Cell plasticity and phenotypic heterogeneity is pervasive amongst tumour cells and is a major driving force behind metastasis and relapse in patients. Many cancer cell lines also display vast cell-to-cell variability, but how this phenotypic diversity and plasticity is established and maintained remains unclear. The cell surface phenotype, CD44+ CD24-, can be used to isolate aggressive, “stem-like” cells that possess enhanced drug resistance and metastatic abilities in various cancer models. Here, we show that distinct phenotypic equilibriums in the proportions of these aggressive cells exist across breast, ovarian, prostate and lung cancer cell lines. This equilibrium is re-established when subpopulations are sorted and replated, suggesting that at a population level, cancer cells can transition through these different cellular states. In contrast, when unperturbed individual cells are sorted and expanded as single-cell clones they exhibit a spectrum of phenotypes, each with their own unique equilibriums. These clonal biases appear to be cell-intrinsic and are stable for over 100 days in culture.
In breast cancer cells, the clonal phenotypes are paralleled by distinct molecular and functional properties. Transcriptional and chromatin accessibility profiling demonstrated that the clonal heterogeneity extends beyond CD44/CD24, and hundreds of genes are differentially regulated across the clones. In vitro, the spectra of clones show distinct differences in cell size, growth rate and their ability to form tumourspheres. In vivo, the clones have dramatic differences in their tumour initiating and metastatic capabilities. Interestingly, while the CD44/CD24 cell surface phenotype predicts in vivo tumourigenesis for some clones, others behave unexpectedly, suggesting another axis of clonal heterogeneity drives in vivo phenotypes.
Our findings give insight into cancer cell plasticity dynamics and evolution, and the remarkable differences in molecular, phenotypic and functional phenotypes across clones in the absence of any genetic manipulations. This has important consequences in the reproducibility and interpretation experiments involving single-cell derived clones such as CRISPR-knockout or reporter cell lines. Ongoing efforts are using this tractable model to identify novel mediators that regulate and maintain cells in this aggressive plastic state, with the aim of identifying novel druggable targets that can be therapeutically leveraged to manipulate cellular phenotypes in cancer.