Abstract
The tumor-initiating cell (TIC) frequency of bulk tumor cell populations is one of the criteria used to distinguish malignancies that follow the cancer stem cell model from those that do not. However, tumor-initiating cell frequencies may be influenced by experimental conditions and the extent to which tumors have progressed, parameters that are not always addressed in studies of these cells. We employed limiting dilution cell transplantation of minimally manipulated tumor cells from mammary tumors of several transgenic mouse models to determine their tumor-initiating cell frequency. We determined whether the tumors that formed following tumor cell transplantation phenocopied the primary tumors from which they were isolated and whether they could be serially transplanted. Finally we investigated whether propagating primary tumor cells in different tissue culture conditions affected their resident tumor-initiating cell frequency. We found that tumor-initiating cells comprised between 15% and 50% of the bulk tumor cell population in multiple independent mammary tumors from three different transgenic mouse models of breast cancer. Culture of primary mammary tumor cells in chemically-defined, serum-free medium as non-adherent tumorspheres preserved TIC frequency to levels similar to that of the primary tumors from which they were established. By contrast, propagating the primary tumor cells in serum-containing medium as adherent populations resulted in a several thousand-fold reduction in their tumor-initiating cell fraction. Our findings suggest that experimental conditions, including the sensitivity of the transplantation assay, can dramatically affect estimates of tumor initiating cell frequency. Moreover, conditional on cell culture conditions, the tumor-initiating cell fraction of bulk mouse mammary tumor cell preparations can either be maintained at high or low frequency in vitro thus permitting comparative studies of tumorigenic and non-tumorigenic cancer cells.