(509f) Development of a Kinetic Model of Lipid Metabolism in Saccharomyces Cerevisiae

Studies on metabolic engineering of lipids have transitioned from push-pull-block approaches, where the lipid network can be treated in a black box fashion, to a more targeted approach, where the interactions within lipid metabolism have been probed, and the effect of network perturbations on overproduction phenotypes assessed. Owing to the tightly constrained and interconnected nature of reactions associated with lipid metabolism, assimilation of the results of such studies into a single compendium has proved to be challenging. To address this limitation, here we report the development of a kinetic metabolic model of lipid metabolism in S. cerevisiae to assimilate the important modules of lipid metabolism that are either known or suspected to play a role in metabolic engineering efforts of lipid-based biochemicals. The components of the model include fatty acid biosynthesis, glycerophospholipid metabolism, sphingolipid metabolism, storage lipids, lumped sterol synthesis, and the synthesis and transport of relevant target-chemicals, such as fatty acids and fatty alcohols. The model was initially trained on timeseries data of a reference yeast strain, which comprised of the complete lipidomic profile (membrane and storage lipids), as well as secreted free fatty acids (FFAs) or fatty alcohols in the growth media.

In order to assimilate more industrially relevant phenotypes and their associated data into the model, we engineered 3 mutants that have been commonly constructed in literature – i) a FFA overproducer with FFA degradation and activation pathways shut off, ii) an intermediate mutant with overexpression in fatty acid biosynthesis flux, and iii) the above overexpression mutant with a heterologous fatty acyl-CoA reductase (FAR) gene that produced fatty alcohols. Timeseries data from these 3 mutants along with the reference strain enabled the construction of a lipid kinetic model that captured the relevant flux distribution through lipid pathways in overproduction mutants. Finally, to probe the lipid network’s response to perturbations, 13 non-lethal single gene deletions in the glycerophospholipid and sphingolipid submodules were constructed in the reference and 3 overproducing mutants, and the lipidomic profiles from these strains were used to train the model thus capturing the lipid system’s response to overproduction-relevant mutations.

We performed a priori structural identifiability analysis on the model and utilized it to systematically reduce pathways and species within the model. Practical identifiability analysis performed on the model along with the generated lipidomics datasets was used to design further rounds of experiments and also shed light on the blind spots in the model’s prediction space. Stability analysis was finally performed on the dynamical system to shed light on the stability of the parameterized mathematical model. Till date, our work represents the richest dataset of lipidomics profiles used to train a kinetic model and has resulted in a model of lipid metabolism that captures engineering-relevant phenotypes. The model is able to successfully emulate the lipid profiles of overproducing mutants reported in earlier studies while providing an explanation for the failure of certain engineering strategies towards overproduction and, to our knowledge, is the first such model to do so. We show how the model can be used to glean insight into the effect of lipid network modifications on overproduction phenotypes and design metabolic engineering strategies for future studies, thus filling a vital gap in the field of lipid metabolic engineering.