Effect of random and hub gene disruptions on environmental and mutational robustness in Escherichia coli

Genome-wide profiling has allowed the regulatory interaction networks of many organisms to be visualised and the pattern of connections between genes to be studied. These networks are non-random, following a power-law distribution with a small number of well-connected 'hubs' and many genes with only one or a few connections. Theoretical work predicts that power-law networks display several unique properties. One of the most biologically interesting of these is an intrinsic robustness to disturbance such that removal of a random gene will have little effect on network function. Conversely, targeted removal of a hub gene is expected to have a large effect.

We compared the response of Escherichia coli to environmental and mutational stress following disruption of random or hub genes. We found that disruption of random genes had less effect on robustness to environmental stress than did the targeted disruption of hub genes. In contrast, random disruption strains were slightly less robust to the effect of mutational stress than were hub disruption strains. When we compared the effect of each disruption on environmental and mutational stress, we found a negative relationship, such that strains that were more environmentally robust tended to be less robust to mutational stress.

Our results demonstrate that mutant strains of E. coli respond differently to stress, depending on whether random or hub genes are disrupted. This difference indicates that the power-law distribution of regulatory interactions has biological significance, making random disruptions less deleterious to organisms facing environmental stress. That E. coli can reduce the effect of environmental stress without reducing the phenotypic effect of additional mutations, indicates that robustness and evolvability need not be antagonistic.