Cell Biology

Deadly Noise

Science Signaling  26 May 2009:
Vol. 2, Issue 72, pp. ec178
DOI: 10.1126/scisignal.272ec178

The great majority of experiments on signaling mechanisms monitor average responses of large numbers of cells and provide little or no information on the cellular variability or noise in the response. Spencer et al. used time-lapse microscopy to follow the response of live individual cells from two human cell lines maintained in culture. This allowed them to explore the source of variation in cellular response to the cytokine TRAIL (tumor necrosis factor–related apoptosis-inducing ligand). Such variation is of vital concern because TRAIL is being used in clinical trials as an agent to kill cancer cells. Cells exposed to TRAIL can either die quickly (within about an hour) or more slowly (within 8 to 12 hours), or can continue to live despite the treatment. When cells were exposed to TRAIL and cycloheximide (to block protein synthesis), Spencer et al. observed that sister cells derived from one mother cell inherited from the mother cell some characteristics that gave them a similar response interval in reacting to TRAIL. This correlation exponentially decayed, however, with a half-life of about 11 hours. This pattern of temporary heritability would be expected if the variation occurred because of the particular concentrations and states of proteins that happened to be present in the mother cell. Indeed, the duration of the similarity in response of daughter cells to TRAIL was highly sensitive to the rate of protein synthesis. The authors confirmed in a mathematical model of TRAIL-induced apoptosis that variation in protein concentrations of signaling components that regulate apoptosis was a likely explanation of the observed cell responses. The modeling also showed that the concentration of any single protein was not sufficient to predict the cellular response. What did appear to be critical was the combinatorial effects of such protein variation on the rate of the reaction in which the protein BID is cleaved by caspases (BID cleavage then leads to permeabilization of the mitochondria and cell death). The authors suggest that, given what we know about signaling pathways, such multivariate control of cell variability is likely to be the rule rather than the exception. The authors also note that, in other instances, cells display robustness in regulatory mechanisms that prevent relatively minor variations in protein concentrations or other properties from influencing cell phenotype. The authors argue that the variability of the cell death response is unlikely to reflect an inability to implement such a robust mechanism for regulation of cell death, but rather may reveal an evolutionary advantage gained from the existing circuitry, in which individual variation at the single-cell level is translated into a graded response at the level of the whole population of cells.

S. L. Spencer, S. Gaudet, J. G. Albeck, J. M. Burke, P. K. Sorger, Non-genetic origins of cell-to-cell variability in TRAIL-induced apoptosis. Nature 459, 428–432 (2009). [PubMed]