Transcriptional master regulator analysis in breast cancer genetic networks

Abstract

Gene regulatory networks account for the delicate mechanisms that control gene expression. Under certain circumstances, gene regulatory programs may give rise to amplification cascades. Such transcriptional cascades are events in which activation of key-responsive transcription factors called master regulators trigger a series of gene expression events. The action of transcriptional master regulators is then important for the establishment of certain programs like cell development and differentiation. However, such cascades have also been related with the onset and maintenance of cancer phenotypes. Here we present a systematic implementation of a series of algorithms aimed at the inference of a gene regulatory network and analysis of transcriptional master regulators in the context of primary breast cancer cells. Such studies were performed in a highly curated database of 880 microarray gene expression experiments on biopsy-captured tissue corresponding to primary breast cancer and healthy controls. Biological function and biochemical pathway enrichment analyses were also performed to study the role that the processes controlled – at the transcriptional level – by such master regulators may have in relation to primary breast cancer. We found that transcription factors such as AGTR2, ZNF132, TFDP3 and others are master regulators in this gene regulatory network. Sets of genes controlled by these regulators are involved in processes that are well-known hallmarks of cancer. This kind of analyses may help to understand the most upstream events in the development of phenotypes, in particular, those regarding cancer biology.

Introduction

Cancer is a pathway-disease (Hanahan and Weinberg, 2000). The main hallmarks of cancer are associated to the action of pathways related to cell proliferation, apoptosis evasion, cell differentiation and in general, to the dysregulation of cell cycle and the alteration of DNA-repairing processes (Hanahan and Weinberg, 2000, Hanahan and Weinberg, 2011). The phenotype of a cell is determined by the activity of a large number of genes and proteins (Basso et al., 2005). Hence, transcriptional regulation lies at the heart of many of the intricate molecular relationships driving the activity of biological pathways (Emmert-Streib et al., 2014).

It has been observed that a number of large scale transcriptional cascades behind such complex cellular processes are actually triggered by the action of a relatively small number of transcription factor molecules that have been called Transcriptional Master Regulators (TMRs) (Han et al., 2004, Sun-Kin Chan and Kyba, 2013, Mullen et al., 2011). It has been argued that these genes control the entire transcriptional regulatory program for specific cellular phenotypes (in eukaryotic cells; Han et al., 2004, Basso et al., 2005, Affara et al., 2013). However, TMRs are also able to act on general cellular processes at the same time (Hinnebusch and Natarajan, 2002, Medvedovic et al., 2011, Affara et al., 2013). A proper understanding of the organization of these TMR-mediated highly-regulated events is thus crucial to elucidate normal cell physiology as well as complex pathological phenotypes (Basso et al., 2005).

Given the complex mechanisms underlying transcriptional regulations on eukaryotes, the identification of TMRs is often based on the (inferred or observed) relationship among them and their cascade of RNA targets in gene regulatory networks (Hernández-Lemus and Siqueiros-García, 2013). Being a primary upstream event in the cell regulatory program, dysregulation of TMRs may have a high impact on cancer-related phenotypes, since under genetic instability conditions, uncontrolled synthesis of these molecules could originate the activation/amplification of several transcriptional cascades (Basso et al., 2005, Baca-Lopez et al., 2014, Baca-López et al., 2012).

A TMR is a transcription factor (TF) that is expressed at the early onset of the development of a particular phenotype or cell type (Sun-Kin Chan and Kyba, 2013). It also participates in the specifications of such a phenotype by regulating multiple downstream genes, either directly or by means of genetic cascades. Transcription factors are hence key cellular components that control gene expression: their activities may determine how cells function and respond to the environment (Vaquerizas et al., 2009).

Transcription factors may act in two opposite directions: either activating or repressing transcriptional activity of their targets. Based on the initial estimations of the whole human genome sequence, it was calculated that the transcriptional machinery could be composed of 200 to 300 genes and there could exist between 2000 to 3000 specific union sites for transcription factors (Lander et al., 2001, Venter et al., 2001). In Vaquerizas et al. (2009) it is stated that in the http://amigo.geneontology.orgGene Ontology database 1052 TFs were defined and just 6% (62 cases) of them had experimental corroboration. Six years later, the same database recognized 1846 TFs and 14% (260) of them had experimental evidence. This is indicative of the fast progress on documenting the transcription mechanisms, but this also points to the overwhelming complexity of the mechanisms of genomic control.

Implementation of computational methods to identify and analyze TMRs is relevant in the context of breast cancer, particularly at its earliest stages. We have probabilistically inferred the gene regulatory network associated with this phenotype, then a computational analysis has uncovered its active TMRs in the context of primary breast cancer. In our study we have considered such an analysis, as well as the resulting TMR-related phenomena in the context of transcriptional regulatory programs. We also discuss here some of the implications of our results in breast cancer biology. The article is structured as follows: Section 2 presents an overview of the materials and methods used in this work. This includes both the experimental datasets used, the network inference strategy and the molecular signature analysis, as well as the algorithm for the discovery of transcriptional master regulators. Section 3 presents some of the main results of the application of this pipeline in primary breast cancer microarray gene expression data. Finally, Section 4 presents some conclusions mainly related with the advantages of implementing a method such as MARINa (Lefebvre et al., 2010) in order to unveil some aspects of regulatory control that may lie behind the establishment of tumor phenotypes.