Metabolism is precisely coordinated around the goal of balancing fluxes to maintain robust growth. Hierarchical regulatory mechanisms for controlling metabolic fluxes, from transcriptional regulation of genes to direct modulation of enzyme activity by small-molecule regulation (SMR) have evolved to achieve this goal. While SMR is ubiquitous in metabolism, the organizational principles of the networks formed by SMR interactions and their generalized role in metabolic coordination is not well understood. In particular, questions such as which metabolites are optimal regulators, which reactions are optimal targets of regulation, and what information is communicated via this form of regulation remain open. Answering these questions is essential to uncovering the fundamental constraints that have shaped metabolic evolution, and a systematic understanding of such constraints is required for forward engineering of metabolic systems.

In this work, a constraint-and structure-based approach is taken to addressing these questions. First, multi-omics data from Escherichia coli and Saccharomyces cerevisiae are used to show that flux sensing arises due to thermodynamic constraints on consuming reactions, so some metabolites always carry information about upstream rates. These data suggest that a standard Gibbs free energy of at least -4 kJ/mol for any consuming reaction is sufficient to make substrate flux sensitivity likely. Since flux-sensitive metabolites carry information about fluxes, they are optimal controllers of the fate of upstream flux via feedforward regulation. Next, a reconstruction of the genome-scale E. coli SMR network (SMRN) is used to show that higher-order regulatory structures, namely incoherent feedforward loops, are conserved. These loops are structured and tuned to use flux-sensitive metabolites to distinguish between flux distributions and provide differential control of supply and demand fluxes in different metabolic states. Finally, the SMRN is re-cast as a thermodynamic network to show that indeed regulators are most often substrates of thermodynamically constrained reactions. Mechanistically, this is because they maximize information transfer in the conditions in which information would normally be lost. Maximization of information transfer is thus hypothesized as an underlying design principle of SMR.