A Hierarchy of Modeling Frameworks for Cyanobacterial Metabolism: From Gsms to ME Models

Cyanobacteria serve as models for more complex photosynthetic organisms as well as promising platforms for bioproduction. Despite their ability to grow without complex carbon substrates, cyanobacteria have previously been limited in their applicability by their relatively slow growth rates. This has been mitigated by the recent discovery of the fast-growing strain Synechococcus UTEX 2973. To help expedite its adoption by the community we have developed stoichiometric, carbon mapping, and metabolism and expression (ME) models for Synechococcus 2973. Stoichiometric models form the foundation for the proposed hierarchy of modeling frameworks. They can deduce maximum theoretical yields for biomass formation, however growth rate prediction is unattainable as the rate of uptake of carbon substrates and cost of the catalytic machinery are fixed. By building upon this foundation, ME models can quantify growth rate as the cost of assembling/disassembling the catalytic resources is directly calculated. This provides the means of identifying the drivers that explain what confers fast growth for Synechococcus 2973 but not for Synechococcus 7942, a related organism that is 99.8% identical in its genome sequence.  In addition, by integrating atom mapping information onto GSM models, improved and expanded flux predictions based on nonstationary MFA can be obtained.

 

Stoichiometric genome-scale metabolic (GSM) models were developed for both Synechococcus 2973 (iSyu627) and Synechococcus 7942 (iSyf686). These two models were developed using a semi-automated method that combined curated gene annotations with the previously developed Synechocystis GSM model iSyn731. The ability to grow on sugar varies across cyanobacteria and was analyzed through comparisons of both the metabolism (via GSM models) and genome of organisms. It has been experimentally shown that under similar conditions, Synechococcus 2973 grows 3.7 times faster than Synechococcus 7942. The inclusion of additional constraints has reconciled this difference in growth rates.

ME models are able to simulate in more detail the metabolic burden caused by enzymatic processes. An ME model of Synechococcus 7942 has also been constructed that expands upon the GSM model to include a complete description of the catalytic machinery. ME models integrate transcriptional and proteomic data and dynamically track metabolites and macromolecular products. The growth rate is calculated as the outcome of metabolism and cellular machinery’s assembly/disassembly costs. An ME model simulation of Synechococcus 2973 was completed that minimized the differences between predicted relative gene expression levels and normalized transcriptomic data.

A genome-scale carbon mapping model in conjunction with 13C labeling data allows for more precise predictions of flux ranges using metabolic flux analysis (MFA) as compared to the ranges determined by a stoichiometric model. We found a 71% similarity between iSyn731 with the fully mapped iAF1260 E. coli model resulting in 221 unique cyanobacterial reactions. The first Genome-scale metabolic mapping (GSMM) models for cyanobacteria, imSyn711 and imSyf608 have been developed for Synechocystis and Synechococcus, respectively. These models reveal 67 novel carbon scrambling reactions unique to cyanobacteria arising from Calvin cycle, photorespiration, expanded glyoxylate metabolism, and corrinoid biosynthetic pathways. These models can be used to carry out non-stationary MFA to uncover cyanobacterial metabolism at a genome-wide level.