C. Khanna, DVM, PhD, DACVIM (Oncology); P. Meltzer, MD, PhD
The development of metastasis is a universally grave development for cancer patients irrespective of specific cancer histology. For both human and canine cancer patients afflicted with osteosarcoma, the most common primary tumor of bone, metastasis to the lungs is the most common cause of death. Lung metastases develop in these patients despite highly effective treatment of the primary tumor. Improving our understanding of the biology of metastasis (verb) as a process and as a clinical entity (noun) is needed to improve outcomes for these patients. Progress towards understanding metastatic disease and its inherent resistance to conventional treatments is limited by many factors. Firstly, the process of metastasis is believed to begin very early in the course of disease progression and has occurred in most patients at the time of their initial presentation. This provides limited opportunities to study these events in patients. Secondly, the genetic aberrations that are responsible for the development of metastasis are complex, heterogeneous and difficult to distinguish from the events responsible for the actual development of cancer. Finally, the access to well described patient samples is limited and often is available only after treatment with cytotoxic chemotherapy. For relatively rare cancers such as osteosarcoma, these problems are amplified.
To define genes and or proteins that contribute to the metastatic phenotype of metastasis in osteosarcoma, we have utilized a cross-species comparative approach that includes murine, canine, and human systems for gene identification and evaluation. This approach has been based on the use of novel non-candidate investigative platforms that are able to survey the expression of genes in normal and diseased tissues and then identify either patterns of gene expression or individual genes responsible for or associated with disease. As an example, we used this comparative approach to identify and then associate the cytoskeleton linker protein, ezrin, with metastasis. The stability and consistency of our findings involving ezrin across species lines has strengthened our belief in this cross-species approach. We have recently extended these efforts to include non-candidate microarray comparisons of gene expression in canine and human osteosarcoma as a means to develop and test new hypotheses regarding the biology and treatment of osteosarcoma metastasis.
The Biology of Metastasis
Metastasis is defined the dissemination of neoplastic cells to distant secondary (or higher order) sites, where they proliferate to form a macroscopic mass. Implicit in this process is the presence of a primary tumor. Metastases are not a direct extension of the primary tumor and are not dependent upon the route of spread (i.e hematogenous vs. lymphatic vs. peritoneal seeding). The process of metastasis is believed to occur through the completion of a series of step-wise events. In order for this process to occur a cancer cell must leave the site of the primary tumor, pass through the tumor basement membrane, and then through or between endothelial cells to enter the circulation (extravasation). While in the circulation tumor cells must be able to resist anoikis (programmed cell death associated with loss of cellular contact), evade immune recognition and physical stress, and eventually arrest at distant organs. At that distant site the cell must leave the circulation and survive in the hostile microenvironment of the foreign tissue. This distant site may be the eventual target organ for metastasis or may be a temporary site. In either case the cancer cell is thought to lie dormant for a protracted period of time before moving to its final location. Following dormancy, cells receive signals to proliferate, create new blood vessels (angiogenesis) or co-opt existing blood vessels and then successfully grow into a measurable metastatic lesion. It is likely that further progression is associated with the repetition of this process and the development of metastases from metastases; as such the steps outlined above continue not only after the detection of the primary tumor but also after the detection metastases. The basic tenants of this model of metastasis have been intact for over 40 years; however, a greater understanding of biological principles associated with the each metastasis process is emerging. A more detailed understanding of each of the steps associated with the metastatic process has emerged from the recent interest and investment from a diversity of disciplines not previously active in this area of cancer biology. The application of mathematic models, bioinformatics, genomic screening, physics and physical chemistry has provided the field with unique "systems" perspectives and investigative tools. The validation of these perspectives and the use of novel tools within the field of metastasis biology have and will require the appropriate use of in vivo models. In the field of cancer, there is a tendency to characterize the value of a model in terms of the similarities shared between the animal and the human cancer that is being modeled. This approach leads to value assignments and an attempt to define the best model of a cancer. It is unlikely; however, that the complexity of cancer in human patients can be entirely modeled by one system alone. By understanding the strengths and weaknesses of a set of models, it becomes possible to choose the appropriate model(s) for study of individual problems or questions. This is true in the use of animal models in preclinical drug development and for studying a complex problem like cancer metastasis. Until recently a significant weakness in the study of cancer biology in canine cancer models has been the availability of reagents. The development of novel technologies for molecular reagents, antibody development, protein expression, and protein-purification has lowered the hurdle for developing canine specific reagents to study spontaneous disease. In 2006 the first public draft of the canine genome sequence was released. This milestone provided the opportunity for dogs with cancer to lend additional insight into the biology of similar human cancer conditions and significantly improves opportunities to evaluate targeted and molecular therapies in canine cancers. Data from the public draft of the canine genome and a recent report by Kirkness et al suggests greater homology between dogs and humans "by several measures" than either species and the mouse. This genetic similarity and the relatively outbred nature of companion animals provide a strong rationale for the use of dogs in biomedical research and more importantly dogs with spontaneous disease (including cancer). For the more commonly studied canine cancers, strong similarities with the same human cancers have been shown. Efforts to validate reagents and further characterize models using more sophisticated techniques has been ongoing within several comparative oncology laboratories around the world. Contributing to this effort, the intramural program of the National Cancer Institute's Center for Cancer Research has recently launched the Comparative Oncology Program. The goals of this program will be to facilitate the use of companion animal cancers in the process of cancer research through the characterization of these models and the design and implementation of preclinical translational trials (http://ccr.nci.nih.gov/resources/cop/).
Metastasis-Associated Genes and Metastasis Suppressor Genes
Cancer cells are not unique in their ability to complete the individual steps required for metastasis. For example, leukocytes and neuronal cells have the ability to invade tissue planes and cross-vascular barriers. Several types of leukocytes demonstrate the phenotype of intermittent adherence to vascular endothelium and are able to resist anoikis. It is also true that stem cells, of various phases of differentiation, are able to perform many of these steps during development and in the adult. What is unique about metastatic cells is that an individual metastatic cell must be able to perform all the steps required for successful metastasis. An extension of this argument is that the genetic changes that permit the metastatic process are not unique to a metastatic cancer cell; however, for the successful navigation through the metastatic cascade a metastatic cancer cell must have an appropriate set of genetic changes. Literally hundreds of genes and their resultant proteins have been suggested to contribute to the development of cancers and to their eventual ability to metastasize. It is possible for a single genetic change in cancer to contribute many of metastasis-associated processes or for several genes to work together towards a single metastasis-associated process. Different metastatic cancers may achieve the metastatic phenotype through distinct constellations of genetic events that in their respective sums complete the list of necessary metastasis-associated processes needed for successful metastasis. Two classes of genes have been broadly defined as contributing to the metastatic phenotype. These include metastasis promoting genes and metastasis suppressors. These genes have functions in normal development and physiology (i.e., cell migration, tissue invasion, and angiogenesis discussed above) that are subverted by the cancer cell in the acquisition of the metastatic phenotype.
The use of high through-put and genome-wide investigations has uncovered many putative metastasis-associated genes in cancer. It should be noted that many metastasis-associated genes have functions that also contribute to tumor formation and progression. Several of these metastasis-associated genes have been validated in canine and feline cancers. Metastasis suppressor genes have been identified in several human cancers. The function of genes has been reviewed elsewhere. Suppressor genes are thought to regulate motility, invasion, angiogenesis, and other processes associated with metastasis. The loss of these genes is, by definition, not thought to be associated with the formation of a primary tumor. Examples of metastasis suppressor genes are listed in table II. The loss of reduced expression of metastasis suppressors has not been documented in canine or feline cancers at this time.
Identification of Ezrin, a Metastasis-Associated Gene
As discussed above to define metastasis-associated genes in osteosarcoma we utilized a comparative approach based on the study of murine, canine, and human cancers for gene identification and evaluation. We have used this comparative approach to define a novel metastasis-associated gene called ezrin. We first identified ezrin using cDNA microarrays and a metastasis based methodology for array evaluation. Through this approach we defined genes most likely to explain differences in the behavior of a high and low metastatic model of murine osteosarcoma. Ezrin's linkage of the cell membrane to the actin cytoskeleton directly allows the cell to interact with its micro-environment and functionally provides an "intracellular scaffolding" that facilitates signal transduction through a number of growth factor receptors and adhesion molecules1. A member of the Band 4.1 superfamily of proteins, ezrin is the best characterized of the ERM (Ezrin-Radixin-Moesin) family. ERM proteins exist in the cytoplasm in an inactive "closed conformation" though N-terminal to C-terminal associations within the protein or with other ERM members. Upon phosphorylation (threonine and tyrosine) ezrin assumes an open "active conformation", moves to the cell membrane, and tethers f-actin directly or indirectly to the cell membrane. We have demonstrated that ezrin is necessary for metastasis in murine transplantable osteosarcoma and genetically engineered rhabdomyosarcoma models, that it is relevant in human-murine sarcoma xenograft models, and that its expression is associated with metastatic progression in pet dogs with naturally occurring osteosarcoma, finally we have found an association between ezrin expression and risk of relapse in pediatric osteosarcoma patients.
Evaluating Antimetastatic Therapies Based on Ezrin Biology
Using similar investigative approaches we have identified an association between ezrin expression and the efficient initiation of translation of proteins by metastatic cells. We have begun to validate these associations in murine, canine and human cancer cells and have conducted preclinical studies in mice that support the therapeutic role of inhibiting translation initiation using rapamycin (and novel analogs). To complete the translational effort we have now completed dose and regimen finding studies in pet dogs with osteosarcoma that will allow the evaluation of the inhibition of translation initiation (by rapamycin) in pet dogs with osteosarcoma.
Extending the Opportunity of the Cross Species Approach
Based on the genomic opportunities provided by the completion of the canine genome, the identical biological behavior of canine and human osteosarcoma, and the increased prevalence and aggressiveness of this disease in dogs we recently used our cross-species comparative gene expression approach to uncover specific genes, gene families/functions, or pathways that were conserved across the dog and human and are commonly linked to metastasis. Using identical oligonucleotide microarray platforms we compared expression signatures for primary tumors and normal tissues from both dogs and humans. Rather than finding clear similarities and differences between the canine and human cancers, the similarities between canine and human osteosarcoma were so strong, that cluster analysis of 265 orthologous transcripts could not distinguish the canine and human cancers. Based on the surprising resemblance between the expression profiles of osteosarcoma in human and dog we then asked whether the aggressive biology of the dog disease could help identify genes important in metastatic progression of osteosarcoma that would have been overlooked if the human disease was studied alone. Four genes were identified from 15, that were consistently overexpressed in canine osteosarcoma ("dog-like" genes) but variably expressed in humans. Using a naïve human expression data set of human osteosarcoma, that was linked to outcome, we found that two of the "dog-like" genes were associated with a more aggressive clinical course in patients. Statistical comparison to random selection approaches suggested this to be unlikely the result of chance alone. This is the first genome-wide comparison of disease conditions between dog and human. The cross-species comparative expression approach demonstrated a remarkable similarity between canine and human osteosarcoma.
Collectively, our comparative approach has provided a novel, necessary and informative perspective for the study of cancer biology. Outcomes of this effort are expected to yield new insights into the understanding of osteosarcoma biology and from their new opportunities to manage this important and life-threatening problem. The approach exemplifies the values of One Medicine model in the study of complex biomedical problems.
1. Paoloni M, Khanna C. Nature Reviews Cancer. 8(2):147-56, 2008.
2. Baird K, et al. Cancer Res. 65(20):9226-35, 2005.
3. Bruce B, et al. Clin Exp Metastasis 24(2):69-78, 2007.
4. Khanna C, Helman L. Annual Reviews Medicine. 57:83-97, 2006.
5. Khanna C, et al Nature Medicine 10(2): 182-186, 2004.
6. Khanna C, et al. Cancer Res. 61(9):3750-9, 2001.