Computational protocol: Model-driven intracellular redox status modulation for increasing isobutanol production in Escherichia coli

Similar protocols

Protocol publication

[…] For systematically investigating redox cofactors (NADH and NADPH) metabolism, the latest GSMM iJO1366 of E. coli MG1655 [] was used as the base model in this study. The transhydrogenase reaction encoded by pntAB (THD2pp) was modified according to the previous report [], and details are listed in Additional file : Table S2. Then, two metabolic reactions involved in Ehrlich pathway from KIV to isobutanol and three isobutanol transport reactions from intracellular to extracellular were also added into the model; the added reactions are given in Additional file : Table S3. The final GSMM proved to be reasonable by the verification, and could be applied for the metabolic simulation of the strain E. coli LA02. The details of model verification were given in Additional file .In order to reduce computational effort, the sets of candidate oxidoreductase reactions available for modification should be determined according to previous reports []. In the GSMM, all reactions, NAD(H) or NADP(H) that participated were located. Then, the reactions carrying no flux, essential reactions, orphan and spontaneous reactions were removed; the reactions involved in the subsystems (cell envelope biosynthesis, glycerophospholipid metabolism, inorganic ion transport and metabolism, ipopolysaccharide biosynthesis and recycling, membrane lipid metabolism, murine biosynthesis and recycling, tRNA charging, and inner/out membranes transports) were excluded. Finally, the remaining sets of reaction were used as candidates for target prediction of rebalancing redox status (the candidate reactions are listed in the Additional file : Table S4). The targets prediction was performed using the COBRA Toolbox v2.0 in MATLAB (The MathWorks, Inc., USA) version 8.1 with Gurobi version 5.6.0 (Gurobi Optimization, Inc., USA) according to Ref. [, ] and details are as follows.The initial fluxes of candidate reactions were first obtained using FBA algorithm with maximization of specific growth rate as the objective function and the experimental constraint (see Additional file ). Secondly, for the knockout simulation, the flux of candidate reaction was set to be zero; for the overexpression simulation, the flux of candidate reaction was amplified to N-fold of the initial flux (N = 1.1, 1.2, 1.3… and 2.0); for cofactor swap simulation, non-native oxidoreductase reaction, obtained from the candidate reaction by cofactor swapping, are added to the system. The flux of native candidate reaction was set to be zero and the flux of the added reaction was set to be M-fold of the initial flux of native reaction (M = 0.1, 0.2, 0.3… and 2.0). The setting values of N and M were derived from the modeling methods reported by Huang et al. [].Finally, the quadratic programming problem was solved by MOMA, which was performed by searching for the minimal Euclidean distance with the same metabolic model. The targets were identified through comparing the phenotypic fraction value, fPH (the ratio of weighted and dimensionless specific growth rate and specific isobutanol production rate) [, , ].fPH=fbiomass×fisobutanol2=vbiomass, modificationvbiomass,initial×visobutanol, modificationvisobutanol,initial2The reaction that had the higher fPH was considered as the better candidate to rebalance redox state for enhancing isobutanol production. […]

Pipeline specifications

Software tools COBRA Toolbox, MOMA
Application Metabolic engineering
Organisms Escherichia coli
Chemicals Ethanol, Glyceraldehyde 3-Phosphate, NADP, Lactic Acid