Quantitative analysis of the cytokine-mediated apoptosis-survival cell decision process

Abstract

How do cells sense their environment and decide whether to live or to die? This question has drawn considerable interest since 1972, when it was first discovered that cells have an intrinsic ability to self-destruct through a process called apoptosis. Since then, apoptosis has been shown to play a critical role in both normal physiology and disease. In addition, many of the basic molecular mechanisms that control apoptosis have been revealed. Yet despite the known list of interactions and regulators, it remains difficult to inspect the network of apoptosis-related proteins and predict how cells will behave. The challenge is even greater when one considers interactions with other networks that are anti-apoptotic, such as growth-factor networks. In this thesis, we develop an approach to measure, analyze, and predict how complex intracellular signaling networks transduce extracellular stimuli into cellular fates. This approach entails three interrelated aims: 1) to develop high-throughput, quantitative techniques that measure key nodes in the intracellular network; 2) to characterize the quantitative changes in network state and cell behavior by exposing cells to diverse fate-changing stimuli; and 3) to use data-driven modeling approaches that analyze large signaling-response datasets to suggest new biological hypotheses.

(cont.) These aims were focused on an apoptosis-survival cell-fate decision process controlled by one prodeath cytokine, tumor necrosis factor (TNF), and two prosurvival stimuli, epidermal growth factor (EGF) and insulin. We first developed radioactive- and fluorescence-based high-throughput assays for quantifying activity changes in the kinases that catalyze key phosphorylation events downstream of TNF, EGF, and insulin. By combining these assays with techniques measuring other important posttranslational modifications, we then compiled over 7000 individual protein measurements of the cytokine-induced network. The signaling measurements were combined with over 1400 measurements of apoptotic responses by using partial least squares (PLS) regression approaches. These signaling-apoptosis regression models predicted apoptotic responses from cytokine-induced signaling patterns alone. Furthermore, the models helped to reveal the importance of previously unrecognized autocrine cytokines in controlling cell fate. This thesis has therefore shown how cell decisions, like apoptosis-versus-survival, can be understood and predicted from the quantitative information contained in the upstream signaling network.