Of mice and (wo)men: Mouse models of breast cancer metastasis to bone

Abstract

Approximately 200,000 new cases of breast cancer and 40,000 breast cancer deaths occur annually in the United States.1 Among breast cancers that become aggressive, metastasis to bone marrow is common.2-5 In many respects, the problem of metastasis to bone is more serious than the original tumor, and its detection often signals the onset of the life-threatening phase of this disease. Long before death, however, a number of other serious complications threaten the well-being of the patient: These include fractures, disability, pain, and hypercalcemia.6 While important advances have been made in understanding and treating the primary malignancy, relatively little progress has been made in understanding the pathogenesis of metastasis and developing effective therapies for these disseminated growths. To date, relatively few validated targets have been identified in breast cancer metastases, precluding efforts to discover metastasis-specific therapies. Such treatments, combined with strategies to eradicate the primary malignancy, would have high clinical impact. These considerations highlight the compelling need to understand the molecular mechanisms by which breast cancer metastasizes to bone and the urgent need for safe and effective agents to prevent or treat breast cancer skeletal metastases. One barrier to identifying the mechanisms of osteotropic metastasis has been the lack of animal models that reflect the complex biology of metastasis in humans.7 A number of experimental models of murine mammary tumor development have been developed, but these models rarely generate bone metastases from the primary site.8 The most commonly used models of breast cancer metastasis employ intravenous administration of cancer cells, a route that bypasses the early steps of the invasion-metastasis cascade, which include local invasion and intravasation. Unfortunately, these experimental models typically generate lung metastases, with few concomitantly developed skeletal metastases. In order to circumvent the trapping of cells in the lung microvasculature, thereby permitting alternative sites of metastasis formation, alternative routes of administration have been employed. Thus human breast cancer cells can be administered via intracadiac injection or are introduced into the mouse bone marrow through direct inoculation.9-11 While these models of human breast cancer to the mouse bone fail to recapitulate the early steps of the invasion-metastatic cascade, they often can generate bone metastases at high frequency and have proven to be valuable for elucidating the later steps of this progression, specifically the process of metastatic colonization, in which initially formed micrometastases succeed in colonizing the marrow and generating osteolytic metastases. These models have been improved further by the use of mice bearing engrafted human bone fragments, allowing migration of human cancer cells to a human target organ.12-15 A major impediment to research on bone metastasis has been the complexity of this multistep process. In order to disseminate from their site of origin, primary carcinoma cells must invade through the adjacent stroma and enter blood or lymphatic vessels (intravasation). Hematogenous spread, which is responsible for the vast majority of life-threatening metastases, results in the deposition of cells in the microvasculature of diverse tissues throughout the body. Specific adhesion by carcinoma cells to the luminal surfaces of endothelial cells may occur in some tissues. However, a more common mechanism of dissemination likely involves the mechanical trapping of 20 to 30 µm diameter cancer cells in the approximately 8 µm wide capillaries. Once lodged in these vessels, carcinoma cells may begin to form a colony intraluminally, eventually rupturing the microvessel around them. Alternatively, cancer cells can invade through the endothelium into the adjacent tissue parenchyma. Following this extravasation, the breast cancer cells experience a novel and quite foreign microenvironment to which they must adapt. Some may survive to form micrometastatic colonies. Very few of these will ever succeed in growing into macroscopic metastases, which, as mentioned earlier, herald the life-threatening phase of the disease. As many as one-third of breast cancer patients, on clinical presentation, carry many thousands of micrometastases in their marrow, but only half of these will ever develop metastatic disease. This provides a dramatic indication that colonization is a highly inefficient process.16-19 Breast cancer metastases generally are osteoclastic, involving demineralization. Rather than resorbing bone directly, the carcinoma cells subvert the physiologic resorptive apparatus of bone, namely, the resident osteoclasts of the marrow, which are normally involved, together with bone-forming osteoblasts, in continuous remodeling of mineralized bone.20-22 This localized activation of the bone-resorption process depends on the ability of breast cancer cells to secrete various factors that perturb the interactions between the osteoclasts and osteoblasts.23-25 The resulting osteolysis releases a number of growth factors and cytokines that are normally sequestered in the collagenous matrix of the bone, which further stimulates the growth and survival of breast cancer cells. The latter then stimulate ever-increasing numbers of osteoclasts, resulting in a self-perpetuating vicious cycle.26 Breast cancer cells secrete parathyroid hormone–related protein (PTHrP), enabling them to recruit a “shell” of osteoblasts around a metastatic colony.20, 22, 36 The latter, by releasing RANKL, stimulate the maturation of osteoclast precursors into osteoclasts. The osteoclasts then resorb mineralized bone, mobilizing factors previously sequestered in its organic extracellular matrix (ECM), among them transforming growth factor β (TGF-β); TGF-β, in turn, stimulates PTHrP release by the tumor cells, fueling the vicious cycle mentioned earlier.26, 37, 38 In addition, the dissolution of the collagenous matrix releases a series of mitogenic and trophic factors, notably bone morphogenetic proteins (BMPs), fibroblast growth factots (FGFs), and platelet-derived growth factor (PDGF). Stromal interactions are important and distinctive for breast cancer osteotropism. Breast cancer cells display cell surface adhesion molecules, such as the αvβ3 integrin, that bind to bone matrix proteins. Osteopontin (OPN) binding and signaling through αvβ3 integrins on breast cancer cells can block apoptosis, providing a survival advantage to breast cancer cells in the bone marrow.27, 28 Breast cancer cells decorate themselves with bone matrix proteins such as bone sialoprotein and OPN, enabling bone stromal cells expressing integrins to trap them.25, 29, 30 There are similarities between stroma of the primary mammary tumors and those of the bone marrow, which may contribute to the ability of breast cancer cells to survive and proliferate in bone. Both microenvironments contain conserved stromal and basement membrane components (i.e., collagen type I, laminin, and fibronectin), as well as distinctive components; for example, many breast carcinomas express OPN.31 Receptor activation of NF-κB signaling is essential for both osteoclasts and mammary epithelial cells during mammary gland development.32 The transcription factor Runx2, involved in osteogenesis, is also present in breast cancer.33-35 Moreover, within the bone marrow, disseminated breast cancer cells encounter additional mitogens and other factors that normally support hematopoiesis and adult stem cells. Together these factors in the bone environment present the disseminated carcinoma cells with an array of chemokines, growth factors, and antiapoptotic factors, too numerous to list here, that support colonization and thus the development of macroscopic metastases.20, 22, 27, 39-43 While some rodents may develop spontaneous mammary tumors, these tumors rarely metastasize or resemble the human histopathology seen in breast carcinoma development. In certain strains of laboratory mice, spontaneous mammary tumors are often the result of retroviral integration of the mouse mammary tumor virus (MMTV) provirus. While a corresponding role for retrovirus integration in human mammary tumor pathogenesis has not been demonstrated, this MMTV promoter and other mouse mammary-specific promoters have proven to be useful experimental tools that can be used to drive tissue-specific expression of oncogenes, among them genes known to be responsible for human malignancies (e.g., ErbB2/Neu, Ras, Myc) and other oncogenes (e.g., pyMT, SV40 LT antigens). The resulting transgenes cause rapid tumor development and, in some cases, metastasis to distant sites.44-48 While these models provide a platform to study oncogene-induced carcinogenesis and soft tissue metastasis within a mouse model, some rely on oncogenes that are rarely or never involved in human breast cancers (e.g., pyMT, SV40 T antigen), and all the resulting mammary tumors generate skeletal metastases at a low frequency, if at all.49 In addition to oncogene-induced carcinogenesis and ensuing metastasis, radiation and applied carcinogenic chemicals have been used to drive mammary tumor formation and dissemination. BALB/c mice have a known polymorphism in the DNA-dependent protein kinase catalytic subunit (Prkdc) that, in the context of a p53 mutation, renders them susceptible to mammary tumors following ionizing radiation exposure.50-52 Similar to MMTV-driven tumorigenesis, these models rarely metastasize to distant sites and almost never disseminate to the skeleton. Additionally, these models share few similarities with the clinical course or histopathology of breast cancer in humans, in part because they activate oncogene expression in inappropriate cells of origin in the mammary gland and express these oncogenes at supraphysiologic levels. The 4T1 mouse model of breast cancer development and metastasis was identified originally as a spontaneous breast cancer model in the BALB/c mouse strain.53 When implanted in syngeneic hosts, the system has been used repeatedly over the past two decades as an experimental model to study tumor development and metastasis to various organs. The parental 4T1 cell line, when injected orthotopically (i.e., into the mammary fat pad), is capable of limited and infrequent metastasis to local lymph nodes, lung, liver, brain, and skeleton, whereas sublines of the 4T1 line (4T1.2 and 4T1.13) have been developed that are highly metastatic and have tropisms to specific organs, including the skeleton.54 These sublines have been used to identify gene expression signatures from different stages of tumor development and metastasis. Moreover, study of the gene expression signature of 4T1 tumor cells relative to other closely related mammary tumors led to the identification of Twist, a transcription factor that serves as an important mediator of the metastatic dissemination of these breast cancer cells.55 The utility of the 4T1 model as a model of breast cancer pathogenesis derives in part from the fact that these tumors can be studied in immunocompetent hosts. The immunocompetence of the host mouse allows the researcher to study the interactions among tumor cells, cells of the immune system, and stromal cells within the marrow. Recent work has demonstrated, for example, the important role of immune cells, namely, T-lymphocytes, dendritic cells, and macrophages, in tumorigenesis and cancer progression.56-59 Consequently, these fully competent models are useful in elucidating the functions of novel pharmacologic agents and antitumor vaccines under development, and in the future, they may enable study of the contribution of various immunocytes to osteotropic metastasis.60 Unfortunately, none of these systems of studying autochthonous breast tumor formation in mice provides a robust skeletal metastasis phenotype. Moreover, these tumors do not closely resemble, at the histopathologic level, breast tumors that are commonly encountered in the oncology clinic. In an effort to develop models that more closely mimic breast carcinoma in humans, mouse xenograft models have been developed that use human breast cancer cell lines within an immunocompromised animal host. While these experimental models do not recapitulate all the interactions between cancer cells and the host tissue microenvironment and immune system, their relative ease, short time to metastasis, and histopathologic similarities to corresponding human tumors make them attractive models to study tumor development and bone metastasis. Initial xenograft models of skeletal metastasis used direct inoculation of cancer cells into the skeletons of immunocompromised mice.61 These models result in near 100% frequency of development of these “experimental metastases” and have been used to assess efficacy of therapeutic agents under development; such agents have been assessed, more specifically, for their ability to block breast cancer growth within the bone and breast cancer–mediated bone resorption.62 They also have provided an in vivo system in which human cancer cells can be passaged through a bone microenvironment, aiding the development of breast cancer cells that have been selected for their osteotropism and thus reproducible and efficient osteotropic metastasis during subsequent rounds of growth in mice. Such cells can be used to study the genetic changes that are required for survival and proliferation of cancer cells within the bone environment, as well as the changes within the bone microenvironment induced by the disseminated cancer cells. More recently, yet other xenograft models of skeletal metastases have used intracardiac injection of human breast cancer cells.11 By injecting MDA-MB-231 human breast cancer cells into the left cardiac ventricle, researchers are able to circumvent trapping of cancer cells in the lung microvasculature, allowing seeding of bone metastases. This model has led to the identification of novel genes and their products that are crucial for efficient growth of tumors within the marrow, including TGF-β, connective tissue growth factor (CTGF), CXC chemokine receptor 4 (CXCR4), matrix metalloproteinases (MMPs), and interleukin 11 (IL-11).11, 63 Additionally, the intracardiac injection model has been exploited as a laboratory model of breast cancer skeletal metastasis that is especially useful in testing of novel therapeutic and even diagnostic strategies, such as the in vivo imaging of signaling pathways active in disseminated cancer cells.64-66 In addition to intracardiac or direct skeletal injection of breast cancer cells, some xenograft models of breast cancer pathogenesis use orthotopic injection of cancer cells, resulting in fully developed primary tumors that give rise to metastases.67 Both MDA-MB-231 and MDA-MB-435 human breast cancer cell lines can be injected into the mammary fat pad and give rise to metastatic breast cancers with increased frequency of detection in the lungs, liver, and lymph nodes.68 MDA-MB-435 tumor tissue from these animal models or tumor explants comprised of human breast cancer cells and mouse stroma can be implanted orthotopically and have been shown to develop skeletal metastases at a near 100% frequency, suggesting that there are important tumor-stroma interactions at the primary tumor site that may promote metastasis to specific organs.69 As useful as they are, these experimental models may fail to recapitulate more subtle aspects of the tumor-marrow interaction that operate in human patients. First, certain signaling interactions between cancer cells and host stromal cells may not occur properly because of interspecies signaling incompatibilities, that is, interactions of ligands of one species with receptors of the other. Second, in the case of intracardiac and skeletal injection models, because these cancer cells do not originate from primary tumors growing in orthotopic sites, they may not undergo certain biologic modifications that metastasizing tumor cells undergo in response to signals received from the activated stroma in these tumors.70 A model of prostate cancer osteotropic metastasis has been developed that is instructive for the development of comparable breast cancer models. Using intravenous administration of prostate cancer cells and an implanted human bone xenograft, researchers were able to demonstrate that human prostate cancer metastasizes to bone in a both a species- and tissue-specific manner.12-14 Similar to the intracardiac and intraskeletal models described earlier, the cells that metastasize to human bone in this model are not required to execute the earlier steps of the invasion-metastasis cascade. Still, they must express genes necessary to home specifically to human bone, whose identities remain unclear. Building off the successful humanized models developed for prostate cancer, researchers have used similar techniques to develop useful breast cancer metastasis models.15, 71 A novel model of breast cancer metastasis has used subcutaneous implantation of human bone fragments followed by orthotopic injection of human breast cancer cells to monitor migration of human cancer cells from the primary tumor environment to a human bone environment.15 This model more closely follows the natural pathology of metastasis development from primary tumor growth and extravasation to homing at a distant site and thus provides a platform to study paired tumor samples (i.e., the primary tumor and subsequent metastasis) in a controlled laboratory setting. Additional modifications of the humanized model have shown that a tissue-engineered bone construct can be used as a metastasis target following orthotopic cancer injection.72 Using silk fibroin protein sponges as scaffolds, human bone marrow–derived mesenchymal stem cells can be seeded and differentiated toward an osteoblast lineage in vitro, creating a biocompatible and 3D porous, silk-based, human bone–like structure.73 The silk fibroin used to create the scaffolds can be coupled with various growth factors (e.g., bone morphogenetic proteins [BMPs] and vascular endothelial growth factor [VEGF], and the bone marrow–derived mesenchymal stem cells can be fluorescently labeled for tracking or genetically manipulated to assess the contribution of specific factors within the bone microenvironment.74 In addition to recent advances in humanizing and manipulating the metastatic microenvironment described earlier, new findings have shown that the primary tumor stromal environment also can be at least partially humanized.75 By implanting human mammary stromal cells from reduction mammoplasties, the mouse mammary fat pad can be converted into a more humanized microenvironment comprised of mouse and human stromal cells; these, in turn, should create a more hospitable microenvironment for the subsequent implantation of human mammary epithelial cells, including those of neoplastic origin. Further modifications of this model eventually may allow the complete humanization of the stromal microenvironment. While, as of this writing, this model has not yet been used to study metastasis, it suggests the possibility of studying the full cascade of steps required in epithelial transformation, tumorigenesis, and metastasis in a fully humanized system. While no animal model can fully recapitulate the progression of breast cancer tumorigenesis and metastasis from initial epithelial cell transformation to extravasation and final arrest in the skeleton, advances in genetic and surgical techniques have combined to move the field progressively closer to using models of human osteotropic metastasis that more precisely mirror disease pathogenesis in humans. The importance of these advanced models is clear, namely, identification of new molecular targets to block metastasis, testing of metastasis-specific therapies in a preclinical setting, and a greater understanding of the interactions of carcinoma cells with the stromal microenvironments that they encounter in the primary tumor and in sites of metastatic dissemination, including the bone. It is our hope that these discoveries and the greater understanding of cancer metastasis that will emerge from these models will begin to address the urgent medical need of the thousands of women who are diagnosed with metastatic breast cancer each year. The authors state that they have no conflicts of interest.