| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
From the Laboratory of Molecular Biology,* Clinical Research Institute of Montréal, Montréal; the Department of Pathology and the Division of Experimental Medicine, McGill University, Montréal; and the Department of Microbiology and Immunology, Université de Montréal, Montréal, Québec, Canada
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionReferences |
|---|
To date, two members of the murine Notch gene family, Notch1 and Notch4, have been implicated in mammary gland development and neoplasia. The Notch4/int3gene was initially identified as a common provirus integration site in mammary tumors of mice infected with mouse mammary tumor virus (MMTV).14 Itwas also found to be truncated, through a similar mechanism, by an intracisternal type A DNA provirus in a mouse mammary tumor.15 Both MMTV and intracisternal type A provirus integrations generated truncated Notch4intra domain that was overexpressed in mammary tumors. Expression of this same Notch4intrain transgenic (Tg) mice under the regulatory sequences of the MMTV LTR also led to mammary tumor formation and impaired mammary ductal and secretory lobule development.16,17 The same gene expressed in Tg mice under the regulation of the whey acidic protein gene promoter (which is active only in the secretory mammary epithelial cells), induced mammary tumors and inhibited growth and differentiation of the secretory lobules without affecting ductal growth.18 In mouse mammary epithelial cells in vitro, activated Notch4intra had a similar effect as in vivo and induced transformation of HC11 cells19 and inhibited the branching morphogenesis induced by HGF and transforming growth factor-ß2 in TAC-2 cells.20 Together, these experiments indicated that the activated form of Notch4 (Notch4intra) has the capacity to transform mammary epithelial cells and to block the growth of epithelial ductal cells and the secretory lobules of the mammary glands.
Similarly, our group also observed MMTV provirus insertional mutagenesis of another Notch family member, Notch1, in MMTV/neu Tg mice infected with MMTV.21 Truncation of the Notch1 gene by the MMTV provirus occurred in a region just upstream of the exons coding for the transmembrane domain and resulted in the activation of the remaining truncated intracellular Notch1 domain. The Notch1intra domain was found to have transforming ability in HC11 mammary epithelial cells in vitro. This was the first indication that the Notch1gene could play an important role in mammary gland tumorigenesis. However, more direct evidence of the oncogenic potential of the truncated murine Notch1intra sequences was lacking. In addition, it remains unknown whether the significant amino acid sequence differences between the intracellular domain of the members of the murine Notch family would affect their oncogenic potential. However, a recent report showed that Tg mice expressing human Notch1intra in the mammary glands developed pregnancy/lactation-dependent mammary tumors.22
We have now directly assessed the oncogenic po-tential of murine Notch1intra and Notch3intra in vivo by generating Tg mice expressing murine Notch1intra or Notch3intra in mammary epithelial cells under the regulation of the MMTV LTR. These regulatory elements can direct expression of the transgene in mammary glands, at all of the different developmental stages as well as in male reproductive organs.23-31 Our results show that murine Notch1intra and Notch3intra, in contrast to human Notch1intra, affect the development and the cellular differentiation of the mammary glands, as well as induce tumors of the mammary glands, which are not pregnancy/lactation-dependent.
|
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
The MMTV/Notch1intra transgene was constructed by ligating the Notch1 mutant A sequences (2.8 kbp)21 to the MMTV LTR from the MMTVneuT construct (2.3 kbp)32 through the Notch1intraSspI restriction site, thus spanning nucleotides 5187 to 8064 of Notch1. The MMTV/Notch3intra transgene was constructed by ligating the polymerase chain reaction-amplified Notch3intra DNA fragment spanning nucleotides 5008 to 7189. One-cell (C3H x C56BL/6) F2 embryos were collected, microinjected, and transferred into pseudopregnant CD1 females as described before.32 The resulting founder mice were analyzed for the presence of the transgene by Southern analysis of tail DNA with a Notch1 (probe K) or SV40-specific probe.21 Notch1 probe K covers the entire intracellular region, from the BamHI (nucleotide 4204) to EcoRI (nucleotide 8064). Tg mice were bred on CD1 background for at least 6 generations and for 10 to 15 generations for most experiments presented. The phenotypes described here have been observed throughout a period of 4 and 3 years, respectively, for Notch1intra and Notch3intra Tg mice.
Tissue Analysis
For mammary gland development study, groups of three Tg and three non-Tg mice were sacrificed at the age of 3, 5, 7, and 12 weeks, at days 6.5 and 18.5 of pregnancy, and at day 4 postnatally (involution). Mammary glands 1, 2, 3, 6, 7, and 8 were removed and frozen with liquid nitrogen for RNA extraction. Gland 9 (right inguinal) was dissected carefully, spread on a clean slide, and processed for whole mounts. Mice were then perfused with 4% paraformaldehyde and gland 4 (left inguinal), and a piece of intestine was postfixed by 1 hour immersion in fixative disection and embedded in paraffin using a Shandon tissue infiltrator (Hypercenter XP, Pittsburgh, PA), sectioned at 5 µm, and stained with hematoxylin and eosin (H&E), as described previously33 or processed for terminal dUTP nick-end labeling (TUNEL) or immunohistochemistry (IHC). For mammary tumorigenesis studies, tumors were excised, immersed in paraformaldehyde, embedded in paraffin blocks, and sections (5 µm) were obtained for further staining or in situ hybridization.
Whole Mount Preparations of Mammary Glands
Mammary glands were processed as described,34 according to the protocol in Biology of the Mammary Gland (http://mammary.nih.gov). The glands were spread on glass slides, fixed in Carnoy’s fixative (six parts 100% ethanol, three parts chloroform, one part glacial acetic acid) for 4 hours at room temperature, washed in 70% ethanol for 15 minutes, changes gradually to distilled water, rinsed in distilled water for 5 minutes, and stained overnight in carmine alum [1 g of carmine (C1022; Sigma, Toronto, ON, Canada)] and 2.5 g of aluminum potassium sulfate (A7167, Sigma) dissolved in 500 ml of H2O. Tissues were dehydrated in an increasing ethanol series (75%, 95%, and 100% for 15 minutes each) followed by xylene (1 hour) and mounted with Permount (Fisher, Ottawa, ON, Canada) and coverslip.
Immunohistochemistry (IHC)
IHC was performed essentially as described.35,36 After dewaxing, sections were treated with 0.3% hydrogen peroxide in phosphate-buffered saline (PBS), then blocked with 1% gelatin in PBS for 3 minutes at room temperature before incubation with rabbit polyclonal antibody to ß-casein (1:1600) for 1 hour at room temperature. This antibody was kindly provided by Dr. Ernst Reichmann (Ludwig Institute for Cancer Research, Laus-anne, Switzerland). This was followed by treatment with Vectastain ABC kit (Vector Laboratories, Inc., Burlingame, CA), as described by the manufacturer. Finally, the positive signal was visualized with 4,4-diaminobenzidine tetrahydrochloride (DAB).
Cell Proliferation
Bromodeoxyuridine (BrdU) (Sigma) was administered by intraperitoneal injection at 200 µg/g mouse weight. Mice were sacrificed by anesthesia with Avertin 90 minutes after BrdU administration and rinsed by intracardiac PBS perfusion and fixed with 4% paraformaldehyde, as described.33 BrdU-positive cells on tissue sections were detected with the BrdU in Situ Detection Kit II (BD Biosciences, Toronto, ON, Canada). The technique was essentially that recommended by the manufacturer. Mammary gland tissue analyzed was included in paraffin blocks together with several other tissues including small intestine and inguinal lymph nodes. These internal positive controls (intestinal epithelial cells in deep crypts and clusters of lymph node follicular cells were BrdU-positive) were checked before analysis and used to exclude a false BrdU signal.
TUNEL Assay
The detection of DNA fragmentation in cells undergoing apoptosis was made by using the TUNEL assay in situ on 5-µm paraffin sections from the same blocks used for BrdU incorporation. The technique was similar as before33 except that Biotin 11-dUTP (Roche, Laval, QC, Canada) and streptavidin-horseradish peroxidase (DAKO, Carpinteria, CA) and DAB substrates were used. To exclude a false TUNEL signal, the internal positive controls (apical intestinal epithelial cells and clusters of lymph node follicular cells were TUNEL-positive) were checked also before analysis.
Northern Blot and Reverse Transcriptase-Polymerase Chain Reaction Analysis
Total RNAs were isolated from different tissues with the Trizol reagent (Invitrogen, Toronto, ON, Canada), and 15 µg was electrophoresed on 1% formaldehyde-agarose gels, transferred by the Northern blot procedure to Hybond-N membranes (Amersham Co., Baie d’Urfé, QC, Canada) and hybridized with 32P-labeled probes as described previously.21
The Notch1 probe K was previously described.21
The Hes1 probe represents a 396-bp fragment at the 5' end of the cDNA, starting at the first ATG (+1). The U5 LTR probe represents a 210-bp SacI-BamHI fragment of the MMTV LTR (Figure 1)
. Probes were prepared by the random primer method, as described previously.21
Reverse transcriptase-polymerase chain reaction analysis was performed as described previously.37
|
Organs to be assessed were fixed with paraformaldehyde or periodate-polylysine-paraformaldehyde fixative, embedded in paraffin, and processed for in situ hybridization, as previously described.21 The Notch1intra riboprobe used represents a 279-nucleotide-long Bru36I fragment from the pMN7 construct.37 This fragment covers the intracellular C-terminal region of Notch1 from nucleotides 5357 to 5636. Tissues from non-Tg control animals hybridized with sense and anti-sense probes, as well as tissues from Tg mice hybridized with a sense probe, failed to exhibit any specific signal. For histopathological analysis, sections were stained with H&E.
Image Analysis
Quantification of mammary gland characteristics was performed on digital images of whole mount tissue, of H&E-stained sections and of TUNEL and IHC sections of inguinal mammary glands, by using Northern Eclipse 6.0 software (Empix Imaging, Mississauga, ON, Canada). The sections were scanned and color images were captured by using Zeiss Axiophot supported with Northern Eclipse. On whole mount tissue, penetration length nipple to front edge, diameter of primary and secondary branches, the number and diameter of terminal end buds (TEBs) were measured in virgin females of different ages. In addition, the length of tertiary branches and the density of alveoli from tertiary branches were counted in 12-week-old virgin females and during early (6.5 days) pregnancy. On H&E-stained sections, the percentage area occupied by fat, by alveoli, and by secretory vesicles was estimated in late (18.5 days) pregnancy and at 4 days of involution, by randomly choosing 10 images at x10 and x20 from each section. For evaluation of BrdU- and TUNEL-positive cells, the percentages of positive cells of the mammary epithelium were counted along with negative cells, by randomly choosing 10 images at x40 from each section. At least 6000 cells were counted for each structure such as TEBs, ducts, and alveoli. For quantitation of ß-casein production during pregnancy (18.5 days), nine images (x20) from each section were randomly taken from each inguinal mammary gland section with ß-casein IHC staining visualized by DAB (without counterstaining). In total, 677 alveoli in Tg and 644 alveoli in non-Tg tissues were quantitated by measuring the dark brown area (µm2) in each alveolus with Northern Eclipse software.
ß-Casein/Luciferase Reporter Assays
The prolactin receptor (PRLR) and the ß-casein-luciferase constructs (containing 2.8 kbp of sequences upstream of the rat ß-casein gene) were given by Dr. J. M. Boutin, Hôtel-Dieu Hospital, Montréal, QC, Canada. The Notch1intra construct has previously been described.21
Notch3intra was constructed similarly in pcDNA3 vector. The dominant-negative human RBP-J
construct38
was provided by Dr. Aly Karsan, BC Cancer Research Centre, Vancouver, BC, Canada. Transfection was performed as described earlier.39,40
Briefly, the human 293T cells were grown in Dulbecco’s modified Eagle’s medium (Life Technologies, Inc., Grand Island, NY) containing 10% fetal calf serum. Two days before transfection, cells were passaged at a density of 2.5 x 105/well, and transfected by using the calcium phosphate technique with a mixture of DNA constructs: the PRLR cDNA (0.2 µg), Notch1intra or Notch3intra (0.8 µg), ß-galactosidase (0.005 or 0.5 µg), and ß-casein reporter gene construct (0.5 µg). Transiently transfected 293T cells were induced with insulin (5 µg/ml), dexamethasone (1 µmol/L), and oPRL (5 µg/ml) for 48 hours. Cells were harvested and the ß-galactosidase activity was measured for normalization of luciferase activity. All experiments were repeated a minimum of three times.
Statistical Analysis
All data from Tg and non-Tg groups were tabulated as means ± SEMs. These include the number and diameter (µm) of TEBs, the length of penetration (mm), the diameter (µm) of primary, secondary, and tertiary branches, the length (µm) of tertiary branches, the density (number/mm) of alveoli from tertiary branches, the area occupied by ß-casein-positivity in each alveolus or fat area on surface area or vesicles (%) relative to that occupied by alveoli (%), the percentage of TUNEL- and BrdU-positive cells. Comparison between Tg and non-Tg groups was performed by using an impaired two-tailed Student’s t-test. Differences were considered significant if P was <0.05.
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
The biological role of mutated truncated murine Notch1intra in mammary gland development and tumorigenesis was assessed in Tg mice expressing this gene in mammary epithelial cells under the regulatory sequences of the MMTV LTR (MMTV/Notch1intra) (Figure 1A)
. Four MMTV/Notch1intra Tg founders (F) were generated (F47400, F39580, F45347, and F51256). With the exception of one female founder 39580 that had a sterile male as offspring, the remaining founders transmitted the transgene to their progeny in a Mendelian ratio. Southern blot analysis showed that the structure of the Tg was grossly intact (data not shown). Mice were bred as heterozygotes with CD1 mice, for at least 6 generations and for 10 to 15 generations for most studies. Mice from two of the four Tg lines (F45347, F51256) did not show any abnormal phenotype, expressed low levels of the transgene, and were not studied further (data not shown). Tg males from lines expressing the transgene (Figure 1B)
were sterile (to be reported separately). Female Tg founder mice F47400 and F39580 were fertile and expressed the transgene (see below and data not shown). They were mated successfully and carried their pups to term, but were unable to feed them. The pups died within 48 to 72 hours with no visible milk in their stomach. In view of this failure of lactation, these Tg lines were thus maintained by foster nursing to non-Tg CD1 mothers.
Impaired Ductal Growth and Ductal Dysplasia in the MMTV/Notch1intra Tg Mammary Glands
The defective milk production observed in MMTV/Notch1intra Tg females was indicative of an impaired development of the mammary glands. We therefore first assessed the growth of the mammary gland ductal tree in these Tg mice. In normal mice, the ductal system does not develop until puberty (5 to 6 weeks). During puberty (5 to 8 weeks) a progressive outgrowth of the primary ducts with secondary and tertiary branching fills the subcutaneous fat pad. This process is driven by the growth of a highly proliferative cell mass, designated the TEBs, which eventually disappear in mature virgin animals (12 weeks), once arborization is completed.
Whole mount and histological analysis of the mammary glands of prepubescent Tg mice (3 weeks) showed that the ducts appear normal, as compared to those from control non-Tg littermates (data not shown), suggesting that overexpression of Notch1intra did not affect the precursor cells giving rise to the primitive ductal tree. A similar whole mount analysis performed in Tg mice at puberty (5, 7, and 12 weeks) revealed no significant differences between Tg and non-Tg females. The structures evaluated include penetration length, diameter of primary and secondary branches, the number and diameter of TEBs, the length of tertiary branches, and density of alveoli from tertiary branches (data not shown). However, in some Tg mice, some ducts were dilated and show small foci of hyperplasia (Figure 2B)
. Older (6 to 12 months) virgin Tg mice were found to have enlarged primary ducts and reduced secondary and tertiary branches (data not shown). The paucity and enlargement of the Tg ducts in older Tg mice was confirmed on histological sections. Moreover, in these older Tg mice, dysplasia of the ducts was evident (Figure 2D)
.
|
or late (18.5 days) (Figure 3F)
pregnancy, including length of tertiary branches (Figure 3C)
and density of alveoli from tertiary branches (Figure 3D)
were much less pronounced in Tg than in non-Tg mice. In older pregnant Tg mice, the ducts became enlarged, distorted, and dysplasic, having two to four layers of epithelial ductal cells, as compared to a single layer in non-Tg mice (data not shown).
|
. Histological observations confirmed these results (data not shown). This phenotype was even worse after a first pregnancy at older age (data not shown) and after the third pregnancy, in which very large Tg ducts were observed (Figure 4, C and F)
. In older Tg mammary tissues, a severe dysplasia, with larger and abnormally formed ducts, was observed and a severe hyperplasia of the ductal epithelial cells was apparent (data not shown). These results indicated that overexpression of murine Notch1intra prevents a normal development of the mammary ducts and induced dysplasia and hyperplasia of the ductal tree, rendering it refractory to the involution process.
|
In normal mice, the lobules and their alveoli are formed at puberty and remain in a primitive form at puberty and in virgin mice. They start to grow significantly in early pregnancy and continue their growth and differentiation during pregnancy to form an epithelial rich glandular structure with mature alveoli.
Whole mount analysis showed that the lobules developed during pregnancy of Tg mice, but these lobules remained small and their numbers were severely reduced as compared to those of non-Tg littermates. This was observed both in early (6.5 days) (Figure 3B)
and late (18.5 days) (Figure 3F)
pregnant Tg females. In pregnant Tg mice, the lobulo-alveolar proliferation never resulted in the complete filling of the fat pad at parturition, as seen in the non-Tg mammary glands (Figure 3, A and B and E and F)
. On histological sections, underdeveloped small lobules with small alveoli could be observed (Figure 3, H and K)
. The Tg alveoli were also undifferentiated and their cells contained less fat droplets and a lower percentage area of secretory vesicles than in non-Tg mice (Figure 3, J–L)
. This resulted in a significant increase in interstitial fat area (Figure 3I)
. During involution, after weaning, after the first pregnancy the underdeveloped lobular/alveolar Tg structure regressed at a slower rate as compared to those of non-Tg mice, as evaluated in the percentage of fat area (Figure 3, G–I
; regression of 35% in non-Tg versus 14% in Tg mice). Together, these results indicated that murine Notch1intra overexpression in mammary glands not only affects ductal development, but also the lobulo-alveolar proliferation associated with pregnancy.
Notch1intra Represses the Production of ß-Casein in Vivo and the Transcriptional Activity of the ß-Casein Gene Promoter
To determine whether this apparent lobulo-alveolar proliferation defect of the mammary glands of MMTV/Notch1intra Tg mice was also reflected in the milk composition, the levels of ß-casein, a major milk protein, were measured by IHC. Production was high during gestation of normal non-Tg mice as expected, but was much lower in Tg mammary glands (Figure 5, A and B)
. A similar decrease in ß-casein protein levels was also documented in murine mammary epithelial HC11 cells overexpressing Notch1intra relative to control HC11 cells (data not shown).
|
, suggesting that the Notch1intra-mediated repression occurs at the level of the ß-casein gene promoter itself. However, although this promoter contains two CSL binding sites, the inhibitory action of Notch1intra on this promoter was not relieved by expression of a dominant-negative mutant of RBP-J/CBF1 (Figure 5C)
, suggesting that this repression is CBF1-independent. Moreover, Hes1 was not significantly enhanced in mammary glands from pregnant (18.5 days) Tg females (data not shown), suggesting that the development of the mammary phenotypes in MMTV/Notch1intra Tg mice may be Hes1-independent. Decreased Proliferation of Ductal and Alveolar Epithelial Cells of MMTV/Notchintra Tg Mammary Glands during Rapid Expansion at Puberty and in Early Pregnancy
The dysplasia observed in the ductal epithelial cells suggested that murine Notch1intra was affecting proliferation of these cells. To determine whether cells of the Tg mammary gland proliferated at a different rate than in non-Tg mice, proliferation was directly measured by counting the percentage of cells stained with anti-BrdU antibodies after a BrdU-labeling period in vivo (Figure 6, A–H)
. This analysis revealed that proliferation of ducts was significantly reduced in young (5 weeks), but not in older (12 weeks) Tg virgin females (Figure 6G)
, whereas the proliferation of TEBs in young (5 weeks) virgin Tg mice was not significantly affected (Figure 6G)
. During pregnancy, the proliferation of ducts was reduced early (6.5 days), but enhanced significantly later (18.5 days) (Figure 6, B and G)
. A similar analysis of alveoli revealed a significant decrease in incorporation of BrdU in Tg females during early pregnancy (Figure 6, D and G)
, but paradoxically an increase in late pregnancy (Figure 6G)
. During involution (4 days) after the first pregnancy, there was a trend for increased proliferation in Tg ducts or lobules as compared to non-Tg ones (Figure 6G
and data not shown), but this did not reach a statistical significance, even after a second experiment (data not shown). However, at 4-day involution after the third pregnancy, proliferation of the epithelial cells of the enlarged Tg ducts (but not those of the normal-size Tg ducts) was significantly increased, as compared to those from the non-Tg ducts (Figure 6, E, F, and H)
. In contrast, proliferation of Tg alveolar epithelial cells remained unchanged relative to that of non-Tg mice at this stage (data not shown).
|
Apoptosis of Ductal and Alveolar Epithelial Cells of Mammary Glands Remains Unchanged in MMTV/Notch1intra Tg Mice
In normal mice, apoptosis is mainly responsible for the regression of the ductal tree and of the lobulo-alveolar structure that occurs during involution. The fact that, during involution, the Tg ducts remained large and dysplastic and that Tg alveoli poorly regressed, suggested that apoptosis may be impaired and/or delayed. We measured the levels of apoptosis in mammary glands of Tg mice and of their non-Tg littermates at the same stage of gland development using the TUNEL assay. The percentage of TUNEL-positive cells in ducts and TEBs of virgin (5 weeks), in alveoli during pregnancy (18.5 days), and during involution of mammary glands after the first pregnancy was not statistically different for Tg and non-Tg females (data not shown). However, at 4-day involution after the third pregnancy, the percentage of apoptotic ductal epithelial cells was significantly increased in the Tg enlarged ducts, but not in the normal-size Tg ducts, as compared to the non-Tg ones (Figure 4, G–I)
. These results indicated that murine Notch1intra does not appear to affect the apoptotic process at different stages of the mammary gland development at an early age, but definitively has a major proapoptotic effect during involution of the gland after successive pregnancies.
Mammary Tumorigenesis in MMTV/Notch1intra Tg Mice
The histological examination of mammary glands from medium-age Tg females, revealed the presence of focal cell proliferation arising next to the normal mammary epithelial tissue (Figure 7D)
(data not shown). These lesions thus suggested a premalignant condition. As these Tg females from lines 47400 (n = 25) and 39580 (n = 1) aged and after a few months of breeding, they developed rapidly growing, solid tumors by 7 to 10 months of age (Figure 7A)
. Three of four mammary tumors (line 47400) injected in nude mice were able to grow (data not shown), confirming their malignant nature. However, there was no evidence of metastases in lungs and liver. These data indicated that mammary tumors developed in Notch1intra-expressing mammary glands of Tg mice after one or a few pregnancies.
|
, indicating that mammary tumors induced by overexpression of murine Notch1intra can be totally pregnancy/lactation-independent.
Northern blot analysis on RNA extracted from these mammary tumors showed that the Tg was expressed at high levels, at the expected 3 kb size (Figure 7C)
. Moreover, in situ hybridization with a Notch1-specific riboprobe revealed that the proliferating mammary epithelial cells in the tumors expressed much higher level of the transgene than the adjacent morphologically nonneoplastic mammary glands (Figure 7D)
.
Histological examination of the mammary gland lesions showed that they were of epithelial origin and ranged from hyperplasia to invasive carcinoma (Figure 8)
. The hyperplastic changes included an increased number of ducts and intraductal epithelial proliferation of various pattern and intensity. In some ducts, the epithelial cells formed tufts and papillary fronds projecting into the lumen (Figure 8A)
. The epithelial proliferation was often more florid filling and distending the ducts with residual lumina being reduced to narrow spaces (Figure 8, B and C)
. Foci of atypical hyperplasia and of carcinoma in situ were present (data not shown). Poorly (Figure 8E)
to moderately (Figure 8D)
differentiated invasive carcinomas were observed. The moderately differentiated carcinomas presented gland-like differentiation and secretion (Figure 8D)
. Marked nuclear polymorphism and numerous mitosis were observed (Figure 8E)
. Areas of necrosis were often found within the tumors (Figure 8F)
. Together, these data indicated that Notch1intra behaves as an oncogene when overexpressed in mammary glands of Tg mice. However, exophthalmia (caused by Harderian gland hyperplasia) and salivary, submandibular, or nasal mucosal/submucosal gland hyperplasia were not observed (data not shown), in contrast to that reported for MMTV/Notch4intra Tg mice.16
|
Because overexpression of truncated murine Notch4intra16,18
or murine Notch1intra (Figures 3 and 8)
were found to prevent mammary gland development and to induce mammary tumors in Tg mice, we investigated whether expression of another member of the murine Notch gene family, Notch3, will also induce similar developmental and oncogenic phenotypes in mammary glands. Notch3 has been reported to be overexpressed in human breast tumors.42
In addition, Notch3intra has already been reported to induce T-cell leukemia when expressed in T cells of Tg mice.43,44
The MMTV/Notch3intra DNA having a structure very similar to that of the MMTV/Notch1intra, was constructed (Figure 9A)
. Eight Tg founders (F85824, F85825, F85826, F85827, F85834, F85835, F85836, F85837) were generated. Among these, three founders (F85827, F85834, F85835) did not give rise to progeny and were considered sterile. One female founder (F85836) was found dead before a line could be established. Among the mice from other founders from which lines could be established, only one (F85824) expressed the transgene in mammary glands, while the three others (F85825, F85826, F85837) did not. In fact, expression of the Notch3intra transgene in the F85824 founder was much higher than that of Notch1intra (F47400), when measured with a probe (U5 LTR) common to both transcripts (Figure 9B)
. Tg mice from this latter founder (F85824) were expanded. Expression of Hes1 in mammary glands of pregnant MMTV/Notch3intra Tg mice was much higher than in non-Tg mice (Figure 9C)
. This contrasted with the absence of detectable induction of Hes1 in mammary glands of MMTV/Notch1intra Tg mice (see above).
|
. Further analysis of the mammary glands at the different stages (virgin, pregnancy at 6.5 and 18.5 days, and involution) showed similar abnormalities in MMTV/Notch3intra Tg mice as those described above for MMTV/Notch1intra Tg mice: those include enlarged primary ducts in older virgin Tg mice, reduced ductal Tg arborization, and decreased density of Tg alveoli in early pregnancy, reduced proliferation in ducts of virgin Tg mice and in ducts and in alveoli of 6.5-day pregnant Tg mice (data not shown), lower number of secretory vesicles (Figure 9, H and J)
, severe decrease of ß-casein production (Figure 9J)
, and significantly enhanced proliferation of alveoli during late pregnancy (Figure 9, L and M)
relative to non-Tg mice (Figure 9K)
. In addition, Notch3intra was as efficient as Notch1intra in repressing the ß-casein promoter (Figure 5C)
. However, in late (18.5 days) pregnancy, some phenotypes were clearly different from those observed in MMTV/Notch1intra Tg mice. In particular, in contrast to the reduced lobulo-alveolar development of MMTV/Notch1intra Tg mice, the Notch3intra-expressing lobules and alveoli were significantly expanded (Figure 9F)
as compared to those of non-Tg mice (Figure 9E)
. The proliferation of ductal epithelial cells was not enhanced, but rather showed a modest decrease (Figure 9M)
. The alveolar epithelial Tg cells were hyperplastic, showing an increased cell density with large, dark nuclei and paucity of cytoplasm (Figure 9H)
. Moreover, apoptosis was significantly enhanced in alveoli and ducts of these pregnant (18.5 days) Tg mice relative to that of non-Tg alveoli (Figure 9, N–P)
. Therefore, it appears that constitutive overexpression of murine Notch3intra, like murine Notch1intra, in mammary glands severely affects mammary gland development and leads to tumor formation. |
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Expression of Murine Notch1intra or Notch3intra Blocks the Development of Mammary Glands
One striking phenotype observed in these MMTV/Notch1intra or MMTV/Notch3intra Tg females is their inability to nurse and feed their pups. Whole mount and histological analysis of MMTV/Notch1intra Tg mice showed inhibition of mammary gland development, affecting both ductal and alveolar growth. This developmental defect resulted in decreased ß-casein milk protein levels and impaired milk production. The small amount of residual milk being produced in the glands may not be secreted either, because pups from Tg females had no visible milk in their stomach.
The role of Notch1 or Notch3 signaling in mammary gland development has not been studied extensively. The fact that Notch1intra or Notch3intra is capable of having such a profound effect on mammary gland development, suggests that the mammary epithelial cells express their downstream effectors. Gain-of-function mutants of Notch1, such as Notch1intra, have previously been shown to affect mammalian cell fate determination and to prevent differentiation of other cells, namely myeloid cells,45-47 CD34+ progenitor cells,48 myoblasts,10,49-51 neurons,10 oligodendrocytes,52 and T-cells.53-56 Our results extend this effect of Notch1 or Notch3 to another tissue, the mammary gland.
A very similar phenotype of impaired mammary gland differentiation has previously been reported in Tg mice expressing murine Notch4intra under the regulation of the whey acidic protein or MMTV promoter,16,18 suggesting that overexpressed Notch1, Notch3, and Notch4 intra-cellular domains are likely to interact through the same effector(s) in mammary epithelial cells. Because Notch4intra is shorter than Notch1intra or Notch3intra, only motifs shared by the three molecules are likely to be involved in such interactions. These results were not unexpected considering that these molecules share relatively a high degree of structural homology. However, they nevertheless remained intriguing in view of a recent report showing that Tg mice expressing human Notch1intra do not exhibit impaired mammary gland development (see below).22
Defective mammary gland development has also been reported in other mutant mice, harboring overexpressed genes, such as ß1,4 galactosyltransferase,57 parathyroid hormone or parathyroid hormone-related protein,58 whey acidic protein,59 mutant p53,60 neuregulin, and hepatocyte growth factor.61 Also, impaired mammary gland differentiation has been reported in some gene-deficient mutant mice. This is notably the case for mice made defective for cyclin D1,62 EGF and amphiregulin,63 progesterone64 and prolactin65 receptors, activin/inhibin RB,66 A-myb,67 C/EBPß,68,69 CSF-1,70 TIMP-1,71 osteoprotegenin ligand/RANKL/TRANCE and RANK,72 Id2,73 or Stat5a74 gene. Although some characteristics of these mutant mice are similar for some aspects to the Notch1intra or Notch3intra phenotype observed here, they nevertheless show distinct features,62,72-74 making comparison between these phenotypes not easy and possibly hazardous. Future work may determine whether Notch1 or Notch3 signaling involves the transcriptional down-regulation of the expression of some of these molecules, as we have documented for ß-casein. These pathways may not be identical in view of some differences observed between Notch1intra and Notch3intra Tg mice: hyperplasia and expansion of Notch3intra-expressing alveoli versus reduced expansion of Notch1intra-expressing alveoli.
Murine Notch1intra and Notch3intra Behave as Oncogenes in Mammary Epithelial Cells
We found that overexpression of Notch1intra or Notch3intra was not only affecting mammary gland development, but also induces mammary tumors at high frequency in Tg mice. Tumor types were diverse and represented the full spectrum of differentiation from hyperplasia, in situ ductal carcinoma, glandular adenocarcinoma, to poorly differentiated adenocarcinoma. These results on the oncogenic function for Notch1intra in vivo confirm and extend our previous data in vitro showing that Notch1intra can transform HC11 mammary epithelial cells in culture.21 Mammary tumor formation occurs in these Tg mice after a relatively long latency (4 to 15 months) and stochastically, suggesting that Notch1intra or Notch3intra is not sufficient by itself to promote transformation and requires the action of other oncogenes or the inactivation of tumor suppressor genes. The transforming ability of Notch1intra or Notch3intra documented here for mammary epithelial cells has previously been reported for T cells.43,44,75 In addition, overexpression of Notch1 and Notch3 in human breast tumors has recently been reported to be associated with poor overall survival.42 However, Notch1intra may have an oncogenic potential more restricted than other oncogenes. In contrast to the activated Ha-RasV12 or the truncated erbB2/neu or int3/Notch4intra oncogenes that induce frequent hyperplasia of the Harderian glands, when expressed under the regulation of the same MMTV LTR,16,30,32,76,77 we have not observed such a phenotype with Notch1intra or Notch3intra. MMTV/Notch1intra did not develop either hyperplasia and adenocarcinoma of salivary glands and hyperplasia of mucosal/submucosal glands described in MMTV/Notch4intra Tg mice.16 It is possible however that levels of Tg expression in these tissues may explain some of these differences.
The mammary tumor phenotype induced by Notch1intra or Notch3intra Tg mice is very similar to the phenotype reported by Jhappan and colleagues16 in MMTV/int3/Notch4intra Tg mice. In both cases, the mammary tumors are diverse and at different stages of differentiation. However, one important difference could be noticed: whereas mammary tumors develop in both male and female MMTV/Notch4intra Tg mice,16 we did not observe mammary tumors in males of our MMTV/Notch1intra or MMTV/Notch3intra Tg mice. This may simply reflect the different background of the mouse used in both experiments. Alternatively, it could suggest that the molecular requirement for transformation of the male mammary glands may be more stringent and that the Notch1intra or Notch3intra protein is lacking elements, present in Notch4intra, to meet these requirements.
In the Notch1intra and Notch3intra Tg mice studied here, the levels of expression of Notch3intra were much higher than those of Notch1intra. In addition, expression of Notch3intra in mammary glands induced high expression of Hes1, whereas Hes1 was not induced in Notch1intra-expressing mammary glands. Yet, the oncogenicity of Notch3intra is not higher, but rather appears lower than that of Notch1intra, suggesting intrinsic differences between these two molecules in their oncogenic potential. This may be reflected in the reported statistically stronger association of high-level Notch1 than Notch3 expression with overall survival in patients with breast cancer.42 Intriguingly, Hes-1 was reported to be up-regulated in lactating Tg mice expressing human Notch1intra in mammary glands.22 Because Hes-1 levels were not measured in that study at the same stages of mammary gland development as we did (late pregnancy), direct comparison of these data remains difficult.
Is the Sequence Evolution of Notch1intra Biologically Relevant?
Surprisingly, phenotypes of impaired mammary gland differentiation and of mammary tumor formation observed here in Tg mice expressing murine Notch1intra contrast significantly with recent data obtained in similar Tg mice expressing human, instead of murine, Notch1 intracellular domain under the control of the same MMTV LTR promoter.22
First, in these Tg mice, alveolar development did not appear to be compromised and absence of lactation was not observed, although some elongated ducts and reduced side branching could be documented in 5-week-old Tg mice. Second, increased ductal proliferation was observed in these young (5 weeks) virgin Tg females, in contrast to the decreased duct proliferation we observed at this stage (Figure 6G)
. Finally, the mammary tumor phenotype reported for the human Notch1intra-expressing Tg mice22
contrasts significantly with the one induced by murine Notch1intra, Notch3intra (this study), and Notch4intra.16,18
In Tg mice expressing any of these murine or human Notch genes, nonregressing mammary tumors were induced after a relatively long latency, suggesting common mechanisms. However, Tg mice expressing human Notch1intra developed a novel phenotype of lactation-dependent regressing mammary tumors, not observed with the other murine Notch1intra, Notch3intra, and Notch4intra transgenes, suggesting perturbation of a novel pathway by the human Notch1intra molecule.
It is not clear why such distinct phenotypes were observed. This may be related to the genetic mouse background used [CD1 (this study) versus mixed C3H x C57BL/6 x FVB], although unlikely. It could also reflect different levels of expression of the transgene, although this is not very likely either in view of the fact that both groups of Tg mice (mouse versus human Notch1intra) developed nonregressing mammary tumors at high incidence. Rather, the different phenotype elicited by human and murine Notch1intra may be directly related to their different amino acid sequences. These changes could prevent or favor the interaction of human Notch1intra with key murine effector(s) required to elicit such a block of differentiation or such pregnancy/lactation-dependent tumor growth. Comparison of murine and human Notch1intra amino acid residues in fact reveals complete identity of the transmembrane and the cdc1, -2, -5, and -6 ankyrin repeats and the PEST domain, very high identity of the cdc3 and -4 domains (with a single conserved amino acid difference in each), as well as of the RAM domain (a single amino acid difference, D1836S). However, the two molecules show significant differences in the amino acid residues of the region C-terminal to the last ankyrin repeat, including several conservative modifications and 18 nonconservative changes, 15 of which involve a proline. However, despite these differences, the human Notch1intra construct still behaves as the murine Notch1intra in the ß-casein/luciferase reporter assay (data not shown).
This comparative analysis suggests that the region C-terminal to the last ankyrin repeat may be involved in the biological differences observed after expression of human or murine Notch1intra in the mammary glands of Tg mice. It appears that this motif has significantly changed during evolution to a point of being incapable of interacting with murine effectors from the mammary glands, and possibly from other organs in a similar way as murine Notch1intra.
In conclusion, the MMTV/Notch1intra and MMTV/Notch3intra Tg mice represent novel animal models to study Notch1 or Notch3 signaling in vivo, especially in mammary epithelial cells, both in the context of differentiation and oncogenesis. This appears relevant in view of the recent observation that their high levels in some human breast tumors is associated with poor overall survival.42
|
Acknowledgements |
|---|
construct. |
Footnotes |
|---|
Supported by the Canadian Institute of Health Research and by the Canadian Breast Cancer Research Alliance through the National Cancer Institute of Canada (to P.J.).
C.H., A.D., and M.L. contributed equally to this work.
Accepted for publication November 29, 2005.
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
