(118d) Tracking Overflow Metabolism for Efficient Growth By Designing a Kalman Estimator for Acid Production in E.coli

Authors: 
Pepper, M., Clemson University
Groff, R., Clemson University
Harcum, S. W., Clemson University
Wang, L., Clemson University
Padmakumar, A., Clemson University

Biopharmaceuticals are grown in cultures of yeast, E.coli, or mammalian cells.  In recent decades, dozens of biopharmaceutical products have been developed and produced, resulting in a $99 billion industry. Bioprocess control and optimization has been a growing field for the past 30 years, with researchers constantly improving the processes and the organisms. To maximize biomass, the culture growth rate must be carefully constrained by the rate of substrate addition. 

Like yeast and mammalian cells, E.coli may be considered to have two main metabolic states, oxidative metabolism and overflow metabolism.  In oxidative metabolism, the organism takes in glucose and oxygen and produces carbon dioxide and some acidic byproducts. Biomass production is most efficient in oxidative metabolism.  In overflow metabolism, the organism takes in more glucose than can be processed by oxidative metabolism, and the excess is converted to a waste metabolite, acetate (E. coli), ethanol (yeast) or lactate (mammalian cells).  The presence of high concentrations of the waste metabolite inhibits metabolism overall.  Thus overfeeding actually results in slower growth. The highest growth rates are achieved by keeping the substrate feed rate as high as possible without pushing the organism into overflow metabolism. In E. coli and mammalian cells, the waste metabolites are acidic. 

In practice, the selection of substrate feed rates is typically based on previous experience with a specific strain.  An open loop exponential feed rate profile often works well, but it can take some time to find the proper exponential rate.  The ultimate goal is to develop control algorithms that use commonly available sensors to actively choose a substrate feed rate that maintains the culture in a state of high oxidative growth.  This abstract specifically will present an estimator of the culture acid production rate.  This estimator will be an important tool for diagnosing the metabolic state of the culture in real time. 

An estimator is described that tracks the acid production rate of a culture.  The pH dynamics of a minimal media culture were characterized by analyzing the output of the bioreactor pH measurements in response to known quantities of acetic acid and ammonia.  The culture media is too complex to be described by a simple Henderson-Hasselbalch equation. Because the pH is controlled to be within a small range, a simple linear model was chosen instead. Under normal culture conditions, a PID loop is used to maintain the culture pH at a specified level by adding base, specifically ammonia.  Using the pH model for the minimal media, a Kalman filter was designed that uses the pH probe measurements and base addition as inputs.  The Kalman filter estimates the acid production rate of the culture based on the inputs.  Since E.coli overflow metabolism produces acidic byproducts, the Kalman filter prediction can be used as an indicator for overflow metabolism. This would enable development of feed rate control algorithms that would maintain a high rate of oxidative metabolism while avoiding overflow.