Tumor Microenvironment Regulates Metastasis and Metastasis Genes of Mouse MMTV-PymT Mammary Cancer Cells In Vivo

Abstract

Metastasis is the primary cause of death in breast cancer patients, yet there are challenges to modeling this process in vivo. The goal of this study was to analyze the effects of injection site on tumor growth and metastasis and gene expression of breast cancer cells in vivo using the MMTV-PymT breast cancer model (Met-1 cells). Met-1 cells were injected into 5 sites (subcutaneous, mammary fat pad, tail vein, intracardiac, and intratibial), and tumors and metastases were monitored using bioluminescent imaging and confirmed with gross necropsy and histopathology. Met-1 tumors were analyzed based on morphology and changes in gene expression in each tissue microenvironment. There were 6 permissible sites of Met-1 tumor growth (mammary gland, subcutis, lung, adrenal gland, ovary, bone). Met-1 cells grew faster in the subcutis compared to mammary fat pad tumors (highest Ki-67 index). Morphologic differences were evident in each tumor microenvironment. Finally, 7 genes were differentially expressed in the Met-1 tumors in the 6 sites of growth or metastasis. This investigation demonstrates that breast cancer progression and metastasis are regulated by not only the tumor cells but also the experimental model and unique molecular signals from the tumor microenvironment.

Metastasis is the most debilitating and lethal complication of breast cancer and has become a critical focus for cancer research. However, the biological nature of metastasis poses challenges for therapeutic intervention. Metastasis of tumor cells is a multistage process that involves loss of cell-to-cell contact, invasive behavior at the primary site, intravasation of tumor cells into vessels, extravasation of tumor cells out of vessels, and tumor cell invasion and tissue modification at a distant site.¹⁸ Unfortunately, a comprehensive understanding of human metastatic disease is incomplete. Much of the current knowledge of cancer metastasis has come from the use of in vivo experimental mouse models of metastasis. Mouse models are invaluable because they allow a unique opportunity to study different aspects of metastasis that play a role in disease progression and include host responses such as immune surveillance and angiogenesis, complex interactions between cancer cells and stromal cells in the tumor microenvironment, the role of hypoxia, pH, and growth factors, changes in the premetastatic niche,³¹ and the bioavailability and metabolism of therapeutics. Nevertheless, knowledge of human metastatic disease is still limited by the availability of model systems that can accurately recapitulate the multistage process of metastasis.

Unfortunately, animal models that accurately replicate metastasis from initial carcinogenesis to end-stage metastatic disease do not exist. Therefore, unique models are used to replicate difference stages of the metastatic cascade, which include subcutaneous, orthotopic, intracardiac, tail vein, heterotopic, and bone (such as the tibia) injections (Fig. 1). Some aspects of model systems can be controversial because they fail to recapitulate all stages of the metastatic cascade. Studies have shown that there is a distinct organ-specific metastatic pattern of individual breast cancer cells and primary tumors.⁴¹ The metastatic patterns are attributed to the biology of the cancer cells, as well as the genetics of the host organism.⁵ The goal of this study was to characterize and evaluate the effect of injection site on tumor growth, metastasis, tumor morphology, and gene expression of mouse mammary cancer cells.

Figure 1.

Multiple injection techniques of MMTV-PymT cells (Met-1) were used recapitulate the multistage process of metastasis. Orthotopic mammary fat pad injections model invasion, extravasation, and tissue modification at metastatic sites. Intravascular injections (intracardiac and intravenous) model extravasation and tissue modification in metastases. Finally, intratibial injections model the ability to grow in the bone microenvironment and modify bone structure, which is a common site for breast cancer metastasis in humans.

There are few transgenic mouse models that accurately recapitulate human metastatic breast cancer. The MMTV-PymT model has been well characterized and shown to accurately recapitulate the stepwise progression of human breast cancer,²⁴ leading to its extensive use and the elucidation of pathways and molecules related to breast cancer tumorigenesis.¹³ The polyoma middle T antigen (PymT) is expressed in mammary epithelial cells under the control of the mouse mammary tumor virus (MMTV) promoter and associates with cellular proteins, resulting in malignant transformation of mammary epithelial cells, as observed in the human disease. Specifically, activation of c-src,⁷ phosphatidylinositol 3′-kinase,⁶² and Raf-Mek-ERK²⁶ leads to the rapid development of multifocal tumors involving all mammary glands,²⁴ with transgenic mice developing lung metastases by 20 weeks of age.² Met-1 cells have been characterized as highly metastatic based on the consistent and repeatable metastasis of cells to the lung following mammary fat pad and tail vein injections.²

The hypothesis of this study is that the type of metastasis model and tissue microenvironment will affect the growth, morphology, and gene expression of Met-1 cells. To test this hypothesis, 5 experimental metastasis injection models were used to study the multistage process of metastasis (Fig. 1). Mammary fat pad injections or subcutaneous injections model the invasion of breast cancer cells into the local microenvironment as well as subsequent stages along the metastatic cascade. Alternatively, cardiac or intravenous injections model the circulating tumor cells, which then extravasate and grow at specific secondary sites. Intravenous tail vein injections predispose to lung metastases, since the tumor cells first pass through the pulmonary capillaries. In contrast, intracardiac injections in the left ventricle result in an initial distribution of tumor cells throughout the entire arterial system and test the ability of tumor cells to colonize all organs. Amazingly, there is usually a high degree of specificity for sites of metastatic growth for individual cell lines.^{1,36,46,52,57,66} Finally, injection of cells into the tibia models the growth of cancer cells in the bone microenvironment. Bone is one of the most common and devastating sites of breast cancer metastasis.^34,42,46,60 Based on the metastasis model and injection site, we found 6 permissible sites for the growth of metastatic Met-1 cells, as well as changes in cellular morphology and gene expression of tumors at each site. This study increased our understanding of the role of the tumor microenvironment in regulating the growth and metastasis of breast cancer cells, and it provides a rational basis for use of the PymT Met-1 cells in mechanistic experiments on breast cancer metastasis in vivo.

Materials and Methods

Murine Mammary Cancer Cell Line (MMTV-PymT Met-1 Cells)

MMTV-PymT cancer cells (Met-1) cells were obtained as a generous donation from Dr Alexander Borowsky (University of California Davis Cancer Center, Sacramento).² Met-1 cells were isolated from a primary mammary tumor in MMTV-PymT; FVB/N mice. The cell line was further selected for high metastatic potential to the lungs when injected into the mammary gland fat pad of FVB/N mice.² Cells were cultured in high glucose (4.5g/l) Dulbecco’s-modified Eagle’s medium (Invitrogen, Carlsband, California) supplemented with 10% fetal bovine serum (Invitrogen), 2 mM of L-glutamine, 10 units/ml of penicillin, and 10 μg/ml of streptomycin. Cells were maintained in humidified incubators at 37°C and 5% CO₂.

Met-1 cells expressing luciferase were generated using a lentiviral vector as previously described.⁵² Transduced cells maintained proliferative properties identical to the parental line (data not shown). Cells were verified mycoplasma-free before mouse injections using a PCR-based method (catalog No. MP-0025, Sigma-Aldrich, St Louis, Missouri). Cells were harvested for injection at 70% to 80% confluence using a mixture of 0.5 mmol/L EDTA and 0.25% trypsin. Viable cells (95%–100%) were counted using a hemocytometer, resuspended in sterile phosphate-buffered saline (PBS), and kept on ice until injection.

Mice and Injection of Met-1 Cells

Female FVB/NCr mice were purchased from NCI (Frederick, Maryland), catalog No. 01F50, and injected at 6 to 8 weeks of age. Mice were housed in ventilated cages in an AAALAC-approved facility. All protocols were approved by The Ohio State University Institutional Animal Care and Use Committee.

Mice were divided into 5 groups. For injections, animals were anesthetized using 3% isoflurane gas induction and maintained at 2% isoflurane on a rodent heating pad. The injection site was sanitized using 70% ethyl alcohol. For both orthotopic (mammary fat pad) and heterotopic (subcutaneous) injections, 5 × 10⁶ Met-1 cells were injected in 50 μl of PBS using a 25-gauge (G) needle (10 mice each group).² Intravascular injections included left ventricle of the heart and the lateral tail vein. For intracardiac injections, mice were placed in dorsal recumbency under anesthesia. Met-1 cells (100 000) were suspended in 100 μl of DPBS for injection. Visualization of the left cardiac ventricle was obtained using a Vevo 660 small animal ultrasound (VisualSonics, Toronto, Ontario, Canada), and cells were injected using a 27G needle (n = 20). Injection of cells into the arterial circulation was confirmed through ultrasound visualization of the cells in the left ventricular chamber of the heart, as well as a pulsing of blood in the needle upon injection.⁴³ Intravenous inoculations of Met-1 cells were performed using the dilated lateral tail vein. Met-1 cells (2 × 10⁶) were suspended in 200 μl of DPBS and injected using a 27G needle.²⁷ Finally, for intratibial injections, Met-1 cells (100 000) were suspended in 20 μl of DPBS and injected through the skin into the proximal left tibia (n = 10) using the tibial crest as a landmark.^56,66 Injections were performed using a 26G needle and a 100-μl Hamilton syringe (Hamilton Co, Reno, Nevada).

Evaluation of Metastases

Mice were weighed and evaluated weekly using bioluminescent imaging, caliper measurements, and gross observation for clinical signs of metastatic disease as described below.

In vivo bioluminescent imaging was performed on a cooled CCD IVIS 100 system equipped with a 50-mm lens as previously described.³⁹ Results were analyzed using LivingImage software, version 2.2 (Caliper Life Sciences, Hopkinton, Massachusetts). Mice were injected intraperitoneally with 4.3 mg D-luciferin dissolved in sterile PBS and imaged while under isoflurane anesthesia. Images were acquired every 3 minutes until the peak signal was achieved for each mouse. Bioluminescent data were compared weekly to evaluate the presence and growth of metastases.

All palpable masses were measured weekly using external calipers. The greatest longitudinal diameter (length) and the greatest transverse diameter (width) were determined and tumor volume calculated by the modified ellipsoidal formula: tumor volume = 1/2 (length × width²).^15,58 Mice remained on study until the mass reached a total volume of 2 cm³ unless ulceration or other complications occurred. Mice were evaluated for clinical signs: cachexia (weight loss exceeding 20% of body weight), dehydration, anorexia, dyspenia, tumor ulceration, or tumor mass greater than 2 cm³. Mice with intratibial tumors were kept on study until they had pain, lameness or limping, or other removable criteria (see above). After reaching any of the previously described criteria, each mouse was euthanized with 100% CO₂ and processed separately as described below.

Postmortem Analysis

After euthanasia, a complete necropsy was performed, and tissues were harvested and sectioned to confirm metastases. Met-1 tumors were divided for both molecular analysis and histopathologic evaluation. Half of each tumor was snap frozen in liquid nitrogen, and the other half was fixed for 48 hours in 10% neutral-buffered formalin, embedded in paraffin, sectioned, and stained with hematoxylin and eosin. All sites were identically processed, with the exception of the tibias and the lungs.

Radiographs were taken of all tibias in situ after euthanasia, and bone loss was evaluated qualitatively using a Faxitron cabinet X-ray system (Hewlett-Packard, McMinnville, Oregon) at 45 kVp for 3.5 minutes. Next, tibias were designated for either molecular analysis (snap frozen in liquid nitrogen) or histopathologic analysis. Tibias for histology were defleshed and decalcified in 10% EDTA pH 7.4 at 4°C for 14 days. They were then embedded in paraffin and sectioned. Lungs were inflated postmortem and evaluated grossly and histologically for the presence of micrometastases. For lung inflation, a skin incision was made along the ventral side of the mouse, exposing the trachea and 1 ml of 10% neutral-buffered formalin was injected into the trachea in situ using a 1-ml syringe and 22G needle. After full inflation, the lungs were then dissected and removed from the chest cavity, placed in formalin, and embedded, with all lobes sectioned for histology. Lung metastases were microdissected from 3 mice for gene expression analysis prior to inflation and fixation.

Histopathology and Immunohistochemistry

In each of the 6 sampled sites, hematoxylin and eosin–stained tumors were examined, and tumor morphology, mitotic index, and percentage necrosis in the tumor were measured. The neoplastic cells were characterized as polygonal (epithelial), spindle-shaped, anaplastic, or mixed morphologies. Characteristics of specific morphologies were as follows: The polygonal (epithelial) morphology was associated with indistinct cell borders and close association to neighboring cells. The spindle cell morphology was characterized by elongated cells with bipolar cytoplasmic boundaries of varying length. The nuclei were spindle to oval shaped and generally centrally placed. Anaplastic cells typically had centrally placed round to oval nuclei. The anaplastic morphology exhibited marked anisocytosis (ranging from 50 to 200 μm in diameter) with abundant eosinophilic and sometimes vacuolated cytoplasm. Anaplastic cells frequently contained multiple nuclei or large nuclei with marked irregular contours. Tumors were classified as mixed when 2 or more morphologic classifications were present. In addition to these classifications, the surrounding normal tissue was also examined for evidence of tumor invasion, inflammation, modification of bone structure (in tibias), and response of the fibrous stroma.

The number of Ki67-postive Met-1 cells were measured at each of the 6 sites of tumor growth by immunohistochemistry. Immunohistochemistry was performed on a Dako Autostainer Universal Staining System (Carpinteria, California). Primary monoclonal rat anti-mouse Ki-67 antibody (catalog No. M7249, Dako) was diluted 1:50 in Dako Antibody Diluent. Biotinylated secondary polyclonal antibody horse anti-mouse (BA-2000, Vector Laboratories, Burlingame, California) was diluted 1:200. Diaminobenzidine was used as the staining reagent, and the slides were counterstained with hematoxylin. The number of Ki67-positive cells per 100 cells was measured in 3 fields of view for 3 mice per site and averaged. Primary monoclonal rabbit anti-mouse E-cadherin antibody (No. 3195, Cell Signaling Technology, Danvers, Massachusetts) was diluted 1:100 and biotinylated secondary polyclonal goat anti-rabbit antibody (BA-1000, Vector) was diluted 1:1000. Positive controls consisted of sections of skin and tonsil. Negative controls were slides stained with a lack of primary antibody.

Gene Expression Analysis

Tumors at each site were microdissected for molecular analysis and consisted of viable tumor tissue and its closely associated tumor stroma. Tumor capsules and normal tissue were avoided. Subcutaneous, mammary fat pad, ovary, and adrenal tumors were grossly encapsulated, and a portion of viable tumor tissue was used for RNA extraction. Lung tumor nodules were microdissected from the lung parenchyma and pulverized whole. The entire tibial metaphysis was harvested from mice with tumor in the tibia (intratibial injection) as confirmed by bioluminescence imaging. All tumor samples were pulverized in liquid nitrogen with a mortar and pestle resulting in Met-1 tumor lysates (Met-1 tumor cells including the internal tumor microenvironment). Met-1 tumor RNA was extracted using Trizol reagent (Invitrogen) and purified using Absolutely RNA (catalog No. 400800, Stratagene, La Jolla, California). RNA quality was assessed using the ratio of absorbance at 260 and 280 nm, and the minimal quality accepted was 1.8; however, most samples were between 1.9 and 2.0. RNA was reverse transcribed using the RT² PCR Array First Strand Kit (SABiosciences, Frederick, Maryland).

Total RNA samples from representative Met-1 tumors were analyzed with the RT² Profiler Cancer PathwayFinder PCR Array (SABiosciences) (n = 3). The PCR Array included 88 representative genes from the following biological pathways involved in oncogenesis: adhesion, angiogenesis, apoptosis, cell cycle control, cell senescence, DNA damage repair, invasion, metastasis, signal transduction molecules, and transcription factors. The user manual and data analysis software is available at http://www.sabiosciences.com.

Orthotopic injections most accurately recapitulate tumor stroma interactions in the primary tumor microenvironment (such as the mammary gland); therefore, mammary fat pad tumors were used as the control for the Superarray analysis. Results were expressed as fold change relative to the gene expression of tumors that grew from Met-1 cells injected into the mammary gland as a control. Genes with less than a 3-fold change in expression were excluded from statistical analysis; however, this may have prevented the identification of relevant genes with less dramatic differences, such as haploinsufficient genes.

Statistical Analysis

Analysis of variance was used for analyzing the mitotic index, and Student t test was used for testing differences between subcutis (treated as control) and other sites. P values were adjusted by the Holm method to account for multiple comparisons. Linear mixed model was used to model the linear trend of tumor volume of mammary and subcutaneous tumors after log transformation of the data. Random effects were assumed on both intercept and slope of the linear line of individual mice. Tumor growth rates were tested by slope difference of 2 tumors. For the data analysis of Ki-67 immunohistochemistry, a linear mixed model was employed to model correlated multiple observations of each of 3 mice for each site. Each mouse was considered a random effect for each site. Contrasts of means of number of Ki-67-positive cells per 100 cells among all 6 sites were calculated. P values were adjusted by the Tukey method to account for multiple comparisons. For PCR array analysis, Student t tests were used to analyze differences in gene expression between individual groups and the mammary fat pad tumors and required P values for statistical significance were adjusted for multiple comparisons (MMP-2, 3 comparisons; E-cadherin, 2 comparisons; FGFR-2, 4 comparisons; and MMP-9, 2 comparisons). Hierarchical clustering was used to cluster tumor samples. Log-rank test was used for comparing survival curves.

Results

Met-1 Tumor Growth Was Dependent on the Tumor Microenvironment

The growth of both subcutaneous and mammary fat pad tumors were measured weekly with calipers until reaching a volume of 2 cm³. All other sites were monitored using bioluminescence imaging and gross evaluation as described, and mice were removed from the study when they reached early removal criteria. Met-1 cells grew significantly faster in the subcutis compared to the mammary gland (Fig. 2) and reached maximum size (2 cm³) in 29 days in comparison to the mammary fat pad (42 days). After tail vein injections, mice survived an average of 42 days, while mice that received left ventricular injections survived on average 85 days before reaching a clinical endpoint (log-rank test, P = .0007). Mice that were inoculated with Met-1 cells in the tibia survived an average of 23 days (Fig. 3). Injection of different cell numbers (mammary, subcutis > intravenous > intracardiac, intratibial) may have played a role in tumor growth rate.

Figure 2.

The growth rate of Met-1 tumors was dependent on site of inoculation. Mammary fat pad tumors reached maximum size at 7 weeks, while subcutaneous tumors grew significantly faster and reached the size endpoint at 4 weeks. ^*P < .001 for difference in tumor growth rates.

Figure 3.

Survival of mice with Met-1 tumors. Met-1 tumors grew at different rates depending on the site and tumor microenvironment, which determined the survival of mice. Clinical endpoints were reached first in the intratibial model, followed by the subcutaneous, mammary fat pad, tail vein, and intracardiac models. The median survival points for each model were as follows: 23 days (intratibial), 29 days (subcutaneous), 42 days (mammary fat pad), 42 days (tail vein), and 85 days (intracardiac). ^*Log-rank test P = .0007.

Incidence and Location of Met-1 Metastases

The 5 models of metastasis resulted in different patterns of growth and metastasis in vivo (Figs. 4–31 and Table 1). All mice injected in the subcutis, mammary fat pad, or tibia developed tumors (10/10 per group). Met-1 cells did not metastasize following orthotopic (mammary fat pad) or heterotopic (subcutis or intratibial) injections, however; tail vein injection of Met-1 cells resulted in 9/11 mice with lung metastases, and 2 of these mice also had accompanying ovarian metastases. Following intracardiac injections, 12 mice had no metastases, 6 had ovarian metastases, 5 had adrenal gland metastases, and 3 had bone metastases (calvaria or humerus). Additionally, 6 mice had multiple metastases (Table 1).

Figure 4.

Mammary fat pad. Mouse. Met-1 tumor growing as an expansile and nonencapsulated mass in the inguinal mammary fat pad. Ruler (mm).

Figure 31.

Percentage necrosis of Met-1 tumors was semiquantitated and categorized as greater than 75%, 50%–75%, 25%–50%, or less than 25% necrotic. Ovary metastases had the greatest amount of necrosis and bone (intratibial) tumors had the least.

Table 1.

Sites of Met-1 Growth and Metastasis.

Injection Site	No Metastases	Lung	Ovary	Ovary + Lung	Ovary + Adrenal	Ovary + Bone	Adrenal	Adrenal + Bone	Ovary + Adrenal + Bone
Mammary	10/10
Subcutis	10/10
Tail vein	2/11	7/11		2/11
Intracardiac	12/20		1/20		1/20	2/20	1/20	1/20	2/20
Intratibial	10/10

Met-1 Tumors Had Different Morphologic Phenotypes Depending on the Site of Metastasis

The Met-1 primary and metastatic tumors were predominantly composed of broad sheets of cells that occasionally formed small discrete nodules and contained regions of spindle-shaped cells (Figs. 4–21). In general, the neoplastic cells were characterized by marked anisocytosis and anisokaryosis (cell diameter ranged from 12 to 100 μm and nuclear diameter ranged from 6 to 90 μm) with significant variation in nuclear shape (smaller nuclei were round to oval or fusiform; larger nuclei had bizarre morphology characterized by irregular contour of the nuclear envelope with frequent multiple lobation). The chromatin pattern varied from clumped to stippled with irregular patterns of heterochromatin present in bizarre nuclei. Multiple nuclei were frequently present, and the nuclei contained prominent nucleoli. Cytoplasm of the neoplastic cells was homogenous and lightly eosinophilic with small clear vacuoles.

Figure 21.

Subcutis. Mouse. Met-1 tumor. Subcutaneous tumors were chiefly mixed phenotype. HE stain.

Tumor morphology (epithelial, spindle shaped, anaplastic, or mixed), percentage necrosis, and mitotic index varied depending on the microenvironment (Fig. 30). Tumors in the mammary fat pad were predominantly of mixed phenotype and were expansile and nonencapsulated but not highly invasive. The mammary tumors expanded into the dermis (with compression of the adnexa) and subcutaneous adipose tissue (Figs. 4–6). Similar to mammary tumors, subcutaneous tumors were chiefly mixed phenotype and were expansile and nonencapsulated with compression of adjacent tissue (Figs. 19–21). In contrast to mammary tumors, subcutaneous tumors invaded adjacent structures (skeletal muscle and panniculus). The cuticular skeletal muscle demonstrated mild to moderate myofiber atrophy with proliferation of immature fibroblasts and collagen (interstitial fibrosis) in response to tumor invasion. Consistent with the high rate of tumor growth and Ki-67 staining, the subcutaneous tumors also demonstrated the highest mitotic index (compared to bone and adrenal gland tumors) (Fig. 29).

Figure 30.

Morphologic categories of Met-1 tumors (anaplastic, mixed, spindle-shaped, epithelial). Ovary tumors (n = 8) had the highest percent of anaplastic cells. Subcutaneous (n = 10), mammary fat pad (n = 10), and adrenal (n = 5) tumors had primarily a mixed cellular morphology. The bone tumors (intratibial injections, n = 10) consisted primarily of spindle-shaped cells, and the lung metastases (n = 9) had the greatest percentage of epithelial-like cells.

Figure 6.

Mammary fat pad. Mouse. Met-1 tumor from the inguinal mammary fat pad with a mixed-cell phenotype. HE stain.

Figure 19.

Subcutis. Mouse. Met-1 tumor after injection of tumor cells in the subcutis (arrow). The inset (left) is a bioluminescent image of the Met-1 tumor in the subcutis. Colors represent relative numbers of viable tumor cells: green > blue.

Figure 29.

The number of mitoses per 400× histologic field in Met-1 tumors. Subcutaneous tumors had the greatest number of mitoses (n = 3.4), followed by ovary (n = 2.8), lung (n = 1.9), mammary fat pad (n = 1.8), bone (n = 1.5), and adrenal gland (n = 1.4) tumors. ^*P < .05.

Tail vein injections of Met-1 cells resulted in lung metastases in 9 of 11 mice (Fig. 7). The lung metastases compressed the pulmonary parenchyma and were composed of nodular highly cellular masses. Lung metastases were predominantly randomly distributed throughout the lung parenchyma. A few of the metastases were located in the subpleural, hilar, or peribronchiolar regions. Some metastases occluded and grew in large arteries and veins (tumor thrombi and thromboemboli). The metastases effaced the pulmonary parenchyma and extended through and over the pleura and into mediastinal connective tissue. Large areas of coagulation and liquifactive necrosis were present in the solid pulmonary and intravascular metastases. In contrast to tumors of any other site, most pulmonary tumors demonstrated an epithelial phenotype (Figs. 8, 9).

Figure 7.

Lungs. Mouse. Multiple Met-1 metastases in the lungs after tail vein injection. Inset (upper right): ex vivo bioluminescent imaging of pulmonary tumors. Colors represent relative numbers of viable tumor cells: red > yellow > green > blue.

Figure 8.

Lung. Mouse. Met-1 metastasis (left) and adjacent pulmonary parenchyma (right). The lung metastases compressed the pulmonary parenchyma and were composed of nodular highly cellular masses. HE stain.

Figure 9.

Lung. Mouse. Met-1 pulmonary metastasis. Most lung metastases had an epithelial phenotype. HE stain.

Intracardiac injection of Met-1 cells resulted in adrenal gland metastases which ranged from 0.1 to 0.5 cm³ (Fig. 16) The metastases were mostly mixed phenotype, with some having an anaplastic phenotype. The tumors completely effaced the adrenal medulla and replaced approximately 90% of the adrenal cortex (Figs. 17, 18). The remaining adrenal cortex was compressed and atrophic. A small portion of compressed zona fasciculata, zona reticularis, and zona glomerulosa was still present and identifiable. In addition, the adjacent renal parenchyma was mildly compressed with distortion of the cortical architecture. The neoplasms invaded the adrenal capsule and the adjacent perirenal tissues as nodular aggregates. Small areas of coagulation necrosis were scattered throughout the neoplasm.

Figure 16.

Adrenal gland (left) and kidney (right). Mouse. The adrenal gland is severely enlarged (1 cm wide) due to a metastatic Met-1 tumor after intracardiac injection of tumor cells.

Figure 17.

Adrenal gland. Mouse. Met-1 metastasis. The adrenal metastases completely effaced the adrenal medulla and replaced approximately 90% of the adrenal cortex. HE stain.

Figure 18.

Adrenal gland. Mouse. Met-1 metastasis. The metastases were mostly mixed phenotype with some having an anaplastic tumor cells. HE stain.

Ovarian metastases occurred after injection of Met-1 cells in either the tail vein or the left cardiac ventricle. The ovaries with metastases were massively enlarged and ranged from 0.5 to 3 cm³ in size (Fig. 10). The entire ovarian medulla was replaced by the neoplasm while a small portion of compressed ovarian cortex remained. Most ovarian metastases were anaplastic and demonstrated a high degree of necrosis (Figs. 11, 12). Several developing follicles were observed in the remaining cortex. Scattered aggregations of granulosa cells were present in the neoplasms with marked disruption of normal follicular structures. Primordial follicles could be identified in the remaining ovarian cortex. In addition, small areas of neutrophilic inflammation were present at the periphery of the remaining ovarian cortex.

Figure 10.

Abdomen. Mouse. Bilateral Met-1 tumors in the ovaries. The ovaries with metastases were massively enlarged and ranged from 0.5 to 3 cm³ in size.

Figure 11.

Ovary. Mouse. Met-1 metastasis. Ovarian metastases were composed of anaplastic tumor cells and demonstrated a high degree of necrosis. HE stain.

Figure 12.

Ovary. Mouse. Met-1 metastasis. Ovarian metastases were composed of anaplastic tumor cells. HE stain.

All tibias injected with Met-1 cells developed bone tumors that induced moderate osteolysis due to osteoclastic bone resorption (Fig. 13). The tumors grew in the proximal diaphysis, metaphysis, and also invaded the epiphyseal marrow spaces (Figs. 14, 15). Trabecular bone was replaced by nonencapsulated, highly cellular mass of neoplastic cells. There was medullary loss of hematopoietic and myeloid precursors with small aggregates of hematopoietic cells present only in the middiaphyseal region. The tumors also invaded and replaced portions of metaphyseal and diaphyseal cortical bone. The remaining endosteal surfaces adjacent to the tumor cells were scalloped, interpreted as previous endosteal bone resorption, although active osteoclasts were not observed. Finally, irregular trabeculae of reactive woven bone were present within some of the tumors. Mild periosteal new bone formation was observed, indicating stimulation of the periosteal surface. In contrast to tumors of other anatomic sites, bone tumors were mostly composed of spindle-shaped cells and demonstrated the lowest degree of necrosis.

Figure 13.

Tibia. Mouse. Image of the hind leg of a nude mouse with a Met-1 tumor in the proximal tibia. Note bone lysis of the cortex and new bone formation in the medullary cavity in the radiograph on the left. The inset (right) is a bioluminescent image of the Met-1 tumor in the tibia. Colors represent relative numbers of viable tumor cells: red > yellow > green > blue.

Figure 14.

Tibia. Mouse. Met-1 tumor in the proximal tibia. Note bone lysis of the cortex and new bone formation in the medullary cavity of the diaphysis. HE stain.

Figure 15.

Tibia. Mouse. Met-1 tumor in the proximal tibia. Met-1 bone tumors were mostly composed of spindle-shaped cells and had minimal necrosis. HE stain.

Proliferation Rate of Met-1 Tumors Was Dependent on the Tissue Microenvironment

The number of Ki67-positive cells was measured at each site of tumor growth and compared to the Met-1 subcutaneous tumors, which had the highest growth rate (Figs. 22–28). Cells growing in the subcutis had the highest percentage of Ki67-positive cells (46% ± 19%), followed by ovary (41% ± 24%), mammary gland (38% ± 14%), adrenal (31% ± 11%), lung (29% ± 9%), and bone (16% ± 7%). The tibial tumors had significantly fewer Ki67-positive cells compared to subcutaneous, mammary fat pad, and ovarian tumors (P < 0.05; Fig. 28). Therefore, the tumor microenvironment affected the rate of cellular proliferation and tumor growth, with the subcutis having the greatest growth advantage for the Met-1 cells.

Figure 22.

Adrenal gland. Mouse. Met-1 metastasis. Thirty-one percent of the Met-1 cells were positive for Ki67. Diaminobenzidine immunohistochemistry with hematoxylin couterstain.

Figure 28.

The percentage of Ki67-positive cells in Met-1 tumors measured using immunohistochemistry. Met-1 cells in the subcutis (46%) had the highest number of Ki67-positive cells, followed by ovary (41%), mammary fat pad (38%), adrenal (31%), lung (29%), and bone (tibia; 16%). ^*P < .05.

Morphologic Pattern, Necrosis, and Mitotic Index of Met-1 Tumors

Met-1 tumors were categorized based on 4 morphologic descriptions (anaplastic, mixed, spindle shaped, epithelial). Ovary tumors had the highest percentage of anaplastic cells. Subcutaneous, mammary, and adrenal tumors had primarily a mixed cellular morphology. The bone tumors (intratibial injections) consisted of primarily of spindle-shaped cells, and the lung metastases had the greatest percentage of epithelial-like cells (Fig. 30).

Percentage necrosis was semiquantitated at each site and categorized as greater than 75%, 50% to 75% necrotic, 25% to 50% necrotic, or less than 25% necrotic. Ovary metastases had the greatest amount of necrosis, and bone tumors (intratibial injections) had the least (Fig. 31).

The number of mitoses per 400× field were measured, and 10 measurements were taken and averaged per site. Subcutaneous tumors had the greatest number of mitoses (n = 3.4), followed by ovary (n = 2.8), lung (n = 1.9), mammary fat pad (n = 1.8), bone (n = 1.5), and adrenal gland (n = 1.4) tumors (P < .05) (Fig. 29). In general, the mitotic index followed a similar trend compared to the Ki67 staining except for the adrenal gland tumors.

Gene Expression of Met-1 Tumors Were Dependent on the Site of Growth

Representative tumors from all 6 sites were harvested and analyzed. Based on a set of 88 pathway-focused genes related to cancer progression, 36 genes were differentially expressed in Met-1 tumors when compared to Met-1 tumors that grew in the mammary fat pad (P < .05). Bone tumors were obtained from tumors that grew from Met-1 cells injected directly into the tibia. Hierarchical clustering revealed that the gene expression of total bone tumors were most distinct from other sites. Clusters of pathways involved in adhesion and metastasis, cell cycle and apoptosis, or angiogenesis were identified (Fig. 32). Upregulation of genes involved in angiogenesis and metastasis were found in bone tumors, while these genes were downregulated at all other sites.

Figure 32.

Expression of 36 genes was different depending on the site of Met-1 tumor growth. Messenger RNA was isolated from Met-1 tumors (n = 3) at 6 sites: bone (B), subcutis (S), mammary fat pad (M), adrenal gland (A), lung (L), and ovary (O). Real-time RT-PCR was used to examine 88 pathway-focused genes related to cancer progression. There were 36 genes that were differentially expressed when compared to Met-1 tumors from the mammary fat pad that served as a control (P < .05). Hierarchical clustering of the genes revealed pathways of importance, including adhesion, cell cycle and apoptosis, angiogenesis, and metastasis. The genetic profile of the intratibial Met-1 tumors were most distinct from the other sites and furthest from the profile of the ovary tumors.

From the original set of 36 genes, 7 genes had a greater-than-threefold difference in gene expression when compared to tumors in the mammary gland (control) and included matrix metalloproteinase-2 (MMP2), matrix metalloproteinase 9 (MMP9), E-cadherin, epidermal growth factor receptor (EGFR), fibroblast growth factor receptor-2 (FGFR2), endothelial-specific receptor tyrosine kinase (Tek/Tie-2), and telomerase reverse transcriptase (Tert) (Table 2). Differences were further evident based on the site of Met-1 growth (Fig. 33). Met-1 bone tumors had the greatest changes in gene expression with significant increases in EGFR (4.1-fold), FGFR2 (27-fold), MMP9 (44-fold), Tek (3.8-fold), and Tert (3.1-fold). Upregulation of MMP2 was also detected in subcutaneous tumors (3.1-fold). The MMP9 and FGFR2 results were confirmed by Q-RT-PCR with identical primers (SA Biosciences) and similar trends were observed (data not shown). E-cadherin was down regulated 5.7-fold in ovary tumors when compared to other sites, and this result was confirmed by immunohistochemistry (data not shown). Therefore, the expression of specific genes were differentially regulated depending on the anatomical location of the Met-1 cells.

Table 2.

Differences in Tumor Gene Expression Compared to Mammary Fat Pad Tumors.

Gene	Microenvironment	Fold Change	P
Matrix metallopeptidase 2	Subcutis	3.1	.007
E-cadherin	Ovary	−5.7	.01
Epidermal growth factor receptor	Bone	4.1	.02
Fibroblast growth factor receptor 2	Bone	27	.02
Matrix metallopeptidase 9	Bone	44	.0005
Endothelial-specific receptor tyrosine kinase	Bone	3.8	.03
Telomerase reverse transcriptase	Bone	3.1	.003

Figure 33.

The gene expression of Met-1 cells was different depending on the tumor site and microenvironment. Seven genes had a greater than threefold difference in expression when compared to the Met-1 tumors in the mammary gland (control), which included matrix metalloproteinase 2 (MMP2), E-cadherin (Cdh1), epidermal growth factor receptor (EGFR), fibroblast growth factor receptor-2 (FGFR2), matrix metalloproteinase 9 (MMP9), endothelial-specific receptor tyrosine kinase (Tek/Tie-2), and telomerase reverse transcriptase (Tert). Met-1 intratibial (bone) tumors had the greatest difference in gene expression with significant increases in EGFR (4.1-fold), FGFR2 (27-fold), MMP9 (44-fold), Tek (3.8-fold), and Tert (3.1-fold). Upregulation of MMP2 was also detected in subcutaneous tumors (3.1-fold). Interestingly, E-cadherin was significantly downregulated in ovary metastases (5.7-fold). ^*P < .05, ^**P < .01, ^***P < 0.001, when compared to tumors in the mammary fat pad.

Discussion

Mouse models are commonly used for metastasis research, but few studies have examined the role of the metastatic model and tumor microenvironment on outcome and metastasis phenotype. Each model of metastasis tests the ability of cancer cells to accomplish different steps of the metastatic cascade. Therefore, use of multiple metastatic models in the same investigation can be used to identify the phenotype and metastatic capabilities of a cancer cell line. Furthermore, the growth of a metastatic tumor in a specific end organ reflects not only the inherent genetic capability of the tumor cells but also their ability to adapt to and modify the tumor microenvironment for successful growth.³ This likely involves genetic contributions to tumor progression and important paracrine signaling between tumor cells and different microenvironmental niches.

An ideal animal model of metastatic breast cancer requires the ability to complete all stages of the metastatic cascade with a similar pathogenesis and pattern as occurs in women.⁴⁵ This is likely unattainable in mice, but mice can be used effectively to model single or multiple steps of the cascade. MMTV-PymT mice develop many large mammary tumors and usually reach early removal criteria prior to morbidity from metastatic lung disease. In addition, bone metastases are not seen in MMTV-PymT mice. The data presented here demonstrated that there were 6 permissible microenvironments for growth of Met-1 cells, although the cells did not grow in all 6 sites following injection in each model. Most notably, the Met-1 cells did not metastasize following mammary fat pad (orthotopic) or subcutaneous injection; however, Met-1 tumor growth rate was significantly greater in the subcutis when compared to the mammary fat pad. Even though the orthotopic tumors grew more slowly than the subcutaneous tumors, the mammary fat pad tumors likely grew too quickly for development of lung metastases. The importance of choosing orthotopic versus subcutaneous implantation is often overlooked in some experimental studies. It has been reported that subcutaneous tumor growth may exaggerate the growth inhibitory effects of tumor suppressor genes when compared to orthotopic models.³² Results from our study confirm that the subcutaneous microenvironment provided a growth advantage for Met-1 cells, which resulted in an accelerated and potentially exaggerated growth rate when compared to the mammary gland (orthotopic site).

For decades, it has been shown that breast cancers have a site-specific predisposition to metastasize to certain organs^5,17 and gene expression in tumor cells regulates the metastatic pattern of cancer cells to seed and grow in various metastatic sites.^30,40 In this study, the Met-1 cells had the capability to seed and grow in a defined set of organs, and some of these sites have not been previously reported. The ovary, adrenal gland, and bone microenvironment are uncommon sites of MMTV-PymT metastasis and were discovered by the use of multiple injection techniques in our investigation (tail vein and intracardiac). These findings may be explained by the inherent genetic profile of the Met-1 cells as well as the genetic contributions from the tumor microenvironment. In addition, bioluminescent imaging of metastases is a sensitive method to identify all sites of metastasis. If bioluminescence is not used, small metastases or bone metastases may be overlooked during gross necropsy.

The site of metastasis and tumor growth significantly affected morphology and behavior of Met-1 cells. For example, ovarian metastases were typically anaplastic and necrotic compared to other sites of hematogenous dissemination (lung and adrenal gland metastases), which had a more prominent epithelial morphology. Subcutaneous tumors had the greatest growth rate and mitotic index, which are markers commonly related to poor prognosis for patients.⁴ There are 2 possible explanations for different phenotypes of the metastases. First, the metastatic process and microenvironment at the site of secondary growth selected subsets of tumor cells with unique characteristics from the injected population of Met-1 cells. Alternatively, the tumor microenvironment actively regulated the variation in phenotype and behavior of the metastases after arrival of the Met-1 cells in the tumor niches. The Met-1 cells that were injected into the metaphyses of the tibias formed tumors with a spindle-shaped morphology, which was different from cells injected in the mammary fat pad and subcutis. This finding supports a direct role of the microenvironment in regulating cellular morphology and behavior of the tumor cells. The bone environment likely induced epithelial-to-mesenchymal transformation of the Met-1 cells. Although there was no difference in E-cadherin expression in the bone tumors, other genes, such as N-cadherin, vimentin, snail, and slug, should be investigated in the future. Bone is an abundant source of TGF-beta, which is known to induce epithelial-to-mesenchymal transformation.

Growth of metastases is regulated by genes and pathways inherent to the cancer cells; however, this is not an autonomous process of the cancer cells. The tumor microenvironment is an important and active partner in tumorigenesis^29,59 and contributes to both growth and metastasis. Stromal-derived factors and genes are responsible for tumor progression including tumor cell proliferation,⁶⁷ metastatic growth,⁴⁷ angiogenesis,¹⁶ and cancer cell survival.³³ Interactions between tumor cells and the stroma modify cell phenotype and differential gene expression of both the tumor cells and the surrounding stroma. In addition, inflammation,^8,10 cytokine secretion,²⁷ tension forces,⁴⁴ or loss of heterozygosity in the stroma of breast cancer cells²⁰ all support the growth of primary and metastatic tumors.

Previous studies have shown that there are important molecular events involved in cell-to-cell and cell-matrix interactions that permit local invasion of cancer cells. These events include (1) regulation of adhesion molecules, such as cadherins, which mediate physiologic interactions of neighboring cells;⁵¹ (2) secretion of MMPs, which degrade and interact with the stromal matrix,^9,61 integrin signaling, and gene regulation;²³ and (3) tumor–stroma interactions.²⁹ Interestingly, once tumors are established at metastatic sites, these factors have also been shown to confer a growth advantage through interaction with the microenvironment.²² For example, cancer cells that secrete MMPs modify the metastatic microenvironment, to permit tissue invasion and confer the ability to grow in organs such as the lung, bone, or brain.^6,38,40,64 Therefore, in this study, the gene expression profiles of Met-1 tumors (Met-1 cells and the included stroma) were compared. Based on differences in gene expression, it is likely that both the inherent genetic capabilities of the Met-1 cells and the tumor microenvironment had important roles in facilitating invasion and stromal modification for growth. This study demonstrated that expression of the invasion genes (MMP2, MMP9, E-cadherin) changed depending on site of tumor growth. Therefore, specific interactions between the Met-1 cells and stroma in each organ microenvironment facilitated the invasion and growth of the Met-1 cells.

This study identified 7 genes (MMP2, E-cadherin, EGFR, FGFR2, Tie-2/Tek, Tert, and MMP9), that were significantly up- or downregulated in Met-1 tumors cells depending on the site of growth. For example, while genes such as FGFR2 were significantly upregulated at multiple sites of metastasis (ovary, adrenal gland, bone, lung), other genes (such as E-cadherin) were significantly downregulated at only 1 site (ovary). Met-1 tumors in the tibias demonstrated the greatest number of differentially expressed genes, which included the upregulation of MMP9, EGFR, FGFR2, Tek/Tie-2, and Tert. This finding suggests that the individual genetic profile of each site may have a site-specific effect on morphology and growth of cancer cells. Bone is the most common site of breast cancer metastasis, which is a significant contributor to patient morbidity and mortality.⁶⁰ It is interesting that the bone microenvironment had a robust effect on the expression of genes known to support tumor cell invasion (MMP2), proliferation (EGFR and FGFR2), angiogenesis (Tek/Tie-2), and survival (Tert) compared to other sites, supporting the conclusion that bone marrow is uniquely capable of supporting breast cancer growth and progression. However, it is unknown why the proliferation index of the bone tumors was lower compared to Met-1 tumors in the mammary fat pad, subcutis, and ovary. It is possible that factors released from bone marrow or resorbing bone (such as TGF-beta) reduced tumor cell proliferation.

Matrix metalloproteinases (MMPs) are secreted by osteoblasts to facilitate bone remodeling by removing nonmineralized osteoid and may play an important role in bone metastasis, invasion, and angiogenesis.²⁸ MMP-2 has been shown to contribute to mammary tumor progression in bone using mouse MMTV-PymT tumors and human breast cancer bone metastases.⁵⁵ Osteoblast-derived MMP-2 protected tumor cells from apoptosis, promoted tumor cell survival, and contributed to tumor-induced osteolysis. MMP-2 was responsible for activating TGF-beta, which increased tumor cell survival. It was concluded that MMP-2 and TGF-beta are important in the vicious cycle of bone metastases in breast cancer.⁵⁵ Inhibition of the mitogen-activated protein kinase, p38, in human MDA-MB-231 breast cancer cells reduced bone metastasis in mice, decreased cell motility, and decreased MMP-9 activity in the tumor cells.⁵⁰ Inhibition of MMP activity has the potential to reduce tumor burden and osteolysis in bone metastases.¹⁴

The fibroblast growth factor receptors (FGFR) may be useful targets for therapy of breast cancer and bone metastases.²⁵ The FGFs and FGFRs function in an autocrine or paracrine manner in breast cancer, and the MMTV commonly integrates near FGF and FGFR genes, including FGFR2.⁵⁴ FGFR1 was important and FGFR2 and FGFR3 less important for human S115 breast cancer cell proliferation, angiogenesis, and tumor growth.⁵³ FGFR tyrosine kinase inhibitors decreased the activity of extracellular signal-regulated kinase 1/2 and AKT and decreased tumor growth and lung metastasis of mouse 4T1 mammary tumors in vivo.¹² In addition, MMP-9 was downregulated, which is necessary for metastasis of 4T1 tumors.

The epidermal growth factor receptor (EGFR) and EGFR ligands are upstream regulators of parathyroid hormone-related protein (PTHrP) in cancers that metastasize to bone, including breast cancer.^19,21,63 PTHrP is an important metastasis gene and is responsible for inducing osteoclastic bone resorption in bone metastases.^11,46,48 Inhibition of PTHrP or EGFR can be used to reduce bone resorption associated with bone metastases or cancer-associated hypercalcemia.³⁵ A tyrosine kinase inhibitor for EGFR has been shown to decrease PTHrP expression and reduce cancer-associated hypercalcemia in mice with human squamous cell carcinoma.³⁷

E-cadherin expression was reduced in all sites of Met-1 tumor growth with the greatest downregulation in ovary metastases compared to Met-1 tumors in the mammary fat pad. Reduced expression of E-cadherin and subsequent reduction in cell:cell adhesion may have contributed to the high degree of anaplasia and necrosis in the ovarian metastases compared to other sites. Interestingly, reduced E-cadherin expression has been associated with necrosis in colon cancer⁴⁹ and with the transformation of differentiated thyroid carcinoma to anaplastic thyroid carcinoma.⁶⁵

This study demonstrated that the Met-1 cells have the ability to grow in the bone microenvironment and metastasize to bone from the arterial circulation. These findings will be important for future studies with the Met-1 cells, which have historically been used only for modeling lung metastasis. It is likely that a bone-selective subline could also be derived from serial passage in vivo in the tibias. The intratibial tumors also showed the greatest degree of genetic variation compared to the mammary fat pad tumors.

In conclusion, this study evaluated the phenotypic, growth, and metastatic patterns of MMTV-PymT cells (Met-1) when challenged in the currently available models of metastasis. The results demonstrated that the type of metastasis model has a significant impact on the sites of metastasis, tumor growth, and tumor morphology of MMTV-PymT cells (Met-1). Changes in gene expression were also identified in tumors, which were different depending on the site of growth. Data from this investigation indicate that breast cancer progression and metastasis are regulated by not only the inherent capabilities of the tumor cells but also the unique molecular signals from tumor microenvironments. Animal models will continue to be indispensable to investigate the pathogenesis of metastasis in vivo; conduct preclinical chemotherapeutic, chemoprevention, and genetic therapy studies; test gene delivery mechanisms; and identify metastasis suppressor and inducer genes.⁴⁶ However, the goal of any animal model is to reliably recapitulate human disease. In this investigation, the type of metastasis model had a significant impact on tumor morphology and metastasis. Therefore, these data must be considered when selecting models for mechanistic studies designed to evaluate the role of specific factors in the pathogenesis of metastatic breast cancer.

Figure 5.

Mammary fat pad. Mouse. Met-1 tumor from the inguinal mammary fat pad with mixed-cell phenotype. Hematoxylin and eosin (HE).

Figure 20.

Subcutis. Mouse. Met-1 tumor. Subcutaneous tumors were chiefly mixed phenotype and were expansile and nonencapsulated with compression and invasion of adjacent tissues. HE stain.

Figure 23.

Tibia. Mouse. Intratibial Met-1 tumor. Sixteen percent of the Met-1 cells were positive for Ki67. Diaminobenzidine immunohistochemistry with hematoxylin couterstain.

Figure 24.

Lung. Mouse. Met-1 metastasis. Twenty-nine percent of the Met-1 cells were positive for Ki67. Diaminobenzidine immunohistochemistry with hematoxylin couterstain.

Figure 25.

Mammary gland. Mouse. Met-1 tumor. Thirty-eight percent of the Met-1 cells were positive for Ki67. Diaminobenzidine immunohistochemistry with hematoxylin couterstain.

Figure 26.

Ovary. Mouse. Met-1 metastasis. Forty-one percent of the Met-1 cells were positive for Ki67. Diaminobenzidine immunohistochemistry with hematoxylin couterstain.

Figure 27.

Subcutis. Mouse. Met-1 tumor. Forty-six percent of the Met-1 cells were positive for Ki67. Diaminobenzidine immunohistochemistry with hematoxylin couterstain.