Fig 1.
An example illustrating that DoL does not reduce the overall metabolic burden.
(A) Basic idea of DoL: a two-step pathway converting a substrate S to a product P via the two reactions R1 (catalyzed by enzyme E1) and R2 (catalyzed by enzyme E2) in a single strain is split into two parts each being performed by one dedicated strain. (B) Assuming an identical total amount of biomass, the metabolic burden (enzyme cost per biomass) is identical for the single strain and the DoL solution.
Fig 2.
Example illustrating how division of labor may lead to a thermodynamic advantage in the production of a target metabolite.
In the left, a metabolic pathway in a cell is considered that synthesizes the target product P. The red values indicate positive values for the standard Gibbs free energy change ( [in kJ/mol]) and thus potential thermodynamic bottlenecks. With an allowed concentration range from 1 M to 10 M for all metabolites except for Pex, where a minimum concentration of 5 M was assumed to consider product synthesis under high external product concentrations, a negative optimal MDF (OptMDF) value would follow, indicating thermodynamic infeasibility of product synthesis in the single strain. In the two-strain community (right), the pathway is divided and an exchange of metabolite B introduced. With this, individual concentrations of metabolite X can be adjusted in the two strains by which thermodynamic feasibility (a positive OptMDF) of the overall transformation is achieved (the blue triangles indicate the direction of the concentrations of X (high/low) when maximizing the driving force). Black arrows in the two-strain solution indicate active and grey arrows inactive reactions.
Fig 3.
Community model structure used in this study and possible exchange directions of metabolites.
Dashed arrows indicate exchange reactions, all other arrows biochemical conversions.
Fig 4.
Pseudo-code of the ASTHERISC algorithm.
For detailed explanations see text.
Fig 5.
Schematic overview of the combined usage of CommModelPy and the ASTHERISC package.
Orange boxes stand for user settings, green boxes for generated or given data files, red boxes for primary program package dependencies, and blue boxes for the programs themselves.
Table 1.
Percentage (and absolute number) of all producible target metabolites for which ASTHERISC found a higher optimal MDF in a multi-strain community compared to a single-strain model.
Note that only those communities were considered, where the MDF was at least 0.2 kJ/mol higher than in the single-strain model.
Table 2.
Statistics about OptMDF advantages (in kJ/mol) with communities for each of the 24 scenarios.
The given numbers stand for the minimal/mean/maximal OptMDF advantage.
Table 3.
Statistics about the number of used extra exchanges in solutions with a community OptMDF advantage for each of the 24 scenarios analyzed.
The given numbers stand for the minimal/mean/maximal number of used extra exchanges.
Fig 6.
Excerpt of selected central reactions of the MDF-optimal single-species and community solution for kdo8p synthesis.
Black arrows indicate active reactions, dashed arrows indicate a sequence of active reactions, and the blue triangles indicate increased or decreased metabolite concentrations of a strain in the community relative to the other strain. The shown ΔrG′ and ΔrG′0 values have all unit kJ/mol. The ΔrG′ values are taken from the specific MDF-optimal solution delivered by ASTHERISC. The red ΔrG′ values indicate thermodynamic bottlenecks, i.e., reactions whose ΔrG′ is fixed under the optimal MDF (and corresponds to the negative value of the OptMDF) All black ΔrG′ values are variable under the given OptMDF. All reaction and metabolite identifiers are based on the definitions in the BiGG database [49].
Table 4.
Concentration ranges of critical metabolites in the ecolicore2double community solution for kdo8p synthesis with optimal MDF.
The concentration ranges do not overlap between the two strains.