The phenotypic and functional differences between cells that initiate human breast tumors (cancer stem cells) and those that comprise the tumor bulk are difficult to study using only primary tumor tissue. We embarked on this study hypothesizing that breast cancer cell lines would contain analogous hierarchical differentiation programs to those found in primary breast tumors.

Eight human breast cell lines (human mammary epithelial cells, and MCF10A, MCF7, SUM149, SUM159, SUM1315 and MDA.MB.231 cells) were analyzed using flow cytometry for CD44, CD24, and epithelial-specific antigen (ESA) expression. Limiting dilution orthotopic injections were used to evaluate tumor initiation, while serial colony-forming unit, reconstitution and tumorsphere assays were performed to assess self-renewal and differentiation. Pulse-chase bromodeoxyuridine (5-bromo-2-deoxyuridine [BrdU]) labeling was used to examine cell cycle and label-retention of cancer stem cells. Cells were treated with paclitaxol and 5-fluorouracil to test selective resistance to chemotherapy, and gene expression profile after chemotherapy were examined.

The percentage of CD44⁺/CD24^- cells within cell lines does not correlate with tumorigenicity, but as few as 100 cells can form tumors when sorted for CD44⁺/CD24^-/low/ESA⁺. Furthermore, CD44⁺/CD24^-/ESA⁺ cells can self-renew, reconstitute the parental cell line, retain BrdU label, and preferentially survive chemotherapy.

These data validate the use of cancer cell lines as models for the development and testing of novel therapeutics aimed at eradicating cancer stem cells.