Understanding breast cancer stem cell heterogeneity: time to move on to a new research paradigm

The clonal expansion and adaptation of BC cells to changing microenvironments [6], and the acquisition of genetic and epigenetic alterations by these cells [11] are well-known dynamic processes contributing to the generation of intra-tumor heterogeneity [12]. In particular, BC heterogeneity can arise from the differentiation of stem-like cells along with the clonal selection that occurs during BC progression, and such heterogeneity represents a major challenge for the design of effective therapies. To make inferences about BC progression, it is important to understand the stem cell origins of the inter-tumor and intra-tumor heterogeneity, which requires more effective BCSC biomarkers.

Two main initial theories have provided some mechanism(s) accounting for BC heterogeneity: 1) the theory of cancer stem cells (CSCs), which suggests that different tumors result from different stem cells, and that all cells within a given tumor are capable of progressing to a higher degree of malignancy [13]; and 2) the theory of clonal evolution, which hypothesizes that different tumors originate from evolution of a single stem cell, and that only the most aggressive clone progresses [14]. A recent study showed that cancer progenitor cells have the capacity to dedifferentiate and acquire a stem cell-like phenotype, indicating that CSC and relatively differentiated progenitors coexist in dynamic equilibrium and are subject to bidirectional conversion [15].

The dominant role of the tumor microenvironment in determining the CSC phenotype characteristics within a malignancy is noteworthy, as it suggests that tumors contain large populations of tumorigenic and non-tumorigenic tumor cells, whose distribution may vary over time [6, 16]. Like normal stem cells, CSCs also display biologically significant phenotypic and functional heterogeneity, and their progeny can show diverse plasticity [17], All these tumor cells need to be therapeutically targeted to improve the cure rate in patients with cancer [8].

Recently, outstanding challenges in identifying CSCs, their dependency on a supportive niche, and their role in metastasis have been addressed by a fluid model [16], in which the quality of stemness, rather than being fixed entity, is a flexible quality of tumor cells that can be lost and gained [18].

The major problems/limitations in BC research are represented by the high flexibility of the CSC system (dictated by the microenvironment) and the difficulty of CSC identification (imposed by the current imperfect biomarkers). Although an extensive compilation of molecular CSC markers for distinct human solid tumor types has been reviewed [18], actually none of the known markers are specific for CSC, and only new cell surface marker combinations may improve and hamper reliability, identification, and enrichment of CSCs, thus new biomarker panels are then urgently needed to recognize and quantify more efficiently both circulating and resident BC CSCs.

It has been previously shown that human BCSCs can be isolated and analyzed based on CD44^high, CD24^low and high aldehyde dehydrogenase (ALDH) activity [19, 20]. Improved profiling and extensive characterization of BCSCs was presented in BMC Cancer by Ghebeh et al., showing the importance of a new biomarker combination for understanding BC carcinogenesis and heterogeneity [21].

The absence of reliable CSC biomarkers continuously stimulates BC research, in order to identify BCSC in ex vivo models, and thus improve their identification and enrichment in the tumor microenvironment [6], and elucidate the biological basis of BC heterogeneity and drug resistance [22].

To better characterize human normal and malignant breast epithelial cell subpopulations, Ghebeh et al., in a research article in BMC Cancer, analyzed a wide panel of breast epithelial stem/progenitor/cancer stem cell markers in normal and malignant breast tissues and BC cell lines, studying subpopulations of cells for mammosphere-forming and colony-forming capacity [21]. These skilful and elegant experiments showed that epithelial population ‘basal A’ progenitor cells (Ep-CAM^-/low/CD49f⁺), ‘luminal B’ progenitor cells (Ep-CAM^high/CD49f⁺), and ‘luminal differentiated C’ cells (Ep-CAM^high/CD49f^-) differ in their ability to form mammospheres and colonies (A>B, while C had no ability) (Figure 1). Although all three populations are found in normal tissue, there is in tumor tissue a shift towards type C and a great decrease in type A, and the majority of the nine BC cell lines analyzed mostly exhibited a population B/C phenotype.

In general, CD44^high/CD24^low cell surface markers were the most efficient panel for selecting normal epithelial progenitors. Further fractionation of CD44^high/CD24^low cells may select for luminal progenitors within Ep-CAM^high/CD49f⁺ cell types, and for basal progenitors within Ep-CAM^{- or low}/CD49f⁺.

Primary BC tissues (mainly luminal Ep-CAM^high) were found to contain CD44^high/CD24^low cells in both CD49f^- and CD49f⁺ cancer cell fractions. Ghebeh et al. showed for the first time that the CD44^high/CD24^low subpopulation within CD49f^high cell types had the highest efficiency compared with other well-known subpopulations (based on MUC-1^-, ALDH⁺, and CD10⁺ expression).

From a cancer biology point of view, Ghebeh et al. have performed an interesting and comprehensive study comparing various subpopulations of cells with stem cell-like properties, supporting the notion that BCSC were predominantly CD49f⁺, and proposing the use of CD44^high/CD24^low in combination with Ep-CAM/CD49f as valuable biomarkers to identify BC cells with enhanced mammosphere-forming and colony-forming ability.

The phenotype of the normal human mammary gland stem/progenitor cells has been previously described in various reports as ALDH^high, CD10⁺, CD44^high/CD24^low or Ep-CAM⁺/MUC1^- and CD49f⁺[18].

Ghebeh et al. found that human mammary epithelial cells with a CD44^high/CD24^low phenotype had the highest progenitor ability, providing a convincing demonstration that, in both normal and malignant breasts, there are multiple CD44^high/CD24^low subpopulations. Within the basal Ep-CAM^-/low/CD49f⁺ cells, the subpopulation of CD44^high/CD24^low has the highest progenitor ability, whereas CD10^- cells have the lowest progenitor ability (that is, the lowest number of differentiated myoepithelial cells).

It is known that luminal mammary epithelial cells have a estrogen receptor-positive (ER⁺) cell population, whereas proliferating normal luminal cells are known to be ER^-[1]. Interestingly, Ghebeh et al. showed a correlation between the CD44^high/CD24^low phenotype and ER^- profile in normal mammary epithelial cells within Ep-CAM^high/CD49f⁺, in full agreement with the progenitor ability of these cells.

As depicted in Figure 1, there is a clear drift in patients with BC towards population C (Ep-CAM^high/CD49f^-) which almost doubled, whereas population A (Ep-CAM^-/low/CD49f⁺) dramatically decreased in BC compared with healthy tissue.

For the first time, it has been shown that the majority of BCSC with CD44^high/CD24^low phenotype exist mainly in the Ep-CAM^high/CD49f⁺ fraction of cancer cells, revealing a significant difference in CD44^high/CD24^low expression (in Ep-CAM^high BC cells) between ER and basal subtypes of BC in CD49f⁺ cancer cells only [21]. In other words, putting together the pieces of the puzzle, these data imply that CD49f (α-6 integrin molecule), if used in combination with CD44^high/CD24^low markers, may be able to link the stem/progenitor cell profile with the heterogeneity of BC subtypes. Thus, BCSCs can best be enriched by selecting for tumor cells with the CD44^high/CD24^low/ALDH^high phenotypes within Ep-CAM^high/CD49f⁺ BC cells.

Improving technological methods, such as single-cell analysis [23] for earlier detection and diagnosis of human BC, in conjunction with the discovery and validation of powerful combinations of BCSC biomarkers, may represent key tools to obtain a significant reduction in morbidity and mortality in BC.

Analysis of the BC microenvironment [6, 7, 23] and the novel identification of pure/specific epithelial stem/progenitor cells [21], will allow detection of alterations within biochemical, morphological, and molecular pathways promoting cancer initiation, progression, invasion, and metastasis, taking into account the different stem/non-stem cell compositions and interactions in the human breast microenvironment [11, 17, 18].

Therefore, although confirmatory studies are needed, it is time to move on to the new paradigm highlighted by Ghebeh et al., namely, that the majority of BC cells have a luminal Ep-CAM^high phenotype with a very small percentage of cancer cells of the Ep-CAM^-/low/CD49f⁺ phenotype.

Further research is required, which should focus on single-cell omic approaches, with particular attention on basal Ep-CAM^low primary cancer cells, as these may correspond to the mammary stem cell-enriched population in the normal breast gland.

Finally, the recently published study in BMC Cancer on CD44^high/CD24^low/CD49f⁺ biomarkers represents a shining example of how the combination of more biomolecules (singularly not perfectly accurate) may significantly improve and strengthen the measurement of BCSCs with significantly higher stem/progenitor ability. These experiments suggest that these biomarkers will be a useful BC biomarker panel and the best phenotype to identify human BCSCs and to better understand BC biology.

Future developments in onco-single-cell-omics [23] will potentially revolutionize cancer biology and clinical practice, providing better understanding of BC heterogeneity, how BCSCs evolve, and which BC cells to target in order to avoid drug resistance [18].