Upgrading the Genome-Scale Model of the Methylotrophic Yeast Pichia Pastoris

The methylotrophic yeast Pichia pastoris become one of the most widely used platforms for recombinant protein production. It also offers a potential as a chassis platform for small molecules production. Its cultivation and physiological characteristics together with a range of well stablished methods to be genetically engineered, make this yeast highly suitable for industrial purposes.

The potential of P. pastoris has boosted research towards a better understanding of its physiology and the development of new tools for systems-level strain improvement (systems metabolic engineering). Along the last 15 years, an intensive effort at studying P. pastoris has resulted in the generation of new information on its physiological behavior under different cultivation conditions as well as the development of new synthetic biology and systems biology tools. In this context, the first genome sequence was published in 2009, and since then, two additional sequences have been published. The availability of these genome sequences has allowed the development of three genome-scale metabolic models.

These three models present certain divergences in some pathways as well as in their global structure. Furthermore, the recently published literature for P. pastoris allows for a further extension of the model. Therefore, we have developed an updated model with improvements in several areas. In short, the procedure included an automatic merging from the three models to obtain an initial paired reaction list. The remaining unpaired reactions were manually revised and paired when possible. Finally, recently characterized or better described pathways such as, glycosylation, sphingolipid biosynthesis and oxidative phosphorylation, among others were added to the model or used to improve the existing pathways. All this resulted in a higher number of gene-protein-reaction relationships as well as total number of reactions.

In addition the model was adapted to the available experimental data for testing. Thus, our consensus model incorporates more detailed and condition-dependent biomass specific composition. The resulting model was validated in several conditions using glucose and different oxygen conditions as well as a range of different combinations of glycerol and methanol. Moreover, the model was also tested against results obtained using recombinant protein expressing strains showing the adequacy of the model to the foreseen tasks with predictions within the precision of the available experimental data.