MEASURING AND MODELING LIFE‐DEATH DECISIONS IN SINGLE CELLS

Abstract

At the level of single living cells, signal transduction involves remarkably complex biochemistry and great variability from one cell to the next. However, cells achieve exquisite control over fate decisions and subtle mutations that interfere with decision making cause disease. How can variability among genetically identical cells be reconciled with an apparent requirement for precise control over biochemical reactons, what is the impact of non‐genetic variability on the evolution of tumors and what are the implications for drug responses in patients?I will begin to address these issues with an emphasis extrinsic apoptosis induced by the TNF family member TRAIL. TRAIL is involved in natural immune surveillance of tumors and is also in development as a therapeutic agent. Some TRAIL‐treated human cells die within ~40 min, some only after 12 hr. and yet others live indefinitely. We have explored three explanations for these differences: (i) genetic or epigenetic variation (ii) the involvement of stochastic fluctuations (iii) transient but deterministic differences in cell state. I will illustrate how all three interact on different time scales to determine those aspects of cellular physiology that are highly invariant and those that are variable.I will also illustrate how studying such problems can be tackled by mathematical analysis of the underlying biochemistry. This involves a continuous interplay between (i) model generation, and variation (ii) calibration of models against experimental data and (iii) model verification through empirical testing of specific molecular hypotheses. I will describe a means to instantiate diverse biochemical hypotheses in networks of differential equations (using a new rules‐based meta‐language PySB) and to rigorously compare predictions to experimental data (via Bayesian parameter estimation). I will also describe strategies for collecting the necessary measurements from single‐cell studies and our attempts to push this into the mouse. I will advance the hypotheses that “systems‐based” approaches are nearing the point that it is possible to replace informal pathway pictures (which dominate molecular biology) with probabilistic mathematical constructs that assign rigorous “degrees of belief” to specific biochemical hypothesis. Finally, I hope to demonstrate that such approaches can be understood and used by a broad community of molecular biologists.