Image-based Bioanalytics To Profile Heterogeneity And Quantify Phenotype: Extracellular Vesicle Size Characterization And High-Content Analysis Of Macropinocytosis

Abstract

Analyzing images has a long history in biology. As imaging technology and image analysis techniques have grown more advanced and more widespread, the applications for image-based tools have dramatically increased. We develop image-based analytical methods for two different biological applications in this thesis. In Chapter 2, we develop Decorrelation Analysis to accurately measure submicron particle size from individual particle trajectories captured during particle tracking experiments, i.e. video ultramicroscopy. Though broadly useful, we intend this method to be used in the physical characterization of heterogeneous biological samples, and in particular extracellular vesicles, which garner substantial attention in immunology, oncology and drug delivery research. In Chapter 3, we develop Spot Sauvola and Latent Heterogeneity Modeling (LHM) for the purpose of exploring macropinocytosis heterogeneity, “phenotypic profiling”, and quantifying macropinocytosis activity, “phenotype scoring”, from fluorescent images of adherent cells. Dramatic improvements in computing power, image analysis methods and the development of fully-automated confocal microscopes have driven the adoption of image-based high-throughput screening for toxicity screening, phenotypic profiling and elucidating drug mechanism-of-action. Because LHM does not heavily rely on huge data sets, we intend LHM to be used by academic researchers analyzing image-based biological experiments for quantifying cellular phenotypes in general and not just for macropinocytosis. A key benefit of LHM is that results are biologically interpretable. KRAS mutations are present in 25-30% of all human cancers, and have been shown to upregulate constitutive macropinocytosis to scavenge extracellular nutrients in pancreatic cancer. Surprisingly, two particular KRAS mutation alleles represent almost 2/3 of all pancreatic tumor cases, which could suggest a particular fitness advantage conferred by these specific mutation alleles and not conferred by alleles found in other KRAS-dependent cancer types, such as non-small cell lung cancer. In Chapter 4 we use the unique analytical tools developed in Chapter 3 to explore the possibility that macropinocytic scavenging is the specific fitness advantage conferred by the mutation alleles KRASG12D and KRASG12V. Our findings support the possibility that macropinocytosis is selectively conferred by G12D and G12V amino acid substitutions but a much larger panel of mutation alleles should be studied.