Transcriptome-based determination of multiple transcription regulator activities in Escherichia coli by using network component analysis

Abstract

Cells adjust gene expression profiles in response to environmental and physiological changes through a series of signal transduction pathways. Upon activation or deactivation, the terminal regulators bind to or dissociate from DNA, respectively, and modulate transcriptional activities on particular promoters. Traditionally, individual reporter genes have been used to detect the activity of the transcription factors. This approach works well for simple, nonoverlapping transcription pathways. For complex transcriptional networks, more sophisticated tools are required to deconvolute the contribution of each regulator. Here, we demonstrate the utility of network component analysis in determining multiple transcription factor activities based on transcriptome profiles and available connectivity information regarding network connectivity. We used Escherichia coli carbon source transition from glucose to acetate as a model system. Key results from this analysis were either consistent with physiology or verified by using independent measurements. B acteria respond to environmental changes through a variety of sensor proteins, which eventually relay the signals to corresponding DNA binding proteins to modulate transcription. The DNA binding transcription regulators, or transcription factors (TFs), typically require posttranscriptional modification or ligand binding to assume an active conformation, which may bind to DNA and either positively or negatively regulate transcription. Here, the activity of a TF is defined as the concentration of its subpopulation capable of DNA binding. The collective activities of TFs can thus be regarded as the physiological state of the cell. Determining these TF activities (TFAs) allows better understanding of how cells respond to changes in the environment. Careful experimental studies in the past few decades have identified conditions that perturb each individual TF independent of others. Although such ideal conditions allowed useful characterization of molecular mechanisms, most environmental perturbations are complex and are likely to provoke multiple regulatory systems simultaneously. Without a proper method of decomposing the regulatory signals, it is difficult to investigate how microorganisms coordinate various regulatory pathways upon an environmental challenge.Here, we report the use of network component analysis (NCA) recently developed in our group (1) to determine the dynamics of the activities of various TFs during a physiological process. This approach uses both DNA microarray data and partial information regarding the membership of regulons as defined by each TF in question. It contrasts with other approaches, such as singular value decomposition (2) or independent component analysis (3), in that it does not depend on orthogonality and statistical independence. Rather, it uses biological information regarding regulatory network topology, even when the topology is incompletely defined. Furthermore, NCA differs from modelbased parameter estimation (4) because it allows deconv...

Reconstructing repressor protein levels from expression of gene targets in Escherichia coli

“…Previous approaches to infer the activity of a transcription factor assumed that the kinetics of gene transcription is adequately described by linear or log-linear models (10)(11)(12)(13)(14). However, it has been noted (15) that gene transcription regulated by a transcription factor resembles the process of enzyme-mediated reactions.…”

Transcription Network Analysis by A Sparse Binary Factor Analysis Algorithm

“…2(d), most of TFAs still preserve the cyclic pattern. We further apply BYY-BFA to temporal gene expression profiles of E. coli during transition from glucose to acetate, with samples taken at 5, 15, 30, 60 min and every hour until 6 h after transition [4]. Similar to [4], the repeated data points are averaged, and we demonstrate the effectiveness of BYY-BFA on 5 of 16 TFs that were analyzed by NCA in [4].…”

“…Therefore, our algorithm directly captures the underlying activation patterns of TFAs, and also extends NCA to the case when the topology of the TF-gene network is not available or not reliable. In addition, results on E. coli data show that BYY-BFA is effective to detect activations of TFAs corresponding to the adaptation to carbon source transition from glucose to acetate [4].…”

“…We further apply BYY-BFA to temporal gene expression profiles of E. coli during transition from glucose to acetate, with samples taken at 5, 15, 30, 60 min and every hour until 6 h after transition [4]. Similar to [4], the repeated data points are averaged, and we demonstrate the effectiveness of BYY-BFA on 5 of 16 TFs that were analyzed by NCA in [4]. Based on the available connectivity data and the applicability for NCA, 22 genes were selected by the NCA toolbox [1], and there were 30 regulations from TFs to genes.…”

“…Three identifiability criteria need to be satisfied in NCA for a unique decomposition [1,3]. NCA has been successfully applied to determine the transcription regulatory activities, on microarray data generated from yeast Saccharamyces cerevisiae during cell-cycle process or Escherichia coli carbon source transition from glucose to acetate [1,4], and so on. Other related methods include the REDUCE [5] that takes normalized motif binding copy number as the regulation strength of the TF for that gene, and obtains TFA profiles from gene expression data by linear regression.…”