Transforming growth factor–β (TGF-β) isoforms are growth factors that function physiologically to regulate development, cellular proliferation, and immune responses. The role of TGF-β signaling in mammary tumorigenesis is complex, as TGF-β has been reported to function as both a tumor suppressor and tumor promoter. To elucidate the role of TGF-β signaling in mammary gland development, tumorigenesis, and metastasis, the gene encoding type II TGF-β receptor, Tgfbr2, was conditionally deleted in the mammary epithelium (Tgfbr2MGKO). Loss of Tgfbr2 in the mammary epithelium results in lobular-alveolar hyperplasia in the developing mammary gland and increased apoptosis. Tgfbr2MGKO mice were mated to the mouse mammary tumor virus-polyomavirus middle T antigen (PyVmT) transgenic mouse model of metastatic breast cancer. Loss of Tgfbr2 in the context of PyVmT expression results in a shortened median tumor latency and an increased formation of pulmonary metastases. Thus, our studies support a tumor-suppressive role for epithelial TGF-β signaling in mammary gland tumorigenesis and show that pulmonary metastases can occur and are even enhanced in the absence of TGF-β signaling in the carcinoma cells.
- mammary gland
Transforming growth factor–β (TGF-β) acts as a potent growth inhibitor of most epithelial cell types (1). It arrests the cell cycle in the G1 phase thereby tightly regulating cell proliferation and potentially exerting a tumor-suppressive role (reviewed in ref. 2). A tumor suppressor role for TGF-β signaling is supported by transgenic mouse studies demonstrating that overexpression of TGF-β1 [mouse mammary tumor virus (MMTV)-TGF-β1] in the mammary epithelium suppresses mammary tumor formation (3). Additionally, overexpression of a dominant-negative mutant form of the type II TGF-β receptor in mammary epithelium results in lobular-alveolar hyperplasia and formation of mammary tumors after a long latency (4–6) . Although inactivating mutations in TGFBR2 or other TGF-β signaling pathway genes, such as Smads, are quite rare in human breast cancers, loss or down-regulated expression of TβRII in human breast cancers has been shown through immunohistochemistry, and this reduced expression correlates with a higher tumor grade (7, 8) . Furthermore, decreased expression of the type II receptor (TβRII) may be an early event in human breast tumorigenesis because women with breast biopsies containing epithelial hyperplasia lacking atypia and a reduced level of TβRII immunostaining cells had a 3.41-fold increased risk for developing invasive breast cancer compared with women with similar lesions having high levels of TβRII immunostaining (9).
Paradoxically, it has been documented that many tumor cells, including breast carcinoma cells, become resistant to TGF-β-mediated growth inhibition, and the tumor cells often increase the production of one or more of the TGF-β isoforms. In response to increased production of TGF-β, tumor cells, which are now resistant to TGF-β's growth inhibitory effects, become more invasive and metastatic (reviewed in refs. 10, 11 ). In addition, several studies using inhibitors of TGF-β have shown that systemic inhibition of TGF-β signaling suppresses metastasis of mammary carcinomas (12, 13) .
To further elucidate the complex role of TGF-β signaling in mammary carcinogenesis, we conditionally inactivated TβRII using Cre/Lox technology (14). We report here that conditional loss of TβRII in the mammary gland epithelium results in lobular-alveolar hyperplasia, similar to the phenotype seen in MMTV-DNIIR animals (5). In addition, MMTV-PyVmT transgene expression in the context of Tgfbr2 knockout results in a shortened latency of tumor formation and an increase in the size of metastatic lung lesions. These studies support the role of TGF-β as a tumor suppressor in mammary gland development and tumorigenesis and show that TGF-β signaling in the mammary epithelium functions to inhibit mammary tumor formation. The data further show that pulmonary metastases are enhanced with loss of TGF-β signaling in carcinoma cells.
Materials and Methods
Generation and Characterization of Tgfbr2flox/flox, Tgfbr2MGKO, MMTV-PyVmT/Tgfbr2flox/flox, and MMTV-PyVmT/Tgfbr2MGKO Mice. Generation of the Tgfbr2flox/flox mice (C57/B6) has been described previously (14). These mice were mated with MMTV-Cre mice (FVB) to generate Tgfbr2mammary gland knockout (Tgfbr2MGKO) mice (FVB/C57/B6; refs. 15, 16 ). The Tgfbr2flox/flox mice were bred 10 generations into the FVB genetic background. Female Tgfbr2flox/flox (FVB) and Tgfbr2MGKO mice (FVB) were mated with MMTV-PyVmT male mice (FVB) to generate MMTV-PyVmT/Tgfbr2flox/flox (FVB) and MMTV-PyVmT/Tgfbr2MGKO mice (FVB). All mice were housed in the Animal Care Facility at Vanderbilt University following the Association for the Assessment and Accreditation of Laboratory Animal Care guidelines.
Identification of Transgenic Mice. Genotypes of Tgfbr2flox/flox, Tgfbr2MGKO, MMTV-PyVmT/Tgfbr2flox/flox, and MMTV-PyVmT/Tgfbr2MGKO animals were determined by using oligonucleotide primers as described previously (14, 17) .
Southern Blot Analysis. Genomic DNA was extracted from mammary tumors and lung metastases and probed as described previously (14).
Isolation and Culture of PyVmT Cells. Mammary tumors were removed and digested at 37°C for 4 hours in serum-free DMEM: F12 + 100 units/mL pen-strep, 250 μg/mL ampho B, 10 mg/mL gentamicin, 2 mg/mL collagenase, and 100 units/mL hyaluronidase. Cells were washed five times with PBS containing 5% adult bovine serum. Cells were plated in flasks coated with 50 μg/mL of collagen in 0.02 N acetic acid. Cells were maintained in DMEM/F12 medium containing 2% adult bovine serum.
Cell Proliferation Assays. For sequential cell-counting experiments, cells were plated at 50,000 cells per well in 6-well plates in complete growth medium and incubated for 24 hours. Cells from three replicate wells were then counted (day 0) using a Coulter Counter (Beckman Coulter, Inc., Fullerton, CA). The remaining wells were treated and counted in triplicate at 24-hour intervals for 7 days.
Reverse Transcription-PCR Analysis for Tumor-Derived Cell Lines. Primers used for PCR were TBRII-R (exon 6) 5′-AATTTCCGGCCGCCCTCGGTCT-3′, TBRII-F (exon 4) 5′-CCCGGGGCATCGCTCATCT-3′, TBRII-F (exon 2) 5′-TAACAGTGATGTCATGGCC-3′, TBRII-R (exon 3) 5′-GGAAGTACTGTGTGAACCC-3′; TBRIII-F 5′-AACTTCTCCTTGACAGCAGA-3′, TBRIII-R 5′-CACTCTCTTTTCCAAAGCC-3′; GAPDH-F 5′-CTGGCATGGCCTTCCGTG-3′, 5′-GAAATGAGCTTGACAAAG-3′.
Histologic Analysis and Immunohistochemistry. Mammary glands and tumors were harvested and immediately fixed in 4% paraformaldehyde/PBS at 4°C overnight. Hematoxylin-stained whole mounts of #4 inguinal glands were sectioned and stained as described previously (18). Apoptosis was visualized with the DeadEnd Colorimetric TUNEL System (Promega Co., Madison, WI). For proliferating cell nuclear antigen staining, sections were incubated with rabbit anti-proliferating cell nuclear antigen (Santa Cruz Biotechnology, Santa Cruz, CA). β-Galactosidase activity assays were modified from a previous report (19).
Western Analyses. Whole cell extracts were prepared as previously described (20). For whole tumor lysates, ice-cold lysis buffer (described above) was added and tissue was homogenized on ice. Tissue lysates were clarified by centrifugation for 15 minutes at 12,000 rpm, 4°C. Protein concentrations were determined as described above. Immunoblotting, for the analysis of Smad2/Smad3 (Santa Cruz Biotechnology), Smad2 phosphorylated on Ser465 (Cell Signaling, Beverly, MA), AKT (Cell Signaling), AKT phosphorylated on Ser473 (Cell Signaling), and β-tubulin was done as previously described (20). The β-tubulin monoclonal antibody developed by Michael Klymkowsky was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Resources and maintained by the Department of Biological Sciences, University of Iowa, Iowa City, IA 52242.
Mice with Conditional Knockout of Tgfbr2 Using MMTV-Cre(Tgfbr2MGKO) Lack Expression of Tgfbr2 in the Mammary Epithelium. MMTV-Cre mice were bred to Tgfbr2flox/flox/Rosa26 reporter mice to generate Tgfbr2MGKO/Rosa26r mice to verify recombination of the mammary epithelium (21). In whole mount preparations and sections of 6-week-old virgin mammary glands, LacZ staining was observed only in the ducts and terminal end buds of Tgfbr2MGKO/Rosa26r mice, whereas no staining was detected in Tgfbr2flox/flox/Rosa26r mammary glands ( Fig. 1A ). Furthermore, recombination in the ductal epithelial cells of the mammary glands in Tgfbr2MGKO mice was determined by laser capture microdissection followed by genomic DNA extraction and PCR. The 241-bp PCR product indicative of the recombined allele is detected only in Tgfbr2MGKO mice, but not in Tgfbr2flox/flox mice ( Fig. 1B and C). Whereas these studies were done in a mixed genetic background, similar results have been obtained in a pure FVB background (data not shown).
Tgfbr2MGKO Mice Exhibit Lobular-Alveolar Hyperplasia. After demonstrating recombination of Tgfbr2 in the mammary epithelium of Tgfbr2MGKO mice, we analyzed the developmental phenotype of the mammary gland. By whole mount preparations and sectioning of the inguinal glands we observed that a majority of the mice exhibit lobular-alveolar hyperplasia between 11 and 19 weeks of age when compared with age-matched Tgfbr2flox/flox mice ( Fig. 2A ). Interestingly, this phenotype seems to regress after 20 weeks of age. Tgfbr2MGKO mice exhibit a higher rate of apoptosis when compared with Tgfbr2flox.flox mice as detected in mammary gland sections by terminal deoxynucleotidyl transferase–mediated nick end labeling analysis when an equal number of cells of each genotype were counted ( Fig. 2B). The significant increase in the rate of apoptosis in Tgfbr2MGKO mice could contribute to the regression of the hyperplastic phenotype, and additional studies are under way investigating this possibility.
Conditional Loss of Tgfbr2 in the Mammary Epithelium Shortens the Latency of Tumor Formation. To determine the effects of complete loss of TGF-β signaling in the mammary gland on mammary tumor metastasis, we crossed the Tgfbr2MGKO mice with the MMTV-PyVmT mice, both in the FVB background and followed the mice until they reached 100 days of age. Conditional loss of TβRII in the mammary gland shortened the median latency of tumor formation from 71 to 47.5 days ( Fig. 3A ). Both the PyVmT/Tgfbr2flox/flox and PyVmT/Tgfbr2MGKO mice develop highly invasive mammary adenocarcinomas with no apparent histologic differences in the primary tumors, and there is no detectable effect of knockout of Tgfbr2 on the number or size of primary mammary carcinomas (data not shown).
Conditional Loss of Tgfbr2 in the Mammary Epithelium Increases Pulmonary Metastases in MMTV-PyVmT Transgenic Mice. Our current understanding of the role of TGF-β signaling in metastasis would predict that knockout of Tgfbr2 in the mammary epithelium would decrease the incidence of pulmonary metastases in MMTV-PyVmT mice. However, the growth of metastatic lesions in the PyVmT/Tgfbr2MGKO mice was strikingly increased when compared with the PyVmT/Tgfbr2flox/flox mice. As stated previously, PyVmT/Tgfbr2flox/flox mice and PyVmT/Tgfbr2MGKO were followed until the mice reached 100 days of age. At this point, the mice were sacrificed and the full necropsies were done. However, to rule out the possibility that the increase in metastatic burden was simply the result of enhanced rate of primary tumorigenesis in the PyVmT/Tgfbr2MGKO mice, mice of both genotypes were followed for 45 days after the first palpable tumor was noted. At either time point, there was a dramatic increase in the number and size of pulmonary metastases present in the PyVmT/Tgfbr2MGKO mice. The massive number of pulmonary metastases in the PyVmT/Tgfbr2MGKO mice made it impossible to accurately count surface metastases; therefore, wet lung weights were recorded and used as a measure of metastatic burden. PyVmT/Tgfbr2flox/flox mice sacrificed 45 days after the first palpable tumor appeared had a mean wet lung weight of 0.169 ± 0.018 g compared with PyVmT/Tgfbr2MGKO mice that had a mean wet lung weight of 0.314 ± 0.082 g (P = 0.0049, n = 5 for each genotype). Similarly, PyVmT/Tgfbr2flox/flox mice sacrificed at 100 days of age had a mean wet lung weight of 0.182 ± 0.056 g compared with PyVmT/Tgfbr2MGKO mice that had a mean wet lung weight of 0.394 ± 0.196 g (P = 0.0371, n = 10 for PyVmT/Tgfbr2MGKO mice and n = 5 for PyVmT/Tgfbr2flox/flox mice). This increase in pulmonary metastases in the PyVmT/Tgfbr2MGKO mice was also evident upon examination of histologic sections of the lungs ( Fig. 3B). Whereas there was no difference in the histology of the metastases with knockout of Tgfbr2, there was a remarkable difference in the size of lesions present in each section.
To rule out the possibility that the metastatic tumors arose from primary tumors in which there was no recombination, we isolated DNA from several mammary tumors, pulmonary metastases, and primary tumor cell cultures from both PyVmT/Tgfbr2flox/flox and PyVmT/Tgfbr2MGKO mice. Southern blot analyses of these samples reveal extensive recombination in all of the PyVmT/Tgfbr2MGKO samples ( Fig. 3C). The small amount of nonrecombined DNA is likely due to stromal and hematopoietic cells present in the tumor tissue. Immunostaining for phosphorylated Smad2 (p-Smad2) was done on sections of primary mammary tumors from PyVmT/Tgfbr2flox/flox and PyVmT/Tgfbr2MGKO as an indicator of TGF-β signaling in vivo ( Fig. 4A and B ). Most PyVmT/Tgfbr2flox/flox tumor cells show nuclear staining for p-Smad2 ( Fig. 4A), whereas PyVmT/Tgfbr2MGKO tumor cells lacked such staining ( Fig. 4B). Immunostaining for p-Smad2 was also done on sections of pulmonary metastases from PyVmT/Tgfbr2flox/flox and PyVmT/Tgfbr2MGKO mice. Interestingly, whereas primary mammary tumors from PyVmT/Tgfbr2flox/flox mice display intense, homogenous p-Smad2 staining, the lung metastases from PyVmT/Tgfbr2flox/flox mice display variable staining. In fact, the majority of the lung metastases display a decreased proportion of cells staining positive for nuclear p-Smad2 compared with the primary tumors ( Fig. 4C and D). As expected, the epithelial cells of primary mammary tumors and lung metastases from PyVmT/Tgfbr2MGKO display a significant loss of p-Smad2 staining; however, fibroblasts and other host immune cells clearly stain positive ( Fig. 4E and F). These data clearly indicate that TGF-β signaling in the mammary epithelium functions to suppress early tumorigenesis and growth of metastasis in the PyVmT transgenic mouse line.
PyVmT/Tgfbr2MGKO Tumor Cell Lines Are Resistant to Transforming Growth Factor–β Growth Inhibitory Effects. We established cultures from primary mammary tumors of MMTV-PyVmT/Tgfbr2flox/flox, MMTV-PyVmT/Tgfbr2MGKO and from the lung metastases of MMTV-PyVmT/Tgfbr2flox/flox and MMTV-PyVmT/Tgfbr2MGKO mice [primary floxed tumor (PF), primary knockout tumor (PK), lung metastases from a floxed mouse (LF) and lung metastases from a knockout mouse (LK), respectively]. Sequential cell count experiments using these cell lines indicate that the PF cells are still sensitive to TGF-β's antiproliferative effects. As anticipated, the PK cells are no longer responsive to TGF-β demonstrating that Tgfbr2 has been successfully deleted ( Fig. 5Aa ). Similarly, neither the LK nor LF cell lines are sensitive to TGF-βs growth inhibitory effects ( Fig. 5Ab). Reverse transcription-PCR analysis of all cell lines using primers specific for exons 2 to 3 expression indicates a complete loss of Tgfbr2 expression in the PK tumor line and decreased expression in the LF and LK cell lines ( Fig. 5B). Primers specific for exons 2 to 3 were used because exon 2 of Tgfbr2 is the region floxed by loxP sites and thus excised by the addition of Cre. This decreased level of expression instead of a complete loss of expression of Tgfbr2 expression in LK cells could be due to the presence a few remaining fibroblasts in the culture. The same may be true for the LF cell line, or it is also possible that, similar to human breast cancers, TGF-β type II receptor levels were down-regulated and not completely lost. As anticipated the PF tumor line displays high levels of Tgfbr2 expression ( Fig. 5B). When the cell lines are treated with TGF-β, the PF line shows strong induction of p-Smad2 whereas the PK, LF, and LK cell lines showed no increase in p-Smad2 and total Smad2 levels remained unchanged ( Fig. 5C).
All of the data indicate a loss of TGF-β responsiveness not only in the PK cells but also in the LF and LK cells and show that TGF-β signaling in carcinoma cells is not required for mammary tumor metastasis. The data also suggest that early loss of TGF-β signaling in the mammary epithelium promotes mammary tumor development and metastasis. Furthermore, we have found that PF and PK lines undergo no morphologic changes or epithelial to mesenchymal transition upon treatment with TGF-β ( Fig. 5D). We also examined protein from four primary mammary tumors from MMTV-PyVmT/Tgfbr2flox/flox, MMTV-PyVmT/Tgfbr2MGKO mice to determine loss of TGF-β signaling and found that tumors from MMTV-PyVmT/Tgfbr2MGKO have decreased expression of p-Smad3 (Supplementary Fig. 1). However, there are no changes in the levels of mitogen-activated protein kinase, p-mitogen-activated protein kinase, AKT, p-AKT or cyclin D1 (Supplementary Fig. 1). In addition, there are no differences detected in the expression of TGF-β1, TGF-β2, or TGF-β3 mRNA in whole tumor extracts from MMTV-PyVmT/Tgfbr2flox/flox and PyVmT/Tgfbr2MGKO mice (Supplementary Fig. 1).
In this study, we conditionally knocked out Tgfbr2 in the mammary epithelium to derive Tgfbr2MGKO mice. These mice exhibit varying degrees of lobular-alveolar hyperplasia, and spontaneous tumor development studies are currently under way. When tumor formation is induced by the PyVmT transgene, loss of Tgfbr2 in the mammary epithelium resulted in formation of mammary tumors with a significantly shortened median latency when compared with PyVmT mice. Contrary to expectations, MMTV-PyVmT/Tgfbr2MGKO mice developed significantly more pulmonary metastases compared with MMTV-PyVmT/Tgfbr2flox/flox mice. We were able to detect very high levels of recombination of Tgfbr2 not only in the primary mammary tumors arising from MMTV-PyVmT/Tgfbr2MGKO mice but also in the pulmonary metastases. Therefore, it is highly unlikely that metastases of the primary mammary tumors were from cells in which recombination had not occurred. The extent of abrogation of the TGF-β signaling pathway was also confirmed although p-Smad2 immunostaining. Interestingly, we derived evidence for diminished TGF-β signaling in vitro and in vivo in lung metastases of PyVmT/Tgfbr2flox/flox mice. These results imply that decrease of TGF-β signaling occurs spontaneously and frequently in the metastatic process of PyVmT/Tgfbr2flox/flox mice. Thus, all of the results generated in this study are consistent with Tgfbr2 acting as a tumor suppressor in the mammary gland. Furthermore, these data show that mammary tumor metastasis can clearly occur in the absence of TGF-β signaling and in fact is enhanced by the loss of TGF-β signaling at least in this model.
Our current findings are in agreement with other recent studies done using colon and prostate mouse models in which disruption of TGF-β signaling results in enhanced tumor progression (22, 23) . Whereas our study contradicts the reports that TGF-β signaling promotes mammary tumor metastasis (reviewed in ref. 24), it is important to analyze the differences in the model systems. Several studies to date clearly show systemic inhibition of TGF-β signaling suppresses mammary tumor metastasis (12, 13) . Systemic inhibition of TGF-β signaling affects activity of the immune system and stroma cells. Several studies point to the importance of the immune system in the regulation of tumor growth and metastasis (reviewed in ref. 25). Importantly, blockade of TGF-β signaling specifically in T cells has been shown to play a role in the enhancement of antitumor immunity and protection against formation of metastases (26, 27) . However, our unique model system allows us to precisely study the effects of abrogation of TGF-β signaling specifically on the mammary epithelium and thus mammary tumor formation and metastasis. It is also possible that the timing of the loss of TGF-β signaling may determine whether or not TGF-β functions as a tumor suppressor or tumor promoter. For example, in human colon cancers loss of TβRII is correlated with the progression of adenomas to invasive carcinomas, but later in tumorigenesis loss of TβRII in colon carcinomas with microsatellite instability is correlated with a better prognosis (28, 29) . Therefore, loss or attenuation of TGF-β signaling later in mammary tumorigenesis may decrease the frequency of metastatic spread. Finally, loss of Tgfbr2 in mammary tumor cells may lead to increased genomic instability (30) and this could represent another mechanism by which loss of TGF-β signaling promotes metastasis.
Here we report that early loss of TGF-β signaling in the mouse mammary gland in the context of expression of the MMTV-PyVmT transgene results in shortened median tumor latency and, more importantly, an increased incidence of metastasis. Our studies suggest that the effects of systemic inhibition of TGF-β may largely be on host cells and not tumor cells. Thus, these results support the model of epithelial TGF-β signaling in mammary tumors being suppressive for both tumor development and metastasis by clearly demonstrating that early loss of Tgfbr2 results in more rapid primary tumor formation and a marked increase in pulmonary metastases.
Grant support: Susan G. Komen Breast Cancer Foundation grant DISS0402357 (E. Forrester), Public Health Service grants CA42572 and CA85492 (H.L. Moses), Vanderbilt-Ingram Cancer Center grant CA68485, and TJ Martell Foundation.
The costs of publication of this article were defrayed in part by the payment of page charges. These articles must therefore be marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Philippe Soriano (Fred Hutchinson Cancer Research Center, Seattle, WA) for generously providing the ROSA26r mice (C57/B6), Dr. Yu Shyr and Bashar Shakhtour for statistical analysis of our data, and Dr. Carlos Arteaga for generously providing us with the MMTV-PyVmT mice and for reviewing this article.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org).
- Received September 10, 2004.
- Revision received November 16, 2004.
- Accepted December 29, 2004.
- ©2005 American Association for Cancer Research.