Flux cost functions and optimal metabolic states

Abstract: The metabolic fluxes in cells follow physical, biochemical, and economic principles. Some flux balance analysis (FBA) methods trade flux benefit against flux cost. However, if flux cost functions are linear and meant to describe underlying enzyme costs, this entails that enzyme efficiencies are constant and ignores the interplay between fluxes, metabolite concentrations and enzyme levels in cells. Here I introduce realistic flux cost functions that describe an "overhead cost", namely the minimum enzyme and metabolite cost associated with the fluxes in a kinetic model. These flux cost functions have general mathematical properties. Enzymatic flux cost functions, which represent enzyme costs, scale proportionally with the flux profile and are concave on the flux polytope. Kinetic flux cost functions represent the sum of enzyme and metabolite costs. If two flux profiles are superimposed, their different demands for metabolite concentrations cause an extra compromise cost, which makes flux cost functions strictly concave in almost all cases. When fluxes change their direction, the enzymatic cost jumps abruptly. Here I propose two methods for flux modelling: Flux Cost Minimisation, a nonlinear variant of FBA with flux minimisation, and Flux Benefit Optimisation, a nonlinear variant of FBA with molecular crowding. The optimal flux profiles, at a given flux benefit, are vertices of the flux polytope. Linear approximations of enzymatic flux cost can be used in FBA. In contrast to flux costs chosen ad hoc, these functions reflect the enzyme kinetics and extracellular concentrations in realistic kinetic models. Based on enzymatic flux costs, we can describe the cell growth rate as a convex function on the flux polytope and derive growth-optimal metabolic states and statistical distributions for the fluxes in cell populations.