The simulations revealed that networks optimized for growth rates acquired hubs. Some scientists have theorized that hubs evolved in part to improve a network's ability to withstand perturbations (i.e., the mutational loss of a metabolite). The new simulations suggest, however, that growth rate alone exerts enough selective pressure. Hub frequency varied somewhat from that found in the entire E. coli metabolic network. But if the authors compared their simulations to smaller natural subnetworks, such as glycolysis, then the hub distributions matched closely.
Natural hubs are especially important in the transfer of biochemical groups (e.g., phosphate or amino acids) between metabolites. According to the new findings, these group transfer reactions are important for the emergence of hubs. When the simulations were run in their absence—groups were added or taken away, but not transferred—hubs were much less prominent.
“Metabolites are hubs because they are key in transfer reactions,” says Pfeiffer. “One metabolite evolves such that it's the best donor for a particular group. The rest of the network benefits by maintaining that donor at a high concentration. And all reactions that require that specific group should specialize to rely on this donor, because it's the best donor around.”