Using high throughput technologies to understand heterogeneous drug response

Heterogeneity is an inherent quality of biological systems, and is one of the properties that facilitates adaptation of organisms to new environments and stimuli. However, the scientific method relies on probabilistic reasoning wherein sample measurements are assumed to represent the population as a whole. In situations where this assumption is incorrect, as when a population is unknowingly made up of multiple subpopulations, the underlying biology may be obscured, resulting in faulty conclusions. Treatments developed based on these faulty conclusions may, at worst, harm people rather than help them. In Part I, I explore heterogeneity at the organismal level by investigating why drug response differs between people. Chapter 1 describes the possible consequences of assessing medical interventions on heterogeneous populations, and introduces pharmacogenomics as the first step towards personalized medicine. For the goals of pharmacogenomics to be realized, the mutational landscape of cytochrome P450 (CYP) enzymes, which catalyze the elimination or activation of most drugs currently in use, must be comprehensively assayed. In Chapter 2, I describe my contribution to this goal in measuring the abundance of nearly all single mutations in one CYP protein, 2C19, and jointly analyzing the dataset with a similar one from its closest neighbor, 2C9. These analyses highlight regions where mutational tolerance differs between CYP2C19 and 2C9, despite their structural and sequence similarity, revealing possible mechanisms underlying their functional differences. In Part II, I investigate how morphological heterogeneity manifests in clonal populations of cells when exposed to drugs. Chapter 3 introduces how visual phenotypes are tightly linked to cell state, emphasizing the paucity of technologies that combine microscopy with mass spectrometry-based proteomics. In Chapter 4, I describe my solution to this problem by integrating Visual Cell Sorting (VCS), which separates cells based on visual phenotypes, with low-input mass spectrometry (MS). I demonstrate that this new technology, VCS-MS, separates cells with similar fidelity to fluorescence-assisted cell sorting, and achieves nearly double the proteomic depth of similar technologies. Taken together, this dissertation demonstrates the critical need to develop and employ technologies capable of investigating heterogeneity at all levels of biological systems.