An evolutionarily significant unicellular strategy in response to starvation in Dictyostelium social amoebae

The social amoeba Dictyostelium discoideum is widely studied for its multicellular development program as a response to starvation. Aggregates of up to 10 (6) cells form fruiting bodies containing (i) dormant spores (~80%) that can persist for months in the absence of nutrients, and (ii) dead stalk cells (~20%) that promote the dispersion of the spores towards nutrient-rich areas. It is often overlooked that not all cells aggregate upon starvation. Using a new quantitative approach based on time-lapse fluorescence microscopy and a low ratio of reporting cells, we have quantified this fraction of non-aggregating cells. In realistic starvation conditions, up to 15% of cells do not aggregate, which makes this third cell fate a significant component of the population-level response of social amoebae to starvation. Non-aggregating cells have an advantage over cells in aggregates since they resume growth earlier upon arrival of new nutrients, but have a shorter lifespan under prolonged starvation. We find that phenotypic heterogeneities linked to cell nutritional state bias the representation of cells in the aggregating vs. non-aggregating fractions, and thus affect population partitioning. Next, we report that the fraction of non-aggregating cells depends on genetic factors that regulate the timing of starvation, signal sensing efficiency and aggregation efficiency. In addition, interactions between clones in mixtures of non-isogenic cells affect the partitioning of each clone into both fractions. We further build a numerical model to test the evolutionary significance of the non-aggregating cell fraction. The partitioning of cells into aggregating and non-aggregating fractions is optimal in fluctuating environments with an unpredictable duration of starvation periods. Our study highlights the unicellular component of the response of social amoebae to starvation, and thus extends its evolutionary and ecological framework.