How to best Control Gene Expression in Cell Populations

in real-­�time.

Gianfranco Fiore*, Giansimone Perrino*, Mario di Bernardo, Diego di Bernardo


Automatic control is an engineering approach to regulate the behaviour of a system in an automated and precise manner. Controllers are used in everyday appliances from thermostats to microwaves, as well as, in complex and large systems such as airplanes and power-plants. The design principle is based on negative feedback, where the quantity to be measured (e.g. the temperature) is measured and compared to the desired value to yield a â??control errorâ?. The controller changes some physical process in the system (e.g. switches the boiler on or off) to reduce the control error to zero. Thanks to advances in molecular biology and biotechnology, control engineering methods have been recently proposed to regulate gene expression in living cells using a variety on computational and experimental approaches. Due to its very recent application to biological systems, it is not yet clear what control strategy works best. Here we provide a computational and experimental comparison of the two main control strategies proposed in literature: proportional-integral (PI) control and model predictive control (MPC). We then propose and test an innovative control strategy, Zero Average Dynamics (ZAD) control, which has been extensively used in electrical power converters but never in biological systems. We used a microfluidics-based control strategy we recently presented [Menolascina F. et al, PLOS Comp. Biolo. 2014; Fiore G. et al, Chaos, 2013; Menolascina F. et al, Automatica, 2011] to control in real-time the expression of a fluorescent reporter protein from the inducible GAL1 promoter in yeast Saccharomyces cerevisiae. In this set-up the computer can monitor cell fluorescence via a time-lapse fluorescent microscope and either feed cells with galactose, thus inducing expression from the GAL1 promoter, or glucose to repress gene expression. The control tasks we investigated are: (i) â??set-pointâ? control in which the protein has to be maintained at 50% of its maximum value in galactose, for 1500 min and (ii) â??tracking-controlâ? in which the protein has to maintain a value of 75% for 500 min, then 50% for another 500 min and 25% for the remaining 500 min. Experiments confirm the numerical simulation, proving that MPC and ZAD strategies can achieve successfully the regulation of gene expression in living cells for both set-point and tracking control, whereas the PI strategy has a worse performance, at least for the â??trackingâ? control task. The ZAD strategy however is much less computationally expensive than the MPC and does not require a detailed mathematical model of the system to be controlled. In conclusion our work shows that it is now possible to achieve precise regulation of gene expression in eukaryotic cells thus making possible to investigate cellular behaviour in completely new ways.