Sébastien Rodrigue - Metabolic modeling and transcriptomic profiling dictate targeted improvement of a genome-reduced Escherichia coli

Genome-reduced cells have a lower number of genes that makes them especially interesting for systems and synthetic biology. Yet, genome reduction entails intense genomic modifications that often result in growth defects or other undesirable phenotypes. Using a suite of tools including adaptive evolution, metabolic modeling, and machine learning, we were able to diagnose and fix a constitutive oxidative stress response occurring in Escherichia coli DGF-298. This strain derives from E. coli K-12 W3110S and only contains 2836 genes with a genome size of 2.99Mbp corresponding to a 36% size reduction. After evolving DGF-298 in the laboratory, we obtained a clone that grows faster than the original DGF-298 and at higher densities than the parental W3110S strain in minimal media. A panel of transcriptomic data for DGF-298 grown under 12 different conditions was generated and analyzed with Independent Component Analysis (ICA), which revealed an overall conservation of the transcriptome profile as compared to the parent W3110S. However, a constitutive activation of the SoxRS regulon was also identified, suggesting constant oxidative stress in the DGF-298 strain. To provide a mechanistic understanding of this activation, a strain-specific genome-scale model (GEM) for DGF-298 was produced by removing the deleted genes in the iML1515 model of E. coli K-12 MG1655. Model-driven analysis revealed an accumulation of glycolaldehyde derived from folate production, a metabolite linked to the activation of SoxRS in E. coli. This systems-level data-driven diagnostic approach could be key for successfully exploiting minimal cells in metabolic engineering.

co-authors: Antoine CHAMPIE, Jean-Christophe LACHANCE, Anand SASTRY, Frédéric GRENIER, Pierre-Étienne JACQUES, Bernhard O. PALSSON