Mutational processes contributing to the development of cancer emerge from various risk factors of the disease and impose specific imprints of somatic alterations in the genomes of cancer patients. These mutational footprints, called “signatures”, can be read from the tumour sequencing data and reveal the main sources of DNA damage driving neoplastic progression. In this sense, they can be considered a form of evidence for historical mutational events that have acted during tumour evolution. I will discuss some of the insights we have obtained into the development and progression of oesophageal adenocarcinoma, an aggressive disease with limited treatment options, by tracking mutational signatures in human cancer tissues as well as 3D cell models of this malignancy. I will also mention some of the methods we have employed to link mutational damage and immune responses in the context of this disease.

Using a stratification strategy based on mutational signatures extracted from the whole-genome sequencing data of 129 tumours, we have previously uncovered three subtypes of oesophageal cancer with distinct aetiologies related to DNA damage repair deficiencies, ageing and oxidative stress, and with different therapeutic options. Further, we have shown that organoids grown in vitro from patients’ tumours effectively recapitulate the genomic and transcriptomic profiles of the tumours of origin, and thus may constitute a suitable (if limited) model for this cancer type. By tracking the evolution of mutational processes during organoid culture growth we were also able to demonstrate a dynamic clonal architecture that mimics well the extensive intratumour heterogeneity observed in this cancer. Tracing mutational signature trajectories from early to later stages of cancer development in both primary tumours and organoid systems unveils a refined picture of evolution in this cancer, with frequent bottlenecks (~60% of cases) where mutational pressures shift. Furthermore, by employing bulk and single cell RNA-seq data we show that the microenvironmental composition of these tumours is indirectly modulated by mutational processes acting on the cancer cells. In particular, oxidative stress appears to be correlated with a fast progressing tumour phenotype and changing dynamics in stromal populations. Finally, we suggest that the observed genomic and immune signatures and their specific temporal dynamics could be further exploited for patient stratification in the clinic.

Using H3K27ac ChIP-seq data coupled with bulk RNA-seq, we have recently discovered two anticorrelated transcriptional programs in neuroblastoma cells: a noradrenergic program governed by transcription factors PHOX2B, GATA3 and HAND2, and a mesenchymal program governed by PRRX1 and AP-1 transcription factors [1].
Here, in we applied 10X single cell RNA-seq to check the presence of the two identities in neuroblastoma cell lines and patient derived xenografts (PDXs). In two neuroblastoma cell lines, we observed the presence of both identities (mesenchymal or noradrenergic) and documented the sporadically occurring transition state. Interestingly, within a “mixed” cell line, cells of mesenchymal identity seemed to be more resistant to doxorubicin and cisplatin than noradrenergic identity ones. We and others have also shown that that in cell lines stably expressing the noradrenergic program, a reprogramming towards the other identity can be achieved though overexpression or silencing of just one of the master transcription factors [2]. Using scRNA-seq data and ChIP-seq data for transitional state cells, we could discover reprogramming marker genes. Further experimental work will show whether blocking the activity of these genes can prevent transition between the two states in neuroblastoma from happening, and therefore contribute to better clinical outcomes.
This work is done in collaboration with the team of I. Janoueix-Lerosey (Curie Institute, Paris) and supported by the INCa grant (2018-1-PL BIO-09-INSERM ADR 5-1).

Recent work has shed light on the extent and relevance of intratumour heterogeneity (ITH) in cancer genome evolution. While this ITH can have an impact on patient response to cancer therapeutics such as targeted therapies, the effect of anti-tumour immunity on cancer evolution and heterogeneity has not been extensively considered.

Here, I will explore the how ITH in genomic and transcriptomic space can be used to shed light on tumour evolution, and the complex interplay between the cancer cell and the immune microenvironment, with a focus on lung cancer.

I will outline how intra-tumour heterogeneity and a tumour’s life history can be deciphered from multi-region sequencing. I will also explore how neoantigens can be identified within homogeneous and heterogeneous tumours, and the extent to which the immune microenvironment is heterogeneous within a tumour. I will explore different mechanisms of immune escape and how these can be elucidated.

Finally, the clinical relevance of both neoantigens and immune escape mechanisms will be considered.

Cancer is characterized by intratumor heterogeneity, where tumor cells contain different collections of somatic mutations, including single-nucleotide variants (SNVs) and copy-number aberrations (CNAs). Single-cell DNA sequencing has emerged as a promising technology to measure this heterogeneity at the level of individual cells and reconstruct the evolutionary process of cancer. Current studies infer single-cell phylogenies using either SNVs or CNAs alone. However, SNVs often overlap with CNAs, and a CNA may delete an SNV later during tumor evolution. Not accounting for such mutation losses may lead to the incorrect inference of single-cell phylogenies for tumors. We describe a loss-supported model for single-cell phylogenies that uses constraints from observed CNAs to inform phylogenetic inference. We use this model to develop a new algorithm, SCARLET (Single-Cell Algorithm for Reconstructing Loss-supported Evolution of Tumors). On simulated data, SCARLET outperforms existing methods in reconstructing the structure of single-cell phylogenies and inferring the presence of mutations. We then use SCARLET to analyze a single-cell dataset from a colorectal cancer patient and show that the loss-supported model yields more plausible phylogenies than existing methods.

Cancer evolves through the emergence and selection of molecular alterations. Cancer genome profiling has revealed that specific events are more or less likely to be co-selected, suggesting that the selection of one event depends on the others. Leveraging large-scale tumor sample cohorts, we have designed algorithmic approaches (SELECT) to infer evolutionary dependencies from non-random patterns of alteration occurrence 1 . By analyzing ~10,000 genomes from multiple tumor types, we constructed a map of oncogenic dependencies associated with cellular pathways, transcriptional readouts, and therapeutic response 1,2 . Furthermore, we propose simple heuristics to map dependencies detected in large pan-cancer cohorts to relatively small patient subsets, where the statistical power is reduced. Using this approach, we could identify evolutionary templates for single tumor types, characterized by alternative and mutually exclusive trajectories of co-occurrent alterations. Finally, by integrating data from high-throughput CRISPR/Cas9 and shRNA screening in cancer cell lines, we demonstrate the functional relevance of individual dependencies within these templates. These results highlight functional redundancies and synergies which alter cancer cell sensitivity to both gene knock-out and drug perturbation. Overall, both statistical and experimental evidence support a role for evolutionary dependencies as key determinants of cancer progression and therapeutic response

1. Mina, M. et al. Conditional Selection of Genomic Alterations Dictates Cancer Evolution and Oncogenic
Dependencies. Cancer Cell 32, 155–168 (2017).

2. Sanchez-Vega, F. et al. Oncogenic Signaling Pathways in The Cancer Genome Atlas. Cell 173, 321–337 (2018).

Cancer heterogeneity is one of the biggest challenges of precision medicine. Within a single patient, different lesions may display different genetic, metabolic and radiological profiles (inter-tumour heterogeneity). In some cases, multi-region sampling has demonstrated that a single tumour may contain heterogeneous, spatially-distinct `habitats' (intra-tumour heterogeneity). A combined computational approach that integrates all the relevant data sources is needed to fully understand how these different physical scales at which tumour heterogeneity manifests itself are related. Crucially, these analyses must contain medical imaging data, as it is the only resource that provides non-invasive and three-dimensional quantitative information about the disease.

In this talk we will discuss how tumours can be divided into distinct `habitats' based on CT and MRI data, and present results that show that quantitative metrics capturing the differences between habitats have prognostic value for diseases such as high-grade serous ovarian cancer. We will also discuss how these habitats can be integrated with the underlying molecular data using imaging-guided biopsies aided by patient-specific 3D-printed moulds. Finally, we will discuss the question of longitudinal multi-scale tracking of tumour heterogeneity.

Heterogeneity is a fundamental feature of complex phenotypes, such as cancer. So far, genomic screenings have profiled thousands of samples providing insights into the transcriptome of the cell. However, disentangling the heterogeneity of these transcriptomic Big Data to identify defective biological processes remains challenging. In my talk I will show you how we addressed this challenge applied to RNA-seq data and introduced a novel concept of discretization of gene expression levels, which we derived from machine learning theory and shaped upon knowledge of RNA biology. I will discuss how we employed our method to detect cellular programs that are altered upon the somatic loss of PTEN. Overall, I will present you how “going discrete”, or using machine learning approaches derived from Big Data analysis combined with RNA biology, can help in disentangling the noise of heterogeneity from biological processes potentially relevant for complex phenotypes.