Stem cells in adult structures have been defined by their ability for self-renewal and for generating a differentiated progeny. In the mammary gland, DeOme and colleagues demonstrated that fragments of different parenchymal portions were able to generate fully functional mammary outgrowths in mice, forming ductal and lobuloalveolar structures composed of epithelial cells and myoepithelial cells 27. This concept was further developed by Kordon and Smith 28, who demonstrated that the progeny from a single cell may comprise the epithelial population of a fully developed lactating mammary outgrowth in mice. The development of the complete mammary tree from a small portion of a duct or from single cells therefore attests to their multifaceted potential.

It was not known, however, whether these progenitor/stem cells would be capable of initiating cancer when exposed to a carcinogenic agent. This issue was addressed by Russo and colleagues 293031, who demonstrated that cancer started in TEBs present in the mammary gland of young virgin rats. The analysis of these structures by electron microscopy allowed one to characterize their cellular composition based upon cell and nuclear size, the nuclear–cytoplasmic ratio, the amount of chromatin condensation, the electron density of the cytoplasm, the number and distribution of organelles, and the presence or absence of Mg²⁺-dependent and Na⁺K⁺-dependent ATPases. Based upon these criteria, in addition to myoepithelial cells, three types of epithelial cells were identified: light cells, intermediate cells, and dark cells 3031. Dark cells were found to be the predominant type in TEBs, intermediate cells and myoepithelial cells were present in significantly lower percentages, and light cells were only occasionally seen so their percentage was combined with that of intermediate cells. The analysis of the DNA labeling index revealed that all the cell types proliferated, although at different rates, depending upon the type of cells and their type of location within the mammary gland tree. Cell proliferation was maximal in intermediate cells located in TEBs, being significantly lower in dark cells and myoepithelial cells found in the same location. High cell proliferation was associated with greater incorporation of H³-DMBA and with a progressive dominance of intermediate cells in DMBA-induced intraductal proliferations and in ductal carcinomas 531. These results indicated that intermediate cells were not only the targets of the carcinogen, but also the stem cells of mammary carcinomas.

Further work by Bennett and colleagues demonstrated that intermediate cells isolated from DMBA-induced mammary tumors originated two cell types in culture 32: the dark cell, representing a terminally differentiated cell or a class in transition to differentiation; and intermediate cells, which could represent an undifferentiated or stem cell, a progenitor of dark cells and myoepithelial cells. Rudland and colleagues 33 isolated and characterized from the normal rat mammary gland and from DMBA-induced mammary adenocarcinomas epithelial cells that were cuboidal and gave rise to a mixture of cuboidal and spindle-shaped cells resembling fibroblasts. In confluent cultures, cuboidal cells acquired the morphology of a third type of cells, which were dark, polygonal and with many small vacuoles, resembling the dark cells ultrastructurally described by Russo and colleagues 31. Chepko and Smith 34 differentiated two division-competent cell populations in the murine mammary epithelium: a subset of 'large light cells' structurally and functionally compatible with early stages of secretory differentiation; and 'small light cells' that were the least differentiated, suggesting that the large light cells were a direct precursor to terminally differentiated cells, both secretory and myoepithelial.

A shift from the pioneering work performed for characterizing, by morphology and by in vitro behavior, the progenitor/stem cells started with the search for immunocytochemical and genomic markers. Smith and colleagues 35 utilized the expression of keratin 6 and keratin 14 in mouse mammary epithelium for defining subsets of morphologically distinct luminal mammary epithelial cells with kinetic properties expected for latent mammogenic stem cells. Keratin 6 was confined to a small number of mammary epithelial cells found in the growing end buds and among the luminal epithelium, whereas keratin 14 was expressed in basally located fusiform cells as the myoepithelial cells. These authors emphasized the usefulness of these markers for identifying mammary epithelium-specific primordial cells.

Stingl and colleagues 3637 utilized new molecular markers (Table 1) for selecting subpopulations of cells with distinct differentiation potential. They described bipotent human mammary epithelial progenitor cells based on the expression of epithelial-specific antigen (ESA), sialomucin 1 (MUC1), common acute lymphoblast antigen (CALLA/CD10), and α-integrin, in combination with exclusion of rhodamine dye. Hebbard and colleagues 38 observed that CD44, a member of the family of cell surface proteins that is expressed in breast carcinomas, is also expressed in the normal mammary gland. CD44 expression in rodents is first detected at puberty, and thereafter it is regulated by the estrous cycle; the expression disappears during lactation and reappears during involution, suggesting that the expression of this protein is a marker of a stem cell. Novel studies in the mouse mammary gland have identified stem cells in TEBs and ducts by pulse-labeling primary mammary epithelial cells with fluorescent TRITC-cell linker membrane label and with BrdU 39. The cells were then transplanted into cleared juvenile syngenic mammary fat pads, in which they were identified as long-lived, label-retaining mammary epithelial cells in mammary ducts that were actively growing or static. That study demonstrated that label-retaining mammary epithelial cells are stem cells and that their progeny (transitional cells) are arranged as transitional units, and also demonstrated that both cells express Zonula Occludens-1 and α-catenin proteins, data suggesting that transitional units retain stem cells.

The study of markers for other stem cells has been useful in the identification of mammary stem/progenitor cells. Stem cell antigen 1 (Sca1) (Table 1) was first described in mice as a hematopoietic stem cell antigen 40. Welm and colleagues detected in the luminal epithelium of mice a Sca1⁺ cell population that is enriched for functional stem/progenitor cells 41. These cells are BrdU label-retaining, lack expression of differentiation markers, and are progesterone receptor-negative. The Sca1⁺ population also shows 'side population' (SP) properties, a characteristic first defined in bone marrow cells 40, as cells with Hoechst dye-effluxing properties that have phenotypic markers of multipotential hematopoietic stem cells. It has been proposed that the protein responsible for that phenotype is breast cancer resistance protein (BCRP1), suggesting that the expression of this protein could serve as a marker for stem cells from various sources 42. Mammary epithelial cells with SP properties were also identified in the human mammary gland. Alvi and colleagues 43 showed that 0.2–0.45% of both human epithelia and mouse epithelia were formed by distinct SP cells. These cells generated ductal and lobuloalveolar structures when transplanted into murine cleared mammary fat pads. The SP cells had a high expression of BCRP, Sca1, and telomerase catalytic subunit, and had low levels of differentiated markers for luminal cell types (epithelial membrane antigen and cytokeratin 19) and myoepithelial cell types (cytokeratin 14). These cells were detected in all human breast samples studied, but their presence was not correlated with age, parity, contraceptive use, or day of menstrual cycle.

Further investigations identified new markers that may be specific for the human stem/progenitor cells. Gudjonsson and colleagues isolated a cell line derived from human mammary cells expressing ESA and lacking MUC expression that could give rise to both luminal epithelial cells and myoepithelial cells in culture 44. One single ESA⁺/MUC^-cell had the ability of generating a terminal ductal-lobular unit-like structure in basement membrane gel, similar to that formed when the cell line was implanted in mice. In contrast, an ESA⁺/MUC⁺ subpopulation was differentiated and luminal epithelial-restricted without stem cell properties (Table 1).

Wicha and colleagues developed a system to enrich the population of human mammary progenitor/stem cells by culturing them in suspension, where they formed 'non-adherent mammospheres' 45. These structures were able to differentiate along all three mammary epithelial lineages and to clonally generate complex functional structures in three-dimensional culture systems. Cytological and immunocytochemical analysis of secondary mammospheres revealed that these structures contained cells positive for α6-integrin, cytokeratin 5 (which was widely expressed), and CD10; ESA-positive and cytokeratin 14-positive cells were less frequently found. Muc1, α-smooth muscle antigen, and cytokeratin 18 were not detected. In addition to cells, mammospheres contained extracellular material. However, immunostains for fibronectin and collagen IV, the classical components of adult gland extracellular material, were negative – although ~20% of the mammospheres stained positive for laminin. In contrast, abundant expression of the embryonic extracellular material components tenascin and decorin, was detected in mammospheres 45. Moreover, the comparison of the genomic profile of undifferentiated cells from mammospheres with that of differentiated cells cultured on collagen identified gene candidates for stem/progenitor cell markers. Some of these genes were already described as involved in stem/progenitor cell-specific functions or in regulation of self-renewal, and abnormal expression of some of the genes has been correlated with breast cancer development such as proliferation, cell survival, and invasion (Table 1).

The identification of the stem cell and of its role in the development and differentiation of the mammary gland from birth to senescence requires an understanding of the effect of estrogen and its cognate ligand receptor alpha (ERα) in these processes. The importance of the role played by the ERα in mammary gland development has been highlighted by the development of the αERKO mouse 46. At birth, the mammary gland of intact animals consists of a rudimentary ductal tree that develops and fills the stroma of the gland in response to increased ovarian estrogen at puberty. The mammary gland of αERKO females does not grow beyond the rudimentary ducts, illustrating the role of estrogens in ductal elongation.

The importance of active ductal growth driven by estrogen has been further emphasized by the higher susceptibility of the breast to be transformed during a 'high-risk' window in the lifespan of a female encompassed between menarche and a first full-term pregnancy 5. This period is characterized by rapid ductal growth and active proliferative activity of the mammary epithelium of Lob 1. These structures are composed of a rapidly proliferating epithelium that has a high content of ERα-positive and progesterone receptor (PR)-positive cells. With the progressive maturation of Lob 1 to Lob 2, Lob 3, and Lob 4 there is a progressive decrease in the percentage of proliferating cells, a reduction in the percentage of cells positive for steroid hormone receptors, and a reduction in the susceptibility of the cells to be transformed by chemical carcinogens 47. These data indicate that the stem cells that originate the mammary tree as well as cancerous lesions are located in a specific compartment of the mammary parenchyma, namely Lob 1 or the terminal ductal lobular unit; these are the cells that were called Stem cells 1 by Russo and Russo 26.

Supporting studies by Petersen and colleagues 48 have shown that a subset of suprabasal breast luminal epithelial cells that are able to generate themselves, as well as differentiated luminal epithelial cells and myoepithelial cells, and are able to form terminal ductal lobular unit-like structures are distinguished by expression of cytokeratin 19. The suprabasal population of breast stem cells consists of undifferentiated 'intermediate' cells with Hoechst dye-effluxing SP properties. These cells lack expression of myoepithelial and luminal apical membrane markers such as CALLA and MUC1. The cells are rich for ERα-positive cells and express several-fold higher levels of the ERα, p21 (CIP1), and Msi1 genes than non-SP cells (Table 1). These cells also form branching structures in matrigel that included cells of both luminal and myoepithelial lineages. These data suggest a model where scattered steroid receptor-positive cells are stem cells that self-renew through asymmetric cell division and generate patches of transit-amplifying and differentiated cells 4950.

ERα /PR⁺ breast cancers exhibit loss of the two key regulators of asymmetric cell division, Musashi-1 and Notch-1, and thus they may arise from symmetric division of the ERα /PR⁺ stem cell 49. These data are supported by the observations of Russo and colleagues that epithelial cells of the Lob 1 co-express ERα, PR, and the proliferation marker Ki67 47, suggesting that these cells could originate ERα-positive tumors. However, these cells represent less than 1% of the total cell population whereas the majority of ERα/PR⁺ cells do not express the proliferation marker, an indication that the cells that contain the receptors are not capable of proliferating. The findings that proliferating cells are different from those that are ERα-positive and PR-positive cells support data that indicate estrogen controls cell proliferation by an indirect mechanism.

Further support is the finding that when Lob 1 of normal breast tissue are placed in culture, they lose the ERα-positive cells, indicating that only proliferating cells that are also ERα-negative can survive, representing a type of stem cell that may originate ER-negative tumors 47. The fact that the majority of proliferating breast epithelial cells do not express ERα and PgR could explain Clayton and colleagues' 51 data that cells characterized as human mammary stem cells present ESA expression, Hoechst dye exclusion, low levels of MUC-1 and CALLA, and lack detectable expression of ERα and ERβ. Cells expressing that phenotype had high cloning efficiency in culture from a single cell, generating mixed colonies containing luminal cells and myoepithelial cells.

As discussed in the previous sections, the identification of a putative breast stem cell has in the past decade reached a significant impulse, and several markers also reported for other tissues have been found in the mammary epithelial cells of both rodents and humans (Table 1). There are, however, four main issues that require further investigation. The first is to determine whether the stem cells or progenitor cells that give origin to a complete mammary gland are the same cells that are affected by a carcinogenic process. Another important point that needs further clarification is the role of ERα as a marker of the stem cells. The third issue is the need to be extremely careful in validating conclusions drawn from in vitro studies by properly confirming them with in vivo data, in which numerous factors (such as age of the donor, reproductive history, number of samples studied, and consideration of intrinsic variability from sample to sample) are so important, but seldom considered in the major publications dealing with the stem cells in the mammary gland. Finally, the data reported in the literature tend to support the concept that the mammary gland contains Stem cells 1 that could be the progenitor of the differentiated breast or could be the site of origin of a neoplastic process. Supporting this concept is the fact that all the genes ascribed to the stem cells in the mammary gland are involved in more than one function of the normal breast as well as the malignant breast.

Epidemiological studies in humans and experimental carcinogenesis models have provided wide evidence of the protective effect of pregnancy from breast cancer development 23456789101112. Russo and colleagues 5101752 have postulated that the mechanism of pregnancy-induced protection is mediated by the induction of mammary gland differentiation driven by the hormonal milieu of pregnancy, which creates a specific genomic signature in the mammary gland that makes this organ permanently refractory to carcinogenesis. Alternative explanations attributed the protective effect of pregnancy to changes in the environmental milieu 53 and/or alterations in the immunological profile of the host 7. A further refinement of the hypothesis of how pregnancy could be affecting cancer susceptibility through induction of differentiation of the mammary gland was first proposed by Russo and Russo 26, who postulated that Lob 1 and the TEB found in the breast of nulliparous women or of young virgin rats, respectively, had not completed their differentiation into Lob 2, Lob 3, and Lob 4, retaining a high concentration of stem cells (Stem cells 1), which are susceptible to undergo neoplastic transformation when exposed to a carcinogenic agent (see previous section and Fig. 2) 26. After the postmenopausal involution of the mammary gland, the architecture of the parous breast is similar to the nulliparous breast, containing predominantly Lob 1 composed of Stem cells 2, an epithelial cell population that is refractory to transformation (Fig. 2).

It was further postulated that the degree of differentiation acquired through early pregnancy permanently changes the 'genomic signature' that differentiate Lob 1 of early parous women from that of nulliparous women, shifting the Stem cells 1 to Stem cells 2 that are refractory to carcinogenesis (Fig. 2). These cells were called Stem cells 2 because, after post-lactational involution, the mammary epithelium remains capable of responding with proliferation and differentiation to the stimulus of a new pregnancy; however, these cells are refractory to carcinogenesis, even though they are stimulated to proliferate and to regenerate the whole mammary gland. Stem cells 2 are characterized by having a genomic signature that has been induced by the first cycle of differentiation (Fig. 2).

Supporting evidence to this hypothesis has been generated during the past 8 years by Russo and colleagues as well as by other researchers. Recent studies by Smith and colleagues 545556 using transgenic whey acidic protein-driven Cre and Rosa 26-fl-stop-fl-LacZ mice provided evidence of a new mammary epithelial cell population that originates from differentiating cells during pregnancy; 5–10% of this parity-induced epithelium survives post-lactational involution after the first pregnancy. With successive pregnancies the population percentage increases, reaching 60% of the total epithelium in multiparous females. The parity-induced mammary epithelial cells (PI-MEC) are equivalent to Stem cells 2 as postulated by Russo and Russo 26 since these cells show capacity for self-renewal and contribute to mammary outgrowth in transplantation studies. PI-MEC can function as alveolar progenitors in subsequent pregnancies, and it is thought that they would be related to differences in response to hormonal stimulation and carcinogenic agents observed between nulliparous females and parous females 545556.

Several authors have focused on finding molecular changes as a mechanism of pregnancy-induced protection 5758596061626364 (Table 2). Russo and colleagues have found that the post-pregnancy involuted mammary gland exhibits a genomic signature characterized by elevated expression of genes involved in the apoptotic pathways, such as testosterone repressed prostate message 2 (TRPM2), interleukin-1β-converting enzyme, bcl-XL, bcl-XS, p53, p21, and c-myc, which can be from threefold to fivefold upregulated 575865 (Table 2). The activation of programmed cell death genes occurs through a p53-dependent process, modulated by c-myc and with partial dependence on the bcl2-family related genes. In addition, inhibin A and inhibin B, heterodimeric non-steroidal secreted glycoproteins with tumor suppressor activity, are also upregulated 57586566. Genes whose level of expression progressively increases with time of pregnancy, reaching their highest levels between 21 and 42 days post-partum, are those coding for a fragment of glycogen phosphorylase, AMP-activated kinase, bone morphogenetic protein 4, and vesicle-associated protein 1. The G/T mismatch-specific thymine DNA glycosylase gene is also increased fivefold in this model (Table 2).

These data indicate that the activation of genes involved in the DNA repair process is part of the signature induced in the mammary gland by pregnancy. These observations confirm previous in vivo findings that the ability of the cells to repair carcinogen-induced damage by unscheduled DNA synthesis and adduct removal is more efficient in the parous and animal mammary gland 17. In concordance with the studies of Srivastava and colleagues 57, Sivaraman and colleagues 61 observed that p53 can be implicated in the protective effect of parity, which can be mimicked by treatment of virgin rats with estrogen and progesterone. Studies by Medina and Kittrell 5960 in the same hormonal model reported that the function of p53 is required for the hormone-mediated protection of DMBA-induced mammary tumorigenesis in mice (Table 2).

Genomic analysis of the mammary gland of virgin rats treated with estrogen and progesterone at doses that have been reported to mimic pregnancy showed downregulation of certain growth-promoting molecules, whereas markers involved in cell cycle control or in the modulation of the transforming growth factor beta signaling pathway were upregulated in the post-treatment involuted mammary gland 62. In that study, an unknown non-coding RNA (designated G.B7) and RbAp46, which has been implicated in a number of complexes involving chromatin remodeling, were found to be persistently upregulated in the lobules of the regressed glands (Table 2). Using gene profile analysis, D'Cruz and colleagues 64 also observed downregulation of growth factors potentially involved in epithelial proliferation, as well as persistent upregulation of transforming growth factor beta 3 and several of its transcripts targets in the involuted gland of parous rats and mice (Table 2).

The proposed model of parity-induced specific changes 26 has been further confirmed by Ginger and Rosen 63, who reported that pregnancy induces multiple changes in the mammary epithelial cells, including nuclear accumulation of p53 and induction of whey acidic protein. During involution a large component of the epithelium is eliminated through apoptosis, and a specific subpopulation of epithelial cells survives this process. The involuted mammary gland has persistent changes in gene expression, nuclear localization of p53, and an altered proliferative capacity in response to a carcinogen. Pregnancy would induce epigenetic changes, such as chromatin remodeling, DNA methylation/demethylation, and histone modifications, affecting cell fate in the parous mammary gland. As depicted in Table 2, all the genes that have been attributed to Stem cells 2 seem to work in different functional pathways than those described for Stem cells 1 (Table 1).

Although further work needs to be carried out in order to better understand the role of Stem cells 2 and their interaction with the genes that confer their specific signature, collectively the data described present evidence that pregnancy, through the process of cell differentiation, shifts Stem cells 1 to Stem cells 2 – cells that exhibit a specific genomic signature that could be responsible for the refractoriness of the mammary gland to carcinogenesis.