Inferring metabolic phenotypes from the exometabolome through a thermodynamic variational principle

TL;DR

This paper introduces a thermodynamic variational principle to infer intracellular metabolic fluxes from extracellular metabolite data, successfully characterizing cellular phenotype switches and aligning with observed human cell data.

Contribution

It presents a novel thermodynamic approach to deduce intracellular fluxes from extracellular measurements, applicable to large-scale human metabolic models.

Findings

The switch from fermentative to oxidative phenotypes can be characterized by metabolite concentrations.

The method accurately reproduces results from a solvable toy model.

Predicted phenotypic maps align with experimental data from human cells.

Abstract

Networks of biochemical reactions, like cellular metabolic networks, are kept in non-equilibrium steady states by the exchange fluxes connecting them to the environment. In most cases, feasible flux configurations can be derived from minimal mass-balance assumptions upon prescribing in- and out-take fluxes. Here we consider the problem of inferring intracellular flux patterns from extracellular metabolite levels. Resorting to a thermodynamic out of equilibrium variational principle to describe the network at steady state, we show that the switch from fermentative to oxidative phenotypes in cells can be characterized in terms of the glucose, lactate, oxygen and carbon dioxide concentrations. Results obtained for an exactly solvable toy model are fully recovered for a large scale reconstruction of human catabolism. Finally we argue that, in spite of the many approximations involved in the theory, available data for several human cell types are well described by the predicted phenotypic map of the problem.