Controlling the composition of bacterial co-cultures via in silico feedback

Communities of microbes play important roles in natural environments and hold great potential for deploying division-of-labor strategies in synthetic biology and bioproduction. However, the difficulty of controlling the composition of microbial consortia over time hinders their optimal use in many applications. In this thesis, I present a fully automated platform that combines real-time measurements and computer-controlled, optogenetic modulation of bacterial growth to implement precise and robust compositional control of a two-strain E. coli community. The platform succeeds in stabilizing the strain ratio of multiple, parallel co-cultures at arbitrary levels and in changing these targets over time. I begin by discussing two technical setups that I optimized for in silico control of light-responsive bacteria. In the first one, the output of mono- or co-cultured populations can be tracked over time under the microscope thanks to a segmentation and tracking pipeline based on convolutional neural networks. To control it, light inputs can be projected onto the culture with spatial and temporal precision. The second platform consists of a set of parallel turbidostat cultures that are interfaced with a flow cytometer for automated sampling, a computer for execution of control algorithms and blue-light LEDs for optogenetic stimulation of the cultures. The result is a platform that is optimally suited for fully-automated, closed-loop co- culture control experiments. Next, I describe how optogenetic control of E. coli growth rate was achieved, by making the expression of an enzyme that confers antibiotic resistance light- dependent and exposing the cells to sublethal concentrations of the antibiotic. Both light-responsive —photophilic and photophobic— strains were developed, as well as light-insensitive strains that display different, fixed growth rates. Co-culturing these different strains results in communities whose compositions can be influenced by external illumination conditions. Additionally, I introduce a general framework for dynamic modeling of synthetic genetic circuits in the physiological context of E. coli. The framework takes into account the bidirectional interactions between synthetic genetic circuit and host organism through the consumption of shared pools of gene-expression resources. It also incorporates how the host regulates the availability of these resource pools as a result of metabolic adaptation to changes in external and internal conditions. Notably, I formulate the framework in such a way that it can be readily applied to any conventional ODE model of a genetic circuit. Using the host-aware modeling framework, I was able to recapitulate the dynamics of the previously-developed, light-responsive cells, both in terms of their growth rate and protein levels. This was used to simulate the behavior of a co-culture between light-responsive and light-insensitive strains subjected to closed-loop, in silico control and, through those simulations, determine the optimal control parameters of a closed-loop compositional control system. Finally, I discuss how such a closed-loop system was implemented experimentally. Computational predictions match the observed behavior of the closed-loop co-culture with remarkable accuracy, highlighting the precision of the developed host-aware model. Moreover, the strain ratio in the closed-loop E. coli co-cultures was stabilized at arbitrary levels for over 80 bacterial generations in fully-automated experiments. I also demonstrate that the strain ratio can be forced to precisely track dynamic references consisting of sudden setpoint changes during a time-course. The strategy for dynamic control over the composition of a bacterial community that I present here opens the door for the implementation of time-varying compositional programs in future applications that rely on synthetic microbial consortia