21ENGBIO Reprogramming bacterial cells using whole-cell models

The recent availability of increasingly precise genome-editing technologies provides synthetic biologists with the ability to modify cells and organisms at the genome-scale level. In the near future, it will be possible to (re-)engineer entire genomes at will, for example, to build minimal cells which can produce useful compounds (e.g. insulin or biofuels).

To guide this process, synthetic biologists usually make use of mathematical models to predict the outcomes of genome engineering and, ultimately, save time and effort of experimental testing. To date, the lack of mathematical models which can represent precisely all functions within a cell is a major bottleneck in developing robust cycles of iterations between design and testing (the so-called "design-build-test-learn cycles") for reengineering living organisms.

Our vision is that whole-cell models can revolutionise design-build-test-learn cycles. Whole-cell models are state-of-the-art computational representations of cells, which account for the function of every gene product and dynamics of every molecule over the entire cell lifecycle. Whole-cell models for two bacterial cells (Mycoplasma genitalium and Escherichia coli) have been developed very recently by our international partner in the US, but have never been used, to date, to guide experiments in synthetic biology.

This programme will integrate, for the first time, whole-cell model predictions and genome-editing technologies within design-build-test-learn cycles to rationally redesign E. coli bacterial cells for the production of useful compounds.

We set up an interdisciplinary team, composed of UK experts in systems and synthetic biology and encompassing the world experts in whole-cell model development and analysis. We are excited to pioneer this adventurous and promising new research programme at the interface of genome engineering, systems and synthetic biology and modern computer science.

Outcomes of this research could transform current synthetic biology approaches by significantly reducing experimental costs, accelerating discovery through automation and promoting whole-cell model impact on broad academic and industrial communities.
Longer-term, as whole-cell models become available for higher species (e.g. human cells), our approach could be relevant for the rational design of patient-tailored treatments.

We will integrate for the first time whole-cell models (WCMs, the most sophisticated mathematical representation of cells) into engineering biology design-build-test-learn cycles, using metabolic engineering as an exemplar. Namely, we will use WCM-based predictions to reengineer E. coli for optimised production of two deoxynucleosides relevant for biomanufacturing: thymidine and deoxycytidine.

Technical aims include:
1) in silico designs of strains for increased thymidine or deoxycytidine yields using gene knockouts. We will attempt two strategies to design genetic perturbations, based on: i) a bilevel optimization algorithm which uses linear programming for the set of deletion targets that maximizes both the production of the compound of interest and cell growth; ii) an evolutionary algorithm that aims at knocking out genes to maximise production of the selected metabolite;
2) strain engineering, using the CRISPR-Cas9 system to perform gene knockout;
3) engineered strain testing, using high-throughput and liquid handling robotics-based methodologies available at Bristol, which enable parallel screening of multiple mutant libraries;
4) benchmarking of model simulations vs experimental data to plan further refinements to the model/strain design algorithm.