Single Cell Dynamics Causes Pareto-Like Effect in Stimulated T Cell Populations

Naive CD4 + T lymphocytes are able to take multiple fate-decisions; they can give rise to various specialized cell types such as T helper effector or regulatory T (Treg) lymphocytes1. They do this in response to stimulations of their T-cell receptors (TCR) and various cytokines. Although studied for decades, the mechanisms of cell fate choice between different options remain elusive. The hypothesis of stochastic fate choice of hematopoietic cells was proposed 50 (fifty!) years ago2. Yet, the debate between the stochastic and deterministic mechanisms is still not settled. Some consider that the acquisition of the differentiated phenotype occurs via a predetermined pathway3, where each signal induces a defined cell fate. Others argue in favour of a stochastic mechanism4. According to this view, a cell responds to a signal by randomly choosing between two or more options. It is the collective action of the individual stochastic choices that creates non-random regularities at the level of the whole cell population. We have previously observed that phenotypic heterogeneity may appear spontaneously and contribute to the fate decisions in a clonal population without the action of external signals5,6,7,8. In the present study we aimed to evaluate the stochastic contribution to T cell differentiation on the basis of single-cell observations obtained in an in vitro system. When the naive T cells are stimulated in vitro with anti-CD3 and anti-CD28 antibody-coated beads, IL-2 and TGF-β, they proliferate and preferentially acquire the Treg phenotype making this artificial system convenient for the study of cell fate decision-making mechanisms9. These conditions are highly selective, because essentially all cells acquire Treg phenotype after a week or so of culture. One can therefore consider that there is little room left for chance during this process. Nevertheless, the cells take at least two different decisions: they divide or they differentiate. It is not known whether these two decisions are independent or whether they are taken in a fixed pre-determined order. Recently, we observed that the majority of the cells are displaying a Treg phenotype after a week of culture, some cells reach this stage after only one or two divisions while others divide up to ten times7. This substantial proliferation heterogeneity is surprising in a cell population where each cell encounters identical conditions. In order to get insight in the origin of this heterogeneous behavior we used a single-cell time-lapse approach coupled to mathematical modeling. Single-cell observations were successfully used to demonstrate the stochastic nature of fate decisions in B-cell differentiation10. Here, we used primary CD4 + cells from Foxp3-GFP knock-in mice so the acquisition of the Treg phenotype could be monitored in living cells using the expression of the GFP protein11. We observed substantial heterogeneity in the proliferation, differentiation and death rates leading to an unequal contribution of clonal cell lineages to the final population. Data-driven modeling of stochastic cell decision allowed us to show that the observed Pareto-like effect essentially results from the cumulative effect of stochastic cell decisions and events. Variations of cell cycle length and cell death rate are the key factors contributing to the phenotypic heterogeneity of the final cell population. The initial differences between the cells in the starting population may reinforce this effect but alone is insufficient to fully account for it. Our observations show that due to the heterogeneity of proliferation and death rate, the final cell population is composed essentially from cells derived from a small number of initial founder cells.