Spatio-Temporal Measurements and Modeling of Genetic Expression

Abstract

Single-molecule, single-cell studies of genetic expression have provided key insights into how cells respond to external stimuli [Munsky, B., et al., Science (2012)]. By directly measuring copy numbers of individual bio-molecules in cells, it is now possible to obtain statistical measures of the spatio-temporal distributions of key signaling and regulatory networks. Such comprehensive datasets can be used to infer system-level models that yield quantitative insight into cellular regulation, predict cellular responses in new experimental conditions, and suggest more revealing experiments to uncover regulatory dynamics. The integration of single-molecule spectroscopy, biochemistry, and numerical modeling is a powerful multi-disciplinary approach to investigating cellular response at the genetic level. A key issue we seek to address is what types of fluctuations are most informative about the underlying gene regulatory process. In other words, how much experimental resources should be spent to measure (i) temporal, (ii) spatial, or (iii) cell-to-cell fluctuations? As an example, we studied Interluekin 1-alpha (IL1α) mRNA expression within human THP-1 cells during stimulus response to lipopolysaccharide (LPS). By spatially resolving individual mRNA using multiplexed single molecule FISH [Femino A.M., et al., Science (1998), Raj A., et al., Nat Meth (2008)] in large populations of single cells at multiple times points, we quantified all three fluctuation types. We expanded the common bursting gene expression model [Peccoud, J., Theoretical Population Biology (1995)] and derived a set of linear ODEs to describe the mean, variance, and co-variance of nuclear and cytoplasmic IL1α mRNA. We fit this model to multiple single-cell datasets. Comparing models inferred from each data set, we are able to draw conclusions on which fluctuation types are most revealing about the underlying system's mechanisms and parameters, providing feedback for new experiments. The approach developed here is applicable to any eukaryotic gene expression pathway.