Le cancer est une maladie du génome. La transformation tumorale résulte de l’acquisition de mutations somatiques via divers processus mutagènes opérant tout au long de la vie du patient. Les mécanismes à l’origine des mutations incluent les erreurs de réplication, les défauts de réparation de l’ADN, les modifications de base spontanées ou catalysées par des enzymes cellulaires, et l’exposition à des agents mutagènes endogènes (ROS) ou exogènes (tabac, UV…). Au cours de ma thèse, j’ai analysé des données de séquençage exome et génome complet de tumeurs hépatiques pour décortiquer les mécanismes à l’origine des mutations dans ces tumeurs, leur interaction avec les facteurs de risque, les processus cellulaires, les gènes drivers, et leur évolution au cours de la maladie. J’ai utilisé des méthodes statistiques existantes et dévoloppé des outils bioinformatiques innovants pour:- extraire les signatures de mutations et de réarrangements structuraux à l’aide de données de séquençage à haut débit- identifier les facteurs de risque et/ou les altérations génétiques à l’origine de chacune- prédire les mécanismes mutagènes à l’origine de chaque mutation somatique- explorer les corrélations entre la densité des mutations et les processus cellulaires comme la réplication et la transcription- reconstruire l’histoire clonale des tumeurs et dater l’apparition des signatures mutationnelles et des aberrations chromosomiques.Ces approches innovantes m’ont permis d’identifier 10 signatures mutationnelles: 5 signatures ubiquitaires à l’œuvre dans toutes les tumeurs hépatiques mais modulées par les facteurs de risque (sexe, alcool, tabac), et 5 signatures sporadiques opérant dans moins de 5% des tumeurs et associées à des étiologies connues (aflatoxine B1, acide aristolochique) ou restant à identifier. J’ai aussi mis en évidence 6 signatures de réarrangements structuraux, notamment des phénotypes duplicateurs et déléteurs, spécifiques de petits groupes de tumeurs. Chaque processus mutagène est modulé différemment par la réplication et la transcription. Les signatures liées à des molécules formant des adducts sur l’ADN (hydrocarbures polycycliques aromatiques, aflatoxine B1, acide aristolochique) sont nettement moins actives dans les gènes fortement exprimés suite à l’action du transcription-coupled repair, alors que la signature 16, liée à l’alcool, présente un motif unique de transcription-coupled damage. Une corrélation étonnante entre la densité des petites insertions et délétions (indels) et l’expression des gènes a été identifiée, conduisant à une accumulation considérable d’indels dans les gènes très forterment exprimés dans les cellules hépatiques. Enfin, l’histoire clonale des tumeurs hépatiques montre l’évolution des signatures mutationnelles au cours du temps et identifie l’accumulation de gains chromosomiques multiples comme un évènement tardif entraînant probablement une croissance de la tumeur jusqu’à une taille détactable en clinique. Ces résultats nous éclairent sur les mécanismes à l’origine des altérations génomiques dans l’histoire naturelle des cancers du foie. / Cancer is a disease of the genome. A normal cell goes rogue and is transformed into a cancerous cell due to acquired somatic mutations in its genome. The catalogue of these somatic mutations observed in the cancer genome is the outcome of multiple mutational processes that have been operative over the lifetime of a patient. These mutational processes that have occurred throughout the development of cancer may be infidelity of the DNA replication machinery, impaired DNA repair system, enzymatic modifications of DNA, or exposures to exogenous or endogenous mutagens. Each mutational process leaves a characteristic pattern – a “mutational signature” on the cancer genome. Various genomic features related to genome architecture, including DNA replication and transcription, modulate these mutational processes. During my PhD, I analyzed whole exome and whole genome sequencing data from liver tumors to understand the mutational processes remodeling these tumor genomes, their interaction with risk factors, cellular processes, and driver genes, and their evolution along the tumor histories. For that aim, I used existing statistical methods and I developed innovative computational tools to:- extract mutational and structural variant signatures from next-generation sequencing data- identify risk factors or genetic alterations underlying each process- predict the mutational process at the origin of each somatic mutation- explore correlations between mutation rates and cellular processes like replication and transcription- reconstruct the clonal history of a tumor and the timing of mutational processes and copy-number changes These innovative analytical strategies allowed me to identify 10 mutational signatures: 5 ubiquitous signatures operative in every liver cancer but modulated by risk factors (gender, alcohol, tobacco), and 5 sporadic signatures operative in <5% of HCC and associated with specific known (aflatoxin B1, aristolochic acid) or unknown mutational processes. I also identified 6 structural variant signatures, including striking duplicator or deletor phenotypes in rare tumors. Each mutational process showed a different relationship with replication and transcription. Signatures of bulky DNA adducts (polycyclic aromatic hydrocarbons, aflatoxin B1, aristolochic acid) strongly decreased in highly expressed genes due to transcription-coupled repair, whereas the alcohol-related signature 16 displayed a unique feature of transcription-coupled damage. A striking positive correlation between indel rate and gene expression was observed, leading to recurrent mutations in very highly expressed tissue-specific genes. Finally, reconstructing the clonal history of HCC revealed the evolution of mutational processes along tumor development and identified synchronous chromosome duplications as late events probably leading to fast tumor growth and clinical detection of the tumor. Together, these findings shed new light on the mechanisms generating DNA alterations along the natural history of liver cancers.

Background: In the management of metastatic colorectal cancer (mCRC), KRAS, NRAS and BRAF mutational status individualizes therapeutic options and identify a cohort of patients (pts) with an aggressive clinical course. We hypothesized that relapsed and refractory mCRC pts develop unique mutational signatures that may guide therapy, predict for a response and highlight key signaling pathways important for clinical decision making. Methods: Relapsed and refractory mCRC pts (N=32) were molecularly profiled utilizing commercially available next generation sequencing (NGS) platforms. Web-based bioinformatics tools (Reactome/Enrichr) were utilized to elucidate mutational profile linked pathways-networks that have the potential to guide therapy. Results: Pts had progressed on fluoropyrimidines, oxaliplatin, irinotecan, bevacizumab, cetuximab and/or panitumumab. Most common histology was adenocarcinoma (colon N=29; rectal N=3). Of the mutations TP53 was the most common, followed by APC, KRAS, PIK3CA, BRAF, SMAD4, SPTA1, FAT1, PDGFRA, ATM, ROS1, ALK, CDKN2A, FBXW7, TGFBR2, NOTCH1 and HER3. Pts had on average had >= 5 unique mutations. The most frequent activated signaling pathways were: HER2, fibroblast growth factor receptor (FGFR), p38 through BRAF-MEK cascade via RIT and RIN, ARMS-mediated activation of MAPK cascade, and VEGFR2. Conclusions: Dominant driver oncogene mutations do not always equate to oncogenic dependence, hence understanding pathogenic ` interactome(s)' in individual pts is key to both clinically relevant targets and in choosing the next best therapy. Mutational signatures derived from corresponding ` pathway-networks' represent a meaningful tool to (I) evaluate functional investigation in the laboratory; (II) predict response to drug therapy; and (III) guide rational drug combinations in relapsed and refractory mCRC pts.

This dissertation proposes tools and analysis of mutational signatures in human cancer and their application to the stratification of patients for drug response. To provide a comprehensive workflow for preprocessing, analysis, and visualization of mutational signatures, I created the Mutational Signature Comprehensive Analysis Toolkit (musicatk) package. musicatk enables users to select different schemas for counting mutation types and easily combine count tables from different schemas. Multiple distinct methods are available to deconvolute signatures and exposures or to predict exposures in individual samples given a pre-existing set of signatures. Additional exploratory features include the ability to compare signatures to the COSMIC database, embed tumors in two dimensions with UMAP, cluster tumors into subgroups based on exposure frequencies, identify differentially active exposures between tumor subgroups, and plot exposure distributions across user-defined annotations such as tumor type. I then use musicatk to analyze the largest tumor sequencing dataset from a Chinese population to date. I identified differences in the levels of signature exposures compared to similar data from a Western cohort. Specifically, COSMIC signature SBS25 was higher in the Chinese dataset for Melanoma and Renal Cell Carcinoma patients and Melanoma patients had lower levels of SBS7a/b (Ultraviolet Light). My analysis also revealed a putative novel signature enriched in pancreatic cancers. Lastly, I assess the ability of mutational signatures to identify patients who may respond to irofulven, a drug for late-stage cancer patients who have defects in the Transcription Coupled Nucleotide Excision Repair (TC-NER) pathway. As the functional understanding of which mutations successfully disrupt this pathway is incomplete, I develop an approach that classifies patients based on evidence of this pathway being disrupted based on levels of mutational signatures. I build a model that successfully predicts patients who will respond to treatment without a known relevant mutation in the TC-NER pathway. The work from this study furthers our understanding of mutational signatures in different populations and demonstrates the feasibility of using mutational signatures to identify patients eligible for drug trials.

CCCTC-binding factor (CTCF) binds DNA, thereby helping to partition the mammalian genome into discrete structural and regulatory domains. In doing so, it insulates chromatin and fine-tunes gene activation, repression, and silencing. Complete removal of CTCF from mammalian cells causes catastrophic genomic dysregulation, most likely due to widespread collapse of 3D chromatin looping within the nucleus. In contrast, Ctcf hemizygous mice with lifelong reduction in CTCF expression are viable but have an increased incidence of spontaneous multi-lineage malignancies. In addition, CTCF is mutated in many human cancers and is thus implicated as a tumour suppressor gene. This study aimed to interrogate the genome-wide consequences of a reduced genomic concentration of Ctcf and its implications for carcinogenesis. In a genetically engineered mouse model, Ctcf hemizygous cells showed modest but robust changes in almost a thousand sites of genomic CTCF occupancy; these were enriched for lower affinity binding events with weaker evolutionary conservation across the mouse lineage. Furthermore, several hundred genes concentrated in cancer-related pathways were dysregulated due to changes in transcriptional regulation. Global chromatin structure was preserved but some loop interactions were destabilised, often around differentially expressed genes and their enhancers. Importantly, these transcriptional alterations were also seen in human cancers. These findings were then examined in a hepatocyte-specific mouse model of Ctcf hemizygosity with diethylnitrosamine-induced liver tumours. Ctcf hemizygous mice had a subtle liver-specific phenotype, although the overall tumour burden in Ctcf hemizygous and wild-type mice was the same. Using whole genome sequencing, the highly reproducible mutational signature caused by DEN exposure was characterised, revealing that Braf(V637E), orthologous to BRAF(V600E) in humans, was the predominant oncogenic driver in these liver tumours. Taken together, while Ctcf loss is partially physiologically compensated, chronic CTCF depletion dysregulates gene expression by subtly altering transcriptional regulation. This study also represents the first comprehensive genome-wide and histopathological characterisation of this commonly used liver cancer model.

Cancer development is driven by somatic genome alterations, ranging from single point mutations to larger structural variants (SV) affecting kilobases to megabases of one or more chromosomes. Studies of somatic rearrangement have previously been limited by a paucity of whole genome sequencing data, and a lack of methods for comprehensive structural classification and downstream analysis. The ICGC project on the Pan-Cancer Analysis of Whole Genomes provides an unprecedented opportunity to analyse somatic SVs at base-pair resolution in more than 2500 samples from 30 common cancer types. In this thesis, I build on a recently developed SV classification pipeline to present a census of rearrangement across the pan-cancer cohort, including chromoplexy, replicative two-jumps, and templated insertions connecting as many as eight distant loci. By identifying the precise structure of individual breakpoint junctions and separating out complex clusters, the classification scheme empowers detailed exploration of all simple SV properties and signatures. After illustrating the various SV classes and their frequency across cancer types and samples, Chapter 2 focuses on structural properties including event size and breakpoint homology. Then, in Chapter 3, I consider the SV distribution across the genome, and show patterns of association with various genome properties. Upon examination of rearrangement hotspot loci, I describe tissue-specific fragile site deletion patterns, and a variety of SV profiles around known cancer genes, including recurrent templated insertion cycles affecting TERT and RB1. Turning to co-occurring alteration patterns, Chapter 4 introduces the Hierarchical Dirichlet Process as a non-parametric Bayesian model of mutational signatures. After developing methods for consensus signature extraction, I detour to the domain of single nucleotide variants to test the HDP method on real and simulated data, and to illustrate its utility for simultaneous signature discovery and matching. Finally, I return to the PCAWG SV dataset, and extract SV signatures delineated by structural class, size, and replication timing. In Chapter 5, I move on to the complex SV clusters (largely set aside throughout Chapters 2—4) , and develop an improved breakpoint clustering method to subdivide the complex rearrangement landscape. I propose a raft of summary metrics for groups of five or more breakpoint junctions, and explore their utility for preliminary classification of chromothripsis and other complex phenomena. This comprehensive study of somatic genome rearrangement provides detailed insight into SV patterns and properties across event classes, genome regions, samples, and cancer types. To extrapolate from the progress made in this thesis, Chapter 6 suggests future strategies for addressing unanswered questions about complex SV mechanisms, annotation of functional consequences, and selection analysis to discover novel drivers of the cancer phenotype.

Contexte et objectif : Plusieurs facteurs de risque de cancer ont été identifiés et il a été estimé que plus de 40% des cas dans les pays développés pourraient être évités en modifiant les facteurs de risque connus. L'objectif général de cette thèse était de démontrer que l’intégration de données génomiques et épigénomiques aux données détaillées sur les expositions environnementales et le mode de vie peut être utile pour identifier des biomarqueurs de ces facteurs et contribuer à augmenter notre connaissance de l'étiologie du cancer. Résultats : Dans un premier temps, nous décrivons comment les signatures génomiques et épigénomiques peuvent être utilisées pour identifier des marqueurs d’exposition et déchiffrer l’étiologie du cancer. Ensuite, nous contribuons au débat relatif à l’hypothèse de la chance dans le développement du cancer et démontrons que les mutations induites par le tabagisme sont plus prédictives du risque de cancer que les mutations aléatoires. Nous introduisons un modèle probabiliste pour la simulation de données mutationnelles et comparons la performance des outils d’identification de ces signatures avec des données réelles et simulées. De plus, nous introduisons une nouvelle méthode pour l’identification des signatures mutationnelles. Enfin, nous utilisons les données de méthylation de la cohorte E3N pour étudier le lien entre l'exposition aux retardateurs de flamme bromés et aux composés perfluorés, deux substances classées parmi les perturbateurs endocriniens, et la méthylation de l’ADN sanguin. Globalement, notre étude ne fournit aucune preuve d'altérations globales du méthylome ou d'altérations à l’échelle des CpGs. Cependant, certains résultats suggèrent l’existence d'altérations de la méthylation de gènes impliqués dans des voies biologiques (ex., la réponse aux androgènes) et nécessitent des recherches supplémentaires.Conclusion : Ce travail contribue à la recherche méthodologique portant sur les signatures mutationnelles en introduisant un protocole de mesure de performance et d’identification des signatures mutationnelles pouvant servir de référence à de futures études méthodologiques ou appliquées. Nos recherches sur les signatures mutationnelles et le méthylome démontrent l'utilité de tels outils pour évaluer les expositions et élucider leur rôle dans l'étiologie du cancer. / Context and aim: Several risks factors have been identified for cancer, and it has been estimated that more than 40% of cases in developed countries are preventable through the modulation of known modifiable risk factors. The overall objective of this thesis was to demonstrate that the analysis of genomic and epigenomic data integrated with well-characterised exposure and lifestyle data may be used to identify markers of environmental exposures and lifestyle and may contribute to increase our understanding of cancer aetiology.Results: We first describe how genomic and epigenomic signatures can be used to identify markers of exposure and decipher the aetiology of cancer. Then, we adopt the mutational signatures framework to contribute to the debate about the “bad luck” hypothesis for cancer and demonstrate that tobacco-related mutations are more strongly correlated with cancer risk than random mutations. We introduce a probabilistic model for the simulation of mutational signature data and compare the performance of the available methods for the identification of mutational signatures using both simulated and real data. Additionally, we introduce a new method for the identification of such signatures. Finally, we use methylation array data in an epidemiological study within the E3N cohort to investigate the association between exposure to Brominated Flame Retardants and Per- and polyfluoroalkyl substances, two organic pollutants that are known endocrine disrupting chemicals, and methylation in DNA from blood. Overall, our study does not provide evidence of methylation alterations at the level of the whole genome, in regions or in single CpGs. Suggestive evidence of alterations in the methylation of genes within plausible biological pathways (e.g. androgen response) warrants further investigations. Conclusion: Our work on the methodological aspects of mutational signature research introduces an original framework for measuring the performance of tools for the identification of mutational signatures that may serve as reference for future methodological or applied research. Our applications of both mutational signature and methylome research demonstrate the usefulness of such tools to assess exposures and elucidate their role in cancer aetiology.

Les cellules cancéreuses sont caractérisées par des altérations de l'ADN causées par des facteurs exogènes, comme l'exposition à des agents environnementaux tels que le tabac ou les UV, ou par des mécanismes endogènes tels que les erreurs de polymérase lors de la réplication de l'ADN. L'analyse des causes et des conséquences de ces altérations permet de mieux comprendre les facteurs et mécanismes à l'origine du développement d'un cancer. Les technologies de séquençages à haut débit offrent l'opportunité d'étudier la nature précise de ces altérations à l'échelle du génome et permettent de révéler des signatures mutationnelles distinctes et spécifiques de cancérigènes, fournissant ainsi des hypothèses sur l'étiologie des cancers.L'objectif de ma thèse a consisté à développer des méthodes et des outils bioinformatiques accessibles et conviviaux permettant de faciliter l'analyse et l'interprétation des signatures mutationnelles à partir de données de séquençage à haut débit. L'application de ces outils et méthodes à des séries originales de tumeurs humaines et de systèmes expérimentaux de mutagénèse et carcinogénèse a permis de mieux caractériser la signature mutationnelle de l'acide aristolochique (AA) ainsi que d'autres cancérigènes d'intérêt / Cellular genomes accumulate alterations following exposures to exogenous factors, like environmental agents such as tobacco smoking or UV, or to endogenous mechanisms such as DNA replication errors. Analysing the causes and consequences of these changes allows a better understanding of the mechanisms underlying cancer development and progression. Next-generation sequencing (NGS) technologies provide the opportunity tostudy the nature of the resulting alterations on a genome-wide scale and started to reveal distinct mutational signatures specific to past carcinogenic exposures providing clues on cancer aetiology.The aim of my thesis was to develop user-friendly bioinformatic tools and methods for facilitating the analysis and interpretation of carcinogen-specific mutational signatures from NGS data. Applying these tools and methods to human tumours and experimental models of mutagenesis led to a better characterisation the mutational signature of aristolochic acid (AA), as well as other carcinogens of interest