Abstract 110: A bioinformatics view of networking in the mouse mammary microenvironment

Abstract

Abstract Applying a novel bioinformatics strategy, the objective of this study was to identify signaling interactions that occur within cell types of the mouse mammary gland. Mining through typical microarray data is quite a challenge, but it is even more difficult to extract biological relevance within the mammary microenvironment. The approach here involves a combinatorial strategy that effectively integrates basic research design, statistics, and bioinformatics. We hypothesized that cells in the microenvironment potentiate tumorigenesis by signaling with each other through critical pathways sharing common ‘network hubs’. Mammary glands were harvested from 8-week-old mice that were wildtype (WT) or had fibroblast-specific Pten deletion (fPten-/-). Fibroblasts, macrophages, endothelial and epithelial cells were selected from the glands through cell sorting and selective cell culture. Replicate samples of their cDNA were applied onto Mouse Exon Arrays. The raw data were processed through Robust Mean Analysis (RMA), and then subjected to Empirical Bayes Arrays (EBArrays) analysis to generate lists of differentially expressed genes between the fPten-/- and WT mice for the four cell types. Each gene was given a probability value as a measure of its true differential expression, and only genes with a value more than or equal to 0.7 were considered for further analyses. Three bioinformatics tools- Database for Annotation, Visualization and Integrated Discovery (DAVID), Biometric Research Branch (BRB) ArrayTools, and Ingenuity Pathway Analysis® (IPA) - were used to analyze the four gene lists. Analysis by DAVID revealed that for the fibroblasts and the macrophages, the major biological machinery activated in the fPten-/- mice related to extracellular matrix remodeling and immune response. The endothelial cells displayed genes involved in complement activation pathway. Interestingly, the genes expressed in epithelial cells related to various aspects of epithelial-mesenchymal transition. Together, this suggests that even in the absence of tumor, the fPten-/- stromal signaling infuses a tumorigenic potential into the microenvironment. A filter on BRB ArrayTools selected genes that had a 2-fold change. Average hierarchical clustering based on Spearman correlation was done to generate heat maps. This refined the list of significant genes between the genotypes for each cell type. The four fPten-/- derived genelists were then uploaded into IPA. This confirmed the output from DAVID, and further revealed that ERBB2, MMP9, TNFα, TGFβ, and NFκB, β-catenin, and Ets, were key network hubs and transcription factors, respectively, through which signaling occurred in the mammary microenvironment. The top merged networks across cell types displayed shared nodes important for communication. In summary, this analytical approach gave an insight into the ‘network players’ and ‘cellular crosstalk’ critical for a tumorigenic environment. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 110.