Strain Design with Constrained Regulatory Minimal Cut Sets in Genome-Scale Metabolic Networks

A variety of important tools for computational strain design in metabolic engineering are based on stoichiometric and constraint-based methods. Among those is the concept of constrained minimal cut sets (cMCSs) which, like many other techniques, applies reaction (or gene) knock-outs to suppress an undesired behavior (e.g. production with low yield) while preserving some desired functionality (e.g. the ability to grow).
The advantage of using knock-outs is that they can be rather easily be implemented in an organism. However, with more intricate genetic engineering techniques like influencing promoter strength it has become also possible to modulate the activity of a given reaction in a more fine-graded manner. To take these possibilities into account the cMCS approach is here extended to constrained regulatory MCSs (cRegMCSs). With cRegCMSs the up/downregulation of reaction rates can be combined with reaction deletions. To achieve this, the metabolic network is augmented with additional pseudo-metabolites and slack reactions as well as constraints for desired and undesired behaviors. Deletion of certain slack reactions leads to the activation of upper/lower limits on other reaction rates thereby emulating a down/upregulation.
In effect, the cMCSs of the augmented network correspond to the cRegMCSs of the base network. Consequently, the search space for cMCSs in the augmented network can be much larger than in the base network. Therefore, a strategy is developed with which it is possible to optionally preselect reactions that may be well-suited for flux regulation. Also, novel algorithmic techniques were developed to improve the efficiency of MCS calculations.
The cRegMCSs approach is illustrated by a simple example network and then applied to the identification of strain designs for ethanol production in a genome-scale metabolic model of Escherichia coli. As a result it is shown that cRegMCSs provide a much larger number of suitable strain designs than cMCSs alone. Furthermore, the number of necessary modifications in a strain design is significantly smaller relative to cMCSs because in addition to reaction deletion a flux up/downregulation is now possible as well. Also, the use of cRegMCSs makes it possible to to fine-tune the desired behaviour in a narrower range.
In this manner, the new cRegMCSs approach may accelerate the implementation of model-based strain designs for the bio-based production of fuels and chemicals by increasing the choices available to metabolic engineers for experimental implementation. In addition, cRegMCSs could also be used for preselecting promising intervention strategies, which can then be analyzed in more detail in combination with kinetic modelling frameworks.