2.2. Modification of genome-scale metabolic models
One of the important effects of the variation in the intracellular pH is the change of the ionization state of metabolites and metabolites contain certain particular functional groups that can lose or obtain protons at various pH levels. To modify the metabolic networks at different pH values, Marvin view software 19.2, 2019, developed by ChemAxon (http://www.chemaxon.com) was used to estimate the charge of metabolites based on their pKa for intended pH value. The mol files of each metabolite, which are generally classified as data files in plain text format containing molecular details, atom, bonds, co-ordinates, and communication detail, were obtained from Chebi (https://www.ebi.ac.uk/chebi) and PubChem (https://pubchem.ncbi.nlm.nih.gov/) databases.
The change of metabolites charge can result in the disruption of the charge balance of reactions and so, the charge balance of the reactions was automatically checked at different pH levels according to the method presented in Figure 1. Initially, the proton was removed from each reaction in the case of presence. Then, charges of the metabolites at the right and left side of each reaction were separately summed, and the total charge of the right side was subtracted from the left side. If it was positive (negative), the proton with the coefficient of the subtracted value was added to the left (right) side of the reaction. The reaction was left unchanged if the subtracted value was zero. The charge balance of transport reactions was not changed because the proton is mechanistically added to the reactions (e.g., simple transport). The charge of the exchange reactions was not changed because they are pseudo reactions that exchange metabolites between the cell and the environment. The accuracy of the method was evaluated by comparing the modified model with the original model at pH=7, and the charge balance of some reactions was not changed due to the fact that these reactions exchange protons between subsystems to transfer materials (e.g., mitochondria to cytosol). S. cerevisiae, Z. mobilis, and E. coli contained 8, 2, and 6 kinds of these reactions, respectively. These reactions were balanced separately, which are shown in Table S1 and S2.
By changing metabolite charges at different pH levels, the charge balance of each reaction would be changed, which affects in proton exchange rates. To determine the sensitivity of growth to the proton exchange rate at different pH levels, developed models could be analyzed with robustness analysis which can assess the optimal value of the objective function by varying the flux of a reaction and indicating the sensitivity of the objective function to changes in a particular reaction. In this regard, the sensitivity of cell growth to proton exchange has been assessed via robustness analysis. Mat files of modified metabolic models are also provided in Supplementary File 3.