Signal-Transduction Based Network Biomarkers for Tumor-Initiating Cells in a p53-Null Mouse Model of Breast Cancer.

Abstract

Abstract Introduction: Tumor initiating cells (TICs), or cancer stem cells, have the capacity to self-renew, and are likely to be responsible for resistance to conventional therapies, disease recurrence, and metastasis. Known TIC biomarkers may allow us to understand which genes are critical to self-renewal of TICs. However, little is known about how these biomarkers manipulate other molecules to enable TICs to self-renew. Here, we develop a systems biology approach to identify the signal-transduction network biomarkers for understanding the mechanisms in the renewal of TICs.Methods: Comparing with conventional biomarker discovery, the network biomarkers are identified by not only differential genes but also protein-protein interaction (PPI) network and available signaling pathways. Depending on the interacting patterns in PPI network and their associated cellular processes in signaling pathways, we can reveal the key signal transductions or subnetworks responsible for self-renewal of TICs. We chose two types of enriched interacting patterns, i.e. triangle and spoon, to indicate the signal transductions around the single biomarkers. A published RNA microarray (GSE8863) was performed on mammary epithelial cells from p53-null tumors sorted by two stem cell markers CD29 and CD24. The differential genes were identified by the Fold-Changes (FCs) comparing the subpopulations of Lin —CD29HCD24H with other subpopulations. Gene Set Enrichment Analysis (GSEA) revealed that the identified differential genes are significantly enriched in 20 regulatory pathways including 'Cell cycle', 'Adherens junction', 'Cell adhesion molecules (CAMs)', 'Tight junction' (P<10-3). A model called Multiple Objective Optimization (MOO) was employed to search the network biomarkers whose component proteins satisfy that (1) be most differentially expressed in Lin —CD29HCD24H cells and (2) pass through a large number of enriched signaling pathways.Results: Four genes (Lsm5, Calm3, Bmi1 and Ezh2) are extremely up-regulated in the Lin —CD29HCD24H subpopulations (FC>4), in which Bmi1 is important in regulating the self-renewal capacity of hematopoietic, as well as human mammary gland stem cells. To understand how these genes manipulate other genes and the enriched signaling pathways, we first analyzed the interacting patterns around these genes. Calm3 and Bmi1 are surrounded by 355 and 22 interacting patterns respectively. The component coding-genes in the interacting patterns are more likely to be differentially expressed (91/127, P <10-61). Then, the protein-paths were identified from the interacting patterns of Calm3 and Bmi1 by MOO. IPA (Ingenuity Pathway Analysis) indicated that these protein paths have many particular cellular functions related to self-renewal, such as 'Cell Growth and Proliferation', 'Hematological System Development and Function', 'Cell-To-Cell Signaling and Interaction', and 'Cell Death' (P <10-3).Conclusion: The network biomarkers revealed that the surrounding interaction patterns and related signaling pathways of single biomarkers are essential to understand the signal-transduction mechanisms of TICs in self-renewal. They may be useful as prognostic or predictive indicators as well as suggest possible targets for novel therapies. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 2138.