Progesterone signals via the PR. The PR is expressed as two isoforms, progesterone receptor A (PRA) and progesterone receptor B (PRB). Both isoforms are derived from a single gene but regulated by two distinct promoters. In humans, E₂ stimulates expression of both PRA and PRB mRNA (31). This is in contrast with ovariectomized mice where PRA expression is stimulated by E₂ and inhibited by P₄. PRB levels are not affected by E₂ alone, but are stimulated by prolonged treatment with P₄ or by P₄ in combination with E₂ (32). Accordingly, in mice, PRA is the main PR isoform expressed during pre-pubertal stages and in adult virgins, while PRB expression increases only during pregnancy (33). Therefore, it has been suggested that in the mouse, the initial proliferative response of the mammary epithelium to P₄, leading to side branching is mediated by PRA, while PRB is needed for a proper lobular alveolar development during pregnancy (32). By contrast, in the normal human breast, both PRA and PRB are coexpressed in the same cells implying species-specific regulation of the isoforms (34). The PR isoforms have isoform-specific transcriptional activities on P₄-responsive gene promoters, resulting in a distinct target gene profile (35). PRs can activate gene transcription in multiple ways and direct binding to progesterone response elements (PREs) a process usually referred to as classical PR signaling. PR can also tether to other transcription factors, such as Stat5 in the regulation of RANKL expression. Finally, PR can mediate so-called “non-genomic” cytoplasmic signaling through interaction with ERα, thereby activating the Rous sarcoma proto-oncogene/(Ras–Raf–MEK–ERK) cSrc/ERK pathway (36).

In classical signaling, both PR isoforms function as ligand-induced transcription factors and contain distinct activation function (AF) domains essential for their transcriptional activity. Two activation domains are common to both PRA and PRB, AF1 and AF2. PRB, however, has an additional activation function domain-3 (AF3) domain localized within the PRB specific N-terminus (37) making it a stronger transcriptional activator than PRA (Figure 1) (38). Sequence motifs essential for AF3 domain function were shown to be highly conserved across mammalian species (39, 40). We have compared the activities of canine PR (cPR) isoforms to human isoforms (hPR) using luciferase constructs containing classical PREs, mouse mammary tumor virus (MMTV)-luciferase, and PRE2-luciferase. We have shown that canine PRA has an expected hPRA-comparable transcriptional activity, whereas canine PRB (cPRB) has low to absent transactivation potential. No differences were found regardless of background cell type such as Chinese hamster ovary cells, canine mammary cells, or human T47D cells in which the endogenous PR was knocked out. The transactivation potential of cPRB could be restored by replacing the cPRB specific N-terminus with the human sequence in a human B-upstream segment (hBUScPRB) chimera. Next, we made canine mammary cell lines with a doxycycline (dox)-inducible expression of cPRB, hPRB, or the hBUScPRB chimera. Transactivation potential on endogenous target genes was then assessed by gene profiling using canine cDNA microarray. In the absence of dox, no effect of P₄ incubation was seen, excluding signaling through P₄ membrane receptors. Only the combination of dox plus P₄ changed the expression of over 600 genes, both in hPRB- and hBUScPRB-expressing cell lines. Only a minority of these genes were influenced by cPRB (41, 42). These results indicate a very limited transactivation potential of cPRB on endogenous genes, thereby questioning its role in mammary gland development and carcinogenesis.

Although the PRA and PRB isoforms are usually equally present in epithelial cells of the human mammary gland, in advanced breast cancer, a predominance of PRA is common, indicating that PRB has a protective function. Patients with PRA-rich tumors or tumors with a high PRA:PRB ratio have a much faster recurrence than patients with PRB-rich tumors (43, 44). Predominance of PRA is especially evident in ductal carcinoma in situ (DCIS) and invasive breast lesions. It has been suggested that P₄ may also lead to transition of tumors from a luminal toward a basal phenotype (45). Germ line mutations in the genes breast cancer 1 (BRCA1) or breast cancer 2 are associated with a predominance of PRA expression (46). BRCA1 physically interacts with PR and inhibits its activity, in part, by preventing binding of the PR to the PRE and promoting the formation of a corepressor complex (47). Because this activity is lost in the context of BRCA mutant proteins, anti-progestins are recommended for tumor prophylaxis in BRCA mutation carriers (48). Mutation of BRCA1 results, moreover, in stabilization of the PR due to the loss of BRCA1-mediated PR ubiquitination and subsequent degradation (49). Interestingly, loss of BRCA1 also results in increased epidermal growth factor 1 (HER1) expression (50). We hypothesize that the combination of stabilization or enhanced PR expression and active HER1 signaling may stimulate phosphorylation of the PR by mitogen-activated protein kinase (MAPK) and specifically stimulate this mode of P₄ signaling.

Within the luminal epithelium of the mammary gland PR-positive cells act as sensors for circulating P₄ concentrations (Figure 2). Upon P₄ exposure, these cells secrete growth factors (RANKL, Wnt) that may stimulate recruitment and differentiation of stem cells (51). P₄ thus induces adult MaSC expansion (52) in mice, and this is also hypothesized to be a major site of carcinogenesis within the human breast (53). As well as an increase in MaSC, P₄ was also suggested to act directly on PR-positive cells and convert them to a hormone-receptor negative, more stem-like state (54). Cluster of differentiation (CD) CD44^HighCD24^Low cells have been reported as breast CSCs since Al-Haij showed that these cells can form tumors in mice (55). In humans, CD44^HighCD24^Low cells seem to have a higher tumorigenic capacity (56) and even in dogs these cells have a tumor-initiating capacity (57) and stem/progenitor cell properties (58). Cell markers for stem cells and progenitors in the human and murine mammary gland (59, 60) are summarized (Table 1).

These stem cells and progenitor cells are necessary in the mammary gland to ensure proper long-term maintenance of mammary tissue structure and function during puberty, pregnancy, and lactation, which may occur multiple times during the reproductive lifespan of an animal (60). Mouse MaSCs have the ability to resist anoikis and form floating colonies, the so-called mammospheres. Progenitor cells from the inner layer, the luminal cells, have high proliferative potential in colony-forming cell (CFC) assays (61, 62) and also have a high level of PR in humans (63). These high-level PR cells occur in poorly developed lobules of the mammary gland and can be characterized by lineage (Lin⁻), epithelial cell adhesion molecule (EpCam⁺), CD49f⁺, aldehyde dehydrogenase 1 (ALDH⁺), and human epidermal growth factor 3 (HER3⁺) (59). ALDH1 is associated with stem/progenitor cell properties in human mammary epithelial cells (HMECs) that lack expression of ER (64, 65). This is in contrast to the mouse where PR⁺ luminal cells are CD29^low, CD49f^low, CD24⁺, and also have ER expression (59). In the canine mammary, tumor cell line (CMT U229) CD49f⁺, CD24^low, CD44⁺ cells have been identified as tentative stem-like cells (66), with unknown ER/PR status.

Most data on hormone receptor status in the literature come from human mature luminal cells that mostly express ER and PR and mediate the proliferative effects of steroid hormones by paracrine signaling. However, E₂ and P₄ treatment also increase the MaSC population and in diestrus, when P₄ levels are the highest, the number of CD24⁺/CD29^hi mammary repopulating units increases 14 times in mice. Thus, steroids do influence stem cells, in both their proliferative and self-renewal abilities (52, 67). Oakes et al. suggest that a paracrine mediator is responsible for the proliferation and maintenance of human MaSCs, and a likely candidate is P₄-regulated RANKL (3).

The PR has two modes of mitogenic action in the mouse mammary gland; a cyclin D1-dependent stimulation of proliferation in PR⁺ cells and a RANKL-mediated paracrine action on nearby PR⁻ mammary epithelial cells (68). Steroid hormone-receptor positive cells release RANKL to neighboring stem cells in a paracrine manner. In a neighboring PR-negative luminal progenitor cell, RANKL can stimulate the transcription factor E74-like factor 5 and thus promote alveolar development (69). P₄ also drives the CD24⁺/CD29^lo/CD61⁺ luminal progenitor cell population to a CD24⁺/CD29^lo/CD61⁻ mature luminal cell differentiation, resulting in mature alveolar lineage expansion. Another possible mediator of paracrine signaling is neuregulin (Nrg1), which is produced in the basal epithelial cells and is a direct transcriptional target of tumor protein 63 (p63). P63 KO mice have several developmental defects such as no limbs, teeth, and mammary glands. Exogenous Nrg1 rescued the lactation through activation of human epidermal growth factor 4/Stat5 in neighboring mouse luminal epithelial cells (70). MaSCs not only receive signals in a paracrine manner indicating paracrine basal to luminal signaling but they can also signal back to regulate the luminal epithelium. This “teamwork” on breast development and homeostasis is only beginning to be unraveled (71).

Similar to steroid hormones P₄ and E₂, pituitary GH is also necessary for mammary gland development. GH deficiency impairs the mammary development in rats and mice. GH influences alveolar and duct development and limits side branching (72). In humans with Laron syndrome, a mutated GH receptor (GHR) affects mammary gland development. Patients with Laron syndrome are shorter, have no diabetes type II and their risk of getting all kinds of cancer is almost 0 (73). By contrast, patients with hypersecretion of GH (acromegaly) and individuals who are taller in height have a higher cancer incidence (18). A sustained exposure to steroid hormones is the best established risk factor for human breast cancer (74), and this also influences canine mammary gland proliferation and carcinogenesis (75, 76). Sustained exposure leads to nodules of lobular hyperplasia and simple or complex adenomas in the mammary gland (76). The proliferative effect on the mammary gland coincides with increased plasma GH concentrations. Dogs with endogenous high plasma P₄ concentrations or treated with exogenous MPA were found to have high plasma GH concentrations. We observed that the canine mammary gland produces GH locally after MPA administration (77). As well as finding immune reactive GH in the mammary gland and a steep decrease in plasma GH concentrations after complete mastectomy, we found expression of GH mRNA in both the canine and feline mammary gland (78). Next, we demonstrated that in the dog, mammary expression of GH was initiated at the same start site as GH from the pituitary (79) but the pituitary-specific POU domain transcription factor was absent from mammary tissue. Analysis of the GH promoter revealed a putative PRE (80) but experiments using a GH-promoter luciferase construct did not show a direct transactivation by P₄-activated PR in canine mammary tumor cell lines (81). In tumors, the staining intensity of the PR varied from no staining, normal nuclear staining to remarkable heterogeneous and perinuclear staining and cytoplasmic staining in spindle cells (82). The major pathways that were activated after prolonged MPA exposure in dogs were determined through gene expression studies. Both mammary tissue and cell lines were used to identify carcinoma-related expression profiles (83–85). Autocrine production of GH signals in human MCF7 cells in vitro are associated with a more invasive phenotype and in vivo with more aggressive tumors (86) indicating that GH activates the same pathways in humans as in dogs.

Growth hormone is a peptide hormone closely related to PRL that exerts its action through GHR present in the cell membrane as constitutively dimerized single transmembrane proteins (87). Upon binding of GH ligands, GHR dimers are transphosphorylated by JAK2 tyrosine kinases that result in activation of multiple downstream signaling pathways, including the STAT pathway, the MAPK pathway, the PI3K pathway, and the protein kinase C (PKC) pathway (88, 89) (Figure 3). Signaling through the JAK/Stat pathway is initiated by phosphorylation of Stat proteins by JAK2. Subsequent dissociation of Stat proteins from GHR is followed by their dimerization and translocation to the nucleus to activate target gene expression. Activation of Stat5A and B isoforms plays an important role in mammary gland development and results in transcriptional activation of multiple target genes including insulin-like growth factor 1 (IGF-1) and serum protease inhibitor 2.1 (spi2.1) (10, 90).

In the canine CMT-U27 cell line, a GHR-mediated growth effect was found as a consequence of increased cell survival by increased p-ERK 1/2 expression. This resulted in proliferation and an increased number of cells in the S and G2M phase (91). When GH binds to the GHR the tyrosine kinase JAK2 is activated, and this activated JAK2 phosphorylates various signaling mediators with the most important being p-ERK1/2 (MAPK). Activated MAPK signals via myelocytomatosis viral oncogene homolog (MYC) and c-AMP response element-binding protein (CREB) in the nucleus to activate the transcription machinery (91). This transcription machinery is also stimulated by the PI3K/protein kinase B (AKT) pathway where GH indirectly signals via IGF-1 enhancing the PI3K/mammalian target of rapamycin (mTOR) pathway (Figure 3).

Expression of GH mRNA has also been observed in human breast cancer specimens (18, 92). This GH may interact with GHR-positive cells that are found in some 90% of human DCIS lesions and in 4–19% of normal breast epithelial cells (18). These GHR⁺ cells can form GH-dependent mammospheres, suggesting that GH may stimulate stem/progenitors to enter the cell cycle. The subpopulation of GHR⁺ cells also contains more progenitor cells with bipotent and myoepithelial differentiation potential compared to the GHR⁻ cell population (18). Mammosphere-initiating cells from both human and mouse mammary gland can repopulate the cleared mammary fat pad in vivo (93) and both the same mammospheres and T47D human cancer cells can be stimulated by P₄ to produce GH. We also found this to be the case in mammary tissue of the dog (18, 82) indicating that there is a link in the mammary gland between P₄ stimulation, GH secretion, and GH/GHR activation with GH having a paracrine role (Figure 2). The effect is most likely paracrine because GHR and PR cells differ in EpCam activity in humans. PR-positive cells are Lin⁻, EpCam⁺, CD49f⁺, ALDH⁺, and HER3⁺, whereas GHR-positive cells are Lin⁻, EpCam^+/−, CD49f^high, and ALDH⁺ (18, 59). EpCam⁺ cells are luminal progenitor cells and EpCam⁻ cells are stem and early progenitor cells (18, 59). There is also a discrepancy for GH and ALDH⁺ status. Only a minority of GH-producing cells were ALDH⁺ in a subset of HMECs, whereas 66% of the sorted ALDH⁺ cells were GHR⁺ and CD49^high (18). Cell sorting by fluorescence-activated cell sorting, however, is not always representative of the regenerative potential of the selected cells. Cell characteristics and stem cell origin can only be reliably gained from lineage tracing analysis. Using lineage trace experiments, Van Keymeulen demonstrated that the mammary gland in the mouse contains different types of long-lived stem cells that are derived from independent precursors during or prior to the onset of puberty (6). Thus, mammary gland stem cells can have a luminal or basal origin, both having a Lin⁻, CD24⁺, CD29⁺ phenotype (59, 60). For stem cell maintenance, the Wnt/β-catenin signaling can be used (60).

Growth hormone is mainly produced in the differentiated pituitary or placental cells but can also be produced locally within the mammary gland under the influence of P₄. GH is responsible for the expansion of mammary stem and progenitor cells when the mammary gland grows, during puberty, pregnancy, and lactation and also in the menstrual cycle when P₄ levels are higher. Development of the mammary gland is cumulatively effected by P₄ and GH, and this regulatory growth process can eventually lead to changed ratios between progenitor and differentiated cells. Progenitor cells, being more proliferative, have a higher risk of oncogenic hits (18). This undesirable consequence of GH signaling in the mammary gland is not only mediated as a direct effect on GHR-positive cells in the MaSC compartment, but also in the stromal compartment, where indirect GH effects are mediated by the synthesis and release of insulin-like growth factor-I. The PR-positive cells within the mammary epithelium act as sensor cells for P₄ signaling and stimulate the local production of not only GH but also RANKL and Wnt4 (17).

In the mammary gland, canonical Wnt activity is essential for both embryonic and postnatal development (109). During puberty and pregnancy, Wnt activity has been linked to P₄ signaling. P₄ has been shown to promote Wnt ligand expression (especially Wnt4) and to activate the downstream signaling in human, mice, and dogs (110, 111). The relevance of Wnt4 is confirmed by the phenotype of conditional knockouts of mammary Wnt4 expression in mice that have impaired ductal side branching, while overexpression of stabilized β-catenin in luminal epithelium results in precocious lobulo-alveolar development, alveologenesis, and neoplasia (112). In addition to its role in normal mammary gland development, deregulation of the canonical Wnt pathway is often associated with tumorigenesis (101). Oncogenic properties of Wnt proteins were first evident in the mammary gland, as Wnt1 and Wnt3 ligand were initially identified as insertion sites for a MMTV (113). Moreover, in human breast cancer, around 60% of examined clinical samples were shown to have elevated levels of nuclear and/or cytoplasmic β-catenin, suggesting an active signaling (114). In addition, in dogs exposed to prolonged MPA, in vivo strong upregulation of Wnt4 mRNA is also found. In dogs with spontaneous mammary carcinomas, we found a 6.8-fold induction of Wnt7a and also 2- to 3-fold changes in Wnt3, 4, 5a, and 5b mRNA (115). In both cases Wnt target genes such as cyclin D1, survivin, axin2, and cMyc were induced. In general, the same common activated pathways were found in canine carcinomas when compared to published human and mouse data (106, 116).

In most human tumors, constitutive activity of the canonical Wnt pathway was shown to be a consequence of mutations in APC or β-catenin (101). By contrast, such mutations are found only rarely in breast cancer (117). Rather, alternative explanations for the canonical pathway activation have been proposed for mammary tumors, including (1) mutations in other components of the pathway, (2) overexpression of Wnt ligands and other activators, (3) loss or downregulation of the antagonists, such as sFRP1, and (4) cross regulation by other deregulated pathways, such as epidermal growth factor (EGF), phosphatase and tensin homolog (PTEN), or tumor protein 53 (p53) signaling (117, 118). Information about canonical Wnt activity in canine mammary tumors is limited. Deregulation of the pathway has been proposed based on elevated β-catenin immunostaining and pathway analysis associated with molecular profiling of normal and tumorous tissue (19, 111, 119–121). None of the studies have, however, quantitatively assessed the activation of canonical Wnt signaling in canine mammary tumors or the underlying mechanism of its activation.

Wnt drives the formation of cells with CSC properties by regulating the expansion and proliferation of MaSCs (60, 122) and Wnt/β-catenin contribute to tumor progression and metastasis and related to all molecular subtypes of invasive breast cancer with a poor clinical outcome (123). The Wnt pathway is therefore a critical component in breast cancer development and a possible therapeutic target across cancer sub types. Wntless (WLS) appears to be required for the release of all Wnt proteins in both the canonical and non-canonical Wnt pathways. WLS is highly overexpressed in 48% in all subtypes of human breast cancer cases, and there is a strong correlation to HER2 overexpression (124). Together with upregulated Wnt signaling, our canine cell lines also have upregulated HER signaling (125). It remains to be shown whether the human and canine high Wnt- and HER-expressing cells point to a comparable stem cell.