Quantitative Design of Perceptive Microbial Communities for Sensing and Biomanufacturing

Abstract

The field of synthetic biology aims to tackle global challenges in medicine and environmental protection by engineering biological components to perform user-defined functions, and thus, create systems that accomplish complex tasks such as signal processing and biomanufacturing. These gene circuit designs have been implemented in cells across organisms, particularly in bacteria due to their simplicity and resilience in diverse environments. Tremendous progress has been made in creating well-characterized microbial circuits to detect and treat diseases, monitor water contamination, and produce therapeutics and consumer goods. Despite this progress, a major challenge in engineering biology is its complex, promiscuous, and evolving nature, which inhibits the ability to create systems that operate orthogonally. Typically, the goal is to isolate components to perform functions that are independent from each other and from native host cell processes, such that precise control of circuit output is maintained. However, unintended interactions with the host, between circuit components, and with other stochastic environmental factors are inevitable. The lack of true orthogonality is exacerbated as we increase the complexity of the tasks we aim to accomplish. Increasing gene circuit size and the number of interactions makes it difficult to maintain orthogonality, contain the circuit within a single population, and predict the resulting behavior. These challenges are particularly evident in multiplexed signal processing, where cells are designed to detect and report multiple signals simultaneously. Thus, the goal of my research was to develop a generalizable approach that enables complex signal processing and biomanufacturing. I combined the use of two promising methods for overcoming the challenges related to increasing circuit complexity. The first is distributing gene circuits across microbial consortia, rather than containing them in a single population, to compartmentalize and insulate circuit components and reduce the growth burden placed on each population. The second is developing computational models to aid in the design and prediction of circuit behavior and enhance our interpretation of circuit outputs to derive meaningful results. I combined microbial community engineering with computational modeling strategies to develop a robust platform for multiplexed signal processing in microbes and to create a responsive engineered living material for tunable protein release. The first aim of my dissertation was to develop a computational platform that could resolve input concentrations from the complex outputs of multi-sensing microbial consortia. Distributing sensor circuits across a microbial community simplified experimental optimization, enabling modular addition of different sensor strains. I developed a computational pipeline composed of a mechanistic model to augment data in a realistic manner and a machine learning model to predict multiple inputs from the microbial community’s response. This pipeline accurately estimated analyte concentrations in 2- and 3-sensor communities with high crosstalk, communities with non-engineered responses to antibiotics, and a 2-sensor community detecting contaminants in hospital sink water. This platform has the potential to enable multiplexed biological signal processing systems to be more widely used in practical applications. To create an engineered living material with controlled protein release, I expanded on a method of encapsulating microbial communities in hydrogel capsules by engineering the community to produce a stimulus-responsive secondary network structure. One population expresses a protein that forms the network within the capsules to trap and release proteins expressed by other populations in the capsules. The protein network formation, capsule material, and cell growth dynamics are responsive to environmental conditions, enabling us to tune protein release with pH, temperature, and glucose. I utilized a mechanistic modeling approach to characterize the relationship between the environmental control knobs and the components of the microbial community to estimate parameters and predict system behavior under new conditions. These two objectives highlight the power of using computational approaches to enhance biological circuit function and enable predictable behavior in microbial community engineering.