This study shows that the expression of K6 is highest during the initial development of the gland and becomes sporadic in the mature gland. K6a expression was confined to cells destined to become the luminal epithelial cells of the mammary gland. Even as early as E16.5, K6a expression was detected in the innermost cells of the mammary bud, as opposed to the outer ring of epithelial cells that are p63-positive (SLG and JMR, unpublished observation). Similarly, the body cells of TEBs, as opposed to the cap cell layer, expressed K6a. K6 expression in the epidermis has previously been shown to be regulated by epidermal growth factor and tumor necrosis factor-α signaling pathways [22
], but nothing is known about what regulates its expression in subpopulations of cells in the mammary anlage or TEBs.
Our original observation of an increased number of K6a-positive cells in the mature mammary glands from C/EBPβ-null mice correlated with a block in development, suggesting an accumulation of more primitive cells [6
]. Several other groups have also observed K6-positive cells in their mouse models of mammary gland development, supporting the idea that K6 may indeed be a putative marker of progenitor cells in the gland. For example, Stingl and colleagues have recently described a population of mouse mammary cells called 'colony-forming cells' or Ma-CFCs that were isolated by fluorescence-activated cell sorting based on a CD24high
profile and occurred with a frequency of about 1 in 63 cells [5
]. Although these cells did not display outgrowth potential in fat pad transplantation experiments, Ma-CFCs were defined by Stingl and colleagues as progenitor cells that are able to grow discrete colonies on low-density adherent cultures. Both mRNA and protein levels of K6 were enriched in these Ma-CFC cells, which also had increased expression of other luminal cell markers, such as K8, K18, and K19.
Notch signaling has been implicated in stem cell self-renewal [25
]. Recombination signal binding protein, J-type (RBP)-J is the common transcriptional mediator of Notch receptors. Targeted deletion of RBP-J in the mammary gland resulted in a transient increase in K6 expression in luminal epithelial cells during early pregnancy [26
]. These results suggested an arrest at an immature stage of mammary gland development, similar to that in the C/EBPβ-null mice.
In addition to being a marker of early mammary gland development and putative progenitor cells, K6 expression has been observed in a subpopulation of cells in both mammary hyperplasias and tumors induced by transgenic expression of Wnt-1, β-catenin, or Myc [7
]. However, hyperplasias and tumors induced by polyoma middle T antigen, Neu or H-Ras are more homogeneous and do not express K6. Li and colleagues therefore suggested that some tumors might arise from the amplification of progenitor cells, whereas the others might promote differentiation of the progenitors, thereby depleting the population.
Lisanti and colleagues have reported increased K6 expression in the hyperplastic mammary ducts of caveolin-1-null mice, along with increased β-catenin expression [27
]. They proposed that activation of the Wnt/β-catenin pathway led to the accumulation of mammary progenitor cell accumulation. Finally, overexpression of the proto-oncogene Met, a receptor tyrosine kinase, under the control of the MSCV (mouse stem cell virus), resulted in non-progressive mammary neoplasms [28
]. These lesions contained large numbers of K6-positive cells, again suggesting that this might represent the expansion of a progenitor cell population. However, when Met was overexpressed under the control of MMTV (mouse mammary tumor virus), no neoplasms were detected and K6 expression was not observed.
Keratinocytes activated by injury or stress express K6 and migrate to the site of wound healing [13
]. Although K6 expression has been associated with hyperproliferation, expression of K6 in the epidermis does not overlap with the incorporation of [3
], supporting our observation that K6-positive cells are mostly quiescent. Instead, it is possible that K6 regulates a migratory function required in progenitor cells so as to permit their dispersal throughout the mammary gland. However, deletion of K6a and K6b did not seem to have any detectable effect on ductal elongation in mammary glands from intact animals or in outgrowths from tissue transplants.
The lack of an overt mammary gland phenotype in the K6a/b double knockout mice was not unexpected, because it has been difficult to demonstrate the functional properties of most markers used to isolate and characterize stem and progenitor cells by means of gene deletion in genetically engineered mice. For example, deletion of one of the best-characterized mammary stem/progenitor cell markers, CD49f (α6-integrin), did not have any reported mammary gland developmental phenotypes in transplants of the null mammary anlage [30
]. This unexpected result might be considered surprising, because integrins have been suggested to have an essential function in adhesion in the stem cell niche [31
]. Furthermore, conditional deletion of CD29 (β1-integrin) resulted in impaired alveolar development and lactation [32
], with no reported effect on ductal morphogenesis [32
]. Although the triple knockout mouse for Brcp1 (Abcg2), Mdr1a, and Mdr1b resulted in the loss of side population (SP) cells, a phenotype defined by the ability of these cells to efflux Hoechst 33342 dye associated with stem/progenitor cells, there was no mention of a mammary gland development phenotype in these mice [33
A key experiment to test the functional effect of loss of progenitor cell markers requires limiting dilution serial transplantation to distinguish long-term versus short-term engraftment of stem/progenitor cells deleted for specific genes. This has only recently been accomplished for CD49f and CD29 [4
]. It is most likely that many of these markers may be only just markers, and may not have critical functions. However, notwithstanding this caveat, it is crucial to identify lineage markers as tools for the characterization of different mammary cell types. K6 seems to be one such useful lineage marker.
In the present study, an increase in the number of proliferating PR-positive cells was observed after the loss of K6a/b. K6a-positive cells co-localized with steroid receptor expression, but rarely with BrdU incorporation, a marker of proliferation. Normally, steroid receptor expression does not overlap with proliferation [19
]. However, inappropriate co-localization of steroid receptors with proliferative markers is often found in pre-neoplastic disease [34
]. A recent study suggests that activated TGF-β acts in an autocrine manner to prevent ERα-positive mammary epithelial cells from proliferating [37
]. In addition to an increased number of K6a-positive MECs, C/EBPβ-null mice have increased levels of activated TGF-β as well as increased expression of PR and decreased proliferation in the mature mammary gland, further supporting this hypothesis [6
]. Interestingly, ERα and PR have also been proposed to be markers of mammary progenitor cells, representing a quiescent population scattered throughout the gland hypothesized to self-renew through asymmetric division [39
]. Human mammary epithelial cells sorted for expression of p21 or Msi-1, suggested to be putative stem cell markers, also had enriched expression of steroid receptors. Additionally, SP cells had sixfold enrichment of ERα-positive cells in this model system; however, K6 expression was not analyzed [40
]. Alternatively, recent studies with a xenograft model of T47D human breast cancer cells expressing ER plus different PR isoforms demonstrated increased K6 mRNA expression in tumors in response to treatment with E + P that was dependent on the expression of PR [41
]. Thus, the precise relationship between steroid receptor and K6 expression in normal mammary epithelial cells, and how this might be altered in breast cancer, remains to be established, but it is interesting to speculate that K6 might have a function in this interaction.