Aneuploidy is a hallmark of human cancer, affecting over 90% of solid tumors. The distinct patterns of aneuploidy across tumor types strongly suggest that some chromosomal changes play a driving role in tumorigenesis. However, aneuploidy is clearly associated with a substantial fitness cost under most circumstances. This “aneuploidy paradox” remains an open question in cancer research, mostly because aneuploidy is challenging to study.
We employ experimental and computational approaches – combining genomics, genome engineering, drug screening and mouse modeling, as well as cutting-edge cell and molecular biology – to better understand this intriguing and important phenomenon.
IDENTIFYING DRIVERS OF ANEUPLOIDY
Our work revealed driver-specific patterns of aneuploidy in genetically-engineered mouse models of cancer, narrowed down the region of interest in one of the most recurrent chromosomal changes in human breast cancer (loss of chromosome 1p), and identified a gene (Sfn) that cooperates with Erbb2 during breast cancer tumorigenesis.
We are applying novel genomic approaches and various techniques for aneuploidy modeling, in order to identify genes that drive the recurrence of specific chromosomal changes in specific cancers.
UNCOVERING VULNERABILITIES OF ANEUPLOID CELLS
Aneuploidy comes with a substantial fitness cost, yet cancer cells evolve to tolerate aneuploidy. The genetic or epigenetic events that lead to aneuploidy, or that allow cancer cells to tolerate it, may also generate unique differential dependencies in aneuploid cells (compared to their euploid counterparts). Such differential dependencies may open a therapeutic window for the selective targeting of aneuploid cells. As cancer cells are almost invariably aneuploid, whereas normal cells are (almost) always diploid, the identification of aneuploidy-targeting drugs has been a long sought-after goal of modern cancer research.
Our work exposed associations between recurrent aneuploidies and drug response in patient-derived xenografts and in human cancer cell lines. We also identified and characterized a novel isogenic system to study cancer aneuploidy in vitro, using naturally-arising genetically-matched aneuploid cell lines. We are performing genetic and chemical screens to identify vulnerabilities of aneuploid cancer cells, and we are using isogenic aneuploidy models to dissect the molecular and cellular basis of these vulnerabilities.
UNDERSTANDING THE SELECTION PRESSURES THAT SHAPE ANEUPLOIDY LANDSCAPES
Different tumor types present distinct patterns of aneuploidy. We previously generated aneuploidy catalogs of common cancer models (mouse models, PDXs, cell lines) and found that hallmark recurrent aneuploidies of human cancer tend to gradually disappear in the model environment, presumably due to distinct selection pressures in the patients and in the models. We are applying computational and experimental approaches to dissect the positive and negative selection pressures that shape the aneuploidy landscapes of tumors.
CANCER MODEL EVOLUTION
Cancer precision medicine is based on the idea that the genomic features of human tumors are associated with their response to drugs. Therefore, assessing the genomics of a tumor could predict its vulnerabilities, thus informing an optimal treatment regimen for the patient. In an attempt to associate genomic features with cellular vulnerabilities, the cancer community heavily relies on cancer models, such as genetically engineered mouse models (GEMMs), patient-derived xenografts (PDXs), and human cancer cell lines. As any biological system, these models evolve, and their evolution can affect their biological traits.
LEVERAGING VARIATION IN CANCER MODEL SYSTEMS
Our work revealed an under-appreciated degree of genomic instability in multiple cancer models, and characterized the implications of this instability for the use of such models in personalized cancer medicine. Specifically, we identified extensive genomic evolution of cancer models, distinct from that seen in patients, which can result in gene expression changes and disparate drug response. The studies of the stability and faithfulness of cancer models improve our biological understanding of these models, and can help guide their proper application.
Importantly, the genomic instability of cancer models presents a unique opportunity to use these models in novel and creative ways, e.g. the study of types of genetic variation that cannot be readily introduced experimentally, such as large chromosomal changes. We continue to study the extent, origin and consequence of the natural evolution of cancer models, and we are leveraging this phenomenon for better modeling of human cancer.
Selected Publications on Cancer Model Evolution
MODELING CANCER WITH STEM CELLS
Human pluripotent stem cells (hPSCs) can self-renew and differentiate into all cell types of the human body, making them an invaluable tool for disease modeling. The same properties also enable them to generate tumors (teratomas) upon their injection into immune-deficient mice. Indeed, hPSCs share many fundamental characteristics with human cancer, and can therefore serve as a good model system for the study of cancer genetics.
APPLYING STEM CELLS IN ANEUPLOIDY RSEARCH
Our previous work characterized the chromosomal changes that frequently arise in stem cells in culture, revealed their functional significance and developed a new approach for their detection. We found that stem cells in culture present the same aneuploidy patterns characteristic of tumors of the same tissue origin. We also revealed that a gain of a single chromosome (trisomy 12) can increase the proliferation and tumorigenicity of hPSCs, essentially “transforming” them. In addition, our work revealed a mechanism through which aneuploidy promotes further genomic instability in hPSCs.
We are currently using hPSCs as a useful tool to dissect the determinants of cellular response to aneuploidy, and to study the role of aneuploidy during early stages of tumorigenesis.
TARGETING STEMNESS IN HUMAN PLURIPOTENT STEM CELLS AND IN CANCER
We deciphered some of the mechanisms that confer tumorigenic traits to hPSCs, and developed new strategies to selectively kill undifferentiated hPSCs. We found that these strategies hold true for undifferentiated cancer cells, thereby enabling the selective elimination of cancer stem cells. A prominent example is the dependence of undifferentiated pluripotent and cancer cells on the synthesis of the monounsaturated fatty acid oleate: SCD1 depletion selectively kills undifferentiated hPSCs and undifferentiated cancer cells across multiple cancer types. We are interested in further interrogating hPSCs to identify additional vulnerabilities of cancer stem cells.