Exploiting contextual effects on synthetic biological constructs for the design of genetic and electrogenetic circuits

Abstract

Synthetic biologists build novel biological networks to perform predefined functions. The majority of work in this field has concentrated on engineering synthetic gene networks in bacteria, which coopt the host’s genetic regulatory machinery for applications including biosensing, bioremediation and biocomputing. The engineering of these synthetic gene networks is an expensive and time-consuming process that involves multiple iterations of a design-build-test-learn (DBTL) cycle. Often the results are impressive, and the synthetic gene network performs very well, but only in the environment in which they were engineered originally. The core problem is that the performance of a synthetic gene network relies fundamentally on factors external to the network itself — the context that the network is placed in. This includes for example the activity of the host’s transcriptional machinery, or the genetic material close to the network, or the specific growth rate of the host. Here I use a case study of a well-known synthetic gene network library to demonstrate that qualitative differences in network performance arise from changes in context and degrade the quality of the library. However I also show advantages to considering the context as a parameter of the library, and a cross-context library is of better quality than the original according to the metrics developed here. At the population level it is often space that dominates context and spatial gradients may exist in any number of important factors. Bacteria such as Escherichia coli have developed chemotaxis systems to provide sensing of spatial gradients and motility to “solve” this problem for individuals. But it is not always desirable or possible for individuals to simply relocate. Natural systems that operate at the level of a populations often operate with space as a parameter, where individuals specialise and differentiate to contribute effectively to the overall function of the system. In this work I take one such system, the electroactive biofilm, and show how two classical synthetic gene networks, the genetic toggle switch and the repressilator, can be designed not only to deal with spatial gradients, but also to fundamentally rely on them. The mathematical models developed are implemented in software and are modular, in the sense that other synthetic gene networks could easily be modeled using the same software. The results provide a foundation for moving toward more sophisticated work in synthetic electrogenetic networks.