It has become clear that there is no single type of breast cancer, but rather at least six sub-types of human cancer have been identified through gene expression profiling. Our recent work has focused on understanding which mouse models of human mammary cancer represent specific sub-types of human breast cancer in order 1) to better understand what determines the mechanisms that underlie the lineage determinants of breast cancer sub-types, and 2) which models best represent particular sub-types of human breast cancer for use in pre-clinical testing for those sub-types. Recently, we have identified in mouse models where functions of both p53 and Rb have been compromised a very comprehensive, integrated genetic network of genes related to replication, DNA synthesis and repair, chromosome maintenance, proliferation, and apoptosis. This network is highly represented in- and predictive of basal-type breast cancer and the most aggressive forms of prostate and lung cancers. We hypothesize that many of these previously unrecognized genes related to basal type tumors may be potentially important targets for anti-cancer therapies and we are pursuing this experimentally. We have also expanded our work related to understanding global genomic changes that occur in GEM mammary tumor models based upon the initiating oncogenic event. We have developed gene expression, array CGH and miRNA datasets from the same tumors from eight mammary tumor models. We hypothesize that cancer may evolve differently depending upon the initiating oncogenic event and that understanding these processes may provide insights into identifying secondary changes that are critical for tumor development and progression. We have determined that models with compromised function of p53, Rb or BRCA1 evolve into a basal-type of mammary cancer, whereas the MMTV-oncogene driven models reflect a luminal phenotype. Interestingly, significant differences in genomic copy number changes and ploidy also distinguish these various mammary tumor models. Analysis of miRNA expression patterns also differ significantly between the models and the functional significance of this is being explored and compared to miRNA expression patterns in human breast cancer. Many of these distinguishing molecular features may be related to the lineage specificity of the mammary tumors that develop. Over the past year, we have developed several new mouse models of mammary cancer that demonstrate the critical interactions of genetic mutations in altering the histologic tumor phenotype that develops. Preliminary analyses indicate that the loss of expression of a particular gene can switch the tumor lineage from an adenosquamous to adencarcinoma phenotype. The molecular mechanisms related to this lineage specificity are being further explored through microarray analyses and functional studies. Crucial to understanding how various sub-types of breast cancer evolve is to explore the potential relationships between mammary cancer cells with stem-cell like properties to tumor development with luminal or basal characteristics. We hypothesize that mammary stem cell like cells can be developed from certain mouse cells through molecular manipulations and we are pursuing these approaches. Establishing such cell lines will enable us to more precisely understand mechanism that transform such cancer precursor cells and expand our ability to find therapies to overcome resistance to cancer treatments. We have begun to focus on two major transcriptional determinants of tumor development from precursor-like cells. Initial work in our lab has demonstrated that the polycomb gene Bmi-1 can greatly increase tumor aggressiveness including metastases in the context of other oncogene activation. Bmi-1 is critical for stem cell renewal and therefore may serve as an important molecular bridge between quiescence, transformation and tumor progression. Our current studies focus on understanding how Bmi is regulated and how it exerts its important biologic effects. Similarly, we are pursuing molecular studies to understand mechanisms that regulate the divergence of basal from luminal tumor types. Related to this work is our identification of a gene signature that appears to distinguish which ER+ tumors will or will not respond to anti-hormone therapies. We are functionally testing the roles of several transcription factors in this signature to determine whether any of these genes are critical determinants of why ER+ tumors do not respond to anti-hormone therapy and whether this could be used for therapeutic advantage. However, it is also clear that tumor cells do not exist as isolated entities, but interact in critical ways with the microenvironment and neighboring stromal cells. We are pursuing studies to explore cross talk between tumor epithelial cells and stromal cells under various conditions using genomic technologies. This will expand our understanding of systems biology where information exchange in a multicellular system will be determined. This will likely have important implications for identifying potential targets to interrupt this important cross-talk between tumor cells and their environment. In order to understand how tumor cells switch from a dormant state to a proliferative metastatic state, we have developed that first in vitro predictive model system to study this phenomenon. Our first major success in following this approach has been the demonstration that the proliferative switch from previously dormant tumor cells can be regulated by changes in the microenvironment, particularly through integrin signaling and cytoskeletal reorganization. We have demonstrated that disruption of integrin signaling or inhibition of cytoskeletal actin reorganization can inhibit the dormant to proloiferative switch and metastatic outgrowth in vivo. This is now a major focus of the work in our lab. Unfortunately, many tumors demonstrate an initial response to chemotherapy, but ultimately become resistant to such therapies. We have been fortunate to have access to a most unusual human set of gastric cancer samples where serial biopsie were obtained from patients prior to therapy, during clinical response and during relapse. This has enabled us to perform the first microarray study on such a patient cohort. We have developed a three-gene signature that predicts whether a patient at time of diagnosis will respond to standard chemotherapy or not. This finding should have high impact in clinical oncology. We have also determined that the global signature that emerges during acquired resistance to chemotherapy shares significant overlap with the poor prognosis signature at initial time of diagnosis. Ongoing studies are functionally confirming the role of the genes in the predictive signature in conferring resistance to standard chemotherapy. Based upon our human gastric cancer studies, we are also building mouse models for gastric cancer to use in preclinical studies for this terrible disease.
Showing the most recent 10 out of 41 publications
