Upgrading a cyanobacterial genome-scale model by inclusion of enzymatic constraints

Over the past decade, several genome-scale metabolic models have been constructed for the model cyanobacteria Synechocystis sp. PCC 6803. These allow us to investigate how metabolism operates at the system level, opening the door for prediction of reaction flux distributions, integration of omics data and exploration of strain designs for production of bio-chemicals. Recently, platforms that facilitate the integration of kinetic data into stoichiometric models have emerged, allowing the construction of enzyme-constrained models (ecModels). Despite the body of work in modelling Synechocystis sp. PCC 6803, the community still lacks an ecModel. To fill this gap, we construct the first ecModel of Synechocystis sp. PCC 6803 and undertake several steps to calibrate, verify and analyse predictions from the model. First, we aligned the simulated and experimental growth rates by calibrating kcat values of reactions associated with glucose uptake and photosynthesis. We used Flux Balance Analysis and Flux Variability Analysis to show a higher degree of accuracy in predicted flux distributions across autotrophic, heterotrophic and mixotrophic growth modes. We documented a 19,985, 32,772 and 340,056–fold decrease in the total flux range for these growth modes, with bounds shrinking around experimentally validated values, elucidating a subset of optimal flux distributions. We then compared lethal gene knockout predictions with an experimental data set and found the model 70.91 % accurate. To analyse the model, we focus on the differences between the stoichiometric model and ecModel, photorespiration, and bio-chemical production. We found that although photorespiration can be part of optimal solutions in some light and carbon conditions, it remained unfavourable for serine production in the context of enzymatic constraints. We compared autotrophic flux distributions between the ecModel and stoichiometric models, and found a minimal set of discrepancies. We demonstrated that genetic knockouts in the ecModel have quantitative differences in growth rate compared to the base stoichiometric model. Finally, the ecModel elucidated a new hypothesis to explain increased carbon productivity in Synechocystis sp. PCC 6803 strains that produce bio-chemicals.