Open Access

ARTICLE

Elucidating functional context within microarray data by integrated transcription factor-focused gene-interaction and regulatory network analysis

Thomas Werner^1,2, Susan M. Dombrowski^3,4, Carlos Zgheib⁵, Fouad A. Zouein⁵, Henry L. Keen⁶, Mazen Kurdi^5,7, George W. Booz⁵

1 Genomatix Software GmbH, Munich, Germany
2 University of Michigan, Internal Medicine-Nephrology Division & Center of Computational Medicine and Bioinformatics (CCMB), Ann Arbor, MI,USA
3 Genomatix Software Inc., Ann Arbor, MI, USA
4 Wayne State University School of Medicine, Detroit, MI, USA
5 Department of Pharmacology and Toxicology, School of Medicine, and the Jackson Center for Heart Research, The University of Mississippi MedicalCenter, Jackson, Mississippi, USA
6 Department of Pharmacology, University of Iowa College of Medicine, Iowa City, Iowa, USA
7 Department of Chemistry and Biochemistry, Faculty of Sciences, Lebanese University, Rafic Hariri Educational Campus, Hadath, Lebanon

European Cytokine Network 2013, 24(2), 75-90. https://doi.org/10.1684/ecn.2013.0336

Abstract

Microarrays do not yield direct evidence for functional connections between genes. However, transcription factors (TFs) and their binding sites (TFBSs) in promoters are important for inducing and coordinating changes in RNA levels, and thus represent the first layer of functional interaction. Similar to genes, TFs act only in context, which is why a TF/TFBS-based promoter analysis of genes needs to be done in the form of gene(TF)-gene networks, not individual TFs or TFBSs. In addition, integration of the literature and various databases (e.g. GO, MeSH, etc) allows the adding of genes relevant for the functional context of the data even if they were initially missed by the microarray as their RNA levels did not change significantly. Here, we outline a TF-TFBSs network-based strategy to assess the involvement of transcription factors in agonist signaling and demonstrate its utility in deciphering the response of human microvascular endothelial cells (HMEC-1) to leukemia inhibitory factor (LIF). Our strategy identified a central core of eight TFs, of which only STAT3 had previously been definitively linked to LIF in endothelial cells. We also found potential molecular mechanisms of gene regulation in HMEC-1 upon stimulation with LIF that allow for the prediction of changes of genes not used in the analysis. Our approach, which is readily applicable to a wide variety of expression microarray and next generation sequencing RNA-seq results, illustrates the power of a TF-gene networking approach for elucidation of the underlying biology.

---

Keywords

---