The generation and maintenance of hybrid epithelial-mesenchymal (E-M) cell phenotypes.

Pancreatic cancer is a fatal disease because of the rapid spread of cancer cells throughout the body leading to metastasis and also because of resistance to therapies which leads to relapse even after apparent successful treatment. The cancer cells spreading around the body are called circulating tumour cells (CTCs). And they act like cancer stem cells (CSCs) because they can produce all cancer cells necessary to form tumours in other organs or in the pancreas in relapse cases. CTCs and CSCs share many attributes of cells in the embryo which indicate that they are regulated by the same proteins. Also, like certain cells that migrate in the embryo in clusters, they have hybrid cell shapes with mixed traits from two major cell types, epithelial and mesenchymal cells, hence they are called cells with hybrid epithelial-mesenchymal (E-M) phenotypes.

That is why analysing these CTCs and CSCs can give us a better picture on the survival of patients, and they hold the potential for developing novel therapies. But at present we are at the earliest stages of exploiting any information we gain from CTCs and CSCs to translate in the clinic.

Recently, CTCs from pancreatic cancer patients were found to express a protein called the Wilms' tumour protein (WT1) which is quite interesting. For a start, our earlier work suggest that it regulates cells with hybrid E-M phenotypes in the embryo. And although it is known to regulate only few cell types in the adult, they all exhibit hybrid E-M phenotypes. And finally our analysis using mouse models for pancreatic cancer supported the notion that WT1 can affect the amount of epithelial and mesenchymal cells in the tumours, probably by controlling how many cells with hybrid E-M phenotypes are made. So in this project, we will test whether WT1 is important for hybrid E-M phenotypes and consequently the generation of CSCs and CTCs in pancreatic cancer.

We will model pancreatic cancer using a human pancreatic cell model that was shown to be faithful in recapitulating normal pancreatic cancer progression. It is also useful because we can use it to study first the impact of forcibly expressing or losing WT1 in cells and how this affect the production of hybrid E-M phenotypes using: cell culture, three dimensional "mini pancreas in a dish" or by transplantation in mice. Second, we will find out how WT1 makes and keeps the cells in the hybrid E-M phenotypes. To achieve this, we will identify the DNAs and RNAs that WT1 binds and regulates using protocols that we previously used in our lab followed by next generation sequencing (NGS). We will also identify targets of a chromatin modifier that we showed to be important for WT1 function in the embryo. The modifier and WT1 affect how genes are expressed not only by binding DNA or RNA but also by affecting the epigenetic environment in the chromosome and in the nucleus. Changing the epigenetic context of genes is important for cells to change fate and become stem-like-cells and WT1 may affect these too.

At the end we will have a comprehensive idea of which genes and epigenetic marks are regulated by WT1, and especially around those genes that affect the hybrid E-M phenotypes and give rise to CSCs or CTCs. Working with our collaborators, we will prioritise the most likely genes and associated epigenetic mechanisms that lead and maintain the hybrid E-M phenotypes downstream of WT1. At the end of the project, we will test whether this is the case in our human pancreatic cancer model. The longer-term objective being to fully appreciate the translational potential of the findings and the data generated.

In this project we aim to provide what we learn from WT1 regulation of the hybrid E-M phenotype as a paradigm to understanding how other embryonic regulators mis-expressed in cancer tissues reawaken dormant embryonic processes to promote metastasis, drug resistance and relapse with fatal consequences.

Cancer cells undergo a partial EMT to generate multipotent progenitors with hybrid E-M phenotypes e.g. CTCs and CSCs, which are critical for metastasis, therapy resistance and relapse. The molecular regulators that retain cells in hybrid E-M states remain uncharacterised; but their transient expression should be necessary and sufficient to maintain cells in hybrid E-M states long enough to generate progenitor cells.

We hypothesise WT1 to be such a molecular regulator.

This is because WT1 regulates cells undergoing EMT/MET in the embryo and maintains few adult cell-types in hybrid E-M phenotypes. And it is strongly expressed in many solid cancers including CTCs from PDAC patients. WT1 is also ranked the top oncofoetal antigen with clinical trials in many cancers showing promising results, including in PDAC. Also, we recently found that WT1 is a) sufficient to bestow potency and derive cells with hybrid E-M states , b) necessary for the E-M balance in primary PDAC in vivo, and c) a molecular switch governing EMT/MET reversibility.

Therefore, we aim to define:

1. The role of WT1 in generating and maintaining the stemness window in normal and neoplastic pancreatic cells.
2. The molecular mechanisms that underlie the role of WT1 in the hybrid E-M phenotypes.

Thus, we are repurposing a human pancreatic cell model known to recapitulate the multistage progression of PDAC in vitro, via organoid culture ex vivo and upon xenograft implantation in vivo. In our model, the expression of two of the commonest drivers of PDAC evolution, K-RasG12D and p53R175H, is inducible. We use xenografts as they explore late tumour stages that remain largely elusive in patients such as minimal residual disease following therapy and homing metastatic cells.

This work will provide direct insights into the molecular mechanisms underlying the functional roles of WT1 and similarly mis-expressed embryonic factors deployed at the nexus of metastasis, therapy resistance and relapse.

To test our hypothesis that WT1 drives and maintains hybrid E-M phenotypes and the stemness window in normal pancreatic cells and PDAC, we will use disease modelling to study WT1's functional roles (Aim 1) and big data-driven delineation of the molecular mechanisms (Aim 2). Thus, we seek as indicated in theme 1 to deliver "better models of disease for pre-clinical research and to establish experimental medicine platforms so that new therapies can be progressed to patients". And as expressed in theme 2 "The biomedical and health research sectors now operate in the 'big data' era where new opportunities exist to extract insights from large and complex datasets. Taking these opportunities will help us ... determine important molecular pathways for the targeting of therapeutics".

Consequently, we strongly believe that the results arising from our research will have a significant impact in many areas including cancer, cell and molecular biology and in particular:

Disease modelling: we will provide more evidence to the feasibility of using human primary cell lines as models for cancer and a way to reduce mice numbers and cost. Moreover, the project will impact synthetic biology-driven development of multifunctional vectors in gain and loss of function (GoF and LoF) experiments in vitro, ex vivo and in vivo to test hypotheses and validate findings from big data (e.g. our pGoldilox Vector).

The hybrid E-M phenotype: this project will provide direct evidence that cells in the hybrid E-M phenotype produce CSCs and CTCs that may act as tumour initiating cells (TICs). It will also provide a comprehensive data-driven picture of the generation and maintenance of this hybrid phenotype in normal and malignant cells. This will impact our understanding of the EMT/MET processes in cancer, wound healing, tissue injury and normal homeostasis.
We will use our analysis of the transcriptome and epigenetic changes after genetic perturbations as large and complex data to extract insights into understanding the stemness window and the E-M phenotypes during and beyond this project. We will delineate the causal regulatory network(s) of the hybrid E-M phenotype. This will be followed by validation of key molecular pathways to be exploited with clinical impact in mind: targeting the pathological WT1 isoforms using therapeutic bispecific T-cell engager antibody targeting specific intracellular WT1 variants.
The project will provide large time-series data sets mapping the DNA and RNA targets of WT1 as it drives the formation of progenitor cells. These will be used to analyse how a single protein controls gene expression from the nucleus to the cytoplasm with direct impact on e.g. cancer subtypes where WT1 is mostly cytoplasmic.

Computational predictor tools: our project will generate high quality datasets that we will use as training sets for machine learning approaches. From basic biology point-of-view, we aim to develop computational tools to define protein function. The images generated of the PDAC stages stained with different markers (Aim 1) will provide a major resource to study the phenotypic heterogeneity in time. In the future, this dataset will impact our development of algorithms that model or predict the evolution of phenotypic heterogeneity (e.g. with CMM colleague Rafael Carazo Salas). Also, development of clinical predictor algorithms (e.g. via deep learning) based on the expression of markers will be shaped by computational analysis of the large and complex data sets (Aim 2, with Ian Overton).

From disease prevention to treatment: Also, understanding how WT1 generates hybrid states impacts therapeutic strategies aimed at blocking the hybrid phenotype formation at different stages of PDAC progression. All these will impact prevention, diagnosis, prognosis, treatment and response to therapy in many diseases.

These insights will be a major step towards implementing the next-generation of precision medicine tools.