Imaging cytometry to measure stochastic gene expression in individual cells
Molecular and cell biologists have uncovered so many molecular interactions involved in the control of various aspects of the cell physiology that it is often no longer possible to comprehend these regulatory networks by relying on intuition and semi-quantitative reasoning. Mathematical modeling and software simulations are now essential to understanding how cells process information and the biological function of individual molecules. Modeling efforts have uncovered the modular architecture of regulatory processes composed of recurrent functional motifs. For instance, the process controlling cellular division in eukaryotes is now understood as a combination of switches and oscillators.
The topology of regulatory networks is not sufficient to build mathematical models that match experimental data; it is also necessary to find parameter values that ensure models behave like the biological system they represent. Unfortunately, very few parameters of specific molecular interactions in living cells have been measured experimentally. As an example, the maturation of fluorescent proteins in vivo remains poorly characterized despite their ubiquitous use as reporters of the dynamics of artificial and natural gene networks. Slow maturation dynamics can obscure faster underlying dynamics at the transcription and translation levels. Moreover, the slow maturation dynamics has rarely been quantified in vivo, despite published protocols and the fact that the mechanisms are well understood. Only a handful of estimates have been published in E.coli and in yeast, leaving most of the fluorescent proteins largely undocumented.
Lack of information about basic properties of fluorescent proteins illustrates the difficulty of estimating rate parameters of molecular networks. Current measurement techniques are only capable of partially observing the dynamics of genetic systems. Only a small fraction of the molecular species represented by variables in a mathematical model of a regulatory network is measurable. Hence, the experimental dataset does not allow the estimation of all the parameters in the underlying model.
Extracting more information than is possible with traditional instruments is one approach to resolving model under-determination. New instruments combining fluorescent microscopy and microfluidics are providing dense-time series of gene expression data in individual cells. In this project, we combine a novel instrument with a method to assess the quality of fit of a stochastic model to single-cell data. Specifically, we are building GenoSIGHT, an automated, adaptive imaging cytometry platform that collects high-time resolution trajectories of single cells in a controllable microfluidics environment. We are using distribution-based method to assess the match between a mathematical model of gene expression and GenoSIGHT data. After fitting the model to experimental data, we can use the model to predict the behavior of a modified system. This approach should lead to more predictive models of simple genetic systems in both systems and synthetic biology.
Sources of Funding
- 2010-2012: National Science Foundation Award DBI-0963988
- 2009-2013: Department of Defense, SMART Graduate Fellowship to Matt Lux.
- Ball DA, Marchand J, Poulet M, Baumann WT, Chen KC, Tyson JJ, Peccoud J. (2011) Oscillatory dynamics of cell cycle proteins in single yeast cells analyzed by imaging cytometry. PLoS ONE 6:e26272
- Ball DA, Lux MW, Adames NR, Peccoud J (2014) Adaptive Imaging Cytometry to Estimate Parameters of Gene Networks Models in Systems and Synthetic Biology PloS one 9 (9), e107087