Metabolic Capabilities of Escherichia coli II. Optimal Growth Patterns

A modeling formalism has been developed recently, based on stoichiometry and linear optimization, that enables determination of the capabilities of metabolic networks. Both single and multiple simultaneous metabolic demands can be studied within this formalism. Here, we use this approach to determine the ability of Escherichia coli's fueling reactions to meet growth demands. "Growth" within this context is defined as a drain of biosynthetic precursors in an appropriate ratio required to produce cellular components. Maximizing biomass yield on glucose in the absence of maintenance requirements, we obtain a yield of 0·588 g DW g-1 glucose, which is higher than experimentally reported values. Inclusion of ATP maintenance costs significantly influences the maximal biomass yield. A maintenance cost of 4-6 ATP molecules per glucose consumed gives maximal biomass yields of 0·4-0·45 g DW g-1 glucose, which correspond to experimental observations of biomass yields. The corresponding metabolic flux distribution shows a remarkable consistency with experimental observations of pathway utilization, including the fraction of glucose metabolized through the pentose pathway, use of the anaplerotic reactions and redox metabolism. A sensitivity analysis demonstrates a robustness of the optimal biomass yield and pathway utilization to changes in metabolic demands. This result is important considering the diversity of metabolic demands that the cell can experience under different growth and environmental conditions. The optimal biomass yield varies significantly, approximately 20%, over the range of P/O ratios found in energy-transducing membranes. A new metric that is based on shadow prices is formulated to measure the relative importance of the biosynthetic precursors to biomass generation. Interestingly, it identifies the sugar monophosphates as the best utilized biosynthetic precursors for growth in the absence of maintenance energy, although they are needed in relatively small amounts. With maintenance costs, ATP assumes a similar importance. With acetate as the sole substrate we show numerical comparisons of the optimal flux distribution with experimental determinations reported in the literature. We find the optimal solution to be in close agreement with experimentally determined pathway utilizations, the measured metabolic flux levels and biomass yields. Taken together, the results presented here argue for the hypothesis that observed metabolic behavior is consistent with stoichiometric optimality of growth and that regulation of metabolism strives to achieve this stoichiometric optimality.