Email updates

Keep up to date with the latest news and content from Breast Cancer Research and BioMed Central.

Open Access Highly Accessed Research article

The gene expression landscape of breast cancer is shaped by tumor protein p53 status and epithelial-mesenchymal transition

Erik Fredlund1, Johan Staaf1, Juha K Rantala2, Olli Kallioniemi3, Åke Borg1 and Markus Ringnér1*

Author Affiliations

1 Department of Oncology, Clinical Sciences and CREATE Health Centre for Translational Cancer Research, Lund University, Lund, Sweden

2 Department of Biomedical Engineering and Knight Cancer Institute, Oregon Health and Science University, Portland, OR, USA

3 Institute for Molecular Medicine Finland (FIMM), University of Helsinki, Helsinki, Finland

For all author emails, please log on.

Breast Cancer Research 2012, 14:R113  doi:10.1186/bcr3236

Published: 27 July 2012

Abstract

Introduction

Gene expression data derived from clinical cancer specimens provide an opportunity to characterize cancer-specific transcriptional programs. Here, we present an analysis delineating a correlation-based gene expression landscape of breast cancer that identifies modules with strong associations to breast cancer-specific and general tumor biology.

Methods

Modules of highly connected genes were extracted from a gene co-expression network that was constructed based on Pearson correlation, and module activities were then calculated using a pathway activity score. Functional annotations of modules were experimentally validated with an siRNA cell spot microarray system using the KPL-4 breast cancer cell line, and by using gene expression data from functional studies. Modules were derived using gene expression data representing 1,608 breast cancer samples and validated in data sets representing 971 independent breast cancer samples as well as 1,231 samples from other cancer forms.

Results

The initial co-expression network analysis resulted in the characterization of eight tightly regulated gene modules. Cell cycle genes were divided into two transcriptional programs, and experimental validation using an siRNA screen showed different functional roles for these programs during proliferation. The division of the two programs was found to act as a marker for tumor protein p53 (TP53) gene status in luminal breast cancer, with the two programs being separated only in luminal tumors with functional p53 (encoded by TP53). Moreover, a module containing fibroblast and stroma-related genes was highly expressed in fibroblasts, but was also up-regulated by overexpression of epithelial-mesenchymal transition factors such as transforming growth factor beta 1 (TGF-beta1) and Snail in immortalized human mammary epithelial cells. Strikingly, the stroma transcriptional program related to less malignant tumors for luminal disease and aggressive lymph node positive disease among basal-like tumors.

Conclusions

We have derived a robust gene expression landscape of breast cancer that reflects known subtypes as well as heterogeneity within these subtypes. By applying the modules to TP53-mutated samples we shed light on the biological consequences of non-functional p53 in otherwise low-proliferating luminal breast cancer. Furthermore, as in the case of the stroma module, we show that the biological and clinical interpretation of a set of co-regulated genes is subtype-dependent.