Abstract:Abstract Cancer precision medicine is based on the idea that specific genomic features of a human tumor are associated with its drug response. Therefore, by assessing the genomic features of a tumor, its vulnerabilities could be predicted, and an optimal treatment could then be tailored. In order to achieve this goal, there is a burning need to develop a comprehensive “dependency map” that would enable to predict the “Achilles heel” of each tumor based on its genomic features. In an attempt to develop such cancer dependency map, the cancer community heavily relies on cancer models, such as genetically engineered mouse models (GEMMs), patient-derived xenografts (PDXs), and human cell lines (CLs). Importantly, xenografts and cell lines are used not just for cancer research, but also to predict the response of a specific patient to anti-cancer drugs (i.e., as tumor “avatars”). As cancer models are invaluable for cancer research, much effort has been recently invested in their genomic and phenotypic profiling. However, like any biological system, cancer models evolve. Whereas tumor heterogeneity and genomic instability have received much attention, the heterogeneity and instability of cancer models - and how the genomic evolution of these models affects their use in cancer research and cancer precision medicine - remain under-explored.In my postdoctoral work, I have studied the extent of genomic evolution in multiple cancer models, its biological origins and its functional implications. This work was comprised of three independent studies, which dissected various aspects of genomic evolution in three major cancer model systems: GEMMs, PDXs and CLs. In this talk, I will describe the main conclusions of my studies, and highlight important themes that emerge from their synthesis.In a first study (Ben-David et al. Nature Communications 2016), we generated the first comprehensive catalogue of copy-number alterations (CNAs) in cancer GEMMs. Mining this resource, we found that chromosomal aberrations accumulated late during breast tumorigenesis, and observed marked differences in CNA prevalence between mouse mammary tumours initiated with distinct drivers. Some aberrations were recurrent and unique to specific GEMMs, suggesting distinct driver-dependent routes to tumorigenesis. Using synteny-based oncogenomics of mouse and human data, we narrowed down the critical region of interest in one of the most recurrent chromosomal changes in breast cancer (loss of chromosome 1p), and identified a gene (Sfn) that cooperates with Erbb2 during breast cancer tumorigenesis. We experimentally validated that loss of Stratifin (SFN) promotes HER2-induced tumorigenesis. This study demonstrates how the natural genomic evolution during tumorigenesis in GEMMs can inform the identification of oncogenic networks that are important for the pathogenesis of human cancer.In a second study (Ben-David et al. Nature Genetics 2017), we followed the genomic evolution of PDXs throughout their derivation and in vivo propagation. We observed extensive evolution of the PDX copy-number landscapes, mostly driven by strong clonal dynamics leading to the expansion of pre-existing minor subclones. Clonal dynamics were strongest during derivation and early propagation of the models, and attenuated at later passages. Reproducible changes were observed across independent PDXs generated from the same primary tumor, indicating the involvement of selection, rather than mere genetic drift. Importantly, the rate of genomic evolution in PDXs was similar to that observed in patient-derived CLs. Tumor evolution in PDXs and in patients followed distinct trajectories. Specifically, aneuploidy events recurrently observed in primary tumors gradually disappeared in PDXs. The genomic stability of PDXs was associated with their response to chemotherapy and targeted drugs, suggesting that the genomic evolution of PDXs could affect the faithfulness of PDX-based prediction of patients' drug response.In a third study (Ben-David et al. Nature 2018), we compared genomic analyses of 106 CLs grown in two laboratories and revealed extensive genetic diversity. Follow-up comprehensive genomic characterization of 27 strains of the common breast cancer cell line MCF7 uncovered rapid genetic diversification. Similar results were obtained with multiple strains of 13 additional CLs. Genetic changes were associated with differential activation of gene expression programs and marked differences in cell morphology and proliferation. Barcoding experiments showed that CL evolution occurred as a result of positive clonal selection that is highly sensitive to culture conditions. Analyses of single cell-derived clones showed that ongoing instability quickly translated into CL heterogeneity. Functionally, testing of the 27 MCF7 strains against 321 anti-cancer compounds uncovered strikingly disparate drug response: over 80% of compounds that strongly inhibited some CL strains were completely inactive in others. These findings have broad implications for cell line use in basic cancer research and cancer precision medicine. Importantly, we suggest practical ways to mitigate the risks posed by CL genomic evolution and to facilitate maximally-reproducible cell line-based research. Moreover, we propose how to constructively build upon this phenomenon in future studies (e.g. to study of types of genetic variation that cannot be readily introduced experimentally, such as large chromosomal changes).In a fourth study (Ben-David et al. Submitted), we performed a deep genomic analysis of human cancer cell lines before and after the introduction of Cas9. We found upregulation of the p53 pathway upon Cas9 introduction, specifically in TP53-WT cell lines. This upregulation was confirmed at the mRNA and protein levels. Moreover, we found elevated levels of DNA repair transcriptional signatures in the Cas9 cell lines, and confirmed that the introduction of Cas9 induced DNA damage using immunofluorescence quantification of γH2AX. Genetic characterization of 42 cell line pairs showed that Cas9 introduction could lead to the emergence and to the expansion of p53-inactivating mutations. Competition experiments with isogenic TP53-WT/TP53-null cell lines confirmed that Cas9 introduction accelerated selection for p53-inactivating mutation. Lastly, we compared Cas9 activity across 719 human cell lines and found that Cas9 was significantly less active in TP53-WT cell lines, in line with functional p53 being a barrier to the efficient expression of Cas9 in human cells. These findings have broad implications for the proper use of CRISPR/Cas9 genome editing in basic research and in clinical applications. They also demonstrate the profound functional consequences of in vitro genomic evolution following a defined, common selection pressure.In summary, my postdoctoral studies revealed a previously under-appreciated genomic instability in multiple cancer models, and characterized the implications of this instability for the use of such models in cancer research. Combined, this body of work exposed extensive genetic evolution of cancer models, distinct from that seen in patients, which results in gene expression changes and disparate drug response. Our studies of the stability and faithfulness of cancer models improve our biological understanding of the advantages and limitations of the models, and can help guide their proper application. Importantly, the genomic instability of cancer models also presents a unique opportunity to use these models in novel and creative ways (Ben-David at al. Nature Reviews Cancer 2019; Ben-David and Amon Nature Reviews Genetics 2019). This work thus highlights both the perils and the opportunities of the natural evolution of cancer models. Citation Format: Uri Ben-David. Genomic evolution of cancer models: Perils and opportunities [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr NG02.

Stepwise-edited, human melanoma models reveal mutations’ effect on tumor and microenvironment

The Landscape of Gain and Loss of Function Mutations in Melanoma

Genetic and Genomic Characterization of 462 Melanoma Patient-Derived Xenografts, Tumor Biopsies, and Cell Lines.

A Comprehensive Catalogue of Somatic Mutations from a Human Cancer Genome

Melanoma models for the next generation of therapies

Whole-genome landscapes of major melanoma subtypes

Multi-modal molecular programs regulate melanoma cell state

Uncovering Minimal Pathways in Melanoma Initiation

Anti-tumor immune response varies among individuals: A gene expression profiling of mouse melanoma

Abstract A01: A comprehensive collection of patient derived xenografts of human melanoma with clinical, genomic, and biological characterization

Patient-derived melanoma models

Targeted, Massively Parallel Sequencing Identifies Novel Genetic Subsets Of Cutaneous Melanoma

Network models of primary melanoma microenvironments identify key melanoma regulators underlying prognosis

A Reconstructed Human Melanoma-in-Skin Model to Study Immune Modulatory and Angiogenic Mechanisms Facilitating Initial Melanoma Growth and Invasion

High-throughput oncogene mutation profiling in human cancer

Transcriptional dissection of melanoma identifies a high-risk subtype underlying TP53 family genes and epigenome deregulation

Human melanocyte development and melanoma dedifferentiation at single-cell resolution

Pre-clinical modeling of cutaneous melanoma

Abstract NG02: Genomic evolution of cancer models: Perils and opportunities

A preclinical model of cutaneous melanoma based on reconstructed human epidermis

Longitudinal assessment of tumor development using cancer avatars derived from genetically engineered pluripotent stem cells