Introduction

The steroid hormone estrogen is an important mediator of the development of the mammary epithelium during puberty [1, 2]. Considering the role estrogen has in promoting mammary gland development, it is not surprising that there is a strong positive correlation between lifetime exposure to estrogen and breast cancer risk [3, 4]. Breast tumors are typically categorized as expressing estrogen receptor-α (ERα+) or as lacking expression of this receptor (ERα−). There are several paradoxical observations regarding the expression of ERα in normal and cancer biology. The first is that the mammary epithelial cells that express ERα in the normal adult mammary gland are typically not proliferating [5–8]. This is counter-intuitive considering the key role estrogen has in promoting the development of the mammary ductal epithelium during puberty; this is particularly so since ERα+ cells are often observed to be proliferating in tumors [6]. The second paradoxical observation is that premenopausal exposure to estrogen appears to have a greater impact on ERα− than on ERα+ tumors since the rate of occurrence of ERα− tumors decreases after menopause whereas the rate of occurrence of ERα+ tumors steadily increases after menopause, albeit at a slightly lower rate than before menopause [9]. This review will present a brief summary of how steroid hormones influence the different components of the mammary epithelial cell hierarchy during mammary gland development, and how this knowledge can be used to understand the origins of different types of breast tumors and the mechanisms that regulate their behavior.

The Mammary Epithelial Cell Hierarchy and the Distribution of Steroid Hormone Receptors

The mammary epithelium is composed of a series of branched ducts that, during lactation, drain milk-producing alveoli. In the virgin state, the ducts in mice are blind-ended structures and are embedded within mammary adipose tissue (the mouse mammary fat pad) whereas in humans, the ducts terminate in terminal ductal lobular units and the entire epithelium is embedded within a collagenous stroma. There are two main lineages of epithelial cells within the mammary epithelium: the luminal cells that line the central lumen and the underlying myoepithelial cells that are adjacent to the basement membrane. During pregnancy, there is the development of a large number of specialized luminal cells that will arrange themselves as alveoli and will secrete milk proteins and lipids during lactation. Our knowledge of the components of the mammary epithelial cell hierarchy has expanded in recent years as a result of the development of strategies to prospectively isolate phenotypically distinct subsets of mammary cells using fluorescence-activated cell sorting and assessing the properties of these cells using in vitro and in vivo functional assays [10–15]. For example, in the mouse, it has been demonstrated that most mammary repopulating units (MRUs; the operational term for cells that can generate multi-lineage ductal-lobular outgrowths when transplanted into cleared mammary fat pads) appear to reside in the basal cell compartment since they express basal specific keratins 5 and 14, but do not express the luminal-specific keratins 8 or 18 [12, 15]. Interestingly, these cells do not express ERα or progesterone receptor (PR) [16]. It has been recently reported that a minor subpopulation of MRUs exists within the luminal cell compartment of the mouse mammary gland, and that these cells can only be detected when co-transplanted with Matrigel™ into cleared mammary fat pads [17]. It is not clear as of yet if these MRUs are comparable to those in the basal cell compartment or if they are more (or less?) restricted in their stem cell potential. These MRUs may also represent more mature progenitor cells that have dedifferentiated to acquire stem cell properties. The presence of stem cells in both the luminal and basal layers has also been described in the mouse prostatic epithelium [18, 19]. MRUs in the human mammary gland have been detected in the basal cell compartment and do not express ERα and PR [20–22].

In vitro colony-forming assays have been used to interrogate the properties of progenitor cells in the mammary glands of both species. In the mouse, most of the progenitor cells reside within the luminal cell compartment [12, 15, 16]. These cells do not express ERα or the luminal cell differentiation marker Gata-3; however, they do express luminal-specific and basal-specific keratins at low levels [12]. These cells also express the alveolar cell-associated transcription factor Elf5 and surprisingly, transcripts associated with milk proteins, even in the virgin state [12, 23]. This phenotype with low expression of both luminal and basal keratins suggests that these cells are a relatively undifferentiated progenitor cell that is an intermediate between basal stem cells and differentiated luminal cells, and the expression of milk transcripts and Elf5 suggests that these cells may represent a pool of progenitors in the virgin mammary gland that will generate alveoli during pregnancy. In vivo lineage tracing to determine the differentiation capacities of these cells has yet to be performed to confirm this hypothesis.

Most of the cells in the luminal cell compartment of the mammary glands of virgin mice appear to be relatively differentiated ERα+ cells. However, there appears to be a small but poorly characterized population of ERα+ progenitor cells [24, 25]. The luminal cell compartment in the human breast also contains a large population of differentiated ERα+ cells, a population of ERα− progenitors, which have a phenotype intermediate between luminal and basal cells and a population of ERα+ progenitors [14, 20–22, 26]. Figure 1 shows a hypothetical mammary epithelial cell hierarchy.

Fig. 1
figure 1

Hypothetical mammary epithelial cell hierarchy with the phenotypes of the different known cellular components in both human (H) and in mouse (M) mammary glands. It is hypothesized that estrogen and progesterone may directly stimulate ERα+/PR+ progenitor cells to proliferate. Estrogen may also induce differentiated ERα+/PR+ cells to secrete amphiregulin, which then induces stromal cells to secrete factor(s) that stimulate stem cells to proliferate. Progesterone induces differentiated ERα+/PR+ cells to secrete RANKL, which stimulates stem cells and putative alveolar progenitors to proliferate

Full size image

The Role of Steroid Hormones in Regulating Normal Mammary Stem and Progenitor Cell Function

Evidence is emerging that estrogen and progesterone exert their effects in the mammary gland primarily via paracrine interactions. For example, estrogen, which is the primary driver of ductal elongation during mammary gland development during puberty, induces secretion of amphiregulin from ERα-expressing cells [27]. Amphiregulin in turn stimulates epidermal growth factor receptor-expressing stromal cells to secrete a factor that then stimulates mammary stem cells and other cells involved in ductal growth [28, 29]. The paracrine factor secreted by the stromal cells has yet to be identified, although a number of candidates such as keratinocyte growth factor have been proposed [30]. In the adult, estrogen seems to have more of a permissive role since it upregulates PR, possibly via the stroma [31]. However, estrogen appears to have a key role in maintaining differentiated ERα+ and stem cell populations in the adult mammary gland since removal of circulating estrogens, either via ovariectomy or by a transient treatment with the aromatase inhibitor Letrozole, results in a large decrease in the number of differentiated ERα+ cells and a several-fold decrease in the absolute number of mouse MRUs and the engraftment potential of these MRUs 5–8 weeks after treatment [32, 33]. The effects of reduced exposure to estrogen on MRUs are thought to be mediated by inducing these cells into a quiescent state [32].

Similarly, progesterone also exerts its proliferative-inducing effects primarily via a paracrine mechanism, although it does require the presence of estrogen for maximal effect, presumably through upregulation of PR [32–34]. In the mouse, it has been demonstrated that progesterone induces receptor activator for nuclear factor κB ligand (RANKL) secretion from ERα+/PR+ mammary cells, and that this factor then directly binds to its receptor RANK and induces side-branching and alveolar development during ductal development and pregnancy [35–37]. RANKL has been demonstrated to regulate MRU function since basal cells express high levels of RANK and treatment of mice with RANKL inhibitor results in decreased clonogenic output from basal cells [32, 33]. However, this is at odds with the observation that PR-B knockout virgin mice have normal ductal development, which is presumably dependent on the proliferation of stem cells [37]. Progesterone has also been demonstrated to mediate secretion of Wnt4 from PR+ mammary cells, which can function as a paracrine factor to regulate branching morphogenesis during early pregnancy [38].

However, progesterone does not solely exert its proliferative effects via indirect mechanisms that target PR- cells since a recent publication has demonstrated that progesterone can induce transient proliferation of PR+ cells in a cyclin D1-dependent fashion in addition to inducing PR− cells to proliferate through a RANKL-dependent mechanism [39]. These results suggest the presence of an ERα+/PR+ progenitor cell that is directly responsive to progesterone. This concept of a distinct ERα+/PR+ progenitor population has also been suggested by previous studies [24–26]. A summary of the direct and indirect effects of estrogen and progesterone on the different mammary epithelial cell subtypes is shown in Fig. 1.

The Cell of Origin of Human Breast Tumors

Human breast tumors are very heterogeneous with approximately five molecular subtypes recognized [40, 41]. It is not known if all breast tumors originate in stem cells and that the genetic mutations acquired during tumor progression drive the cells down different differentiation pathways, or if the heterogeneity reflects the different cellular origin of the tumors, or if it is a combination of the two mechanisms. However, it is important to keep in mind is that the gene signature of a tumor represents the average of all the cells sampled within the tumor. It is conceivable that the cancer stem cell component of the tumor is a small proportion of the total tumor cell population, and thus, its gene signature may be diluted by the bulk of the non-stem tumor cells. Evidence is emerging that metaplastic/claudinlow tumors, which are a subtype of basal breast cancer, may have a stem cell origin due to their mesenchymal and squamous metaplastic elements, their very aggressive nature, and their gene expression patterns resembling those obtained from CD44+CD24−/low breast cancer stem cells [42]. However, metaplastic breast cancer is a relatively rare type of tumor [42, 43], and it appears that most basal-like breast tumors have a gene expression profile that resembles those obtained from luminal progenitor cells, thus, suggesting that basal-like breast tumors may have their origin in this cell type [21]. If so, this would resolve the confusion regarding basal-like breast tumors being categorized as “basal” when they also express luminal-specific keratins [44]. Gene expression profiling of normal differentiated ERα+ luminal cells reveals that these cells have a gene signature that is similar to those obtained from Luminal A and Luminal B tumors [21]. The cell of origin of ErbB2-amplified tumors is not known, although MMTV-Neu mouse models suggest ERα− luminal progenitors could be the target in this type of cancer [45, 46].

Influence of Steroid Hormones and Tamoxifen on Human Breast Tumor Cells

As stated previously, estrogen appears to have a greater impact on ERα− than on ERα+ tumors since the rate of occurrence of ERα− tumors decreases after menopause whereas the rate of occurrence of ERα+ tumors steadily increases after menopause [9]. Considering the emerging evidence that estrogen and progesterone have strong indirect proliferation-inducing effects on mammary stem cells and putative alveolar progenitors, populations which are both ERα−, it is not surprising then that the rate of incidence of ERα− tumors decreases at menopause should these tumors be derived from these subpopulations. One can hypothesize that ERα+ tumors that originate in ERα+ luminal progenitor cells become tumorigenic via mechanisms independent of estrogen-related signaling; in such cases, estrogen may be permissive for tumorigenesis, but is not the key driver of the malignancy. As a result, tumors of this type could occur even in the low estrogen environment after menopause.

A large proportion of women with ERα+ tumors will become resistant to tamoxifen treatment, typically within 15 months of treatment [5, 47]. There are several possible mechanisms for the emergence of tamoxifen resistance in ERα+ tumors aside from host-associated factors (e.g., drug pharmacokinetics). The first is that there is the presence of a tamoxifen-resistant clone within the starting tumor cell population that, by its inherent nature, makes it insensitive to tamoxifen treatment [48]. Experiments using ERα+ breast cancer cell lines support this concept since they have demonstrated the presence of an ERα− subpopulation within ERα+ cell lines, and that these populations are tamoxifen- and chemotherapeutic-insensitive [49, 50]. The second potential mechanism is that ER+ cancer (stem) cells could undergo a phenotypic shift from an ERα+ state to an ERα− state. Phenotypic plasticity has been described in human breast and melanoma cancer cell lines [51–53], in mouse hematopoietic cell lines [54], and in human melanoma tumors serial-passaged in immune-deficient mice [51]. It remains to be established if a similar phenomenon is occurring in primary human breast tumors in vivo; if so, then it raises the worrying question if cancer stem cells can be effectively targeted [55]. Another mechanism which could account for the emergence of resistance to tamoxifen is a change in ERα+ cancer stem cells such that they still remain ERα+, but they have acquired intrinsic mutations of such that they have ligand-independent activation of ERα-related signaling and are resistant to this drug or an agonistic response to tamoxifen (reviewed in [47, 56]). This mechanism has been the predominant model under consideration over the years, although in hindsight, it may be difficult to distinguish this mechanism from the other two if the studies are not performed at a clonal level. However, not all ERα+ tumors become resistant to tamoxifen, and one could hypothesize that tumors that remain sensitive to tamoxifen originate in and are driven by ERα+ luminal progenitor cells [57]. It is known that approximately two thirds of ERα+ tumors contain a high proportion of proliferating ERα+ cells [6] and that estrogen has direct mitogenic effects on ERα+ breast cancer cell lines [58, 59]. Just as progesterone has a direct mitogenic effect on normal ERα+/PR+ progenitor cells [39], one can hypothesize that estrogen as well can have a direct mitogenic effect on this cell population, and that this population is expanded in cancer.

Conclusion

The information presented above summarizes how estrogen and progesterone appear to influence mammary epithelial stem and progenitor cells via indirect and possibly direct mechanisms, respectively, and how this underlying biology may explain the differences in a woman’s risk of being diagnosed with either an ERα− or ERα+ breast tumor at different stages of her lifetime. The largest unknowns in this model are the properties of the ERα+/PR+ progenitor cell in the normal mammary epithelium that may be a target for malignant transformation. It is essential that new markers that may be of use to prospectively isolate these cells from normal mammary tissue be identified. This information can then be used to further characterize cancer stem cells in ERα− and ERα+ breast tumors.