Functional architecture of Escherichia coli: new insights provided by a natural decomposition approach

TL;DR

This paper reveals the natural modular and hierarchical organization of Escherichia coli's transcriptional regulatory network using a novel decomposition approach, identifying new network elements and classifying genes based on topological features.

Contribution

It introduces a mathematical criterion for classifying network elements into hierarchical or modular genes and uncovers intermodular genes as key integrators in the network.

Findings

Modular genes form physiologically correlated groups.

Hierarchical genes encode global regulators.

Intermodular genes integrate signals from different modules.

Abstract

Background: Previous studies have used different methods in an effort to extract the modular organization of transcriptional regulatory networks (TRNs). However, these approaches are not natural, as they try to cluster highly connected genes into a module or locate known pleiotropic transcription factors in lower hierarchical layers. Here, we unravel the TRN of Escherichia coli by separating it into its key elements, thus revealing its natural organization. We also present a mathematical criterion, based on the topological features of the TRN, to classify the network elements into one of two possible classes: hierarchical or modular genes. Results: We found that modular genes are clustered into physiologically correlated groups validated by a statistical analysis of the enrichment of the functional classes. Hierarchical genes encode transcription factors responsible for coordinating module responses based on general interest signals. Hierarchical elements correlate highly with the previously studied global regulators, suggesting that this could be the first mathematical method to identify global regulators. We identified a new element in TRNs never described before: intermodular genes. These are structural genes that integrate, at the promoter level, signals coming from different modules, and therefore from different physiological responses. Using the concept of pleiotropy, we have reconstructed the hierarchy of the network and discuss the role of feedforward motifs in shaping the hierarchical backbone of the TRN. Conclusions: This study sheds new light on the design principles underpinning the organization of TRNs, showing a novel nonpyramidal architecture composed of independent modules globally governed by hierarchical transcription factors, whose responses are integrated by intermodular genes.