EColiCore2: a reference network model of the central metabolism of Escherichia coli and relationships to its genome-scale parent model

Abstract

Genome-scale metabolic modeling has become an invaluable tool to analyze properties and capabilities of metabolic networks and has been particularly successful for the model organism Escherichia coli. However, for several applications, smaller metabolic (core) models are needed. Using a recently introduced reduction algorithm and the latest E. coli genome-scale reconstruction iJO1366, we derived EColiCore2, a model of the central metabolism of E. coli. EColiCore2 is a subnetwork of iJO1366 and preserves predefined phenotypes including optimal growth on different substrates. The network comprises 486 metabolites and 499 reactions, is accessible for elementary-modes analysis and can, if required, be further compressed to a network with 82 reactions and 54 metabolites having an identical solution space as EColiCore2. A systematic comparison of EColiCore2 with its genome-scale parent model iJO1366 reveals that several key properties (flux ranges, reaction essentialities, production envelopes) of the central metabolism are preserved in EColiCore2 while it neglects redundancies along biosynthetic routes. We also compare calculated metabolic engineering strategies in both models and demonstrate, as a general result, how intervention strategies found in a core model allow the identification of valid strategies in a genome-scale model. Overall, EColiCore2 holds promise to become a reference model of E. coli’s central metabolism.

When Do Two‐Stage Processes Outperform One‐Stage Processes?

“…The parameters for the OSF model are shown in Table . Some of them (e.g., ) have been reported in the reference, others such as or are either well‐known from biochemistry or have been calculated from the EColiCore2 model . For the TSF model we assume that the growth phase of KBM10111 takes place under aerobic conditions without any lactate production.…”

“…), and, moreover, that, under growth‐coupled product synthesis, the biomass yield and the growth‐coupled product yield cannot be varied independently of each other. We therefore used a stoichiometric model of the central metabolism of E. coli to calculate the yield space diagram for lactate and biomass synthesis (Figure S2, Supporting Information). This diagram can be used to determine the maximal (true) lactate yield feasible under a given biomass yield.…”

“…The low specific substrate conversion rate of non‐growing cells during the production phase is the key reason for this unfavorable performance. Stoichiometric calculations with the EColiCore2 model indicate that the minimum glucose uptake rate of non‐growing E. coli cells with an ATP demand for NGAM of up to 8.39 mmol/gDW/h (used in Ref …”

“…For the parameter values, it is indicated whether they have been measured (m) in Ref., are known (k), or have been calculated (c) from a stoichiometric model of E. coli 's central carbon metabolism. In some cases (m+c), measured fluxes were used together with the stoichiometric model to calculate unknown fluxes (e.g., NGAM ATP demand).…”

“…Some of them (e.g., r S;max ) have been reported in the reference, [30] others such as Y true ATP=S or Y true X=S are either well-known from biochemistry or have been calculated from the EColiCore2 model. [32] For the TSF model we assume that the growth phase of KBM10111 takes place under aerobic conditions without any lactate production. Again, the TSF parameters (Table 2) are either well-known (including the maximum substrate uptake rate of glucose, r S;max , which is usually set to 10 mmol/gDW/h under aerobic conditions) or have been calculated from the EColiCore2 model.…”