Quantitative analysis of the receptor-induced apoptotic decision network

Cells use a complex web of protein signaling pathways to interpret extracellular cues and decide and execute cell fates such as survival, apoptosis, differentiation, and proliferation. Cell decisions can be triggered by subtle, transient signals that are context specific, making them hard to study by conventional experimental methods. In this thesis, we use a systems approach combining quantitative experiments with computational modeling and analysis to understand the regulation of the survival-vsdeath decision. A second goal of this thesis was to develop modeling and analysis methods that enable study of signals that are transient or at intermediate activation levels. We addressed the challenge of balancing mechanistic detail and ease of interpretation in modeling by adapting fuzzy logic to analyze a previously published experimental dataset characterizing the dynamic behavior of kinase pathways governing apoptosis in human colon carcinoma cells. Simulations of our fuzzy logic model recapitulated most features of the data and generated several predictions involving pathway crosstalk and regulation. Fuzzy logic models are flexible, able to incorporate qualitative and noisy data, and powerful enough to generate not only quantitative predictions but also biological insights concerning operation of signaling networks. To study transient signals in differential-equation based models, we employed direct Lyapunov exponents (DLEs) to identify phase-space domains of high sensitivity to initial conditions. These domains delineate regions exhibiting qualitatively different transient activities that would be indistinguishable using steady-state analysis but which correspond to different outcomes. We combine DLE analysis of a physicochemical model of receptor-mediated apoptosis with single cell data obtained by flow cytometry and FRET-based reporters in live-cell microscopy to classify conditions that alter the usage of two apoptosis pathways (Type I/II apoptosis). While it is generally thought that the control point for Type I/II occurs at the level of initiator caspase activation, we find that Type II cells can be converted to Type I by removal of XIAP, a regulator of effector caspases. Our study suggests that the classification of cells as Type I or II obscures a third variable category of cells that are highly sensitive to changes in the concentrations of key apoptotic network proteins. Thesis Supervisor: Douglas A. Lauffenburger Title: Professor of Biological Engineering, Chemical Engineering, and Biology Thesis Supervisor: Peter K. Sorger Title: Professor of Biological Engineering Acknowledgements I have been immensely lucky to have had the opportunity to complete this thesis in the company of fantastic friends and colleagues. I would especially like to thank my advisors, Doug Lauffenburger and Peter Sorger for giving me the opportunity to join their research groups, and especially for their support and advice during my graduate career. They have gently guided me while giving me freedom to pursue my favorite research topics as a result, I have learned more than I expected along the way. I thank my committee members, Bruce Tidor and Alexander van Oudenaarden, who have also offered inspiration and useful research and career advice. I would also like to thank Suzanne Gaudet for her mentorship. Her positive outlook, keen scientific insights, and friendship have been invaluable to me. The work presented in this thesis would not have possible without the help of labmates and collaborators. I would like to thank the members of the Sorger lab and the Cell Decision Process Center, and the DOE Computation Science Graduate Fellows for their help and advice. I would like to acknowledge John Burke, Debbie Flusberg, Suzanne Gaudet, Sabrina Spencer, and especially John Albeck (the "death" group in the Sorger lab) for establishing a creative and helpful working-group to quantitatively study apoptosis. I am grateful to have had the opportunity to work with George Haller and Robert Szalai on the dynamical systems and DLE analysis; Viji Draviam on developing new metrics to study kinetochores; Julio Saez-Rodriguez and Jeremy Muhlich on automatic fitting of logic models; Jaydeep Bardhan, Kevin Chu and Jacob White on model order reduction; and H. Steven Wiley on modeling methods to study signaling. Finally, I would to thank my support network of family and friends. I am very grateful to be surrounded by BE classmates, many of whom have become very good friends and helped me through the trials and tribulations of both work and play. I would like to thank Reshma Shetty and Kathryn Loving for helping me study for the qualifying exams. I am also thankful for the ongoing support from my childhood friend Megan Boyle. I am thankful to have parents and parents-in-law, Brent and Deeporn Beardsley, and Don and Becky Aldridge, who have been so supportive. I would especially like to thank my husband John for his unwavering love and support, and his ongoing commitment to making sure that I am (outside) having fun.