Identifying and manipulating molecular mechanisms controlling cancer stem cell metastatic potential in a human oral cancer model.

Tumour metastasis, which seeds secondary tumours in distant organs, causes the majority of cancer deaths. Cancer stem cells drive tumour metastasis. To do this, they undergo epithelial-mesenchymal transition (EMT) to adopt a migratory mesenchymal phenotype that can disseminate from the primary tumour and migrate to secondary sites. Upon reaching secondary sites, they then undergo mesenchymal-epithelial transition (MET) to re-gain a proliferative epithelial phenotype that can form a secondary tumour. This process is known as the metastatic cascade. For this to occur, epithelial cancer stem cells within the primary tumour must possess the ability to undergo EMT into a mesenchymal phenotype that in turn retains the ability to undergo MET at a secondary site - this ability to switch phenotype is termed 'plasticity'. The ability to therapeutically control the plasticity of epithelial cancer stem cells to prevent initiation of the metastatic cascade would stop tumour metastasis at its root. We hypothesise that there is variation within the epithelial cancer stem cell population in the primary tumour - only some of these cancer stem cells possess the plasticity required to initiate the metastatic cascade. We further hypothesise that it is possible to therapeutically manipulate the plasticity of epithelial cancer stem cells, and thereby prevent metastatic transitions. We propose to test these hypotheses in a human oral cancer model treated with the well-characterised EMT-inducer TGFbeta.

New single cell approaches are a powerful tool for dissecting variation within cell populations, and provide a means to probe the existence of discrete epithelial cancer stem cell sub-populations with differing responses to TGFbeta. We will use single cell RNAseq to identify distinct epithelial cancer stem cell sub-populations and infer the ability of these sub-populations to undergo EMT into a mesenchymal phenotype that in turn retains the ability to undergo MET. We will use this information to identify candidate molecular pathways controlling these cancer stem cell sub-populations, and inactivate these pathways using a CRISPR screening methodology combined with single cell RNAseq in order to determine molecular targets whose inactivation can prevent initiation of the metastatic cascade. In selecting targets, we will focus on druggable nodes within key pathways. We will then test the importance of these targets in metastasis using human pathological specimens, and whether inactivating these targets can prevent metastasis in new engineered metastasis models that we have developed in our lab.

Oral cancer is one of the top ten cancers worldwide, with over 300,000 cases annually, and incidence is increasing both worldwide and in the UK (in the UK, incidence has increased by 23% over the past decade). Oral cancer is a deadly disease with frequent metastatic spread, which is the single most important predictor of poor outcome. This research project will generate important knowledge of the molecular pathways controlling metastasis in oral cancer and, given the central role of EMT in metastasis, this may be generalizable to other tumour types. Targets emerging from this study will be taken forward for the development of new targeted therapies to prevent metastasis.

We hypothesise that there is variation within the epithelial cancer stem cell population in a primary tumour - only some of these cancer stem cells possess the plasticity required to initiate the metastatic cascade - that is, to undergo EMT into a mesenchymal phenotype that in turn retains the ability to undergo MET at a secondary site. We propose to test this hypothesis in a human HPV-negative oral cancer model treated with the well-characterised EMT-inducing growth factor TGFbeta.

We will use single cell CITE-seq (single cell RNAseq with barcoded antibodies to cell surface markers used in cancer stem cell phenotyping) combined with advanced downstream bioinformatic techniques to identify cell clusters and infer the transitions between these clusters. In this way we will identify epithelial cancer stem cell sub-populations that are able to undergo EMT into a mesenchymal phenotype that in turn retains the ability to undergo MET. Bulk RNAseq on clonal sub-lines with differing plasticity and FACS sorted sub-populations using known markers will support this analysis. We will use this information to identify candidate molecular pathways controlling these cancer stem cell sub-populations. We will then inactivate these pathways using CROP-seq, a CRISPR loss-of-function screening methodology that combines with single cell RNAseq to enable determination of molecular targets whose inactivation can induce loss of key cancer stem cell sub-populations and thus prevent initiation of the metastatic cascade. In selecting candidate targets for the screen, we will focus on druggable nodes within key pathways. We will then test whether target inactivation can prevent metastasis in new engineered metastasis models that we have developed in our lab. Finally, we will test the association of targets with metastasis through immunofluorescent antibody staining of human pathological specimens stratified on metastatic status and bioinformatic analysis of TCGA data.