(547e) Using Hierarchical Modularity to Identify Substrate Cycles in Metabolic Networks

An important motif in metabolic networks is the substrate cycle, which describes a set of reactions that serve to replenish any metabolite participating in the cycle without a net catabolic or anabolic function. These cycles would be thermodynamically infeasible without the net consumption of an energy cofactor or the net transfer of ions across membranes. The most commonly known substrate cycles involve reactions from glycolysis and gluconeogenesis that are simultaneously active: inter-conversion of glucose and glucose 6-phosphate; fructose 6-phosphate and fructose 1,6-bisphosphate; and phosphoenolpyruvate and pyruvate. Once thought to be futile, and hence the synonymous terminology in literature, these cycles have more recently been ascribed roles in cellular functions such as thermogenesis and metabolic control. In this regard, substrate cycles in adipose tissue, namely the esterification and hydrolysis of triglycerides, have been investigated as bioenergetics targets for the treatment of obesity [1]. Differential expression of substrate cycle enzymes has also been explored as a possible approach to alter cancer cell metabolism [2].

With the advent of large-scale model reconstructions of metabolic networks, computational tools can lead to the systematic discovery of novel substrate cycles previously unexplored. Gebauer and coworkers recently pointed that substrate cycles and cyclical elementary flux modes (EFMs) are equivalent, and showed that the latter can be identified by applying existing EFM enumeration algorithms, provided that the model network has been appropriately reduced to remove exchange reactions and associated metabolites [3]. However, complete and exhaustive EFM enumeration on large scale networks is not yet computationally feasible, and the prior work limited the EFM search to only those substrate cycles involving ATP consumption. Indeed, many substrate cycles previously described in literature also involve NADH, NADPH or ion transfer across membranes.

We propose in this study a more targeted approach for the computational identification of potential substrate cycles in the context of the hierarchical modularity of the metabolic network. We apply ShReD-based partitioning [4] (our previously published network partition algorithm based on Shortest Retroactive Distances) on a large scale liver network (hepatonet1) [5], and conduct an exhaustive EFM analysis on individual modules rather than the whole network using EFMtool [6]. Since ShReD-based modules are identified based on cyclical interactions, they are in principle conducive to identifying substrate cycles. The EFM search is exhaustive for all but a small number of modules at the top of the hierarchy, which contained too many reactions for EFMtool to complete the enumeration within a reasonable time. In this manner, the discovered cyclical EFMs, which are possible substrate cycles, are placed in the context of the overall functional organization of the network. We identify cyclical EFMs involved in transport, lipid synthesis, folate metabolism, sugar metabolism, and amino acid metabolism, with the net transformation of several different cofactors including NADH, NADPH, as well as other hub metabolites such as sulfate, phosphate, and hydronium ions. 

1.        Tseng Y-H, Cypess AM, Kahn CR (2010) Cellular bioenergetics as a target for obesity therapy. Nature reviews Drug discovery 9: 465–482. 

2.        Locasale JW, Cantley LC (2010) Altered metabolism in cancer. BMC biology 8: 88. 

3.        Gebauer J, Schuster S, De Figueiredo LF, Kaleta C (2012) Detecting and investigating substrate cycles in a genome-scale human metabolic network. The FEBS journal 279: 3192–3202. 

4.        Sridharan GV, Hassoun S, Lee K (2011) Identification of Biochemical Network Modules Based on Shortest Retroactive Distances. PLoS Computational Biology 7. 

5.        Gille C, Bölling C, Hoppe A, Bulik S, Hoffmann S, et al. (2010) HepatoNet1: a comprehensive metabolic reconstruction of the human hepatocyte for the analysis of liver physiology. Molecular systems biology 6: 411. 

6.        Terzer M, Stelling J (2008) Large-scale computation of elementary flux modes with bit pattern trees. Bioinformatics (Oxford, England) 24: 2229–2235