Analyzing Life-Death Decisions in Human Cells using a Systems Approach

Peter K. Sorger

Dept. of Biology and Biological Engineering, MIT Cambridge MA,02139
Dept. of Systems Biology, Harvard Medical School

Cytokines, growth factors and their receptors activate complex signaling cascades that regulate cell proliferation, death and differentiation. We are using a systems biology approach to study cytokine signaling in human cells with an eye towards understanding cell-type variation and the differences between normal and diseased states. We focus on signaling downstream of pro-apoptotic cytokines such as Tumor Necrosis Factor (TNF) and TRAIL and pro-survival growth factors such as EGF and the insulin-like growth factors (IGF). I will describe the development, training and testing of a physico-chemical model describing apoptotic signaling initiated by TNF and TRAIL. We have regressed our compartmentalized ODE-based model against data obtained from single cells and analyzed the findings using singular perturbation theory. To validate the model experimentally, we have examined cells exposed to a wide range of death ligand concentrations and cells in which specific apoptotic proteins have been depleted by RNAi.

We find that receptor mediated cell death constitutes a variable delay-snap action switch in which cells can wait for many hours before committing to death, but that once the process starts, it proceeds extremely rapidly and always to completion. A well-studied caspase cascade is involved in linking receptor activation to death, but it is a feedforward pathway involving the mitochondrion that gives it its snap-action behavior. Numerical analysis reveals the quantitative details of amplification and protein compartmentalization necessary for this feed-forward control. Modeling also suggests that under certain conditions in which caspase inhibitors (IAP proteins) are partially depleted and mitochondrial depolarization delayed, a state might be reached in which caspases are only partly activated and substrates incompletely cleaved. We have generated this state experimentally using a variety of RNAi, protein overexpression and pharmacological interventions and shown that a cells can survive partial caspase activation. However, cells in which the apoptotic switch has fired incompletely appear to suffer the type of genomic damage that predisposes cells to oncogenic transformation. Thus, the snap-action behavior of receptor-mediated caspase activation revealed by modeling and experimentation is likely to be a critical feature of programmed cell death.

Back