There are several ways in which one might study how cells behave in a fluctuating environment. At one extreme, one might look at average properties of the population, for instance, the extent to which a certain protein is phosphorylated. At the other extreme, one might look at single cells, for instance, at the signal of a fluorescent reporter expressed by the cells. Studying populations of dividing cells at the single-cell level, however, remains a challenge, although there is growing interest in cellular heterogeneity and its biological consequences.

In recent theoretical work, Edo Kussell at New York University and Stan Leibler at the Rockefeller University describe a framework for measuring the effective strength of selection acting on populations of cells. The makeup of a cellular population may change in response to environmental fluctuation both via individual cells sensing and responding to the changing conditions and via stochastic fluctuations upon which selection can act. Could they disentangle these two effects, Kussell and Leibler wondered, and thus measure the effective strength of selection in a growing population of cells?

The researchers describe a thought experiment that should allow one to do this. “Imagine that you grow cells in the condition you're interested in, and in parallel you grow them in the same conditions except you now increase the fitness differences between cells by some constant multiple,” explains Kussell. This should now accentuate the fitness differences between the cells so that selection can act on them. “At the end of the day, if the population structure is different between these two experiments, then you know selection is in play under those conditions.” In other words, you disentangled the effect of selection from that of cells sensing and responding to environmental change.

The problem is, it is difficult if not impossible to actually translate this experiment into reality. How would one practically increase the fitness of all individuals in a population by a small constant factor? What Kussell and Leibler found, however, is that by measuring a property that they call the variance of the historical fitness of cells, the results of the above thought experiment can be inferred. This property can be measured by examining independent individual lineage histories of cells in a growing population and by simply recording the number of times a cell divides over time under different conditions of interest.

Not only does the formulation of Kussell and Leibler provide a tool to study the effective strength of selection in a particular system, it also allows one to infer whether a cell or unicellular organism is using a stochastic switch to adapt to a particular environmental fluctuation. “For stochastic switches, there is a very pronounced peak in the historical fitness variance, which occurs at fluctuation periods that are longer than the generation time,” says Kussell. “So you can infer the existence and the type of an internal mechanism when you essentially don't know what the organism is doing, and you don't know what genes are important.” Undoubtedly, it will be of interest to begin to dissect the contribution of particular genes to such predicted stochastic switches.

In this work, Kussell and Leibler used lineage data obtained from simulations of growing Escherichia coli. However, experimental data in which growing cells, either microorganisms or mammalian cells, are imaged over time by live-cell video microscopy are increasingly becoming available. The theory developed in this work may well prove useful for studying the mechanisms by which populations of dynamic cells, stem cells or tumor cells, for instance, change in response to the environment, in contexts in which there is a selective advantage for the cells to be in one state or another.