Introduction
The mammary gland comprises a ductal epithelial network embedded in a stromal matrix. The ducts are composed of an inner layer of luminal cells and an outer layer of myoepithelial cells. Pregnancy is accompanied by the expansion and differentiation of alveolar luminal cells, resulting in secretory cells that produce and secrete milk. Although the function of the mammary gland is preserved across species, marked anatomic differences exist between human and mouse mammary tissue. The human mammary gland is characterised by a branching network of ducts that terminate in clusters of small ductules that constitute the terminal ductal lobular units (TDLUs). In contrast, the mouse mammary epithelial tree does not contain TDLUs, although small alveolar buds are formed during each estrous cycle. Moreover, the human breast parenchyma is significantly more fibrous than the mouse stroma, which contains predominantly adipocytes. Despite these architectural differences, accumulating evidence suggests that remarkable parallels are found between the hierarchy of epithelial cells that exist in the mammary glands of humans and mice [
1].
Distinct epithelial subtypes have been prospectively isolated from both mouse [
2‐
5] and human mammary glands [
6‐
10]. Functionally analogous subpopulations have been identified: the MaSC-enriched/bipotent progenitor, committed luminal progenitor and mature luminal cell subsets. In the mouse, MaSCs are found within the basal CD49f
hiCD29
hiCD24
+Sca1
- subset (referred to as MaSC-enriched), whereas committed luminal progenitor cells exhibit a CD29
loCD24
+CD61
+ (or Sca-1
-CD24
+) phenotype, and mature luminal cells display a CD29
loCD24
+CD61
-phenotype [
2,
3]. In human mammary tissue, the CD49f
hiEpCAM
-/lo subpopulation has been demonstrated to be enriched for MaSCs, based on
in vivo transplantation either into the mouse mammary fat pad [
7] or under the renal capsule [
6]. Luminal progenitor and differentiated cells prospectively isolated from human breast tissue are characterized by CD49f
hiEpCAM
+ and CD49f
-EpCAM
+ phenotypes, respectively.
There are similarities as well as species-specific differences in the expression of cell-surface markers on the epithelial subsets. Both the mouse and human MaSC-enriched populations express high levels of CD49f. However, CD24 is a marker of epithelial cells in the mouse mammary gland, but not in human breast tissue, where it specifically marks luminal cells [
3‐
5,
7,
11]. Significantly, both the human and mammary MaSC-enriched populations lack expression of the steroid hormone receptors ERα and PR [
7,
12]. Moreover, these MaSCs do not express detectable levels of ERBB2/HER2, reminiscent of the triple-negative receptor phenotype that characterizes many basal cancers [
13].
Understanding the relation between normal epithelial cell types and the different molecular subtypes of breast cancer is fundamental to gaining insight into cell types predisposed to carcinogenesis. At least six distinct subtypes of breast tumors have been defined on the basis of gene expression profiling. These include the luminal A and B, basal-like, claudin-low, HER2/ERBB2-overexpressing, and normal breast-like subtypes [
14]. We recently used the emerging human mammary hierarchy as a framework for understanding aberrant cell subsets that may arise during breast oncogenesis [
7]. The claudin-low subtype was found to be most closely associated with the gene signature of the MaSC-enriched population, whereas the molecular profiles of the basal-like subtype of breast cancer showed remarkable concordance with the luminal progenitor gene signature. Not surprisingly, the expression profiles of the luminal A and B subtypes were closest to that of mature luminal epithelial cells. Interestingly, the molecular portrait of premalignant tissue from
BRCA1 mutation carriers, who usually develop basal-like breast cancers, showed striking similarity to the luminal progenitor signature [
7].
In the context of the mouse mammary gland, transcriptome analyses of epithelial cells have highlighted the differences between basal and luminal cells and revealed a number of potential regulators [
5,
15]. Here we performed genome-wide transcriptome analyses of three different mouse epithelial subpopulations and established pathways that are conserved in functionally equivalent subsets in humans by using specific gene signatures. We further used these signatures to interrogate mouse models of mammary tumors, providing insight into cell types that contribute to breast oncogenesis.
Discussion
In this study, we describe a comparative transcriptome analysis of functionally analogous human and mouse mammary cell populations using an Illumina platform. Four prospectively isolated populations were evaluated, corresponding to those enriched for basal/mammary stem cells, committed luminal progenitor, mature luminal epithelial, and stromal cells. Distinct gene signatures were apparent for the mouse subpopulations, reminiscent of that found for human mammary cell subsets [
7]. Comparison of the mammary epithelial signatures across human and mouse, combined with Ingenuity pathway analysis, revealed a number of conserved genes and pathways that are likely to regulate key processes during mammary ontogeny.
The MaSC-enriched subset exhibited the largest number of genes conserved across species. This subset comprises stem cells, likely basal progenitor cells, as well as mature myoepithelial cells. These cells share many common cell-surface markers that have impeded efforts to fractionate this population. Multiple transcriptional regulators (
Irx4, Mef2C, Slug, Egr2, Twist2, Tbx2) were found to be highly expressed in this basal subset. Interestingly, the leucine-rich repeat-containing G protein-coupled receptor
Lgr6 [
31], which belongs to the same subgroup as Lgr5, a stem cell marker of small intestine, colon, and hair follicles [
32], was identified as a component of the MaSC-gene signature. A prominent integrin network also emerged; these proteins play an important role in mediating interactions between basal cells (including MaSCs) and the underlying extracellular matrix. Of relevance, several genes attributed to cells that have undergone an EMT [
33], such as slug, vimentin, and absence of E-cadherin expression, also characterize basal cells in the mammary gland. Therefore, the expression of these genes in breast tumor cells may indicate the acquisition of basal cell characteristics rather than an EMT. The recently described link between Wnt signaling and the EMT [
34] may also reflect an active Wnt pathway in MaSCs or other cells in this basal population.
Kit, Cyp24A1, and Elf5 appear to be defining markers of committed luminal progenitor cells in both species. Interestingly, the tyrosine kinase KIT was reported to be overexpressed in basal breast cancers [
35] and
BRCA1-associated basal cancers [
7], suggesting that it may serve as a useful prognostic marker or therapeutic target. Elf5 has been demonstrated to be important for driving alveolar cell differentiation during pregnancy [
36] but may play an earlier role in regulating luminal cell-fate decisions. Interestingly, triple-negative breast cancer patients have been shown to have lower serum vitamin D levels, and Cyp24A1 is known to catabolize both 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D. It is therefore tempting to speculate that higher levels of CYP24A1 might be linked to increased breast cancer risk [
37]. Other interesting candidates include CXCR4, a receptor implicated in mediating metastasis of breast cancer cells through its ligand SDF-1 [
38], and CD14 and lipopolysaccharide-binding protein (LBP), which are implicated in Toll-like receptor signaling and LPS-mediated inhibition. In the mature luminal population, active pathways identified by IPA include the Wnt ligands (Wnt4, 5A, 7B), which may act on MaSCs to enhance their self-renewal or proliferation. Expression of the transcriptional regulator
Lmo4 was downregulated in the mature luminal subset, consistent with findings that this oncoprotein is important for promoting mammary epithelial cell proliferation and inhibiting differentiation [
39,
40]. Conversely, the expression of other transcriptional regulators (
ERα,
Myb,
PR, and
Cited1) was significantly upregulated in mature luminal cells.
A high degree of concordance was found between the expression profiles of the basal and mature luminal cell subsets in the mouse mammary gland described here and those previously reported [
5,
15], although the luminal progenitor expression profiles proved to be more divergent. For the basal/MaSC-enriched population, conserved pathways such as the Ephrin, Wnt, and extracellular matrix networks were also identified as nodes in interaction mapping of the basal subset by Kendrick
et al. [
15]. Similarly, the gene expression profiles of the mature luminal subset (reported here) shared substantial overlap with that of the ER
+ population described by Kendrick
et al. [
15], with the ER/glucocorticoid receptor signalling network emerging as one of the predominant nodes. The expression profile of the luminal progenitor subpopulation, however, exhibited substantial differences from that of the ER
- [
15] and Ma-CFC subsets [
5], indicating that they may represent distinct or heterogeneous cell populations. Nevertheless, the Kit and TLR signaling pathways identified here using Ingenuity Pathway analysis were also revealed as distinct modules in the ER
- network by ROCK analysis [
15]. The gene profiles determined for the same three epithelial subsets isolated from human mammary tissue by Raouf
et al. [
11] show similarities but also differences that likely reflect short-term culture of their cells before gene expression studies [
11].
Interrogation of breast cancer subtypes with the gene signatures of normal human epithelial cell subsets has revealed striking relations. Intriguingly, the luminal progenitor gene signature shared marked similarity with the basal subtype of breast cancer and preneoplastic breast tissue from
BRCA1 mutation carriers [
7]. Moreover, aberrant luminal progenitor cells were detected in
BRCA1 mutation carriers, suggesting that they serve as a target population for further oncogenic events [
7]. To extend these studies and identify candidate cell types that might contribute to oncogenesis in mouse models, we explored the link between the mouse mammary epithelial hierarchy and models of mammary tumorigenesis. The MaSC-enriched transcriptional signature was highest in
MMTV-Wnt-1 and
p53-/- tumors, indicating that cells within these tumors exhibit similarities with MaSCs or basal progenitors. Although cancer stem cell populations have been identified in these tumors [
41‐
43], one cannot conclude that these bear resemblance to MaSCs based on expression profiling studies. Rather, the molecular profiles may indicate a cell type that has been expanded during tumor progression. It is notable that preneoplastic tissue from
MMTV-Wnt-1 transgenic mice has been shown to harbor an expanded mammary stem cell pool as well as aberrant bipotent progenitor cells, indicating that more than one cell of origin may exist in the
Wnt-1 model [
42].
The luminal progenitor signature was highest in
MMTV-
Neu and
MMTV-PyMT tumors. Compatible with this observation for the
MMTV-Neu model, FACS analyses of these tumors has indicated a homogeneous population of cells expressing high levels of the luminal progenitor marker CD61
+ [
42]. Thus, luminal progenitor cells may have undergone expansion in these tumors. The
MMTV-
Neu strain, however, does not accurately recapitulate HER2-overexpressing cancers arising in women, because
MMTV-Neu tumors do not show significant gene overlap with the HER2-positive subtype but are more similar to human "luminal" tumors [
14]. Interestingly, the mature luminal signature was highest in
MMTV-Myc tumors. The small progenitor subset in the 'mature' population [
2,
28], rather than the differentiated luminal cells, is likely to contribute to tumorigenesis in this model.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
EL contributed to the conception and design, collection, assembly of data, and manuscript writing. DW and GS contributed to data analysis and interpretation and manuscript writing. BP, TB, MA, and FV contributed to the collection and assembly of data. HY provided clone HMbeta1-1 hybridoma to CD29 and advice. JEV and GJL contributed to the study conception, provision of study materials, data analysis, and manuscript writing. All authors read and approved the final manuscript.