Long-term model predictive control of gene expression at the population and single-cell levels

Abstract

Gene expression plays a central role in the orchestration of cellular processes. The use of inducible promoters to change the expression level of a gene from its physiological level has significantly contributed to the understanding of the functioning of regulatory networks. However, from a quantitative point of view, their use is limited to short-term, population-scale studies to average out cell-to-cell variability and gene expression noise and limit the nonpredictable effects of internal feedback loops that may antagonize the inducer action. Here, we show that, by implementing an external feedback loop, one can tightly control the expression of a gene over many cell generations with quantitative accuracy. To reach this goal, we developed a platform for real-time, closed-loop control of gene expression in yeast that integrates microscopy for monitoring gene expression at the cell level, microfluidics to manipulate the cells' environment, and original software for automated imaging, quantification, and model predictive control. By using an endogenous osmostress responsive promoter and playing with the osmolarity of the cells environment, we show that long-term control can, indeed, be achieved for both time-constant and time-varying target profiles at the population and even the single-cell levels. Importantly, we provide evidence that real-time control can dynamically limit the effects of gene expression stochasticity. We anticipate that our method will be useful to quantitatively probe the dynamic properties of cellular processes and drive complex, synthetically engineered networks.model based control | computational biology | high osmolarity glycerol pathway | quantitative systems biology U nderstanding the information processing abilities of biological systems is a central problem for systems and synthetic biology (1-6). The properties of a living system are often inferred from the observation of its response to static perturbations. Timevarying perturbations have the potential to be much more informative regarding the dynamics of cellular functions (7-12). Currently, it is not possible to precisely perturb protein levels in an analogous manner, even though this perturbation would be instrumental in our understanding of gene regulatory networks. Indeed, despite the development of novel regulatory systems, including various RNA-based solutions (13), transcriptional control by means of inducible promoters is still the preferred method for manipulating protein levels (14, 15). Unfortunately, inducible promoters have several generic limitations. First, there is a significant delay between gene expression activation and effective protein synthesis. Second, many cellular processes can interfere with gene expression through internal feedback loops whose effects are hard to predict. Third, the process of gene expression shows significant levels of noise (16)(17)(18). Given these limitations, novel experimental strategies are required to gain quantitative, real-time control of gene expression in vivo.Here, we see the...

Automatic Control of Gene Expression in Mammalian Cells

“…A more advanced fully automated strategy was described in 2012, to control the expression of a reporter protein from the Hog1-responsive promoter using a microudics platform (3 ) with osmotic pressure as control input. We reported the rst control strategy to regulate gene expression from the GAL1 promoter using galactose and glucose as control inputs (4 ) by means of an ad-hoc microuidic platform.…”

Iterative experiment design guides the characterization of a light-inducible gene expression circuit

“…5 were obtained by applying precomputed light induction patterns in an open loop fashion. Although the use of feedback control (20,24) is always recommended to compensate for unpredictable disturbances or unknown initial conditions, the success of our open loop experiments is proof that the remaining mismatch in our model is practically negligible, contrary to what was observed for the deterministic model used in ref. 20.…”

Entrainment and Control of Bacterial Populations: An in Silico Study over a Spatially Extended Agent Based Model

“…For example Milias-Argeitis et al [12] used a fourth-order linear model describing a light-responsive genetic network in order to control the gene expression of a microbial population, around a reference value, using optogenetics. Uhlendorf et al [13] used a two-variable delay differential equation (DDE) model to capture the dynamics of the yeast hyperosmotic stress response and compute inputs to make a population of cells to follow a time-varying signal. In Menolascina et al [14], a switching control strategy is developed and implemented in a five-variable time-delayed dynamical system that models a synthetic network in yeast in order to control the output of one of its gene products to a specific value.…”

“…To study the feasibility of linear control techniques over a population of bacterial cells part of a microfluidics platform we extend the spatially resolved ABM presented in Mina et al [21] to study open loop (nonfeedback) and closed loop (feedback) control strategies in silico. Thus, unlike most work in control of cellular populations [12][13][14], we also consider the spatial dependencies of coupled cells undergoing global control; in this case classical P-control, PI-control and PID-control [1].…”