| 總結: | Bacteria, the most abundant organisms supporting life on Earth, inhabit various niches, from soil and aquatic life, to residing in hosts. Bacteria frequently experience oxidative stress during aerobic growth, host immune attack, microbiome interactions, and under antibiotic treatment. Here, we study bacteria under hydrogen peroxide (H2O2) and observe that individual cells exhibit variability in their oxidative stress response. Such diversification of cellular phenotype is often observed in response to changing environments, but the underlying causes of cell-cell variability and its implications for cellular function in the context of stress adaptation remain unclear. Cellular response variability is typically attributed to stochasticity in biochemical reactions inside a cell. Our machine learning algorithm reveals that the variability in cellular response is not stochastic but determined by the precise response of cells to dynamic levels of H2O2, created by short-ranged cell-cell interactions. We complement single-cell time-lapse imaging with mathematical modeling to probe the consequences of single-cell variability on population adaptation. We find that the H2O2 scavenging activity of cells creates strong H2O2 gradients, thereby protecting a large proportion of the population. We show that the gene expression fluctuations of individual cells during constant H2O2 treatment are in fact driven by chaos. Although it has been suggested that chaos may play a role in many biological phenomena, it has remained challenging to disentangle chaos from noise as a source of variability in biological data. The close correspondence between our experiments and model allowed us to show that chaos emerges from deterministic feedbacks between cells and their environment. These feedbacks amplify small differences in initial conditions, resulting in diverging stress response dynamics that lead to seemingly random phenotypic variability. Next, we investigate how individual cells manage to regulate the levels of many diverse genes that encode oxidative stress tolerance factors. We find that the dynamics of over two dozen genes create a diversity of spatiotemporal expression patterns that benefit the stress adaptation of a cell population. Overall, we show that bacterial stress responses can generate variability under stress via deterministic factors without noise. Our work provides a general approach for uncovering the hidden variables that drive variability in cellular responses to environmental changes and for probing the regulatory mechanisms from molecular, single-cell, and population perspectives.
|
|---|