rearrangement of the cytoskeleton, rarely have the two been studied on the level of an individual cell and as an integrated process.;This thesis employs a multidisciplinary approach involving materials science, classical cellular biology and machine learning techniques to rigorously quantify the dependence of fibroblast morphology and migration on substratum stiffness.;First, thin film pAAM hydrogel substratum were generated with different crosslinker densities. Various aspects of the substratum were rigorously quantified including: elastic modulus, thickness and the density of surface ligand. The characterization of the substratum demonstrates the ability to accurately reproduce hydrogels over a range of elastic moduli (4.1 kPa to 136.2 kPa).;Introduction of BALB/c fibroblasts to these substrata allowed for the analysis both fibroblast morphological and migratory behaviors. Findings demonstrate that within a population, complexity in cellular architecture increases with the elastic modulus of the substratum and that cellular speed and persistence, as determined by the Random Cell Walk Model (RCWM), are biphasic with increased substratum stiffness. In addition, a stiffness-dependent increase in the diversity of both behaviors within the context of a population and an individual cell are reported. Finally, through the development of a morphological classification system, aided by the use of machine vision and learning techniques, experimental evidence is presented showing the interplay between cellular dynamic changes in morphology and the resulting migration pattern.