Targeting Tumor Heterogeneity

Current models of carcinogenesis are dominated by the assumption that oncogenic mutations have defined advantageous fitness effects on recipient stem and progenitor cells, promoting and rate-limiting somatic evolution. However, this assumption is markedly discrepant with evolutionary theory, whereby fitness is a dynamic property of a phenotype imposed upon and widely modulated by environment. We use mouse models of cancer initiation, mathematical models of clonal evolution, and analyses of tissue-level evolutionary processes in non-cancerous human tissues to better understand the evolutionary forces that control somatic cell evolution and thus cancer risk. These studies support a model whereby the primary causes of cancer, such as old age and cigarette smoking, associate with increased risk for cancer through their impacts on tissue microenvironments. In particular, our studies support a key role for aging or insult mediated increases in inflammation in altering selection for cancer-associated mutations. These studies indicate that strategies to prevent or treat cancers will need to incorporate interventions that alter tissue microenvironments. While we cannot prevent many if not most of the mutations that accumulate through our lives, we do have the ability to manipulate tissue microenvironments so as to change the evolutionary trajectories of oncogenically-mutated cells.

Coauthors: Catherine Pham-Danis, Andrii I. Rozhok, Kelly C. Higa, Hannah A. Scarborough, Curtis J. Henry, Nathaniel Little, Travis Nemkov, Kirk C. Hansen, Angelo D’Alessandro, and Charles Dinarello.

As cancers progress, the tumor cells come into contact with a wide variety of cells in the tumor microenvironment (TME). These TME cell types can influence tumor cell behavior through cell-cell communication. To study this process in vivo, our lab has developed a zebrafish model of melanoma driven by the BRAFV600E oncogene. This model allows us to simultaneously image and perturb the TME cell types to understand their role in progression and metastasis. Using this model, we have recently uncovered novel interactions between melanoma cells and keratinocytes and adipocytes, two highly abundant cells in the TME. Adipocytes can transfer fatty acids from their cytoplasm into the melanoma cells. These fatty acids are metabolized and lead to epigenetic rewiring of the melanoma cell state. In contrast, keratinocytes are the recipient of cytoplasmic contents from the melanoma cell, which leads to their reprogramming to a more plastic, tumor-promoting state. These data suggest a complex network by which there is bidirectional cross-talk between tumor cells and a variety of heterogeneous TME cell types. The mechanisms by which these cells communicate, and ways of therapeutically targeting this cross-talk, will be the focus of the talk.

Increasing evidence supports complex subclonal relationships in solid tumours, manifested as intratumour heterogeneity. Parallel evolution of subclones, with distinct somatic events occurring in the same gene, signal transduction pathway or protein complex, suggests constraints to tumour evolution that might be therapeutically exploitable. Emerging data from TRACERx, a longitudinal lung cancer evolution study will be presented. Drivers of tumour heterogeneity change during the disease course and contribute to the temporally distinct origins of lung cancer driver events. APOBEC driven mutagenesis appears to be enriched in subclones in multiple tumour types. Oncogene, tumour suppressor gene and drug induced DNA replication stress are found to drive APOBEC mutagenesis. Evidence that intratumour heterogeneity and chromosomal instability is finely tuned will be presented, to create sufficient diversity for adaptation mitigating the risks of excessive genome instability resulting in cell autonomous lethality. On-going chromosomal instability, manifested as Mirrored Subclonal Allelic Imbalance (MSAI) is found to be a major driver of intratumour heterogeneity in non-small cell lung cancer, contributing to parallel evolution and selection. The finding of subclonal driver events, evidence of ongoing selection within subclones, combined with genome instability driving cell-to-cell variation is likely to limit the efficacy of targeted monotherapies, suggesting the need for new approaches to drug development and clinical trial design and integration of cancer immunotherapeutic approaches. The clonal neo-antigenic architecture may act as a tumour vulnerability, targeting multiple clonal neo-antigens present in each tumour to mitigate resistance and treatment failure. The role of cancer genome instability driving immune evasion and HLA/MHC loss and immune escape will be presented​.

Treatment of TNBC patients has been challenging due to the heterogeneity of the disease and the absence of well-defined targets. Cancer genome sequencing studies focusing on TNBC have failed to identify novel recurrently mutated cancer-driving genes precluding immediate opportunities for targeted therapeutic development. Due to lack of appropriate targeted therapy we have focused our efforts on studying epigenetic heterogeneity in TNBC. Using a combination of experimental and computational approaches, we have characterized the histone H3 lysine 27 acetyl (H3K27ac) profiles of a large panel of TNBC cell lines and found very few recurrent super-enhancers (SE) implying a high degree of epigenetic heterogeneity and transcriptional dependencies. Based on our preliminary data we hypothesized that (1) the epigenetic profiles of TNBCs are better predictors of the functional properties and thus clinical behavior of the cells and (2) SE-associated genes will reveal novel transcriptional dependencies and thus therapeutic vulnerabilities.

To test these hypotheses, in addition to H3K27ac ChIP-seq, we have performed Histone Mass Spec, RNA-seq, and DNA methylation and Metabolomics profiling as to have a comprehensive characterization of the epigenetic landscape of TNBC. Based on both ChIP and RNA-seq analysis we have observed three distinct clusters of TNBC cell lines. Since SE are often found approximate to genes that are important for defining cell identity we are currently investigating transcription factors as to potentially explain drivers behind these three distinct TNBC clusters observed.

Coauthors: Nicholas W. Harper[1], Veerle Daniels[1], Xintao Qui[1], Anne Fass[l]. Daniel Temko[1], Jennifer Y. Ge[1,2], Peter Sicinski[1], Jacob Jaffe[3], Franziska Michor[1], Anthony Letai[1], Henry Long[1], Myles Brown[1], and Kornelia Polyak[1,2,3].

1. Dana-Farber Cancer Institute, Boston.
2. Harvard Medical School, Boston.
3. Broad Institute of Harvard and MIT, Boston.

Diffuse gliomas have long been noted to exhibit special heterogeneity, with multiple subclones of the original cell population making up the tumor. To further understand the evolution of gliomas, we looked at the temporal heterogeneity of adult diffuse glioma by collecting sequencing and clinical data on 210 matched tumors across multiple time points. The samples were divided based on the IDH1 mutation and 1p/19q codeletion status into 3 subtypes. We found that patients whose tumors followed a neutral evolutionary trajectory have better outcomes compared to cases with evidence of clonal selection. Evolutionary profile and prior therapy did not show association, suggesting that standard treatment does not drive the selective process. We observed a hypermutator phenotype in samples spread out between all glioma subtypes. This phenotype was associated with prior temozolomide treatment in all but one case. Patients found to have a hypermutator glioma did not show a change in overall survival compared to their non-hypermutator counterparts. Clinical factors that we found to be predictive of survival include age and number of temozolomide cycles at initial diagnosis. Other clinical factors considered include radiotherapy, gender, tumor location and extent of resection.

Coauthors: Floris P. Barthel, Kevin C. Johnson, Hoon Kim, Roel Verhaak, Jackson Laboratory for Genomic Medicine, Farmington, Connecticut; and GLASS Consortium.

It was thought that the ability of cancer cells to metastasize was an exclusive trait of tumor cells in late stages of progression. However, clinical and experimental evidence revealed that dissemination can occur from early lesions (EL). In early lesions the HER2 oncogene activates a branching morphogenesis program that restricts p38 signaling, disrupts E-cadherin localization and promotes an EMT-like program that fuels early dissemination. Importantly, this program allows early disseminated tumor cells (eDTCs) to adopt a dormant phenotype.

Bulk and single cell RNA sequencing (scSeq) of EL, overt primary tumors (PT), as well as of early DTCs (eDTCs) and early and late DTCs (e/lDTCs) confirmed that EL and eDTCs display an enrichment in a mesenchymal/basal phenotype vs. low abundance of epithelial markers. Computation analysis of bulk and scSeq data revealed that eDTCs display an enrichment in pluripotency regulating genes and CDK inhibitor expression, supporting a dormant phenotype. Further, the basal and quiescence program was linked to ZFP281, a key transcriptional regulator of primed pluripotency in embryonic stem cells and a barrier of naive pluripotency. Functional analysis of ZFP281 strongly suggests that ZFP281 has an important role driving the dormant and stem-like phenotype of eDTCs.

Together our data supports the hypothesis that dormant eDTCs are not only growth arrested, but also maintain pluripotency-like plasticity and survival that enables later reactivation.

Coauthors: Ephraim Kenigsberg[2], Xin Huang[3], Jianlong Wang[3], Julio A. Aguirre-Ghiso[1].

1. Department of Hematology and Medical Oncology, Icahn School of Medicine at Mount Sinai, New York.
2. Genetics and Genomic Sciences, Icahn School of Medicine at Mount Sinai, New York.
3. Cell, Developmental & Regenerative Biology, Icahn School of Medicine at Mount Sinai, New York.

There are no therapies that specifically target metastatic disease. Many clinical studies have failed to identify genetic driver mutations specific to metastatic nodules, suggesting that targetable biological differences between primary and metastatic tumors are driven by epigenetic, metabolic and other reversible mechanisms.

Prior work from our lab has shown that metastasizing melanoma cells experience high levels of oxidative stress both in circulation and visceral organs. Successful metastasizes undergo reversible metabolic changes to enhance their survival by increasing their cellular antioxidant capacity. One of the adaptations relies on metastatic nodules increasing the levels of NADP+as well as NADPH compared to primary tumors. Since, NADPH is essential for efficient regeneration of cellular antioxidants such as glutathione,metastatic cells may be producing more NADP+ via NAD+ kinase (NADK) activity during metastasis to withstand oxidative stress.

Our preliminary data shows upregulation of NADK in metastatic nodules compared to primary tumors using a patient-derived xenograft of melanoma metastasis. Induction of oxidative stress in vitro in melanoma cell lines lead to an increase in NADK levels. Genetic depletion of NADK decreased survival of melanoma cells under oxidative stress, while overexpression made them more resistant, suggesting a protective role of NADK in metastasis. Finally, we have found that acute and chronic induction of oxidative stress causes expression of different isoforms of NADK, a switch also observed between subcutaneous tumors and metastatic nodules. We are exploring the molecular mechanisms of isoform switching as a mechanism of survival and adaptation to increased oxidative stress during the metastatic cascade.

Advances in microfluidic technologies and molecular profiling have propelled the rapid growth and interest in achieving a ‘liquid biopsy’ in cancer. As tumors grow, they will aggressively invade surrounding tissue due to rapidly dividing cancer cells. The tumor is nourished by an ample blood supply that provides an avenue for these cells to enter the peripheral blood stream. Individual circulating tumor cells (CTCs) are released at very low numbers, but are highly desirable due to their molecular cargo. Larger aggregates or clusters of CTCs are thought to break off from the most aggressive cancers and can also be found in the blood. In addition to these rare circulating tumor cells and clusters, billions of tiny (~nm) particles from the tumor will evade the blood stream, referred to as extracellular vesicles, which also contain genetic information about the tumor. Through a collaborative effort between bioengineers, biologists, and clinicians, my laboratory at Massachusetts General Hospital has developed microfluidic devices to isolate and characterize these rare circulating biomarkers from whole blood. Data from these devices will be presented with a focus on our recent effort to characterize CTCs and extracellular vesicles from the blood of patients with highly aggressive brain tumors. Patients were monitored throughout treatment, from their initial diagnosis to end of treatment. Through the microfluidic isolation of blood based biomarkers from patients, our goal is to obtain complementary data to the current standard of care to help better guide treatment and identify new biomarkers and putative therapeutic targets.

Coauthor: Brian V. Nahed, Massachusetts General Hospital.