Breast cancer is a highly heterogeneous disease that includes HER2/NEU+,basal-like and ER+ tumors.The most aggressive subtypes,basal-like and HER2/NEU+,are treated with chemotherapy or chemotherapy plus Trastuzumab/Herceptin,respectively,yet metastatic disease that often develop in these patients is incurable.To identify therapeutics that targets tumor-initiating cells (TICs) in HER2/NEU+ and basal-like breast cancer,we are analyzing mouse models for these breast cancer subtypes.Using a mouse model for HER-2/NEU+ breast cancer,MMTV-neu,we have identified Her-2/Neu TICs by cell sorting and transplantation into isogenic mice and have shown that TICs are indistinguishable from cells that induce tumorsphere growth under non-adherent conditions (Cancer Res.2007; Clinical Cancer Res.2009).While tumorsheres are enriched for TICs,the monolayer cultures quickly exhaust the TIC fraction.Using the tumorspheres as surrogate for TICs and lend-viral shRNA technology to knockdown the kinasome,we are screening for kinases that are required for proliferation or survival of Neu tumorsphere but not monolayer cells.Several clones that specifically block sphere but not monolayer growth have been identified and will be subjected to various validation criteria.Similar analysis for basal-like breast cancer has been hampered by the lack of good mouse models.The tumor suppressors Rb is commonly lost in sporadic cases of basal-like breast cancer.We have deleted Rb specifically in the mammary epithelium and found that the mice develop distinct mammary tumors with features of either luminalB or basal-like breast cancer.The basal-like mammary tumors express basal cytokeratins,epithelial-to-mesenchymal markers,high levels of pl 6ink4a and mutant p53,and are negative for ER (submitted).This mouse model provides a powerful preclinical platform to screen for novel therapeutics and test potential drugs for basal-like breast cancer.