P4-03-04: Identification of Molecular Targets for Cancer-Initiating Cells Using a Triple-Negative Breast Cancer Mouse Model.

Abstract

Abstract Background Triple-negative breast cancer (TNBC), which is negative for estrogen receptor, progesterone receptor, and HER2, cannot be targeted with hormonal or anti-HER2 agents and frequently relapses. Thus, innovative molecular targeted therapies for TNBC are needed. Cancer-initiating cells (CICs) play a crucial role in tumor recurrences. TNBC has a much higher proportion of CICs compared with the other breast cancer subtypes, reflecting TNBC's clinical aggressiveness. However, qualitative or molecular traits of TNBC CICs have not been fully elucidated. We established a TNBC mouse model that successfully recapitulated human-like TNBC in vivo (SABCS 2010 abstract P1-03-02). We here report the phenotypes and molecular features of CICs in our TNBC mouse model. Methods: We created this animal model by transferring HRAS(G12V) into Ink4a/Arf-knockout (KO) mouse mammary epithelial cells (MECs) in vitro and inducing tumors in mammary fat pads of recipient mice. The induced TNBCs were digested, and single suspended cells were reacted with antibodies to candidate CIC markers CD29, CD24, CD44, Sca1, CD61, and CD49f. Expression patterns of the tumor cells were analyzed by FACS. Tumor cells were sorted according to their expression levels and inoculated into recipient mouse mammary fat pads to determine tumor-initiating activity. To analyze the cell-cycle status of tumor cells, we stained them with Hoechst 33342 (for DNA staining) and pyronin Y (for RNA staining) and analyzed them by FACS. The CICs’ genome-wide molecular profiles were determined by mouse microarray expression analysis. Results: In FACS and the in vivo tumor-initiating assay, the CD49f expression level was the most significant for defining tumor-initiating activity. That is, CD49f− tumor cells had much higher tumor-initiating activity than did CD49f+ cells at 100-cell inoculations (CD49f−, 5 tumors/8 mice; CD49f+, 0/8). We hypothesized that tumor initiation by CD49f− cells depends on their quiescent profile, which is a trait of CICs. To test this hypothesis, we further fractionated CD49f− cells into CD49f− quiescent (CD49f−q) cells and CD49f− dividing cells and inoculated these fractions into recipient mice. CD49f−q cells had higher tumor-initiating activity than that of CD49f− dividing cells. We thus determined that CD49f−q cells were CICs in our model. Genome-wide expression patterns for CD49f−q cells and CD49f+ cells were compared by paired t test, and over- and underexpressed genes in CD49f−q cells were functionally annotated by Ingenuity pathway analysis. The top-ranked bioprocess of overexpressed genes in CD49f−q cells was hepatic satellite cell activation (HSCA), followed by pantothenale and Coenzyme A biosynthesis, and HER2 signaling in breast cancer. HSCA is known to result in liver fibrosis due to collagen deposits in response to liver damage. In CD49f−q cells in our model, collagen family proteins and the upstream endothelin-A receptor were overexpressed and significantly correlated with HSCA. Conclusion: We identified CICs in a mouse TNBC model and revealed their molecular profile. We plan to analyze expression of HSCA-associated proteins in human TNBC. Our first step is to determine whether endothelin-A receptor is a therapeutic target in human TNBC. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P4-03-04.