Functional organisation of Escherichia coli transcriptional regulatory network

Abstract

Taking advantage of available functional data associated with 115 transcription and 7 sigma factors, we have performed a structural analysis of the regulatory network of Escherichia coli. While the mode of regulatory interaction between transcription factors (TFs) is predominantly positive, TFs are frequently negatively autoregulated. Furthermore, feedback loops, regulatory motifs and regulatory pathways are unevenly distributed in this network. Short pathways, multiple feed-forward loops and negative autoregulatory interactions are particularly predominant in the subnetwork controlling metabolic functions such as the use of alternative carbon sources. In contrast, long hierarchical cascades and positive autoregulatory loops are overrepresented in the subnetworks controlling developmental processes for biofilm and chemotaxis. We propose that these long transcriptional cascades coupled with regulatory switches (positive loops) for external sensing enable the coexistence of multiple bacterial phenotypes. In contrast, short regulatory pathways and negative autoregulatory loops enable an efficient homeostatic control of crucial metabolites despite external variations. TFs at the core of the network coordinate the most basic endogenous processes by passing information onto multi-element circuits. Transcriptional expression data support broader and higher transcription of global TFs compared to specific ones. Global regulators are also more broadly conserved than specific regulators in bacteria, pointing to varying functional constraints.

Fig. 1

Core transcriptional regulatory network of E. coli. Blue and pink nodes represent genes encoding for TFs and sigma factors, respectively; each node label is accompanied with its connectivity showing the number of regulatory targets. Edges represent cross-regulatory interactions (green for activation, red for repression, blue for dual interactions and yellow for sigma transcription), whereas loops represent transcriptional autoregulations. Specific subnetworks, such as the one associated with the regulation of carbon sources, are delineated with dashed lines to distinguish different regulatory modules. This figure was generated using Cytoscape.

Fig. 2

Functional organisation of E. coli core transcriptional network. (a) selection of carbon source (group A), global regulation (group B), and regulation of developmental processes (biofilm and chemotaxis, group C). (b) Average mRNA levels per cell for each TF group defined in (a), together with standard deviations. Levels of mRNA were recovered for 13 TFs (56%) of members of group A, all TFs of group B, and 13 TFs (54%) of group C. Nodes in light blue represent members of two-component systems.

Fig. 3

Multi-element regulatory circuits found in the cross-regulatory network of E. coli. Extent of informational flux between two nodes is denoted by the thickness of the edges.

Fig. 4

Conservation of E. coli sigma and transcription factors across 216 non-redundant bacterial genomes. Node sizes are proportional to the corresponding conservation interval and to the number of genomes in which an ortholog was found (shown in parentheses).