(149d) Model-Guided Engineering of Cyanobacteria for Improved C4-C5 Alcohol Production

Cyanobacteria are promising hosts for fuel and chemical production as their main metabolic inputs are minimal and renewable (i.e. light, carbon dioxide, water, and a nitrogen source). However, cyanobacteria engineered to produce compounds of interest frequently suffer from low yields and/or are genetically unstable. As such, metabolically coupling chemical production to growth may be important in engineering cyanobacteria because it can generate robust production characteristics. Additionally, a growth-coupled production approach may help to circumvent gaps in our knowledge about cyanobacterial metabolic regulation, which likely factors into the poor production characteristics that are often reported relative to heterotrophic strains. In this work, a genome-scale metabolic model is being used to guide engineering of the cyanobacterium Synechococcus sp. PCC 7002 for production of short- to mid-chain alcohols (e.g. n-butanol) that are potential biofuels. PCC 7002 is of particular interest as a cyanobacterial production strain due to its relatively rapid growth-rate and tolerance to saline and high-light conditions. Metabolic engineering strategies for growth-coupling the production of various alcohols in PCC 7002 were investigated via the metabolic model iSyp708, which was previously developed by our lab. Using this approach, a potential strategy for enhanced production of n-butanol and 2-methyl-1-butanol in PCC 7002 was successfully identified. This strategy hinges on coupling the overproduction of the alcohols’ metabolic precursors to nitrate assimilation by rewiring PCC 7002’s native NADH-cycling pathways. Strains have been constructed to test this approach, with initial results indicating that alcohol production in the engineered strains is improved relative to alcohol production in a WT background. To fully capitalize upon this improvement in strain background (engineered NADH-cycling vs. WT), we are currently working to optimize the expression of the heterologous alcohol production pathway in PCC 7002. To accomplish this, we are employing an active learning/machine learning approach developed by our lab (ActiveOpt), which guides experiments to arrive at an optimal production phenotype in a small number of design-build-test cycles. This study demonstrates that genome-scale metabolic models and machine learning approaches are useful in guiding cyanobacterial metabolic engineering and helps identify genetic alterations for optimizing the production of biofuel compounds in cyanobacteria.