The Role of Steroid Hormones in Breast and Effects on Cancer Stem Cells

This review will discuss how the steroid hormones, estrogen and progesterone, as well as treatments that target steroid receptors, can regulate cancer stem cell (CSC) activity. The CSC theory proposes a hierarchical organization in tumors where at its apex lies a subpopulation of cancer cells endowed with self-renewal and differentiation capacity. In breast cancer (BC), CSCs have been suggested to play a key role in tumor maintenance, disease progression, and the formation of metastases. In preclinical models of BC, only a few CSCs are required sustain tumor re-growth, especially after conventional anti-endocrine treatments. CSCs include therapy-resistant clones that survive standard of care treatments like chemotherapy, irradiation, and hormonal therapy. The relevance of hormones for both normal mammary gland and BC development is well described, but it was only recently that the activities of hormones on CSCs have been investigated, opening new directions for future BC treatments and CSCs.


Introduction
The cancer stem cell (CSC) concept proposes a hierarchical organization of the cells within a tumor, where only a small subset of cells, the CSCs, drives and sustains tumor growth. In preclinical studies using breast cancer models, CSCs are defined as self-renewing cells that can propagate the tumor, which makes them very important in the processes of tumor recurrence, metastasis, and resistance to therapy. These roles make them an important therapeutic target [1].
Other strategies have also been used to identify BCSC enriched populations. Mammosphere formation, high aldehyde dehydrogenase (ALDH) activity, and CD49f or CD133 expression are properties that have been utilized to isolate CSCs ( Table 1). The mammosphere colony assay tests the capacity of BCSCs to survive in non-adherent culture conditions and to form spherical colonies, called mammospheres [17][18][19]. The activity of ALDH1, which retinaldehyde to retinoic acid, is detected by an enzymatic assay (ALDEFLUOR) and flow cytometry [20]. The proportion of cells with ALDH1 expression in breast cancer has been shown to correlate with poor prognosis [20][21][22]. Finally, CD49f and CD133 (Table 1) have recently been shown to enrich for CSCs in chemotherapy resistant triple negative and endocrine- At the present time, the most robust enrichment for BCSCs is achieved through the use of CD44 + /CD24 −/lo and ALDH+. These two cell populations have been demonstrated to mark BCSCs in different states and with gene expression resembling either mesenchymal (CD44 + /CD24 −/lo cells) or epithelial characteristics (ALDH+ cells) [24]. A small overlapping population of cells which is both CD44 + /CD24 −/lo and ALDH+ was identified, which suggested that BCSCs possess cellular plasticity and can dynamically switch between mesenchymal and epithelial states. The epithelial-mesenchymal transition and vice-versa (mesenchymal-epithelial transition) can be driven by the tumor microenvironment, with hypoxia or transforming growth factor beta playing key roles in this [25,26]. It is likely that other signaling factors that have been reported to regulate BCSC activity, such as hormones, will influence this. Herein, we discuss the regulation of BCSC function by the steroid hormones, particularly estrogen and progesterone, and their antagonists [22].

Estrogen and BCSCs
Estrogen promotes mammary epithelial cell proliferation and is therefore critical for normal breast development, but it also stimulates breast tumor growth through the estrogen receptor (ER) [27]. Estrogen binds to its receptors, ERα and ERβ, which are nuclear ligand-activated transcription factors, to modulate the transcription of target genes [28]. The effects of estrogen in the breast epithelium are mainly mediated by ERα, which has a higher affinity to 17β-estradiol, the physiological form of estrogen, than does ERβ [29]. Transcription factors need nuclear receptor co-regulators to mediate their action on target DNA sequences; in this case, ER signaling is dependent on FOXA1 expression, which promotes local DNA unwinding facilitating the access of ER to DNA [30].
Around three out of four breast tumors express ERα. Its expression is associated with luminal differentiation markers and with a more favorable breast cancer prognosis and is the most important breast cancer predictive factor for endocrine responsiveness [31,32]. Exposure to high levels of estrogen during women's lifetime is established to be associated with increased risk of postmenopausal breast cancer [33]. However, exogenous estrogen used as hormone replacement therapy may reduce the risk of invasive breast cancer and breast cancer-specific mortality in postmenopausal women [34]. This paradoxal effect of the role of estrogen in breast cancer initiation and progression might be explained by the different impacts of estrogen on different breast cancer cell types. On one hand, the pro-proliferative function of estrogen in ERα-positive breast cancer cells has been well characterized, but on the other hand, literature detailing the effects of estrogen on breast cancer stem cell (BCSCs) remains relatively scarce [35].
These effects are proposed to occur indirectly via paracrine mechanisms since BCSCs (CD44 + CD24 −/lo and ALDH + cells) are mostly ERα-negative [36][37][38]. It has been reported that treatment of CSC-enriched mammosphere population with estrogen decreases the proportion of BCSCs in ERαpositive breast cancer cells as a result of downregulation of embryonic stem cell genes [39]. This observation could in theory explain the better prognosis of ERα-positive tumors [40]. But it also has been shown that ERα-positive breast cancer cells can secrete FGFR and EGFR ligands in response to estrogen, which can act as paracrine mediators to promote CSC activity and expand the fraction of CD44 + CD24 −/lo cells [41,42]. In contrast, Axlund et al. reported that estrogen does not change cancer stem/progenitor cell properties on its own [43]. The reasons why some data show a protective effect of estrogen whereas others show that it can enhance cancer cell growth are not yet clear, but likely are related to other underlying differences in the tumor models used in the studies. More recently, despite the fact that BCSCs do not express the classical ERα, estrogens have been suggested to act directly on BCSCs through the ERα36 variant and ERβ. ERα36 variant activates mitogenic signaling via the AKT/ GSK3β pathway and is essential for the maintenance of CD44 + CD24 −/lo cells of two ER-positive breast cancer cell lines [44]. Ma and colleagues identified ERβ expression to be associated with stem cell markers CD44 and ALDH1 and also to be important for mammospheres formation [45]. Interestingly, ERβ gene expression has been reported to be upregulated in FACS sorted human breast stem cells [46••] compared to the total tumor cell population.
The complex implications of estrogen signaling in human breast cancer cells with stem-like characteristics indicate that further studies are needed to fully elucidate the effects of estrogens on BCSCs. Standardization of experimental conditions is warranted since the use of different BCSC markers, models, or culture conditions alters the analysis of CSC activity.

Anti-estrogen Drugs and BCSCs
Anti-estrogen therapies are used for breast cancer treatment of ER-positive tumors in both the adjuvant and metastatic settings. The principal classes of drugs are selective estrogen receptor modulators (SERMs, e.g., tamoxifen) and downregulators (SERDs, e.g., fulvestrant) as well as aromatase inhibitors, that reduce estrogen synthesis [47]. Since BCSCs are mostly ER-negative, they are not targeted by anti-estrogen therapies and several publications have reported that these therapies enrich for cells with BCSC characteristics.
Tumors treated with letrozole (aromatase inhibitor) increased in CD44 + CD24 −/lo mammosphere forming cells [48]. Piva and colleagues reported that tamoxifen-resistant MCF-7 cells have increased CD44 + CD24 −/lo and ALDH+ populations and form more mammospheres than the parental cells. In addition, they established that expression of the embryonic stem cell marker SOX2 and consequent activation of WNT signaling pathway was key for BCSCs survival after tamoxifen treatment [49]. Another study showed that ERα36 promotes tamoxifen resistance by increasing the proportion of CD44 + CD24 −/lo cells and mammosphere-forming cells [44]. Our group has shown that BCSCs (ALDH+ cells) are enriched following anti-estrogen treatment of breast cancer cells both in vitro using patient samples and in vivo using patient-derived xenografts. We also found that ALDH+ cells have high expression of JAG1 ligand and NOTCH4 receptor and that high ALDH1 expression predicts anti-estrogen resistance in women treated with tamoxifen [38]. Recently, two different studies from Sansone and colleagues demonstrated how the transfer of miR-221 or full mitochondrial DNA from cancer-associated fibroblasts to breast cancer cells through circulating extracellular vesicles could promote an exit from dormancy of BCSCs (CD133+) leading to endocrine therapy resistance [50,51].
Together, these findings suggest that inhibition of estrogen signaling in breast cancer cells may lead to an increase of the proportion of BCSCs. It still needs to be addressed whether this phenomenon occurs through selection and survival of the ER negative BCSCs, through induction of BCSCs characteristics in the ER+ cells, or by both processes. Either way, we hypothesize that BCSCs that survive anti-estrogen treatments can enter a dormant state and eventually re-initiate tumor growth, sometimes several years after the therapy. Further interest in this field has given rise to several clinical trials directly targeting CSCs (via recognized markers like ALDH) or using different signaling pathways linked to CSCs ( Table 2).

Progesterone and BCSCs
Progesterone plays a pivotal role in lobuloalveolar development of the mouse mammary gland during pregnancy [52,53]. In premenopausal women, breast epithelial cell proliferation is highest in the luteal phase of the menstrual cycle during maximum progesterone secretion [54,55]. Progesterone signaling is mediated by the progesterone receptor (PR), expressed as two isoforms (PRA and PRB) that are only different by a third activation domain on the 5′ end of PRB [56]. Importantly, the ratio of these two isoforms is key in the normal development of the mammary gland [57]. Further evidence from isoform-specific murine mutants demonstrates that mammary gland morphogenesis is linked to PRB, whereas PRA plays a prominent role in the ovarian homeostasis [58,59]. The gene expression patterns while largely overlapping indicate that PRB can regulate gene expression of more genes in comparison to its counterparts [60].
In normal human breast cells, stimulation with progesterone in matrix-embedded culture increases bi-potent cell numbers [61]. Evidence in mouse models corroborates that progesterone and PR signaling drives mammary gland development by expansion of the mammary stem cell population; this signaling is also appropriated in carcinogen-induced mammary tumor formation [62][63][64]. In established cancer cell lines, progestin administration leads to an increase of progenitor cells and CSC markers [65]. In the normal breast epithelium, the ratio of the PR isoforms remains balanced, but this is disrupted in the cancer setting, favoring the expression of PRA [61,66]. The increased risk of developing breast cancer has been linked to atypical hyperplasia [67] which often exhibits loss of PRB, equally, altered ER expression, sole PRA expression, and preferential PRB loss is also reported in the normal breast tissue of women with germline BRCA1/2 mutations [68]. Such women demonstrate PR isoform imbalance and double the circulating progesterone levels compared to matched controls; however, the cause and significance of these findings remain obscure [69].
During mammary gland expansion, PR mediates proliferation via paracrine signals, including RANKL (receptor activator of nuclear factor-κB ligand) and Wnt4. These signals are secreted from PR+ sensor cells and act on PR− progenitor cells, expressing RANK and the Wnt receptors Frizzled and LRP5/6 [63,64]. In multiple rodent models, deletion or inhibition of PR or the RANK/RANKL pathway results in significant reduction in mammary carcinogenesis [62,63,70,71]. More recently, Nolan and colleagues have also shown the potential of RANKL as a therapeutic target in a Brca1deficient mouse model, while in normal breast tissue of BRCA1-mutation carriers, identifying luminal RANK(+) progenitors that are highly proliferative and bear a molecular signature similar to that of basal-like BC, this indicates RANKL inhibition as a promising strategy in the prevention setting [72•].
One of the mediators of progesterone-induced stem/ progenitor cell functions in normal mammary gland is CXCR4/CXCL12 [73]. Signaling by progesterone occurs in a paracrine manner on luminal cells expressing CXCL12 while CXCR4 expression is also induced in both basal and luminal PR− cells. Inhibition of CXCR4-CXCL12 signaling axis can arrest the progesterone induced expansion of mammary stem/progenitor cells. Ginestier and colleagues translated the inhibition of CXCR1 with either a specific blocking antibody or by methanesulfonamide (a CXCR inhibitor known as Reparixin), in which both depleted the CSC population of two BC cell lines in vitro and in vivo [74]. This approach is currently under evaluation in the clinic, using reparixin in both early and advance breast cancer ( Table 2). A further effect of progesterone is the secretion of growth hormone (GH) in human breast epithelial cells, driving proliferation of the stem/progenitor breast cells expressing growth hormone receptor (GHR) [75].
Similarly, the increase in CK5 + cell population (linked to tumor-initiating properties and therapy resistance) and CD44 hi or CD44 + CD24 − BCSCs has been linked to progesterone in several ER+PR+ cell lines but particularly in T47D cells, which have high PR levels, through gene amplification, even in the absence of estrogen [65,[76][77][78]. In cell lines where PR expression is still dependent on estrogen, co-stimulation with CBR clinical benefit rate, BORR best overall response rate estrogen and progesterone is required, while estrogen alone was not able to induce BCSCs. In terms of potential mechanisms, reports have shown that PR signaling inhibits the expression of miR-29 and miR-141, while de-repressing KLF4 and STAT5A, respectively [77,79]. Both studies showed expansion of the CK5+/CD44+ CSC population with an increase in colony formation and in vivo tumor initiating capacity. KLF4 is a transcription factor required for maintenance of both BCSCs [80] and pluripotency in embryonic stem cells [81] whereas STAT5A is a transcription factor that regulates the mammary luminal progenitor population [82]. The maintenance of leukemic stem cells heavily depends of BCL6 expression while also essential for progesterone-induction of CK5+ cells in luminal breast cancer [83]. This progesterone-induced expression of BCL6 is suppressed by prolactin, further demonstrating the interplay taking place in hormonal signaling in the regulation of BCSCs [84]. The paracrine signaling taking place in the normal mammary gland between PR+ and PR− cells may indeed be acting in the same fashion with PR-BCSCs. Furthermore paracrine signaling of non-endogenous overexpressed RANKL in human breast cell lines increases the CD44+CD24− BCSC pool, promoting tumor initiation and metastasis [85]. However, despite strong preclinical data, clinical trials of denosumab, a monoclonal antibody targeting RANKL, have not translated to any improvement in cancer specific survival despite their valuable role in reducing skeletal complications from bone metastases. Altogether, this large body of evidence indicates that the expansion of both normal and BCSC is largely or in part driven by progesterone, although the exact mechanisms remain to be elucidated. Inhibition of PR directly or its paracrine/downstream mediators could translate to rational drug targets for breast cancer prevention and therapy.

Anti-progesterone Drugs and BCSCs
The Women's Health Initiative study reports that combination of estrogen with progestin (synthetic progesterone derivative), but not estrogen alone, was associated with an increased breast cancer incidence and mortality [86]. Tumorigenesis in the mammary gland can be attributed to the effects of progesterone signaling expanding the stem cell pool, which may transform to BCSCs and eventually lead to the formation of ER + / PR + tumors [87]. Recent reports in vitro have shown that natural and synthetic progestins can increase CSC-related markers ALDH high and CD44 high (Table 1) and that this enrichment of a subpopulation of cancer cells may be of functional significance in the development of BC in vivo [65]. The potential of progestin modulators as anti-tumor agents has recently been addressed using a patient derived xenograft model of breast cancer; investigators showed ulipristal acetate, a selective progesterone receptor modulator translated to significant anti-tumor effect, with reduction in Ki67 and Cyclin D1 [88].
During much of the 1990s, a great investment in antiprogestins as therapeutic agents was seen. Several trials were initiated for BC and other indications, as monotherapies and or in combinations. However, despite much interest, no antiprogestin is currently used as the recommended standard of care in any cancer setting either through lack of activity or tolerability. An example of these was the onapristone phase I trials which showed liver function test abnormalities, halting its clinical development [89,90]. Recent years have seen a new series of clinical trials (Table 3) using anti-progestin drugs like mifepristone and onapristone (the latter, now administered in a new formulation to avoid previous observed hepatoxicity) in breast cancer and other solid tumors [91][92][93][94][95]. Based on recent research literature, these drugs may target BCSCs in ER + /PR + tumors; although hypothetical, this merits further investigation. Trials in the prevention setting are also investigating the effects of ulipristal acetate, a selective progesterone receptor modulator, assessing proliferation and CSC markers in normal breast tissue [94]. New evidence on the complex interaction of estrogen and progesterone now elucidates the co-regulation of proliferative signaling under both steroid hormones. Progesterone acting through PR is able to ameliorate the effects of estrogen by reducing it activation of downstream effectors [96••]. This may offer and explanation as to why double positive endocrine tumors (ER + /PR + ) are classified as less aggressive than single ER + breast cancers, translating to better prognosis [96••, 97]. A hypothesis currently being evaluated in clinical trials investigates the potential benefits of PR agonists, as single agents in improving and prolonging response or in combination with cholecalciferol or letrozole ( Table 3). The renewed interest in PR as potential therapeutic target may hold clinical benefits by either modulating ER co-regulations by progestins or by reducing the progenitor pool via PR signaling. Despite new advances and insights into the effects of PR, more clinical work is need to validate the preclinical data.

Conclusions
In breast cancer, both estrogen and progesterone signaling have effects on CSC activity. BCSCs are reported to be low or negative for steroid hormone receptors, and therefore, effects must be indirect, mediated through paracrine or juxtacrine cell-cell signaling (Fig. 1). It remains possible that there is a degree of autocrine signaling downstream of hormones that regulates some BCSCs. The effects of estrogen and progesterone have only been partly described in cancer, for progesterone particularly; there is more data from normal mammary epithelium than from cancer. For estrogen, there is evidence that following in vitro treatment of serum-starved DLT dose limiting toxicity, maximum tolerated dose, PFS progression-free survival, CBR clinical benefit rate, BORR best overall response rate, ORR objective response rate, OS overall survival. * Trial designed in the prevention setting breast cancer cells, CSC activity is upregulated and that this is regulated by EGF, FGF, or Notch1 receptors, indicating indirect, paracrine or juxtacrine signaling between cells (Fig. 1). In contrast, anti-estrogens, such as tamoxifen or fulvestrant, block direct estrogenic effects on cell proliferation, and indirect signals to the ER− BCSCs. Surprisingly, however, tamoxifen can actually increase BCSC activity in mammosphere colony culture [38,39,98], and more recently, the same has been confirmed for both tamoxifen and fulvestrant in vivo [38]. In breast cancer, both estrogen and progesterone signaling have effects on CSC activity. BCSCs are reported to be low or negative for steroid hormone receptors, and therefore, effects must be indirect, mediated through paracrine or juxtacrine cell-cell signaling (Fig. 1). It remains possible that there is a degree of autocrine signaling downstream of hormones that regulates some BCSCs. The effects of estrogen and progesterone have only been partly described in cancer, for progesterone particularly; there is more data from normal mammary epithelium than from cancer. For estrogen, there is evidence that following in vitro treatment of serum-starved breast cancer cells, CSC activity is upregulated and that this is regulated by EGF, FG,F or Notch1 receptors, indicating indirect, paracrine or juxtacrine signaling between cells (Fig.  1). In contrast, anti-estrogens, such as tamoxifen or fulvestrant, block direct estrogenic effects on cell proliferation and indirect signals to the ER− BCSCs. Nevertheless, as mentioned previously, tamoxifen can actually increase BCSC activity [38].
Acknowledgments The authors would like to apologize to those authors whose work was not cited due to space limitations.
Funding Information The authors' work is supported by funding from Breast Cancer Now.

Compliance with Ethical Standards
Conflict of Interest Dr. Denis G. Alferez, Dr. Bruno M. Simões, Dr. Sacha J. Howell, and Dr. Robert B. Clarke declare that they have no conflict of interest.
Human and Animal Rights and Informed Consent This article does not contain any studies with human or animal subjects performed by any of the authors. Fig. 1 Representation of juxtacrine and paracrine signals involved in estrogen and progesterone regulation of BCSCs. Estrogen (E2) and progesterone (Pg) bind to their receptors along with nuclear transcription factors, respectively, regulating expression of downstream target genes. Estrogen sensor cells (non-BCSCs) increase transcription of EGF (epidermal growth factor), AREG (amphiregulin), TGFα (transforming growth factor α), and FGF (fibroblast growth factor), which will signal to the BCSCs through the EGFR and FGFR receptors. Non-BCSCs can also signal with BCSCs via Notch signaling. Progesterone sensor cells (non-BCSCs) upregulate the transcription of several key signaling factors. Regulation of BCSCs via Pg may occur via activation of RANK/RANKL, Wnt receptors/Wnt4, CXCR4/CXCL12, and GHR/ GH paracrine signaling (dashed lines). Estrogen and progesteroneinduced signals can be blocked by anti-estrogens (e.g., tamoxifen and fulvestrant) and anti-progesterone drugs (e.g., mifepristone and onapristone) Open Access This article is distributed under the terms of the Creative Comm ons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.