Flux-based classification of reactions reveals a functional bow-tie organization of complex metabolic networks

Abstract

Unraveling the structure of complex biological networks and relating it to their functional role is an important task in systems biology. Here we attempt to characterize the functional organization of the large-scale metabolic networks of three microorganisms. We apply flux balance analysis to study the optimal growth states of these organisms in different environments. By investigating the differential usage of reactions across flux patterns for different environments, we observe a striking bimodal distribution in the activity of reactions. Motivated by this, we propose a simple algorithm to decompose the metabolic network into three subnetworks. It turns out that our reaction classifier, which is blind to the biochemical role of pathways, leads to three functionally relevant subnetworks that correspond to input, output, and intermediate parts of the metabolic network with distinct structural characteristics. Our decomposition method unveils a functional bow-tie organization of metabolic networks that is different from the bow-tie structure determined by graph-theoretic methods that do not incorporate functionality.

Structure and dynamics of core/periphery networks

“…Cores extend to a larger segment of the small, specialized metabolism of organisms living in constant environmental conditions (such as symbiont bacteria), while a smaller ratio of core reactions represents organisms experiencing a large variability of environmental changes requiring a large variability of "fan-in" reactions (such as free living bacteria). Enzymes catalyzing the core reactions had extended mRNA half-lives, and had a considerably higher evolutionary conservation -in agreement with similar observations in interactomes [130,[138][139][140]. Bow-tie structures reveal vulnerable connections, especially in their areas connecting the core and the fan-in/fan-out components, which may be used for drug targeting [130,131,139].…”

Drivers of structural features in gene regulatory networks: From biophysical constraints to biological function

“…Can phenotypic constraints also shape their structural features? The answer is again "yes", as has been shown in both metabolic [113,200,201] and signaling networks [202]. In both of these types of networks, the large scale structure tends to be organized in terms of a core part and two peripheral parts, one feeding into the core and another feeding out of it.…”

FMO rewires metabolism to promote longevity through tryptophan and one carbon metabolism

“…To help determine the biological relevance of the changes we observed in the OCM network following fmo-2 expression, we applied a computational model (Supplementary Data reported to be important for influencing the aging process 11,16,17 and are potential key targets for the fmo-2-mediated longevity response. The stoichiometric coefficients for the reaction and transport processes in this system are stored in the matrix S (Supplementary Data 8), where under steady-state conditions S*J = 0, where J is the vector of fluxes 34,35 . The entries in the vector J and matrix S are defined in Figure 4A.…”