389053 Autonomous Control of Metabolism with Synthetic Sensor-Circuits
Micro-organisms have been engineered to manufacture diverse chemical products, including drugs, materials, and fuels; however, production titers are often anti-coupled to the pathway’s biosynthesis rate, due to competition for essential metabolites and the accumulation of toxic byproducts. As a result, a key challenge has been to maximize production titers with changing bioreactor conditions and feedstock compositions. To solve this challenge, microbial sensors have been employed to dynamically control pathway flux; for example, using stress-responsive promoters to control isoprenoid biosynthesis (Dahl, Zhang, et. al., 2013) and fatty acid-based products (Zhang, Carothers, & Keasling, 2012). Similar to the chemicals industry, the development of sophisticated metabolic controllers has the potential to adaptively optimize diverse biosynthesis processes, though quantitative design principles have yet to be elucidated.
Here, we developed, experimentally characterized, and computationally modeled sensor-circuit-pathway systems to develop a design approach for selecting the ideal sensor and circuit to control a target pathway. We built sensors that take advantage of the cell’s natural stress response to detect growth phase changes, starvation conditions, the presence of inhibitory feedstock components (furfural), and accumulation of toxic byproducts. We experimentally combined sensors with signal amplifier genetic circuits to increase the sensors’ control ranges. We then experimentally connected sensor-circuits to a carotenoid biosynthesis pathway that directly competes with cell growth. Sensors, circuits, and the pathway are individually modeled, and combined, to predict how different sensors and circuits control pathway flux dynamics. We propose an initial set of design rules to maximize production titers.
Specifically, our characterized sensors increase protein expression by up to 5-fold, with respect to control, in response to farnesyl diphosphate depletion as a result of activation of an optimally balanced neurosporene biosynthesis pathway with a maximum productivity of 517 ug/gDCW/hour (Farasat et. al., 2014). These sensors are combined with a two-repressor signal amplification circuit to further widen the input-output range by 10-fold. Each component’s model has been experimentally validated; sensors are modeled using state-dependent differential equations, circuits are modeled using statistical thermodynamics, and the pathway is modeled using elementary mass action kinetics. We present both graphical and numerical solutions to this classic control problem, applied to biosynthesis pathways, and highlight experimental sensor-circuit-pathway results indicative of typical control behaviors. Our results are the first steps towards developing comprehensive control principles for metabolic systems, while providing re-usable sensors and circuits to dynamically regulate diverse pathways.