Theses - Cancer Research UK Cambridge Institute
Permanent URI for this collection
Browse
Recent Submissions
Item Embargo Tumour ecology of breast cancer patient-derived xenograftsLerda, GiuliaTumours are complex ecosystems where several neoplastic and non-transformed cell populations (the tumour microenvironment – TME) interact in heterogeneous and dynamic ways. Descriptive studies in patient material, although extremely informative, often lack the temporal and spatial resolution needed to fully recapitulate this variability. Mouse models represent a natural complement to these correlative analyses, allowing longitudinal and spatially-extensive studies as well as perturbation experiments. Among pre-clinical models, patient-derived tumour xenografts (PDTXs) have been largely excluded from ecological descriptions due to the absence of a human TME and to the immune-compromised status of the mouse host. Evidence however suggests that PDTXs might be good candidates for ecosystem analyses because they contain composite tumour and TME compartments where cells may form functionally relevant interactions. To demonstrate this, we used the extensive collection of breast cancer PDTXs established in our laboratory and performed the first in-depth description of ecosystem complexity in these model systems, confirming their potential as study tools for tumour ecology. After a global survey of all 100 available models, we thoroughly analysed a set of 15 xenografts using single cell RNA- and DNA-sequencing (scRNA/DNA-seq), three-dimensional imaging with serial two-photon tomography (STPT) and two-dimensional transcriptomic and proteomic profiling (the latter using imaging mass cytometry - IMC). In this thesis, we will present part of this work, focusing in particular on the phenotypic and spatial complexity revealed by scRNA-seq, STPT, IMC and *in situ* transcriptomics. These analyses detail for the first time the features of the PDTX TME, which contains multiple cell types that significantly overlap with phenotypes described in other model systems and clinical samples. These cells have varying frequencies in different PDTXs and include multiple innate immune populations with tumour-reactive phenotypes. Expanding previous descriptions, we also show pervasive inter- and intra-model transcriptional heterogeneity in the PDTX tumour compartment. By applying ligand-receptor analysis, we study how different populations interact with each other and generate complex ecological dynamics. Using STPT, IMC and *in situ* transcriptomics we show that the geographical distribution of different neoplastic and TME cells is largely model-specific, indicating a concerted organisation of PDTX spatial ecosystems by the malignant cell-autonomous compartment. In agreement with this, we predict multiple exclusive interactions between cancer and stroma and show that regional variations in tumour phenotypes result in heterogeneous TME features within specific samples. Overall, this thesis unveils the phenotypic and spatial heterogeneity of PDTX ecosystems, showing their variable organisation between and within models. By providing evidence of potential interactions between different cell populations, our study suggests the functional relevance of this variation. In general, these findings support a wider application of PDTXs for the study of tumour ecology and have strong implications for the interpretation of xenograft studies and for the characterisation of tumour-host interactions *in vivo*. The future integration of other data modalities (scDNA-seq) and extension to patient data will likely confirm these observations and offer complementary information on the processes controlling phenotypic and spatial cellular variation in cancer.Item Open Access Genetic Subclones Identify Phenotypic Variation in Diffuse Large B-Cell LymphomaWang, BoyaDiffuse large B cell lymphoma (DLBCL) is the most common form of non-Hodgkin’s lymphoma. Many efforts have characterised DLBCL subtypes that account for some of the heterogeneity in clinical outcome following standard therapy. However, not much is known regarding intratumoural heterogeneity. Case studies show that subclones can emerge after targeted therapy and have different drug responses. Thus, our question is: what are the genotypes and phenotypes of DLBCL subclones? We hypothesise that DLBCL tumours contain genetic subclones, with phenotypic variation that reflects B cell development. Here, I present the largest single-cell DLBCL dataset to date, consisting of 91 cases with paired single-cell RNA (scRNA) and single-cell ATAC (scATAC) sequencing. This work identified genetic subclones based on scRNA copy number. Interestingly, 85% of tumours have genetic subclones (mean 2, range 1–6). We identified seven clusters of gene expression signatures, or themes, that characterise intratumoural subclone diversity. Four themes (memory B cell, plasma cell, germinal centre identity, and B cell identity) recapitulate B cells blocked in differentiation and account for most of the phenotypic heterogeneity between subclones. Three additional themes (MYC, cell cycle, metabolic) characterise transitory cell states. I used an independent discovery dataset (n = 562 samples) to associate genetic alterations with theme expression. A key observation was that amplification of REL (which encodes c-REL, a NF-κB transcription factor subunit), is negatively associated with expression of the memory B cell theme in the discovery dataset (t-test, p = 1×10−11) and this was validated in scRNA of genetic subclones (t-test, p = 0.03). Amplification of REL is a recurrent alteration in germinal centre B cell subtype (GCB) DLBCL, and its functional consequences have not yet been characterised. NF-κB transcription factors (c-REL, RELA) have unique functions in regulating normal B cell development. I hypothesise that REL amplification is an oncogenic event that restricts B cells to the germinal centre and prevents memory cell differentiation. I have designed cell-line and mouse models that will be used to test this hypothesis. In summary, my work has identified DLBCL genetic subclones with highly variable phenotypes that may have distinct therapeutic vulnerabilities. It offers a multi-platform single cell resource for the DLBCL community.Item Open Access Tracing clonal cell plasticity in the microenvironment of pancreatic ductal adenocarcinomaKoplev, Simon; Koplev, Simon [0000-0002-8586-5614]The pancreas may regenerate without stem cells by induction of cellular plasticity between differentiated cells. Although essential for the remarkable capacity for pancreas regeneration, this property entails a vulnerability to cancer through acinar-to-ductal metaplasia and dedifferentiation by epithelial-mesenchymal transition, which contributes in largely unknown ways to metastasis in pancreatic cancer. As with most solid tumours, pancreatic ductal adenocarcinoma develops at the interface of cellular plasticity, genomic instability, and tumour microenvironments. However, current methodologies of analysing tumour tissue are unable to measure these modalities jointly. Here, by developing a method for epitope barcoding measured by cyclical imaging of antibody binding with confocal microscopes, this thesis explores the utility of multicolour cell labelling to investigate how polyclonal cancer evolution relates to epithelial-mesenchymal transition. The assembled barcodes consist of combinations of epitopes attached to a fluorophore and a nuclear-localisation signal, which enables cell sorting, quantitative image analysis, and effective decoding by statistical analysis. Using subcutaneous mouse models of pancreatic cancer, this thesis demonstrates that isogenic epitope labelling can detect epithelial clonal expansion events co-localised with fibroblasts in the tumour microenvironment. The epitope barcode constructs were stably expressed in mouse tumours for at least 2-3 weeks, proving feasibility of multicolour lineage tracing with cyclical imaging. By amending the barcodes with 8bp tags sequenced with single-cell RNA-seq, transcriptomic scores for epithelial-mesenchymal transition showed evidence for selectional acquisition of epithelial lineages during subcutaneous colonisation. Such labelled isogenic cell populations were amenable to Luria–Delbrück fluctuation analysis detecting lineages with epigenetic inheritance of cell states. Altogether, epitope barcoding and cyclical imaging form a compelling experimental strategy for investigating the clonality of transdifferentiation in the microenvironment of pancreatic cancer.Item Embargo Targeting the adenosine pathway reshapes a myeloid-dependent tumour microenvironment in pancreatic ductal adenocarcinoma. An opportunity for therapeutic interventionGraziano, VincenzoBackground: The prognosis for patients with pancreatic ductal adenocarcinoma (PDAC) remains extremely poor with less than 10% of patients surviving more than 5 years from the diagnosis. PDAC is considered relatively resistant to conventional therapies and recently lack of efficacy of immunotherapy has been shown for this aggressive malignancy. Recently a novel immunosuppressive pathway, called adenosine pathway, has been suggested to contribute to the ability of PDAC to evade the immune system and its resistance to immunotherapies (IOT) and cytotoxic treatments, by generating extracellular adenosine (eAdo) as a consequence of the conversion of a molecule of ATP by enzymes called CD39 and CD73. The stimulation of the adenosine receptors by extracellular adenosine on the immune cells has been shown to favour a pro-tumorigenic and tolerogenic microenvironment. Results: Using syngeneic, *in vivo* murine models of PDAC with differential immune infiltration and response to IOT and the KPC mouse model, CD39 and CD73 were shown to be highly expressed on tumour-infiltrating immune cells. The distribution of extracellular adenosine appeared heterogeneous in murine PDAC, with high concentrations in hypoxic areas, enriched for myeloid infiltration. M2 macrophages among other myeloid populations, highly expressed the Adora2a receptor, particularly in the IOT-resistant model. Targeting the adenosine pathway (Adoi) delayed tumour growth and reduced the occurrence of lung metastases in IOT-resistant tumours. Further, Adoi improved the efficacy of combinations of cytotoxic agents or immunotherapy. Genes related to immune modulation, hypoxia response and tumour stroma were downregulated following Adoi. Further tumour-infiltrating M2 macrophages were found reduced following Adoi. Finally, an adenosine-related gene dataset generated by RNAseq was shown to have similarities with the squamous subtype of PDAC identified by Bailey in humans (Bailey P. et al. 2016) and a specific adenosine signature was generated associating the adenosine pathway to a poorer prognosis in human PDAC. Conclusions: The formation of eAdo appears to favour the development of the immunosuppressive and pro-tumorigenic TME in PDAC, contributing to its resistance to conventional and novel therapies. The presence of the adenosine pathway in human PDAC seems to associate with biological features of aggressiveness and be related to a poorer prognosis. Therefore, inhibition of the adenosine pathway may represent a strategy to modulate the TME and improve response to therapy in patients with PDAC.Item Open Access Development of nanobodies as tools to investigate G-quadruplex DNA secondary structuresGalli, SilviaGuanine-rich sequences in DNA can fold into four-stranded alternative secondary structures called G-quadruplexes (G4). G4s are suggested to be involved in key genome functions such as transcription, genome stability and epigenetic regulation, together with numerous connections to cancer biology. The detection G4s in cells has been facilitated by the use of structure-specific small molecules or antibody probes and much work has relied on G4 recognition by a single-chain variable fragment antibody, BG4. Other G4 probes have been reported but generally their usage has been limited to certain applications and in many cases the probes have limited selectivity for G4s. The aim of this thesis was to develop an alternative probe with high affinity and good selectivity for G4s for use in applications for which BG4 in not suitable and to provide an important orthogonal validation of results previously obtained with BG4. A phage-display screen was used to select nanobodies, antibody fragments derived from of heavy-chain only antibodies of Camelid family, that have high affinity and selectivity for the G4 structure from the human *MYC* oncogene promoter (MycG4). Biophysical characterisation confirmed that one nanobody (B11) binds with high affinity to a wide range of DNA G4 conformations *in vitro*. B11 successfully confirmed the folding of G4s in chromatin isolated from cancer cells. Furthermore, mutational analysis coupled with molecular dynamics simulation identified specific arginines contributing to the interaction of B11 with MycG4. Lastly, I demonstrated that B11 can be expressed in human cells for use in experiments to map G4 genomic locations *in situ*. The results from these experiments are significant to the field as they have demonstrated the validation of earlier data describing the location of folded G4 structures in the human genome. I have also extended previous studies by gaining molecular insights in the mode of action of an antibody-G4 interaction. Finally, I have provided initial proof-of-principle that a G4 antibody can be deployed in a cellular context to probe G4s in living cells. In sum, I have successfully generated a novel tool for the future study of G4s in key cellular processes, such as transcription, and in diseases such as cancer.Item Embargo Investigating Tumour-Immune Cell Interactions in Paediatric MalignanciesMasih, Katherine; Masih, Katherine [0000-0002-0809-668X]There have been significant successes in the treatment of childhood cancer, with the overall survival rate reaching nearly 80%. However, certain leukaemias and solid tumours, as well as metastatic and relapsed cancers, remain resistant to current therapeutic approaches. These patients endure aggressive therapies and severe treatment-related toxicities, often with little chance of a cure. Therefore, novel, less toxic therapeutics with increased efficacies are greatly needed. Immunotherapies have emerged as potential solutions to these challenges. However, not all patients respond to these treatments, and reliable biomarkers that could guide the accurate use of therapy are lacking. In particular, we lack a complete understanding of interactions between the patient’s immune system and tumour. We hypothesised that biological properties of tumours that do not respond to immunotherapy can be detected prior to treatment and serve as biomarkers for patient selection. Furthermore, we hypothesise that tumour cell phenotype, including the diagnosis and genomic background, and prior exposure to therapy can shape the tumour immune microenvironment (TME) and that deep profiling of the patient TME could unmask clinically relevant biomarkers of response and inform future trials. This thesis aimed to both understand non-response to immunotherapies and to deeply profile the TME using established and novel multi-omic analysis platforms on patient samples. To test these hypotheses, we employed three cancer models from paediatric patients. To identify tumour-intrinsic mechanisms of resistance to cellular immunotherapy, we studied B-cell acute lymphoblastic leukaemia treated with CD19 CAR T-cells. To investigate the TME we utilised bulk RNA-seq from patient medulloblastomas and performed highly multiplexed immune focused imaging of paediatric neuroblastoma. We discovered a distinctive methylation pattern and associated stem-cell epigenome as a novel candidate pre-treatment biomarker of non-response to CD19 CAR T-cell therapy in paediatric acute lymphoblastic leukaemia. We next utilised well established deconvolution methods to investigate the TME in medulloblastoma and discovered five transcriptionally distinct clusters, including one that was composed of WNT, SHH, Group 3, and Group 4 tumours, which were associated with differential TME landscapes. We next developed novel immune focused imaging panels to investigate the spatial TME in neuroblastoma. We found that the frequency and spatial distribution of immune cells is influenced both by tumour genotype and exposure to chemotherapy. Finally, we found that high pre-treatment expression of PTEN, TIM-3, CD127, and markers of NK-cells and M1 macrophages and low expression of VISTA and CD40L were associated with improved overall survival. Collectively, this work provides novel insights into tumour-immune interactions across paediatric malignancies. Furthermore, we identify promising biomarkers to be further validated for risk stratification, for predicting response to immunotherapy, and to be utilised as novel therapeutic targets in the TME, which could improve outcomes for children with cancer.Item Embargo Targeting the DNA damage response in hormone-driven cancersKolyvas, EmilyGenomic instability is a hallmark of cancers including breast cancer (BC) and prostate cancer (PCa). Defects in the DNA damage response (DDR) as well as replication stress are responsible, in part, for driving genomic instability and accumulation of DNA damage. While the resulting alterations can be beneficial for tumour development and progression, it also presents a vulnerability for pharmacologic inhibition. This has been demonstrated previously by PARP inhibitor sensitivity in BRCA-deficient cancers. This has led to the discovery and development of several DDR pathway inhibitors. Understanding the biology and genomic context that could lead to tumour responses to these inhibitors will be instrumental for identifying the patient populations that will benefit most. In this thesis, the DDR is explored as a therapeutic target in hormone-driven cancers. Hormone signalling has been shown to drive DNA repair, and is therefore a worthwhile model to study mechanisms of sensitisation to DNA damaging therapeutics. The androgen receptor (AR) has been linked to radio-resistance in PCa and BC by regulating the expression of DNA repair genes. The interaction between AR and DNA repair cofactors, such as DNA-PKcs, is under active exploration as a potential target for sensitising AR-driven cancers to radiotherapy (RT). Chapter 3 aimed to investigate the intersection between AR and DDR signalling pathways and explore the potential of targeting AR and DNA-PKcs for radio-sensitisation. Results indicate that single-agent treatment with inhibition of AR or DNA-PKcs led to sensitisation, and combination treatment had an additive effect. Proteomics data to determine AR cofactors suggested an interaction between DNA repair proteins including DNA-PKcs and AR. Gene expression analyses following combination treatments revealed various signalling pathways implicated in sensitising these models to radiation. Further analysis will be instrumental in validating these pathways and associated TFs with a mechanism for radio-sensitisation in BC and PCa. Furthermore, in Chapter 4, PTEN loss was assessed in BC and PCa as a sensitiser to DNA damage and DDR inhibitors. Nuclear PTEN has been implicated in maintenance of genome stability, and this was tested in PTEN knock-out cell line models. The results show that PTEN loss results in increased sensitivity to ionising radiation along with impaired recovery from exogenously induced replication stress. When screened against a panel of DDR inhibitors, PTEN loss proved to be a predictor of response to inhibitors that sense and respond to replication stress. Interestingly, when assessing DDR inhibitor response in BC PDXs, nuclear PTEN protein expression, but not mRNA expression, is a predictor of response. This suggests that nuclear PTEN could be a useful biomarker of response to these inhibitors. Additionally, RNA sequencing of a PTEN knock-out model suggested increased immune activation. This could suggest a role for combination of DDR inhibitors with immune checkpoint blockade in this model. To test these hypotheses, a CRISPR knock-out of PTEN in a BC patient derived xenograft (PDX) was developed and characterised. Given the known advantages of PDXs for preclinical drug development, generation of a CRISPR knock-out in a PDX is a valuable tool for studying specific effects of single gene loss. Results showed that a stable knock-out was developed that was maintained with passaging and with metastasis, providing a useful model for downstream analysis of DDR effects of PTEN loss.Item Embargo Mechanisms of piRNA-guided co-transcriptional gene silencingEastwood, Evelyn LouiseEukaryotic genomes are populated by repetitive sequences and transposable elements (TEs), mobile DNA sequences that are able to change their position and replicate within a host genome. Heterochromatin establishment at these sequences prevents ectopic recombination and TE mobilisation. In *Drosophila* ovaries, this depends on Piwi, which is guided by a bound small RNA to nascent TE transcripts where it regulates the formation of transcriptionally repressive chromatin states. As well as numerous chromatin modifying proteins, this mechanism depends on the Panx-induced co-transcriptional silencing (PICTS) complex, which contains Panx, Nxf2 and Nxt1. PICTS operates at the interface between piRNA-guided Piwi and heterochromatin effectors. While the N-terminal region of Panx is known to deliver the transcriptional silencing signal, the molecular mechanism by which PICTS induces heterochromatin formation at TE insertions is not fully resolved. In the first part of this thesis, I identified a fourth component of PICTS, Cut-up (Ctp), which I showed is essential for co-transcriptional transposon repression. While initially identified as a component of cytoplasmic dynein, Ctp has been shown to function as a dimerisation hub protein, mediating the assembly and stabilisation of many dynein-independent protein complexes. I demonstrated that the primary function of Ctp within PICTS is to support the assembly of a higher-order complex through dimerisation of the Panx C-terminus. Next, I sought to further explore the molecular function of PICTS using biochemical and structural approaches. I established the expression and purification of full-length and mutant PICTS complexes and showed that PICTS exists as a constitutive homodimer that is able to associate with single-stranded RNA. While structural studies are still ongoing, together this work has contributed to a detailed understanding of the molecular assembly of PICTS. Understanding transcriptional silencing and chromatin regulation more broadly relies on sophisticated tools to profile the genome-wide distribution of histone modifications. In the final part of this work, I established a modified CUT&RUN (Cleavage Under Targets & Release Using Nuclease) protocol that relies on chromatin reader domains fused directly to Micrococcal Nuclease (MNase) and allows rapid and specific antibody-free profiling of histone marks. Together, this work has contributed to a broader understanding of how small RNAs regulate chromatin structure.Item Open Access Prevalence and fate of pro-oncogenic clones in the human colonSkoufou Papoutsaki, Maria NefeliNormal aged tissues are thought to exist as a patchwork of mutations. The detection of cancer-driver mutations in the normal human colonic epithelium had been so far limited. Using immunohistochemistry and sequencing methods, and by screening a large number of patients and crypts, pro-oncogenic clones were detected for eight colorectal cancer driver genes; ARID1A, APC, CTNNB1, FBXW7, KRAS, PTEN, TP53 and SMAD4. Losses or gain of function in these genes were associated with biases in the processes of clone fixation and expansion, which determined their event burden in the tissue. However, by comparing the event burden in the normal tissue and the frequency of the same mutations in colorectal cancer, it was observed that a high mutational burden does not dictate the Availability of clones for Neoplastic Conversion (ANC). For example, TP53 clones had the highest event burden in the normal tissue but KRAS clones had a higher ANC. Some of the factors that could be contributing to the differential fate of pro-oncogenic clones were investigated. An increase in cell-proliferation seemed necessary to induce biases in clonal dynamics but not sufficient to explain the different fates of clones in terms of their ANC. In addition, larger KRAS patches seemed to possess different characteristics from smaller ones, often exhibiting abnormal crypt morphology and containing a reduced frequency of CD4+ immune cells. Finally, a human colonic KO organoid model was used to study the clonal behaviours in vitro and gain insight into the potential mechanisms and pathways that could be mediating mutation-associated advantages. The normal human colon appears to be more mosaic in terms of cancer-driver mutations than originally thought. This means that the status of certain cancer-associated mutations can be reclassified based on their degree of representation in the normal tissue and mutations with a higher probability of initiating cancer can be identified. Thousands of invisible pro-oncogenic clones are inferred to exist in the aged colon but not all of them may have an equal chance for neoplastic transformation. Differential fates were found to exist between clones with mutations in different genes, different amino acid changes in the same gene or even the same amino acid change. This information could in time be used for early-detection efforts by stratifying individuals based on the risk of the mutations that they might possess.Item Open Access Development and application of an imaging and spatially-driven toolkit to elucidate vasculogenic mimicry in breast cancer mouse modelsQosaj, FatimeVasculogenic mimicry (VM) is one form of tumour vascularisation that describes tumour cells that have acquired endothelial-like features, endowing them with the ability to form vessel-like structures. VM networks have been postulated to be functional, enabling them to play a critical role tumour survival and metastatic disease. These de novo vessels are independent of angiogenesis processes, in which host endothelial cells form vessels from pre-existing vasculature, and instead have been shown to underpin therapeutic resistance to anti-angiogenic therapies (AATs). Although molecular data largely supported by in vitro work has immensely contributed to our understanding of VM, the spatial features that define the VM phenotype remain largely unknown and understudied. How these vessels present in space and in 3D, has, until recently, been unknown. This underpins the scarcity of in vivo evidence and subsequent imaging and spatial data which would help further illuminate VM. Therefore, the overarching goal of this dissertation was to curate a novel toolkit that enables a revolutionary approach to better capturing and understanding VM, in vivo. The toolkit has largely entailed the optimisation of a vascular perfusion assay and the integration of two state-of-the-art technologies: a 3D two-photon imaging modality and a single cell, multiplexed immune-labelling proteomics platform. The former has greatly improved our ability to confidently capture genuine VM networks in their natural environment whilst the latter has enabled a novel approach to better resolving the spatial features of these networks and VM-tagged tumours more broadly. Upon successful development, optimisation and validation of the VM toolkit, the final phase of this project was to apply it first to mouse models of VM, followed by human cancer cell line-derived mouse models, all in the triple negative breast cancer setting. The culmination of this PhD has yielded three impactful achievements. First is an optimised and validated novel toolkit that enables VM networks to be confidently and reliably captured and better understood, spatially. 3D evidence for VM can now be directly interrogated with intricate spatial technologies for further molecular and spatial characterisation. Second is the application of this toolkit to mouse models of VM, illuminating a complex vasculature across 3D models and the prominent role of anti-angiogenic pathways in these VM-tagged tumours. Third is the application of this tool kit to VM-competent human cancer cell line-derived mouse models of VM, enabling additional in vivo models to be established. In these models, genuine VM networks were captured, encapsulating some of the most convincing in vivo and 3D evidence for VM across all models supporting this PhD and arguably across much of the current VM literature. This PhD has enabled the elusive VM phenotype to be more robustly captured and comprehensively resolved spatially, using a bespoke toolkit in addition to identifying and exploring additional in vivo models of VM. These are pivotal accomplishments that will directly impact the field and enable the biological importance and relevance of this mechanism to be further supported. The implications that the toolkit developed and the insights gathered in support of this project can be clearly defined and are highlighted throughout this dissertation.Item Embargo Investigating the role of innate lymphoid cells in targeted radiotherapy in lung cancerPng Ren Jie, ShaunRadiotherapy is a mainstay of lung cancer treatment, and its role in regulating immune responses to promote anti-cancer response is increasingly appreciated. However, it is still not fully understood how radiotherapy modulates the immune microenvironment in lung cancer. In the lung, ILC2s are critical innate lymphocytes that modulate Type 2 immunopathology, but their role in modulating the immune response to lung cancer is not fully understood. Furthermore, it is unclear how ILC2s respond to targeted lung radiotherapy. In this thesis, I characterised the response of ILC2s to lung radiotherapy and elucidate their response alongside the general immune response in targeted radiotherapy to lung cancer. Finally, I aimed to establish a model to study the role of ILC2s in modulating the adaptive immune response in the primary lung cancer context. I established a model of targeted lung irradiation using a preclinical treatment platform known as the Small Animal Radiation Research Platform (SARRP) and monitored immune responses. In the naïve setting, targeted lung irradiation was observed to deplete ILC2s in the long term, but this was not observed in the majority of immune cells studied. Irradiated ILC2s expressed the activation marker OX40L, though this was not accompanied by increased production of Type 2 cytokines. In the tumour setting, ILC2s also express OX40L but do not appear to contribute to anti-tumour immunity. Probing the overall immune response to targeted lung radiotherapy in the tumour context demonstrated the importance of both the innate and adaptive immune response in radiotherapy-induced anti-tumour immunity. Dendritic cells (DCs) and adaptive CD4+ and CD8+ T-cells were activated and are crucial for anti-tumour immune responses. Innate NK cells are also important as their depletion reduced the effectiveness of radiotherapy. To study the role of ILC2s in the regulation of adaptive immune responses in cancer, I validated an inducible antigen-expressing cassette and confirmed its functionality both in vitro and in vivo. Expression of the antigen by tumour cells elicited adaptive CD4+ and CD8+ T-cell responses. I have also derived primary lung cancer cell lines to be used in combination with the cassette and demonstrate their capacity to engraft in the lung upon orthotopic implantation. Further work is needed to refine the orthotopic model of primary lung cancer, and to incorporate the cassette into the cell line. This could then be used alongside ILC2-deficient mouse models to study the role of ILC2s in modulating the adaptive immune response in lung cancer.Item Embargo Deep learning on whole-slide images for early detection and risk prediction of oesophageal cancerBerman, Adam; Berman, Adam [0000-0003-1438-4802]This dissertation introduces novel computational techniques to identify patients at particularly high risk for progressing from Barrett’s oesophagus (BE) to oesophageal adenocarcinoma (EAC) earlier and more accurately using data from a minimally-invasive cell collection device. It also introduces a new software library for efficiently handling computational pathology tasks. Oesophageal adenocarcinoma is usually diagnosed late, leading to a five-year mortality rate of only 13%. The identification of its precursor, Barrett’s oesophagus (BE), is thus a crucial early detection goal. Identifying the cancer at an early stage drastically increases patient five-year survival to 80%. A new minimally-invasive screening device for BE detection called the Cytosponge presents a solution. Despite this advance, only 0.3% of BE cases progress to cancer per patient-year, leading to a large number of costly and invasive follow-up procedures. I have therefore developed machine learning systems to predict which precursors become deadly by identifying two features prognostic of progression to EAC: atypia, a kind of cellular irregularity, and P53 aberrance. These models automate patient stratification and drastically reduce the time it takes to screen for these progression markers. I have also identified a clinically relevant correlation between the automatically detected quantity of BE in a pathology slide and the length of the BE segment identified from endoscopy. Beyond oesophageal cancer, the inspection of stained tissue slides by pathologists is essential for the early detection, diagnosis, and monitoring of disease. However, WSIs present a number of unique challenges for analysis, requiring special consideration of image annotations, slide and image artefacts, and evaluation of model performance. I have therefore developed SliDL, a Python library for performing pre- and post-processing of WSIs for deep learning. SliDL allows users to perform essential processing tasks in a few simple lines of code, bridging the gap between standard image analysis and WSI analysis. By providing a framework in which deep learning methods for WSI analysis can be developed and applied, SliDL increases the accessibility of an important application of deep learning. Digital pathology is rapidly growing as a topic salient to both computer science and medicine. My work aims to contribute to both fields, including a software library to democratise access while applying it to a pressing issue in cancer early detection.Item Embargo The role of innate immunity in pancreatic cancer progression and treatmentLee, ShengImmune cells are highly abundant in the tumour microenvironment of pancreatic ductal adenocarcinoma (PDAC) and they significantly influence the entire process of PDAC tumourigenesis. While the anti-tumour response is typically mediated by adaptive immunity, cells of the innate immune system can significantly influence this anti-tumour response and in some cases dictate response to treatment. This thesis presents work on two projects – the first project is focussed on characterizing the immunomodulatory effects of gemcitabine in combination with an Ataxia Telangiectasia and Rad3-related (ATR) inhibitor (Gem/ATRi) in pancreatic cancer, whereas the second project describes the pro-tumourigenic role of type 2 innate lymphoid cells (ILC2s) in PDAC. In chapter 3, Gem/ATRi showed preclinical efficacy in the ‘T cell high’ 2838c3 tumour model and induced dendritic cell (DC) activation in both the tumour and draining lymph node (LN). This was associated with a substantial depletion of all intratumoural DC subsets in the tumour and a selective depletion of LN-resident DCs in the dLN. Gem/ATRi similarly caused a depletion of intratumoural CD8+ T cells, but of the remaining population there was a decrease in the proportion of exhausted cells along with an increased proportion of proliferating cells. Experiments using the ‘T cell low’ 6419c5 tumour model revealed that cDC1 in these tumours were dysfunctional and unresponsive to stimulation compared to those in 2838c3 tumours. In chapter 4, I demonstrated the pro-tumourigenic role of ILC2s in PDAC and investigated mechanisms that potentially underlie this observation. ILC2 deletion significantly extended the survival of 2838c3-bearing mice and altered the intratumoural immune infiltrate. Attempts to pinpoint the mechanism(s) underlying the pro-tumourigenic role of ILC2s (i.e. IL-33, IL-13, eosinophils, NK cells and the ILC2-OX40L-Treg axis) did not yield any positive results, although it is clear that they modulate tumour growth via an effect on adaptive immunity. Further experiments involving CD8+ T cell depletion in ILC2-deficient mice revealed that ILC2s influence tumour growth via both CD8+ T cell-dependent and independent mechanisms. Finally, the impact of ILC2 deletion on PDAC tumour growth was found to be dependent on tumour cell-intrinsic factors, possibly on those that dictate the strength of the baseline anti-tumour CD8+ T cell response.Item Embargo Validation of photoacoustic imaging biomarkers in cancer biologyHacker, LinaPerfusion-limited hypoxia, referring to spatial and temporal fluctuations in oxygen levels, has emerged as a key driver of malignant disease progression in cancer, promoting higher metastatic potential, increased therapy resistance and poorer patient outcome1,2. However, the origin and impact of these hypoxic fluctuations on tumour progression have not yet been fully understood. Harnessing the photoacoustic effect, photoacoustic imaging (PAI) holds significant potential to elucidate these dynamics, but to fulfil this potential, a need for a thorough technical and biological validation arises. This thesis presents technical validation of PAI systems, which builds confidence in the subsequent biological validation performed using relevant biomarkers in the studies of perfusion-limited hypoxia. First, the current state-of-the art phantom materials in biophotonics are surveyed and general design considerations for preparation of tissue-mimicking phantoms are discussed to guide the development of a stable phantom material in PAI. Acoustic and optical material characterisation systems are then established and validated to enable thorough characterisation of phantom materials. Building up on this groundwork, a phantom material is developed for use in PAI, which exhibits stable acoustic, optical and mechanical properties with tuneable tissue-mimicking characteristics. Using custom phantom setups, a thorough technical characterisation study of a commercial mesoscopic PAI system is then conducted, outlining strengths and limitations of the system in characterising vessel-related biomarkers in tissues. Following these technical validation studies, the thesis embarks on applying photoacoustic mesoscopy and macroscopy in studies of perfusion-limited hypoxia in two distinct murine xenograft models, and validating this work using histopathological and transcriptomic analyses. These studies indicate that tumour vasculature undergoes rapid fluctuations in perfusion that are impacted by the underlying maturity of the vascular network, leading to variations in tumour oxygenation. Building up on an extensive validation framework, this thesis highlights the promise of PAI to advance our understanding on perfusion- and oxygenation dynamics in tumour tissues, thereby assisting the development of targeted treatment regimes in future.Item Embargo Oncogenic KRAS drives heterogeneity in metabolic and signalling pathways through genetic and non-genetic mechanismsHowitt, AnniePancreatic adenocarcinoma (PDAC) is an aggressive disease that is typically diagnosed at a late stage. PDAC development occurs over approximately 20 years and the initiating event is often an activating mutation in key cell signalling hub KRAS. Despite extensive research into PDAC and KRAS, the impact of both genetic and non-genetic variability during the earliest steps of carcinogenesis are scantly described. We hypothesise that both forms of heterogeneity might alter the efficiency of PDAC initiation. On the one hand, oncogenic KRAS alleles might rewire cellular signalling and metabolic networks causing phenotypes and adaptation in an allelespecific manner. On the other hand, cell-to-cell variability within mutant clones might permit a population of cells to escape tumour suppressive mechanisms such as senescence and adapt to oncogenic signalling, therefore supporting early cancer development. We have combined cell genetics, single-cell biochemistry and a multiomics approach to investigate both genetic and non-genetic heterogeneity during initiation of PDAC. We generated an isogenic panel of cell lines harbouring the three most common mutations of KRAS in PDAC (G12D, G12R, G12V). This system consisted of inducible expression of KRAS mutants in immortalized but non-transformed pancreatic cells to mimic PDAC initiation. In Chapter 2, we explore the short-term cellular response to KRAS induction and adaptation of cells to oncogenic KRAS. Using western blotting, metabolomics and RNAseq we show how specific KRAS mutant alleles influence the cell signalling, nucleotide metabolism and cell morphology. We show that the G12D and G12V mutants, frequently observed in KRAS-driven cancers, differ significantly from the PDAC-specific G12R mutant. The mutants display allele-specific growth factor signalling, purine metabolism rewiring and an epithelial to mesenchymal transition. In Chapter 3, we looked in greater temporal resolution and at a single-cell level at the immediate signalling and metabolic response to KRAS mutations, focusing mainly on G12D. We used a FRET sensor reporting mitochondrial ATP levels in live cells to study G12D cells at multiple time points. We found an early drop in mitochondrial ATP which was only identified when considering intracellular heterogeneity. This drop was followed by a peak in mitochondrial ATP at a later time point, caused by a combination of purine synthesis and mitochondrial ATP production. We also observed an increase in activated and total levels of mitochondrial dynamics protein DRP1 and the presence of an anti-oxidant signature. The involvement of DRP1 suggests an increase in mitochondrial homeostasis early during adaptation to oncogenic KRAS. Using techniques such as metabolomics, phosphoflow and scRNAseq we were able to further dissect this adaptation process. We also found heterogeneity in cells exploring broader levels of phosphorylated ERK upon G12D induction, which further varied over the studied time course. The recorded variability of pERK levels supports the hypothesis of an oncogenic signalling sweetspot. Taken together, this study sheds light on the process of adaptation to oncogenic KRAS. Adaptation to mutant KRAS occurs over multiple stages even within this phase of early adaptation. These stages include variation in pERK signalling, nucleotide metabolism and likely mitochondrial homeostasis. Our results suggest that both genetic and non-genetic heterogeneity play important roles in these processes and thus, the earliest events of PDAC initiation. Understanding signalling and metabolic rewiring in a KRAS mutant specific manner may help us untangle allele-specific dependencies that could inform patient stratification and disease management. Furthermore, identifying processes by which normal cells can adapt to oncogenic KRAS at a single-cell level reveals the necessary steps for disease initiation, and may provide options for preventative intervention.Item Embargo Development of imaging techniques to assess tumour cell deathHesse, Friederike; Hesse, Friederike [0000-0001-5427-7564]Cell death is an important imaging target for assessing early tumour treatment response and the effectiveness of therapy. Traditionally, anatomical imaging modalities measure changes in tumour size (Response Evaluation Criteria in Solid Tumours (RECIST)). More recently these assessments have also included functional and molecular information, as these parameters can show earlier responses to chemotherapies, e.g. using PET measurements of 2-[¹⁸F]-fluoro-2-deoxy D-glucose uptake (FDG-PET). PET Response Criteria in Solid Tumours (PERCIST) was introduced as a potentially more sensitive method of assessing treatment response, particularly with therapies that stabilise disease. While metabolic imaging may indicate drug target engagement, and in some cases tumour cell death, there is a need for imaging methods that detect tumour cell death more directly post-treatment and that can give an indication of longer term treatment outcomes. In this project ²H magnetic resonance spectroscopic imaging with ²H-fumarate and PET-CT with ¹⁸F-C2Am are investigated as methods for the early detection of tumour cell death post-treatment. Fumarate, an intermediate in the tricarboxylic acid cycle, is hydrated in the reaction catalysed by the enzyme fumarase to produce malate. With loss of plasma membrane integrity in dying cells fumarate can rapidly gain access to the enzyme. By measuring increased malate production in treated tumours after fumarate injection, cell necrosis can be assessed. It could be established that ²H MRS and MRSI measurements at 7 T of [2,3-²H₂]fumarate conversion to ²H-labelled malate can be used to detect early cell death in various tumour models. ¹⁸F-C2Am is an ¹⁸F-labelled derivative of a phosphatidylserine-binding protein, the C2A domain of synaptotagmin-I (C2Am), which can be used to image tumour cell death in vivo using PET. C2Am binds in a Ca²⁺ -dependent manner with nanomolar affinity to phosphatidylserine (PS) on the outer leaflet of the plasma membrane of apoptotic cells or to PS on the inner leaflet following disruption of plasma membrane integrity in necrotic cells. The agent was tested in mice bearing human xenografts treated with a TRAIL-R2 agonist, MEDI3039, or with conventional chemotherapy. Both imaging markers may provide more specific and sensitive methods for detecting cell death, which can be used to rapidly assess treatment response and effectiveness at an early stage.Item Open Access Dissecting cell-autonomous and microenvironmental factors governing lymphatic metastasis in breast cancerJauset Gonzalez, CristinaMetastasis is the primary cause of deaths in cancer patients. Tumour dissemination can occur through the bloodstream or lymphatic system. The presence of tumour cells in the draining lymph node (lymph node involvement) represents the most important prognostic factor for breast cancer patients. Despite its clinical implications, the mechanisms governing lymphatic metastasis remain poorly understood. This thesis aims to identify novel determinants of lymphatic dissemination in breast cancer. It focuses on the influences of the tumour immune microenvironment and the cancer cell-autonomous metastatic drivers associated with lymph node involvement. Using transplantable syngeneic mammary carcinoma 4T1 model, a differential preference for using the lymphatic system as a dissemination route was observed in immunocompetent mice when compared to immunodeficient mice. Characterisation of the differences in the tumour microenvironment revealed a higher infiltration of neutrophils in the tumours with lymph node metastasis. Depletion of neutrophils reduced lymphatic dissemination in 4T1 cells. In vitro, tumour-associated neutrophils were seen to promote tumour migration. Finally, single-cell RNA-seq analysis of lymph nodes suggests that neutrophils could have an immunosuppressive role in lymph nodes. Tumour cell-intrinsic gene signatures driving lymph node organotropism were investigated in subclonal lines derived from the 4T1 model with enhanced ability to seed lymph node metastasis. Transcriptomic profiling of these lines highlighted the role of metabolic adaptation and immunosuppression as central determinants of lymphatic metastasis. Finally, other syngeneic breast cancer models were used to assess the validity of the above-mentioned findings across breast cancer subtypes. Molecular characterisation of these indicated that immunosuppression, lipid metabolism and cell plasticity are common denominators associated with lymphatic metastasis among breast cancer models. This thesis advances our understanding of the involvement of neutrophils and tumour cell-intrinsic mechanisms that contribute to lymphatic metastasis in breast cancer. This work suggests potential gene and pathway targets that could be exploited therapeutically for the management of lymph node involvement.Item Open Access Compositional models for mutational signature analysisMorrill Gavarro, LenaBackground: Cancer is a process whereby the accumulation of mutations leads to a clonal expansion of cells, forming a tumour mass with the capacity of expanding to surrounding tissues. Understanding how these mutations come to be in the first place is important from an epidemiological and evolutionary point of view. Several external – or exogenous – factors, such as UV light, tobacco smoke, or ionising radiation, have the capacity to create mutations. Moreover, the process of DNA replication which is necessary to create two cells from one is, like any other copying mechanism, not entirely faithful. Therefore, spontaneous – endogenous – mutations are created each time the genetic material of a cell is replicated. Some mechanisms of mutation lead to more mutations than others, and can create them either with varying intensity or at the same rate throughout life, or even throughout tumour development. Mutational signatures were introduced as a proxy to quantify the number of mutations created by each mutational process. This thesis focuses on statistical methods for the analysis of such mutational signatures. Crucially, mutational signatures, in the type of questions that I address, have a characteristic: they are compositional data, because we are interested in studying their relative contribution to the total mutation load. Because of this, they have to be analysed in a multivariate way and in relative terms. Approach: I introduce appropriate compositional models to analyse two types of mutational signature: single-nucleotide polymorphism signatures, and copy number signatures. The Dirichlet-multinomial mixed effects model for single-nucleotide data is shown to have good sensitivity compared to existing fixed-effects alternatives, and higher specificity than compositional models that do not allow any overdispersion. Similarly, the incorporation of random intercepts in the logistic-normal model for copy number data increases sensitivity with respect to the fixed-effects version. The models are publicly available on github and are readily applicable to other types of compositional data. Results: Firstly, I use the mixed-effects Dirichlet-multinomial model to characterise the differential abundance patterns between clonal and subclonal mutations across 23 cancer types of the PCAWG cohort. There is ubiquitous change, which can be detected already at nucleotide level. There is higher dispersion – higher variability between samples – of signatures in the subclonal group, indicating, possibly, the presence of di↵erent clones with distinct active mutational processes. The signatures of clearest differential abundance are signatures of low abundance, many of them with the tendency to be, to some extent, the result of bleeding from other signatures, and of unknown aetiology. Although we should be wary of these signatures, differential abundance persists despite excluding them, and the relative changes between clonal and subclonal mutations, in the form of ranked coeffi cients of the signatures of highest confidence, are robust to the subset of signatures used. Secondly, I explore the use of similar models for the study of copy number signatures, and to answer three questions about the mutation dynamics in high grade serous ovarian cancer: whether the relative contribution of mutational processes changes from early-stage to late-stage samples, whether it changes from diagnosis to relapse, and whether it changes from whole-genome-duplicated (WGD) to non-whole-genome-duplicated samples. The CN signature landscape differs significantly between early and late stage samples in that there are much higher rates of WGD in late samples. However, there is no noticeable coordinated difference between matched archival and relapsed samples, although a few patients experience WGD between archival and relapse. Overall, the results indicate that there are large levels of heterogeneity in copy number signatures between patients, but less so within patients, with the exception of punctual cases, and therefore suggest that, although the first response to therapy is dictated by the mechanisms of repair, the relapse occurs not at the level of the genome at 0.1x resolution. Moreover, by use of a Support Vector Machine, I show that copy number signatures can be used to categorise WGD from non-WGD samples at 95% accuracy, using two independently labeled cohorts of TCGA and ICGC samples. Outlook: This thesis introduces the use of compositional models to study the dynamics of mutational signatures in the comparison of two groups of samples, and they can be readily applied to several other regression settings in this discipline or others in which compositional data arises. From both the copy number and point mutation standpoint the models indicate that signatures are dynamic. Further work is needed to better elucidate which mutational signatures are behind the changes, and which mutational processes are behind the signatures. Besides the biological insight into DNA mutation and repair, these results have potential clinical relevance, as cancer treatment often targets or takes advantage of impaired mechanisms of repair.Item Open Access DNA G-quadruplex structures in human cancer cellsHui, Wai In; Hui, Wai In [0000-0001-8041-0218]DNA structures alternative to the double helix have emerged as key features of the genome for the understanding of genetics and diseases. In particular, G-quadruplexes (G4s), four-stranded structures formed in guanine-rich sequences of cellular chromatin, are implicated in transcription, replication and genome stability. G4s may also present new opportunities for targeting in anti-cancer therapeutic interventions with small molecules. In this thesis, I expand the G4-profiling toolkit to better understand the biological roles of G4s, and potentially offer practical insights for G4-targeting drug development. First, quantitative G4-chromatin immunoprecipitation with sequencing (qG4-ChIP-seq), a method for mapping and comparing G4 landscapes between samples, was used to study G4s in cell lines and patient-derived tumour xenografts from different breast cancer subtypes. Differentially enriched G4s in each cancer model were associated with copy number aberrations and single-nucleotide variants, as well as common breast cancer driver regions, suggesting a link between cancer genome instability and G4 structure formation. Subsequently, to increase the versatility of G4 profiling, I developed G4-Cleavage Under Targets and Tagmentation (G4-CUT&Tag), a more efficient method to profile G4s with higher signal-to-noise ratio and 100-fold lower cellular input than G4-ChIP-seq. Further pushing the detection limit, I optimised G4-CUT&Tag for the first mapping of G4s at single-cell resolution. I demonstrated that individual cell identity can be discerned within a mixed cellular population based solely on single-cell G4 profiles. This result demonstrates that G4 signatures in individual cells relate to the fundamental identity of a cell. Next, I developed single-nuclei G4&RNA-seq, a multiomic method to simultaneously profile G4s and poly(A)-tailed RNA within the same single nucleus. Preliminary data provides proof-of-principle to directly associate G4 formation at individual loci with their transcriptional output within individual cells. Using this approach, I then showed its potential applications in discerning G4 landscapes in different cellular states with reference to cell cycle transcriptomic data within a mixed cell population. My work now enables future genomic investigations on cell-to-cell variation of a DNA secondary structure relative to other chromatin features that were previously not possible. Overall, this thesis demonstrates advancements in G4-profiling methodologies and enables a high resolution and multi-dimensional exploration of the incidence of G4s and their functions.Item Open Access Elucidating age and pregnancy related alterations in the murine mammary glandKovačević, TatjanaBreast cancer is a heterogeneous disease that is the prevailing cause of cancer-related deaths in women. However, many studies have shown that a full-term pregnancy early in life can decrease the chance of developing cancer by up to 50% during menopause [1]. This is one of the highest protective effects known for any cancer type, and understand- ing the fundamental changes that happen in the mammary gland during pregnancy could be greatly beneficial for tackling this disease with large social implications. In this thesis, we used the mouse model system to investigate the extent and nature of the changes inputted into the mammary gland through ageing and pregnancy and characterised the larger implications of the change in the context of protection against breast cancer using single-cell sequencing as the primary tool. In this investigation we focused on the epithelial cell compartment and found that age and parity affect the luminal progenitor cells most profoundly, with both of them leading to the expansion of alveolar-like cells. Ageing leads to alveolar maturation without prior parity stimulation, and although these cells are similar to cells acquired during parity, they do show some distinguishing qualities. On the other hand, the differences introduced into the mammary gland through parity at a different age are subtle, and in most cases can only be observed when we look at gene set enrichment analysis rather than the effect on individual genes. We found a specific subset of alveolar cells that are retained after all parity-related signals have ceased. Although we did not find conclusive evidence of different levels of retention of these cells post parity at different ages, gene expression in these suggests a higher activity of the P53 pathway, DNA damage response and lower activity of MYC in the young cohort. This suggests that parity at a younger age could have an effect of establishing an environment where the cells are less likely to accumulate tumorigenic insults. Through this work, we created a scRNA sequencing catalogue that can be used for investigating the effects of ageing on the mammary gland composition and gene expression as well as the effect of parity on differently aged animals, both in the context of normal development as well as in the context of disease