Hierarchy and feedback in the evolution of the Escherichia coli transcription network

doi:10.1073/pnas.0609023104

Hierarchy and feedback in the evolution of the Escherichia coli transcription network

*Unité Mixte de Recherche 168/Institut Curie, 26 rue d'Ulm, 75005 Paris, France;
^†Università degli Studi di Milano, Dipartimento di Fisica, Via Celoria 16, 20133 Milano, Italy;
^§Politecnico di Milano, Dipartimento di Fisica, Pza Leonardo Da Vinci 32, 20133 Milano, Italy; and
^¶Istituto Nazionale di Fisica Nucleare, 20133 Milano, Italy

Edited by Michael S. Waterman, University of Southern California, Los Angeles, CA, and approved January 22, 2007 (received for review October 12, 2006)

Abstract

The Escherichia coli transcription network has an essentially feedforward structure, with abundant feedback at the level of self-regulations. Here, we investigate how these properties emerged during evolution. An assessment of the role of gene duplication based on protein domain architecture shows that (i) transcriptional autoregulators have mostly arisen through duplication, whereas (ii) the expected feedback loops stemming from their initial cross-regulation are strongly selected against. This requires a divergent coevolution of the transcription factor DNA-binding sites and their respective DNA cis-regulatory regions. Moreover, we find that the network tends to grow by expansion of the existing hierarchical layers of computation, rather than by addition of new layers. We also argue that rewiring of regulatory links due to mutation/selection of novel transcription factor/DNA binding interactions appears not to significantly affect the network global hierarchy, and that horizontally transferred genes are mainly added at the bottom, as new target nodes. These findings highlight the important evolutionary roles of both duplication and selective deletion of cross-talks between autoregulators in the emergence of the hierarchical transcription network of E. coli.

Footnotes

^‡To whom correspondence may be addressed. E-mail: mcl{at}curie.fr or herve.isambert{at}curie.fr
Author contributions: M.C.L., B.B., and H.I. designed research; M.C.L. and P.J. performed research; and M.C.L. and H.I. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS direct submission.
This article contains supporting information online at www.pnas.org/cgi/content/full/0609023104/DC1.
↵ ^‖This corresponds to a conservative view of homology where no domains are acquired or lost after duplication. More flexible and realistic definitions of homologs yield essentially the same results (Note 2 in SI Appendix).
↵**This is not strictly true for the more recent RegulonDB 5.5 data set, where a few of these two-node feedback loops are observable, though the signature for negative selection remains (see Note 5 in SI Appendix).
Abbreviations:

TF,

transcription factor;

AR,

autoregulator;

TG,

target gene.