|Home | About | Journals | Submit | Contact Us | Français|
Notch genes play a critical role in mammary gland growth, development and tumorigenesis. In the present study we have quantitatively determined the levels and mRNA expression patterns of the Notch receptor genes, their ligands and target genes in the postnatal mouse mammary gland. The steady state levels of Notch3 mRNA are the highest among receptor genes, Jagged1 and Dll3 mRNA levels are the highest among ligand genes and Hey2 mRNA levels are highest among expressed Hes/Hey target genes analyzed during different stages of postnatal mammary gland development. Using an immunohistochemical approach with antibodies specific for each Notch receptor, we show that Notch proteins are temporally regulated in mammary epithelial cells during normal mammary gland development in the FVB/N mouse. The loss of ovarian hormones is associated with changes in the levels of Notch receptor mRNAs (Notch2 higher and Notch3 lower) and ligand mRNAs (Dll1 and Dll4 are higher, whereas Dll3 and Jagged1 are lower) in the mammary gland of ovariectomized mice compared to intact mice. These data define expression of the Notch ligand/receptor system throughout development of the mouse mammary gland and help set the stage for genetic analysis of Notch in this context.
Mammary gland development is governed by a variety of signaling pathways involved in cell fate specification and differentiation. The mouse gland is a powerful model for the study of mammary epithelial growth, morphogenesis, differentiation and neoplastic transformation (Daniel and Smith, 1999). The majority of growth and development takes place after birth and progresses through defined stages, including puberty and pregnancy leading to lactation and subsequently involution. At birth, female mice possess rudimentary mammary glands, each composed of a simple system of branching ducts opening at the nipple. At approximately 3 weeks of age, the ducts begin to grow rapidly due to intense mitotic activity within terminal end buds (TEBs). The end buds contain undifferentiated stem cells or cap cells that give rise to the epithelial cells (Smith and Boulanger, 2003). The epithelial ducts invade, elongate rapidly and branch. By 10 weeks of age in nulliparous females, the fat pad is filled with a highly organized ductal structure, and the end buds have disappeared. During pregnancy, secretory alveolar lobules develop from the branching duct structure until they completely fill the fatty stroma. At the end of lactation massive apoptosis occurs leading to mammary gland involution (Daniel and Smith, 1999).
The mesodermally derived fatty stroma of the mammary gland is an understudied but important structure. These cells regulate growth, patterning and differentiation of ventral ectodermally derived mammary epithelial cells. In mice, white adipose tissue appears to be required for normal mammary development. Transplantation of mammary epithelium into other types of organ stroma does not support mammary epithelial growth, or results in abnormal growth (Parmar and Cunha, 2004; Wiseman and Werb, 2002). Stromal cells appear to influence epithelial cell behavior through secretion of growth factors and/or by altering the composition of the extracellular matrix in which mammary epithelial cells reside.
The Notch gene family encodes four related transmembrane receptors that interact with five membrane-bound ligands encoded by the Delta/Jagged gene families (reviewed in Callahan and Egan, 2004). Ligand binding stimulates signaling by first inducing proteolytic cleavage of Notch receptors, followed by nuclear translocation of the Notch intracellular domain (ICD) (Kopan and Ilagan, 2009). The Notch-ICD enters the nucleus and serves as a transcriptional activator, although it does not possess DNA binding activity. Rather, Notch-ICD associates with the transcription factor Rbpj (CBF1), the primary transcriptional mediator of canonical Notch signaling. The Notch-ICD/Rbpj complex trans-activates promoters containing Rbpj binding sites, such as those that control expression of Hes and Hey bHLH-family transcriptional repressors (Kato et al., 1996; Kopan and Ilagan, 2009). Conditional knockout of the Rbpj gene in mammary progenitors disrupts cell fate specification and differentiation during pregnancy (Buono et al, 2006). In addition, Notch activation can directly stimulate luminal cell fate specification in purified progenitor cells (Raouf et al., 2008 and Boras et al., 2008). Interestingly, these latter studies also included analysis of Notch pathway gene expression on sorted populations of mammary epithelial cells from non-pregnant humans and mice, respectively. However, little is known about Notch receptor, ligand and target gene expression in the developing mammary gland during puberty, pregnancy, and involution as well as the effect of estrogen on the expression of the Notch pathway. The aim of the present study was to determine timing and levels of mRNA expression of the different Notch receptors, their ligands and their canonical target genes during different stages of mammary gland development in FVB/N inbred mice. In addition, we have examined the pattern of Notch receptors and Hey2 expression in mouse mammary gland by immunohistochemistry using well-characterized Notch-specific antibodies.
Total RNA from two independent pools of mammary tissue was collected from the number 4 inguinal mammary gland (Brill et al., 2008). Each pool was from five FVB/N females, at indicated developmental stages and extracted as previously described (Gallahan et al., 1996). Briefly, RNA was prepared using Trizol reagent (Invitrogen, Carlsbad, CA) followed by treatment with RQ1 DNAse-I (Promega, Madison, Wisconsin), according to the manufacturer’s recommendations. Also, nine-week old FVB/N female mice, from our colony, underwent bilateral ovariectomy as previously described (Raafat et al., 1999). Mammary tissue was collected one week after ovariectomy and total RNA was prepared and subjected to DNAse-I treatment as described above. This study was approved by the Institutional Ethics Committee for Laboratory Animals use in Experimental Research. Mice were kept under standard laboratory conditions according to guidelines of the National Cancer Institute. DNAse treated RNA was subjected to PCR analysis to ensure successful DNA degradation. The quality and quantity of RNA was measured by Agilent bioanalyzer-2100 (Agilent Technologies, CA, USA) according to the manufacture’s instructions, with a cut-off value of 1.5. Mammary gland cDNA synthesis was performed using SuperscriptII reverse transcriptase (Invitrogen, Carlsbad, Ca, USA) with 1 μg of DNAse-I treated total RNA used as template in a 20-μl-reaction volume. For mammary gland quantitative gene expression analysis 1 μl cDNA was subjected to PCR amplification using TaqMan Universal PCR Master Mix Reagents from Applied Biosystems (Foster City, CA, USA). qPCR primers were obtained from Applied Biosystems. For each gene, a standard curve was created using a specific cDNA clone containing the region to be amplified. Quantitative-PCR (qPCR) efficiency was calculated from the slope of the standard curve. The cut-off value of the slope was −3.58, which corresponds to 90% efficiency. This approach insures equal efficiency among qPCR runs, thereby allowing comparison of gene-specific expression during development as well as comparison among different genes. Standard curve slope and amplification plots were analyzed using MxPro PCR software (Stratagene). The relative abundance of mammary gland target mRNA was calculated as the ratio of the copy number of target mRNA normalized to the copy number of Gapdh mRNA, a housekeeping gene in mammary tissue. Reactions were run in duplicates and repeated at least three times in optical 96-well plates with optical caps using the MX3000 multiplex quantitative PCR system (Stratagene). We used the recommended MX3000P optimized cycling program: 95°C, 10 min (1 cycle); 40 cycles of 95°C, 15 s; 60°C, 1 min.
HC11, a clonal mouse mammary epithelial cell line derived from spontaneously immortalized COMMA-1D cells (Ball et al., 1988) were grown in RPMI-1640 supplemented with 10% fetal calf serum (FCS), 10 ng/ml EGF (recombinant human, Sigma) and 6 μg/ml insulin (Sigma). All culture media were supplemented with penicillin and streptomycin. HC11 cells were transfected with control-shRNA plasmid (Santa Cruz, SC-108060), Notch1-shRNA plasmid (Santa Cruz, SC-36096-SH), or Notch2-shRNA plasmid (Santa Cruz, SC-40136-SH) using attractene transfection reagent (Qiagen, 301004) according to the manufacturer’s recommendations. For stable cell lines, transfected cells were selected by adding 5μg/ml puromycin (Sigma, P9620) to the medium. The medium was changed every 48 h.
Notch3 mutant mice were derived from the PST033 ES cell line, which was obtained in a gene trap screen for targeting into genes coding for transmembrane proteins (Mitchell KJ et al., 2001; Arboleda-Velasquez JF et al., 2008). The specific allele disrupts Notch3 at a site just downstream of the exon coding for EGF repeat 21 (Xu K et al., 2010). For Notch4, mammary tissue was collected from 10 wk-old Notch4/Int3 knockout mice (Krebs et al., 2000) and age matching FVB/N mice.
HC11- Notch1-shRNA and HC11- Notch2-shRNA cells were grown on chamber slides and then fixed for 5 min in 4% paraformaldehyde, followed by brief washing in PBS (pH 7.4). After 60 min blocking in PBS-1% bovine serum albumin, cells were incubated for 1 h with anti-Notch1 or anti-Notch2 antibodies diluted in PBS-1% bovine serum albumin. After being washed twice with PBS and stained with the proper fluorochrome-coupled secondary antibodies, samples were mounted for microscopic inspection in mounting medium (DAPI, Vector Lab. Inc.).
Immunohistochemical analysis of Notch receptors in paraffin embedded sections of mammary gland tissue, was carried out using the ABC method according to the manufacturer’s recommendations (Vector Laboratories Inc., Burlingame, CA, USA). Primary antibody incubation was carried out overnight at 4°C, followed by incubation with secondary antibodies. The Hey2 antibody (Millpore, Billerica, Ma) was diluted 1000-fold and anti-rabbit secondary antibody (PK-6101, Vector Laboratories, Inc., Burlingame, CA, USA) was diluted according to the manufacturer’s recommendations. Since active Notch signaling requires the localiztion of the Notch receptor ICD and its migration to the nucleus, we focused our discussion on the cells showing nuclear staining. Tissue sections were counterstained with hematoxylin. Staining was considered strictly nuclear if the nucleus is positive and the cytoplasm is negative and was considered cytoplasmic only if the nucleus is negative and the staining is strictly cytoplasmic. Labeling index was determined in at least a total of 3000 cells for each target protein.
Ten-week old FVB/N female mice, from our colony, underwent bilateral ovariectomy as previously described (Raafat et al., 1999). Mammary tissue was collected from intact females and one week ovariectomized female mice. Protein extracts were subjected to electrophoretic separation on polyacrylamide gels, after blotting, western analysis was performed using Notch2 antibody (Cell Signaling, 2420). The blots were stripped with Restore Western Blot Stripping Buffer (Pierce, Rockford, IL) according to the manufacturer’s instructions and blotted using Notch3 antibody (Santa Cruz SC-5593) followed by β-actin antibody (Sigma, A3854)
Quantitative values are represented as the mean of at least three experiments. All in vivo experiments were repeated at least twice and at least five mice were used in each experiment. All data are expressed as the mean ±SEM.
To compare steady state mRNA levels for the four different Notch receptors, five Dll and Jagged ligands as well as Hes/Hey target genes, mRNA was harvested from two independent pools of mammary tissue at different stages of development. The levels of these mRNAs were normalized to the levels of glyceraldehyde-3-phosphate dehydrogenase (Gapdh) mRNA, and reported as the ratio of mRNA copy number for the gene of interest to mRNA copy number for Gapdh (Figure 1). To validate this approach we examined levels of the epithelial cell marker Keratin-8 (Krt8) mRNA. The relative levels of Krt8 mRNA increased through mammary gland development to the first day of lactation and then decreased during involution. By day 14 of involution Krt8 mRNA levels have significantly decreased and are most similar to levels in the 10 wk old mouse mammary gland. These results are well correlated with the epithelial content of the mouse mammary gland (Daniel and Smith, 1999), validating use of the quantitative approach for gene expression analysis.
The levels of Notch1, Notch2 and Notch3 mRNA in the mammary gland increased as the mammary gland developed from 5 weeks of age through early pregnancy (Figure 2). Notch1-3 mRNAs levels decreased in the mammary gland of late pregnant mice. Expression of Notch receptors is at its lowest levels in the apoptotic stage (day1 involution) and in the quiescent mammary gland (day 14 involution), respectively. Quantification of Notch receptor mRNA expression in the mouse mammary gland during development revealed that Notch3 expression is most abundant, whereas Notch4 mRNA was the lowest. Bouras et al (2008) also reported low levels of Notch4 expression as compared to Notch1, -2 and -3.
Expression levels for Notch ligand mRNA varied dramatically with developmental stage and gene (Figure 3). Of the ligands, levels of Jagged1 and Dll3 mRNA were highest throughout mammary development. However, unlike Dll3, Jagged1 expression decreased markedly during late pregnancy after reaching peak expression in early pregnancy. Both Dll3 and Jagged1 decreased markedly during lactation in comparison to the levels of the early and late pregnancy levels. Dll1, Dll4 and Jagged2 mRNAs were at very low levels in mammary glands of nulliparous and pregnant mice. Higher levels of Dll1 mRNA, but not Dll4 mRNA, were detected in early involuting mammary glands. The highest levels of Jagged 2 mRNA were detected in mammary glands of pregnant mice.
Canonical Notch signaling is mediated through interaction of Notch-ICD with Rbpj; the primary transcriptional mediator of Notch signaling. Mammary expression of Rbpj mRNA (Figure 4) peaked at 5 days of pregnancy. Activation of Notch target genes should represent a reliable readout of Notch/Rbpj signaling activity. Known targets of Notch signaling are members of Hes/Hey gene families that encode transcriptional repressors (Iso et al., 2001). We analyzed expression of Hes1, Hes5, Hey1, and Hey2 mRNA in the mammary gland during postnatal development (Figure 4). Hes1 mRNA levels were high in mammary glands of 10 week old nulliparous and day 5 pregnant mice. These levels began to drop during late pregnancy and then rose again at day 1 and 14 of involution, where Hes1 mRNA was at its highest levels. Hes5 mRNA was least abundant of the four Notch target genes analyzed, and was expressed at a constant level throughout postnatal mammary development. The highest levels of Hey1 mRNA were detected in mammary glands of pregnant mice and early involuting mammary glands. Expression of this gene then decreased through lactation and late involution, with the lowest levels occurring in mammary gland at involution day 14. High levels of Hey2 mRNA were observed in mammary glands of 5 and 10 week old nulliparous mice. These levels did not change during pregnancy, but decreased during lactation and then increased again during involution. Levels of Hey2 mRNA, at any given point during postnatal development of the mammary gland, were at least two orders of magnitude higher than Hey1.
The high degree of homology among Notch receptors (Raafat and Callahan; 2001) makes it important to confirm specificity of Notch antibodies. Three Notch1 antibodies were tested, Cell Signaling (3268), Cell Signaling (4147) and Epitomics (1935-1). In addition, three Notch2 antibodies were tested, abcam (ab52302), Santa Cruz (SC117423) and Cell Signaling (2420). Since deletion of Notch1 or Notch2 is embryonic lethal (Conlon et al., 1995; Hamada et al., 1999), we developed two HC11 stable cell lines where Notch1 (HC11-Notch1-shRNA) or Notch2 (HC11-Notch2- shRNA) expression was blocked. Real time qRT-PCR analysis of RNA from HC11-Notch1-shRNA and HC11-Notch2-shRNA cells showed that these lines have reduced RNA levels of Notch1 (Figure 5Aa) and Notch2 (Figure 5Ab), respectively, compared to wild type and control transfected HC11 cell lines. We used these lines to screen several Notch1 and Notch2 primary antibodies at several dilutions. The Cell Signaling (4147) Notch1 antibody showed the most specific staining at 1000X dilution (Figure 5B, a, b and c). For Notch2, ab-cam (ab52302) antibody showed the most specific signal at 2000X dilution (Figure 5B, d, e and f). Fluorescence signal was observed in control transfected HC11 cells (Figure 5B, b and e) but not in HC11-Notch1-shRNA and HC11-Notch2-shRNA stained with anti-Notch1 and anti-Notch2 antibodies, respectively (figure 5B, c and f). We interpreted these staining patterns as confirming specificity of 4147 and ab52303 antibodies to detect only Notch1 and Notch2 respectively.
To confirm specificity of Notch3 and Notch4 antibodies we used sections of mammary tissue from Notch3 and Notch4 knockout mice (Krebs et al., 2000) as negative tissue for each receptor. Sections of mammary tissue from female FVB/N mice were used as positive tissue for Notch3 and Notch4. Three different Notch3 primary antibodies were tested, abcam (ab23426), Santa Cruz (M-134:SC-5593) and Santa-Cruz, (M-20:SC-7424). Also, three different Notch4 primary antibodies were tested: Santa Cruz (SC-8646) and (SC-8644) and Millipore (07-189). Once again, several dilutions of these primary antibodies were tested. The Santa-Cruz M-134 antibody showed the most specific Notch3 staining at 1000X dilution (Figure 5C, compare a and b). No Notch3 staining was observed in Notch3 knockout mammary tissue (Figure 5C, b), whereas mammary sections from 10wk old FVB mice showed a high degree of positivity (Figure 5C, a). The Millipore (07-189) antibody showed the most specific Notch4 staining at 1200X dilution. No Notch4 staining was observed in Notch4 knockout mammary sections (Figure 5C, d) but positive-nuclear staining was observed in the ducts, and also in the myoepithelial cells in mammary glands of 10wk old wild type FVB mice (Figure 5C, c).
At 5 weeks of age, Notch-positive cells were observed in terminal end buds (Fig. 6A, ad). Notch1-positive end bud cells were localized to the body cell layer. The cap cell layer was negative for Notch1 (Fig. 6A, a). Notch3-positive cells were detected throughout the cap cells layer and many body cells were positive for Notch3 staining as well (Fig 6A, c). A few cap and body cells stained for Notch4 (Fig 6A, d). In addition, Notch1, -2, and -3 positive cells were detected in basal and luminal layers of the more mature ducts. Notch2-positive cells showed cytoplasmic staining. The percentage of Notch1 and Notch3 positive cells (3% and 32% respectively) in mammary epithelium of 5 wk- old mice was higher than the percentage of Notch2 and Notch4 positive cells (1% each) (Figure 6B).
In the mammary glands of 10 wk-old virgin mice, the ductal tree had grown to the limits of the mammary fat pad. Notch2 staining was mainly cytoplasmic (Figure 6A, f). Notch3 showed nuclear staining in both the basal and luminal cells of the ducts (Figure 6A, g). Interestingly, Notch4 showed a strong myoepithelial staining and low nuclear staining agreeing with the low mRNA levels of this receptor (Figure 6A, h). Notch1 showed weak nuclear staining (Figure 6A, e), The percentage of Notch1, Notch2 and Notch4 positive cells remained low (1%, 3% and 1% respectively) in 10 wk mammary sections, whereas, 20% of ductal cells expressed Notch3 (Figure 6B).
In early pregnancy, Notch receptor staining was observed in alveolar buds (Figure 6A. i-l). Notch1 and 2 showed cytoplasmic and nuclear staining (Figure. 6A, i and j). The percentage of Notch1, Notch2 and Notch4 nuclear positive cells remained low (2, 1 and 0.5 % respectively) (Figure 6B). The percentage of Notch3-positive cells decreased significantly to 12%, compared to levels detected in the nulliparous female gland (Figure 6B).
In the lactating mammary gland, Notch2 showed nuclear staining (Figure 6A, n). Notably, the percentage of Notch2-positive cells was significantly higher in lactating mice (16%), compared with virgin (1% in 5 week old mice and 3% in 10 week old mice) and pregnant mice (1%)(Figure 6B). In contrast, the percentage of Notch3-positive cells in glands of lactating mice decreased significantly to 10%, compared to the glands from 5 week old (32%) and 10 week old (20%) mice (Figure 6B). The FVB lactating mammary gland (Figure 6A, p), was almost devoid of Notch4 positive cells. In contrast 5% of cells were Notch1-positive in the lactating gland (Figure 6B).
To examine Notch signaling, immunohistochemical analysis of Hey2 was also performed on sections of mammary tissue corresponding to the stages tested for Notch expression (Figure 6A, q to t). Intense Hey2 nuclear staining was observed in TEB cap cells and basal cells of ducts in 5 wk old mammary glands (Figure 6A, q). Nuclear staining was also observed in basal and luminal cells of mammary ducts from 10 wk-old mice (Figure 6A, r). In the early pregnant mouse mammary gland (Figure 6A, s), Hey2 was detected in luminal epithelial cells of newly formed and proliferating alveoli. In the lactating mammary gland (Figure 6A, t), Hey2 was detected in basal epithelial cells of secretory alveoli. Hey2 staining decreased as mammary gland development progressed through pregnancy and lactation (Figure 6B) and was at its lowest level during involution (data not shown).
Postnatal mammary gland development and function are highly dependent upon ovarian hormones; estrogen and progesterone (Fendrick et al., 1998). Rizzo et al (2008) reported on cross-talk between Notch and estrogen receptors in breast cancer. To determine whether estrogen affects Notch receptor and ligand mRNA expression we measured their levels in mammary glands of FVB/N mice that were ovariectomized at 9 weeks of age. Ovariectomy positively affected expression of Notch2 and negatively affected Notch3, whereas Notch1 and Notch-4 mRNA levels were unaffected by ovariectomy (Figure 7A). The levels of Notch2 mRNA were increased by10 fold, whereas the levels of Notch3 mRNA were down regulated by 15 fold in mammary glands from ovariectomized mice. Western blot analysis of mammary protein extracts, confirmed the effect of ovariectomy on Notch 2 and Notch3 (Figure 7B). Similarly, Dll1 and Dll4 mRNA levels were higher (50 and 18 fold, respectively) whereas Dll3 and Jagged1 levels were lower (90 and 2 fold, respectively) in mammary glands of ovariectomized mice (figure 7C). In contrast, Jagged2 mRNA levels were unaffected by ovariectomy.
Sequence analysis of 4500 nucleotides 5′ to the ATG start codon of mouse Notch receptors and their ligands showed that only Jagged1 does not contain an estrogen response element (ERE) within the tested region. In contrast, all four Notch receptor genes as well as genes coding for Dll3, Dll4 and Jagged2 had imperfect ERE binding sites (Table 1). Dll1 had a near perfect ERE (GGTCAN5TGACC). Analysis of intron-1 also showed an imperfect ERE in Notch1, Notch2, Notch3, Notch4, Dll4 and Jagged1 genes.
Earlier work demonstrating involvement of Notch signaling in mammary development and tumorigenesis (Jhapan et al., 1992; Gallahan et al. 1996; Hu et al. 2006; Buono et al., 2006) led us to investigate expression of Notch receptors, ligands and target genes at multiple developmental stages. The mammary gland contains a mixture of cell populations that move geographically within the gland during postnatal development. At birth, female mice possess rudimentary mammary glands composed of a simple system of branched ducts opening at the nipple. Normal mammary development can be divided into several stages that differ in morphology, function, and hormonal responsiveness.
Relative to Notch3 mRNA levels, only low steady state levels of other Notch receptors mRNAs were detected at different stages of postnatal mammary development. These levels are a function of cell type complexity in this tissue, with a mixture of expressing and non-expressing cells noted for each gene. In addition, absolute expression levels within expressing cells are likely controlled in response to physiological and developmental status of the mouse. To address this point, we validated the specificity of several antibody preparations against mouse Notch receptors by demonstrating the absence of staining if expression of a specific Notch receptor was inhibited or disrupted (Figure 5B and C). Whereas, Notch receptor specific staining was observed in wild type HC11 cells, FVB/N mammary tissue or Notch transgenic mammary tumor tissue. Notch receptors and Notch target gene expression was observed in mammary epithelial cells at all developmental stages analyzed.
In the peripubertal mammary gland, TEBs that drive ductal morphogenesis, are comprised of an outer cap cell layer enriched in mammary stem cells and inner body cells fated to the luminal compartment (Daniel et al., 1987). High levels of Notch3 and Hey2 expression were detected in TEB cap cells, which are the most highly proliferative population at this stage (Fendrick et al., 1998). The cap cell layer was devoid of Notch1 and Notch2 suggesting that Notch 3 and Notch4-positive cells constitute the major pool of proliferating cells at this developmental stage. Proliferation of cap cells gives rise to a multilayeredinternal mass of body -cells below the cap (Hovey et al., 2002). Formation of the ductal lumen requires apoptotic cell death within the body cell layer (Hovey et al., 2002; Mailleux et al., 2007). Notch activity affects proliferation and apoptotic programs (Artavanis-Tsakonas et al., 1999). Data from two groups indicate that Notch signaling blocks apoptosis during thymocyte maturation (Jehn et al., 1999; Deftos et al., 1998). It is possible that Notch expression in end bud body cells, protects these cells from apoptosis. As a result mammary ductal luminal cells are Notch-positive. Alternatively, Notch signaling could be involved in inducing their death (Miller and Cagan, 1998). Expression of Notch by cells in the TEBs also suggests a role for Notch signaling in regulating proliferation during peripuberty. Stem cell units have been identified in TEBs and ducts of the mouse mammary gland (Kenney et al., 2001).
Deletion of Notch signaling in a mouse mammary stem cell (MaSC) enriched population, resulted in increased stem cell activity in vivo, as well as formation of aberrant end buds, indicating a role for endogenous Notch signaling in limiting MaSC expansion (Bouras et al., 2008).
Notch positive cells are evenly distributed in mammary ducts of post-pubertal mice. Inhibition of Notch signaling in both basal and luminal cells of the postnatal mouse mammary gland can lead to an imbalance in the proportion of basal and luminal cells resulting in basal cell expansion (Buono et al., 2006). This suggests that Notch can regulate epithelial cell fate and may impact stem cell activity. Notch signaling also targets luminal progenitor cells for expansion, leading to hyperplasia and tumor development (Bouras et al., 2008). These data indicate that signaling by one or more members of the Notch family of receptors promotes luminal cell fate commitment. Mammary alveolar units originate from alveolar luminal epithelial cells that line the ducts. Targeted disruption of Notch expression or Notch signaling in the mouse mammary gland during pregnancy has also revealed that Notch signaling regulates alveolar cell maintenance (Buono et al., 2006; Raafat et al., 2009). Interestingly, the mammary gland appears grossly normal in Notch3 and Notch4 knockout mice, suggesting that Notch1, Notch2 or multiple Notch genes function redundantly to control cell fate specification, proliferation and differentiation in this tissue.
In the normal mammary gland, the percentage of Notch1 and Notch3 expressing cells was lowest during early pregnancy and lactation (Figure 6B). It is interesting, therefore, that transgenic mice overexpressing activated Notch1 or Notch3 were unable to feed their pups because of impaired ductal and lobulo-alveolar mammary gland development. In addition the β-casein gene promoter was repressed in these mice (Hu et al., 2006). Thus it is possible that the reduced number of cells expressing Notch1 and Notch3 in the mammary glands of pregnant and lactating mice is important for normal lobular alveolar development and lactation. Low expression levels and positive cell numbers of Notch4 in the mammary gland during development, suggest that the Notch4 levels are very low per cell. Like Notch1 and Notch3 transgenic mice, Notch4 transgenic mice expressing the intracellular domain of Notch4 were also incapable of lactation due a block in mammary lobular-alveolar development. This suggests that in addition to the number of cells expressing a particular activated Notch receptor, expression levels per cell is also important in determining the consequence of Notch signaling. Consistent with this hypothesis is the observation that in Drosophila severity of a Notch phenotype is influenced by the level of Notch ICD in the nucleus (Artavanis-Tsakonas et al, 1999). Of the ligands, levels of Jagged1 and Dll3 mRNA were highest throughout mammary development. DLL3 has been found to play a Notch-inhibitory role (Ladi et al., 2005). Jagged1 mRNA levels decreased markedly during late pregnancy after reaching peak expression in early pregnancy. The other ligands were expressed at lower levels throughout postnatal mammary gland development. In one study of human mammary tissue, using mRNA in situ hybridization, Jagged1 expression was detected in myoepithelium and high levels of Notch3 mRNA in adjacent luminal cells (Reedijk et al, 2005). In another study, Dll4, Jagged1, Jagged2, Notch1 and Notch3 mRNAs were detected by RT-PCR in tissue from reduction mammoplasty (expression of Dll1, Dll3, Notch2 and Notch4 mRNAs could not be detected) (Stylianou et al, 2006). In this study, Notch3 mRNA levels were higher than those for Notch1, while Jagged1 mRNA levels were higher than Jagged2, which is consistent with our data in the developing mouse mammary gland. This similarity is likely due to a conservation of Notch gene function during mammary gland development.
Complete development of the mammary gland takes place after puberty, it depends on hormones and is closely regulated by the interaction of mammary epithelium with its stroma (Parmar and Cunha, 2004; Wiseman and Werb, 2002). Lower levels of Notch receptor and ligand mRNAs can also be detected in epithelium-divested (cleared) mammary fat pads (data not shown). Mammary gland development is highly dependent on ovarian hormonal stimuli. Mammary ductal elongation is directed by estrogen as well as by growth factors such as insulin-like growth factor-I and epidermal growth factor (Fendrick et al., 1998). The requirement for estrogen is supported by the absence of ductal elongation in estrogen receptor knockout mice (Bocchinfuso and Korach., 1997). Also, ovariectomy of mice at 2 months of age, retards mammary ductal and lobular alveolar development as compared to non-ovariectomized control mice (Nandi, 1958). Since Notch signaling is required for mammary lobular alveolar development (Buono et al., 2006), we tested whether ovariectomy would affect expression of Notch receptors and ligands. Ovariectomy positively affected levels of Notch2, Dll1 and Dll4 mRNA and negatively affected levels of Notch3, Dll3 and Jagged1 mRNAs. Notch1, Notch4 and Jagged2 mRNA levels were relatively unaffected by ovariectomy. The decrease in Notch3 and increase in Notch2 in the OVX mice mammary glands is notable but Notch 3 levels are still the highest among the Notch receptors. However, the changes in ligand expression in the OVX mice do change the balances significantly compared to the intact mice. A primary consequence of ovariectomy is a loss of estrogen production, a known positive, or in some cases negative, regulator of gene transcription. Based on the effects of ovariectomy on transcription of some Notch receptor and ligand genes, we examined promoter regions for the presence of estrogen responsive elements (ERE). Although the EREs that were detected in Notch receptor and ligand genes were imperfect as compared to the perfect ERE palindrome (5′-GGTCANNNTGACC-3′) (Lin et al., 2004), there is strong evidence that EREs can function in more loosely structured sequence motifs (Klein-Hitpass et al., 1986). For example, half-ERE palindromes (TGACC), that are more than 100 bp apart, can act synergistically to confer sensitivity to estrogen induction, either to a proximal promoter or to heterologous promoters (Kato et al., 1992). Even a single, non-canonical ERE is capable of conferring transcriptional regulation by estrogen (Berry et al., 1989). Also, duplication of ERE half-palindromes in Notch1, Notch2, Notch3, Dll3, Dll4 and Jagged2 promoters provide a structural basis for estrogenic regulation of these genes. Down regulation of Notch2, Dll1 and Dll4 in the intact mammary gland compared to mammary glands from ovariectomized mice suggests that certain regulatory sequences and/or chromatin structures are playing a role in regulation of these genes (Muller et al., 2009, Strom et al., 2000). In addition estrogen receptor binding can lead to recruitment of coactivators or co-repressors (Heldring et al., 2007). In this regard, about half of ERE regulated genes are down regulated by estrogen (Carroll et al., 2006; Lin et al., 2007). However, induction of Notch receptor or ligand gene expression in mammary glands of ovarectomized mice may not translate into induction of Notch signaling, since in estrogen receptor positive breast cancer cell lines, estradiol induced Notch1 expression but inhibited Notch signaling (Rizzo et al., 2008). Long and short-term ovariectomy affect both estrogens and progesterone and their receptors (Raafat et al., 1999). Currently we are investigating the relationship to progesterone receptor and its signaling pathways.
Expression levels of mRNA for multiple Hes and Hey Notch target genes were found to increase as the mammary gland develops from 5 weeks in the nulliparous mouse to day 5 in pregnant mice. This reflects the same pattern observed in expression of Notch receptor mRNAs in the mammary gland. Conditional ablation of Rbpj in the mouse mammary gland causes a profound cell fate switch when it occurs in the early pregnant or pubescent gland (Buono et al., 2006). However, when ablation of Rbpj occurs at day 15 of pregnancy, mammary gland development occurs normally (Buono et al., 2006, Raafat et al 2009). This is consistent with there being a window of time in which canonical Notch signaling can affect lobular alveolar development. The Hes and Hey gene families are the best-characterized canonical Notch/Rbpj target genes. Activation of Notch signaling up regulates Hes and Hey transcription (Iso et al., 2003; Raafat et al., 2009). In addition, Hes and Hey promoters contain Rbpj binding sites suggesting direct induction by Notch ICD/RBPJ complexes (Fischer and Gessler, 2003). Dimers of Hes and/or Hey proteins repress transcription of a variety of genes by interacting with transcriptional co-repressors or sequestering transcriptional activators (Iso et al., 2003, Shi and Harris, 2006). Hes1 and Hes5 double deficient mice phenocopy the neurogenic characteristics of Notch1 null mice (Kageyama and Ohtsuka, 1999), whereas Hey1 and Hey2 double deficient mice replicate the angiogenic phenotype observed in the Notch1 deficient mice (Fischer et al., 2004). Interestingly, the relative abundance of Hes1, Hes5, Hey1 and Hey2 mRNA levels varied over three orders of magnitude in the developing mammary gland, consistent with a complex pattern of expression and function of Notch receptors and their ligands. Adding complexity to this situation are reports of Notch-independent regulation of Hes/Hey gene transcription (Leimeister et al., 2000; Zavadil et al., 2004; Katoh and Katoh, 2007, and Doetzlhofer et al., 2009). In this regard, Doetzlhofer et al., 2009 have shown the Fgf3 can upregulate Hey2 independent of Notch signaling.
One confounding problem affecting interpretation of Notch receptor and ligand mRNA expression is our inability to evaluate the impact of Notch signaling, or the lack of it, from the epithelium to the surrounding stroma, and how this might affect the level of Notch and ligand gene expression in the cleared fat pad. The functional significance of our Notch receptor, ligand and target gene expression data await development of mouse strains wherein individual or combinations of Notch receptor and ligand genes are conditionally ablated, or in which expression of a reporter gene is put under transcriptional control of a particular Notch gene or Notch ligand gene.
This research was supported by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research. S. E. Egan’s lab has been supported by funds from Genome Canada through the Ontario Genomics Institute.