Quantitative approaches to understand and modulate single-cell heterogeneity

Abstract

Natural biological variability is essential for adaptation and survival in dynamic environments, but this heterogeneity can also hinder therapeutic efficacy (e.g., antibiotic resistance in bacteria) and robust cell engineering (e.g., undesirable cell types in stem cell differentiation). Quantitative analysis of single cell behavior can aid in designing better strategies for achieving desirable outcomes despite cellular heterogeneity. However, the complexity of the underlying non-linear signaling networks combined with multistability and biomolecular noise pose a significant challenge to better understanding and characterizing these phenomena. Here, we developed a framework to analyze single-cell heterogeneity in order to understand biological variability as well as design strategies to homogenize cellular responses. Specifically, we developed and validated a facile energy landscape generation methodology that offers intuitive and faithful visualizations of cell phenotype and facilitates a more quantitative understanding of cellular dynamics.We then applied this framework to the p21-/Cdk2-dependent quiescence-proliferation decision in breast cancer dormancy to examine single-cell dynamics on the landscape when perturbed by hypoxia, a dormancy-inducing stress. Using single-cell time-lapse microscopy, we found that initial position on a p21/Cdk2 landscape did not fully explain the observed cell-fate heterogeneity under hypoxia. Instead, cells with higher cell state velocities prior to hypoxia, influenced by epigenetic parameters, tended to remain proliferative under hypoxia. Thus, the fate decision on this landscape is significantly influenced by “inertia”, a velocity-dependent ability to resist directional changes despite reshaping of the underlying landscape, superseding positional effects. Such inertial effects may markedly influence cell-fate trajectories in tumors and other dynamically changing microenvironments. Finally, we devised a strategy to control cell-to-cell heterogeneity in p53 tumor suppressor dynamics in response to radiotherapy. Using a mathematical model-guided approach aided by live cell microscopy and single-cell tracking, we exploited the oscillatory nature of the p53 irradiation response to entrain and synchronize p53 dynamics in silico and in vitro. This synchronized population is expected to respond more uniformly to treatment. Further, there is a change in p53 frequency due to entrainment that translates to altered mRNA expression dynamics of downstream targets implying that we could drive these cells to a desired fate outcome. Interestingly, the range of frequencies across which the system could be entrained was significantly wider than the range suggested by the mathematical model. Energy landscape analysis showed, somewhat non-intuitively, that this was due to a weaker than predicted p53 oscillator in cells and, in general, that a weak endogenous oscillator that has a wider and shallower stable steady state is more entrainable. Overall, these studies exemplify the potential of utilizing a quantitative energy landscape-based approach to analyze cell-to-cell variability in decision-making and evaluate potential interventions to modulate this heterogeneity.