Thèses sur le sujet « Multi-omic »

The advent of advanced high-throughput biological technologies provides opportunities to measure the whole genome at different molecular levels in biological systems, which produces different types of omic data such as genome, epigenome, transcriptome, translatome, proteome, metabolome and interactome. Biological systems are highly dynamic and complex mechanisms which involve not only the within-level functionality but also the between-level regulation. In order to uncover the complexity of biological systems, it is desirable to integrate multi-omic data to transform the multiple level data into biological knowledge about the underlying mechanisms. Due to the heterogeneity and high-dimension of multi-omic data, it is necessary to develop effective and efficient methods for multi-omic data integration. This thesis aims to develop efficient approaches for multi-omic data integration using machine learning methods and network theory. We assume that a biological system can be represented by a network with nodes denoting molecules and edges indicating functional links between molecules, in which multi-omic data can be integrated as attributes of nodes and edges. We propose four network-based approaches for multi-omic data integration using machine learning methods. Firstly, we propose an approach for gene module detection by integrating multi-condition transcriptome data and interactome data using network overlapping module detection method. We apply the approach to study the transcriptome data of human pre-implantation embryos across multiple development stages, and identify several stage-specific dynamic functional modules and genes which provide interesting biological insights. We evaluate the reproducibility of the modules by comparing with some other widely used methods and show that the intra-module genes are significantly overlapped between the different methods. Secondly, we propose an approach for gene module detection by integrating transcriptome, translatome, and interactome data using multilayer network. We apply the approach to study the ribosome profiling data of mTOR perturbed human prostate cancer cells and mine several translation efficiency regulated modules associated with mTOR perturbation. We develop an R package, TERM, for implementation of the proposed approach which offers a useful tool for the research field. Next, we propose an approach for feature selection by integrating transcriptome and interactome data using network-constrained regression. We develop a more efficient network-constrained regression method eGBL. We evaluate its performance in term of variable selection and prediction, and show that eGBL outperforms the other related regression methods. With application on the transcriptome data of human blastocysts, we select several interested genes associated with time-lapse parameters. Finally, we propose an approach for classification by integrating epigenome and transcriptome data using neural networks. We introduce a superlayer neural network (SNN) model which learns DNA methylation and gene expression data parallelly in superlayers but with cross-connections allowing crosstalks between them. We evaluate its performance on human breast cancer classification. The SNN provides superior performances and outperforms several other common machine learning methods. The approaches proposed in this thesis offer effective and efficient solutions for integration of heterogeneous high-dimensional datasets, which can be easily applied to other datasets presenting the similar structures. They are therefore applicable to many fields including but not limited to Bioinformatics and Computer Science.

Asthma is a respiratory syndrome associated with airflow limitation, bronchial hyperresponsiveness and inflammation of the airways in the lungs. Despite the ongoing research efforts, the outstanding heterogeneity displayed by the multiple forms in which this condition presents often hampers the attempts to determine and classify the phenotypic and endotypic biological structures at play, even when considering a limited assembly of asthmatic subjects. To increase our understanding of the molecular mechanisms and functional pathways that govern asthma from a systems medicine perspective, a computational workflow focused on the identification of personalized transcriptomic modules from the U-BIOPRED study cohorts, by the use of the novel MODifieR integrated R package, was designed and applied. A feature selection of candidate asthma biomarkers was implemented, accompanied by the detection of differentially expressed genes across sample categories, the production of patient-specific gene modules and the subsequent construction of a set of core disease modules of asthma, which were validated with genomic data and analyzed for pathway and disease enrichment. The results indicate that the approach utilized is able to reveal the presence of components and signaling routes known to be crucially involved in asthma pathogenesis, while simultaneously uncovering candidate genes closely linked to the latter. The present project establishes a valuable pipeline for the module-driven study of asthma and other related conditions, which can provide new potential targets for therapeutic intervention and contribute to the development of individualized treatment strategies.

Con il termine di –omica spaziale si intende l’insieme di diverse tecniche che consentono di rilevare alterazioni significative delle biomolecole all’interno dei loro tessuti d’origine o delle strutture cellulari, permettendo quindi di integrare ed ampliare la comprensione dei cambiamenti biologici che si verificano in tessuti patologici complessi ed eterogenei, come il cancro. Tuttavia, per comprendere appieno la complessità e le dinamiche al di là delle condizioni patologiche, è necessario studiare e integrare diverse analisi molecolari, come quelle di lipidi e glicani, in modo da ottenere un’istantanea molecolare il più completa ed estesa possibile della malattia. Tra le tecniche di -omica spaziale, quella di desorbimento e ionizzazione laser assistiti da matrice (MALDI) abbinata alla spettrometria di massa imaging (MSI), permette lo studio della componente molecolare del tessuto patologico tramite un approccio multiplex, che permette di esaminare diverse centinaia di biomolecole in una singola analisi. Pertanto, l’analisi MALDI-MSI viene utilizzata per studi -omici spaziali di proteine, peptidi e N-glicani su campioni di tessuti clinici fissati in formalina e inclusi in paraffina (FFPE). Per quanto riguarda i lipidi, invece, questo tipo di analisi è sempre stato considerato poco efficace su campioni FFPE a causa della perdita di una grande quantità di contenuto lipidico durante le fasi di lavaggio con solventi organici, mentre i restanti lipidi resistenti ai solventi sono inaccessibili poiché trattenuti nei legami incrociati della formalina. In questi tre anni di dottorato, abbiamo sviluppato nuovi approcci MALDI-MSI per l'analisi spaziale multi-omica su campioni di tessuto clinico FFPE. Le prime tre pubblicazioni riportate in questa tesi si sono concentrate sullo sviluppo di protocolli MALDI-MSI per lipidi in campioni FFPE. In particolare, due di essi descrivono il metodo di preparazione del campione per la rilevazione di ioni di fosfolipidi carichi positivamente, principalmente fosfatidilcoline (PC), in campioni clinici di carcinoma renale a cellule chiare (ccRCC) e in un modello di xenotrapianto di cancro al seno. La terza pubblicazione riporta la possibilità di utilizzare ioni di fosfolipidi carichi negativamente, principalmente fosfatidilinositoli (PI), per definire firme lipidiche in grado di distinguere i gradi di tumore del colon-retto che presentano diverse quantità di linfociti infiltranti il tumore (TIL). Il lavoro finale propone un originale metodo MALDI-MSI multi-omico per l'analisi sequenziale di lipidi, N-glicani e peptidi triptici su una singola sezione FFPE. In particolare, il metodo è stato inizialmente implementato su replicati tecnici di cervello murino e successivamente utilizzato su campioni di ccRCC, come ulteriore prova, ottenendo una caratterizzazione più completa del tessuto tumorale grazie alla combinazione delle informazioni molecolari. Complessivamente, questi risultati aprono la strada a un nuovo approccio multi-omico spaziale basato sulla spettrometria di massa imaging (MSI) che è in grado di restituire un ritratto molecolare più ampio e più preciso della malattia.
The field of spatial omics defines the gathering of different techniques that allow the detection of significant alterations of biomolecules in the context of their native tissue or cellular structures. As such, they extend the landscape of biological changes occurring in complex and heterogeneous pathological tissues, such as cancer. However, additional molecular levels, such as lipids and glycans, must be studied to define a more comprehensive molecular snapshot of disease and fully understand the complexity and dynamics beyond pathological condition. Among the spatial-omics techniques, matrix-assisted laser desorption/ionisation (MALDI)-mass spectrometry imaging (MSI) offers a powerful insight into the chemical biology of pathological tissues in a multiplexed approach where several hundreds of biomolecules can be examined within a single experiment. Thus, MALDI-MSI has been readily employed for spatial omics studies of proteins, peptides and N-Glycans on clinical formalin-fixed paraffin-embedded (FFPE) tissue samples. Conversely, MALDI-MSI analysis of lipids has always been considered not feasible on FFPE samples due to the loss of a great amount of lipid content during washing steps with organic solvents, with the remaining solvent-resistant lipids being involved in the formalin cross-links. In this three-year thesis work, novel MALDI-MSI approaches for spatial multi-omics analysis on clinical FFPE tissue samples were developed. The first three publications reported in this thesis focused on the development of protocols for MALDI-MSI of lipids in FFPE samples. In particular, two of them describe a sample preparation method for the detection of positively charged phospholipids ions, mainly phosphatidylcholines (PCs), in clinical clear cell Renal Cell Carcinoma (ccRCC) samples and in a xenograft model of breast cancer. The third publication reports the possibility to use negatively charged phospholipids ions, mainly phosphatidylinositols (PIs), to define lipid signatures able to distinguish colorectal cancers with different amount of tumour infiltrating lymphocytes (TILs). The final work proposes a unique multi-omic MALDI-MSI method for the sequential analysis of lipids, N-Glycans and tryptic peptides on a single FFPE section. Specifically, the method feasibility was first established on murine brain technical replicates. The method was consequently used on ccRCC samples, as a proof of concept, assessing a more comprehensive characterisation of the tumour tissue when combining the multi-level molecular information. Altogether, these findings pave the way for new MSI-based spatial multi-omics approach aiming at an extensive and more precise molecular portrait of disease.

Approaches to the identification of genetic variants associated with complex brain diseases have evolved in recent decades. This evolution was supported by advancements in medical imaging and genotyping technologies that result in rich data production in the field of imaging genetics and radiogenomics. Studies in these fields have taken different designs and directions from genomewide associations to studying the complex interplay between genetics and structural connectivity of a wide range of brain-related diseases. Nevertheless, such combinations of heterogeneous, high dimensional and inter-related data has introduced new challenges which cannot be handled with traditional statistical methods. In this thesis, we proposed analysis pipelines and methodologies to study the causal relationship between neuroimaging features, including tumour characteristics and connectomics, genetics and clinical factors in brain-related diseases. In doing so, we adopted two longitudinal study designs and modelled the association between Alzheimer's disease progression and genetic factors, utilising local and global brain connectivity networks. In addition to that, we performed a multi-stage radiogenomic analysis in glioblastoma using non-parametric statistical methods. To address some limitations in the methods, we adopted the Structural Equation Model and developed a mathematical model to examine the inter-correlation between neuroimaging and multi-omic characteristics of brain-related diseases. Our findings have successfully identified risk genes that were previously reported in the literature of Alzheimer's and glioblastoma diseases, and discovered potential risk variants which associate with disease progression. More specifically, we found some loci in the genes CDH18, ANTXR2 and IGF1, located in Chromosomes 5, 4 and 12, to have effect on the brain connectivity over time in Alzheimer's disease. We also found that the expression of APP, HFE, PLAU and BLMH have significant effects on the structural connectivity of local areas in the brain, these are the left Heschl gyrus, right anterior cingulate gyrus, left fusiform gyrus and left Heschl gyrus, respectively. These potential association patterns could be useful for early disease diagnosis, treatment and neurodegeneration prediction. More importantly, we identified gaps in the imaging genetics methodologies, we proposed a mathematical model accounting for these limitations and evaluated the model which produced promising results. Our proposed flexible model, BiGen, addresses the gaps in the existing tools by combining neuroimaging, genetics, environmental, and phenotype information to a single complex analysis, accounting for the heterogeneity, inter-correlation, and non-linearity of the variables. Moreover, BiGen adopts an important assumption which is hardly met in the literature of imaging genetics, and that is, all the four variables are assumed to be latent constructs, that means they can not be observed directly from the data, and are measured through observed indicators. This is an important assumption in both neuroimaging, behavioural and genetic studies, and it is one of the reasons why BiGen is flexible and can easily be extended to include more indicators and latent constructs in the context of brain-related diseases.

Neuroblastoma is the first neurogenic-extracranial solid cancer occurring in infancy and childhood. The genetic aberration most commonly associated with a poor prognosis is MYCN gene’s amplification. We hypothesize that effective anti-MYC therapeutics can be developed by understanding the regulation and function of N-MYC in neuroblastoma. Since N-MYC is an intrinsically disordered protein, it is still challenging to target this transcription factor, however, the model is shifting significantly after discovering novel therapeutic targets that impact MYC-driven tumorigenesis. The following work explores how MYCN expression affects the induction and maintenance of neuroblastoma. By using different multi-omic approaches and many promising innovative techniques, we were able to identify and characterize new potential vulnerabilities of this pathology, which may work in concert with N-MYC for the instruction of a high-risk neuroblastoma phenotype. My studies’ first objective was to investigate whether and how N-MYC can regulate transcription of lncRNAs by comparing transcriptional profiles between non-amplified and MYCN-amplified neuroblastoma cells. Here, we singled out lncNB1, which is selectively higher expressed in high MYCN cells only and it is also firmly and almost uniquely transcribed in neuroblastoma among all types of cancers. Our data showed that N-MYC directly activates transcription of lncNB1, instructing a complex network of molecular interactions, ultimately resulting in increased N-MYC protein stability, reinforcing the N-MYC oncogenetic program. The second objective was to assess how high N-MYC expression may cooperate to establish a dynamic regulatory axis with the E2F3 transcription factor, impacting the development of the high-risk cancer phenotype. Taken together, our unbias screenings uncovered potential candidates that help to fill the knowledge gap in understanding what is the impact of N-MYC in childhood neuroblastoma, providing new opportunities for the development of specific treatments able to target the function of MYC oncoproteins in a context of MYCN gene amplification.

Type 2 diabetes mellitus (T2DM) is the most rapidly growing disease worldwide, with a more than four-fold increase in diagnosed people in the last 30 years. Almost 500 million people are affected, and many more are thought to be undiagnosed. T2DM is a disorder of human metabolism resulting from the resistance of peripheral tissues to the hormone insulin, which is produced by the β-cells of the pancreas. Insulin regulates blood glucose levels and therefore insulin resistance can generate a profound hyperglycemia that itself has major health risks. T2DM is largely considered therefore to be a ‘lifestyle’ disease, involving energetic excess from poor diet and physical inactivity. T2DM is a major risk factor for several comorbidities including cardiovascular disease (atherosclerosis and stroke), non-alcoholic fatty liver disease (NAFLD) and ocular and neurological disorders (retinopathy and neuropathy). Another major potential consequence of T2DM is renal disease known as ‘diabetic nephropathy’ (DN), which is a microvascular complication eventually leading to end-stage renal disease (ESRD). Despite this, the vast majority of molecular studies aimed at better understanding the basis of pathogenesis in T2DM have examined cell lines or other tissues (e.g. the liver). Given that the first major symptoms of T2DM include polydipsia (increased thirst) and polyuria (increased urination), which both have some basis in the regulation of blood pressure in the kidney glomerulus, a complete analysis of the T2DM kidney is somewhat overdue. Human studies of T2DM are plagued with reproducibility issues due to person-person differences, including age, diet, current drug treatment and genetics. We therefore employed a reproducible rodent (rat) model of T2DM that utilises a combination of high fat (HF) diet and low-dose streptozotocin (STZ) injection, which induces pancreatic insufficiency via β-cell dysfunction. Animals subjected to a single treatment (HF diet or STZ injection) or to neither were used as controls. Biochemical and physiological testing showed the T2DM animals showed all traits of human disease, including weight gain, elevated blood glucose levels and reduced insulin tolerance. Histological and EchoMRI analysis of the T2DM kidney demonstrated both morphological defects and physiological alterations consistent with human T2DM, including changes to glomerular health and the formation of structures akin to Kimmelstein-Wilson nodules seen in DN. We demonstrated that application of large-scale high-throughput profiling of the proteome permits systematic assessment of proteins from T2DM kidney tissue. Major changes were observed in pathways associated with metabolism, including the tricarboxylic acid (TCA) cycle and fatty acid biosynthesis / metabolism. We also performed a comparative analysis of the urine from these animals, which showed changes to urinary albumin (increased abundance) and major urinary protein Mup1 (decreased abundance) consistent with human disease. Furthermore, we identified 4 proteins, including apolipoprotein A2, alpha amylase and regenerating islet-derived protein 3, which could be putative urinary biomarkers for T2DM-associated DN. Proteome-level analysis highlighted pathways associated with oxidative stress and signal transduction as altered in animals subjected to HF diet and STZ injection (T2DM). Given the known role of phosphorylation-based signalling in transmitting the insulin response and the dysregulation of insulin signalling in T2DM liver, we next examined renal signalling in T2DM animals by phosphoproteomics. Additional experiments examined the role of reactive oxygen species (ROS) in protein post-translational modifications (PTMs) in these animals. More than 20,000 sites of PTM were identified and quantified across the 4 biological groups. While many pathways were broadly influenced by HF diet and β-cell dysfunction, significant alterations in T2DM animals across both the phospho- and redox-proteome were observed in the mitochondrial TCA cycle, pyruvate metabolism and glycolysis / gluconeogenesis pathways. For example, multiple cysteine redox PTMs were observed as significantly regulated in T2DM-like animals on phosphoenolpyruvate carboxylase (PEPCK), which lies at the nexus of those three pathways, and confirms a role for ROS in mediating cross-talk between metabolic pathways resulting in altered cell signalling and pathway flux. We further speculate that the identified pathways may be linked to structural changes in glomerular podocytes, as alterations in these pathways correlate with previously reported abnormalities in podocyte biology and are characteristic of DN. This thesis has provided a comprehensive molecular multi-omic analysis of the kidney in a rodent model of T2DM. Future studies will be needed to further validate our data, using metabolomics and lipidomics style approaches, coupled with functional studies to determine the role of protein and protein PTM changes in enzyme catalysis and pathway flux. Furthermore, translation of these results into the clinic will require testing of large human cohorts, for example to determine the efficacy of putative urinary markers identified here. By further characterising the roles of proteins and their PTMs, altered protein interactions and related pathways, and how they go on to form an integrated network, we have endeavoured to better understand the molecular mechanisms underlying T2DM-induced DN. Given the complexity and multi-organ involvement in T2DM, better understanding the renal proteome provides a useful resource in enabling stratification of DN diagnosis and improved options for interventional therapies.

To paraphrase Stan Ulam, a Polish mathematician who became a leading figure in the Manhattan Project, in this dissertation I focus not only on how computer science can help biologists, but also on how biology can inspire computer scientists. On one hand, computer science provides powerful abstraction tools for metabolic networks. Cell metabolism is the set of chemical reactions taking place in a cell, with the aim of maintaining the living state of the cell. Due to the intrinsic complexity of metabolic networks, predicting the phenotypic traits resulting from a given genotype and metabolic structure is a challenging task. To this end, mathematical models of metabolic networks, called genome-scale metabolic models, contain all known metabolic reactions in an organism and can be analyzed with computational methods. In this dissertation, I propose a set of methods to investigate models of metabolic networks. These include multi-objective optimization, sensitivity, robustness and identifiability analysis, and are applied to a set of genome-scale models. Then, I augment the framework to predict metabolic adaptation to a changing environment. The adaptation of a microorganism to new environmental conditions involves shifts in its biochemical network and in the gene expression level. However, gene expression profiles do not provide a comprehensive understanding of the cellular behavior. Examples are the cases in which similar profiles may cause different phenotypic outcomes, while different profiles may give rise to similar behaviors. In fact, my idea is to study the metabolic response to diverse environmental conditions by predicting and analyzing changes in the internal molecular environment and in the underlying multi-omic networks. I also adapt statistical and mathematical methods (including principal component analysis and hypervolume) to evaluate short term metabolic evolution and perform comparative analysis of metabolic conditions. On the other hand, my vision is that a biomolecular system can be cast as a ?biological computer?, therefore providing insights into computational processes. I therefore study how computation can be performed in a biological system by proposing a map between a biological organism and the von Neumann architecture, where metabolism executes reactions mapped to instructions of a Turing machine. A Boolean string represents the genetic knockout strategy and also the executable program stored in the ?memory? of the organism. I use this framework to investigate scenarios of communication among cells, gene duplication, and lateral gene transfer. Remarkably, this mapping allows estimating the computational capability of an organism, taking into account also transmission events and communication outcomes.

Recent literature has highlighted that mutations causing neurodevelopmental syndromes are particularly enriched in genes related to chromatin regulation and synaptic functions. While the latter could be easily predicted, the former shed seeds to the flourishing of epigenomic studies focused on this type of disorders. Intriguingly, most of such disorders couple different shades of intellectual disabilities with peculiar cranio-facial features and systemic defects which are shared, opposite or unique across them. I have set up a dynamic framework of analysis, that encompasses the comparison of multiple disorders, cell types and cultures, to highlight cell-type specific and developmentally-relevant paths of transcriptional deregulation. Building on my lab’s expertise to harness potency and stability of induced pluripotent stem cells (iPSCs), I identified two main axes of development through which we could characterize on the one hand cerebral cortex related dysregulations and on the other hand cranio-facial features associated traits, peripheral nervous system- and cardiovascular system-related dysregulations. The former is based on the production of adult glutamatergic cortical neurons through ectopic expression of NGN2 in iPSCs and, in parallel, through production of brain organoids: 3D cultures obtained by neuronal differentiation and patterning via sequential exposure to small molecules. The latter is based on differentiation of iPSCs to neural crest stem cells (NCSCs) and mesenchymal stem cells (MSCs). I collected and standardised transcriptomic data coming from controls- and patient-derived iPSCs accounting for six disorders, NCSCs for five disorders and MSCs for two disorders; NGN2 neurons for two disorders and brain organoid for one disorder. During my research I helped define new standards for RNA-seq experiments tailored for differential-expression analysis and developed or implemented tools to make cross-disorder and cross-tissue comparisons in a connectable way. This work let me identify regulatory circuitries shared by all disorders or by subgroups characterized by shared phenotypes; symmetric deregulations in disorders caused by mutation of opposite histone modifiers; unexpected subgroups that will require further investigation. For most disorders, my work confirms previously published evidence that dysregulations identified at the pluripotent stage can be inherited and amplified in disease-relevant tissues in a tissue-specific fashion. Thus, I was capable of identifying disease-specific dysregulations at the pluripotent stage and in disease-relevant tissues; I drew conclusions on iPSCs cross-disorder transcriptional dysregulations through the definition of transcriptional modules; I implemented an analytical framework to boost the ability of identifying the effect of knocking down a certain gene on transcriptional and epigenetic landscapes; I identified sets of genes whose deregulation at the pluripotent stage reverberates and amplifies along development, funnelling and filtering several analyses to converge on a small set of actionable targets; I identified a small set of potential direct targets of PRC2 complex involved in brain development and on the onset of Weaver Syndrome; I identified BAZ1B-specific transcriptional dysregulations in NCSCs that confirm its importance for migration and craniofacial morphogenesis but more in general for chromatin remodelling and human evolution; I helped in the molecular characterization of YY1 mutations, which led to the identification of Gabriele-de Vries Syndrome; I contributed to the molecular characterization of Kabuki Syndrome in neural crest and adult cortical neurons.

ADNP encodes Activity-Dependent Neuroprotective Protein, whose de novo heterozygous mutations cause Helsmoortel-Van der Aa Syndrome (HVDAS), a rare developmental syndrome affecting brain formation and neuronal functions, involving autism spectrum disorder and intellectual disability. Although ADNP is one of the single-gene most frequently mutated in ASD, its precise role in the syndrome onset has yet to be clarified. ADNP is the DNA-binding component of the newly identified chromatin remodeler complex ChAHP in mESC. It recognizes euchromatin regions to establish less accessible local chromatin domains and has also been recently identified as a new player in the regulation of genomic topology, competing with CTCF in the organization of chromatin architecture. Our aim is to understand the genetic and epigenetic implications of ADNP underlying this neurodevelopmental condition; we harnessed cell reprogramming to establish a highly informative cohort of patient-specific iPSCs and use it as a platform to develop meaningful model for the pathology, thus enabling the assessment of the ADNP pivotal relevance in both pluripotent and neuronally-patterned stages. We discovered an altered gene expression program associated with cell fate decision and neuronal lineage commitment, highlighting a neurodevelopmental disruption elicited by ADNP mutations already at the pluripotent stage. Employing CRISPR/Cas9-engineering, we FLAG-tagged the endogenous ADNP to assess its genomic occupancy and revealed a genome-wide distribution of ADNP at gene-regulatory elements and a predominant presence at transposable elements, Alu sequences in particular. We decoupled ADNP and CTCF interplay in our human iPSCs model, and found a global redistribution of active enhancer histone marks signature, which sustain upregulation with the intervention of EZH2-mediated derepression. Finally, HVDAS cortical organoid models show morpho-functional impairment in the early stages of neuronal differentiation, with decreased size and lower mitotic activity, coupled with accelerated maturation phenotype assessed through single-cell transcriptomic analysis. Altogether, with these results we delineate how ADNP deficiency affects pluripotent regulatory landscape and disease-relevant mechanisms that ultimately impact neuronal development and functionality.

L’identification des mutations constitutionnelles à l’origine d’une prédisposition génétique au cancer est essentielle à la prise en charge médicale des patients et de leurs familles. Depuis l’implémentation des technologies de séquençage à haut-débit dans les laboratoires diagnostiques, le principal défi n’est plus la détection des variations génétiques mais leur interprétation et leur classification. La question de l’interprétation de la variation est particulièrement cruciale lorsqu’elle conditionne la stratégie thérapeutique. Ainsi, il est essentiel de disposer de tests simples adaptables en routine diagnostique pour faciliter l’interprétation des variations génétiques. Dans ce contexte, nous avons utilisé un test fonctionnel développé par notre équipe pour classer des variations dans le gène TP53 à l’origine du syndrome de Li-Fraumeni et pour appréhender la corrélation génotype - phénotype chez les patients LFS. Dans un deuxième temps, nous avons évalué la pertinence d’une approche multi-omique (RNA-Seq et métabolomique) pour discriminer les cellules sauvages des cellules avec mutation hétérozygote du gène TP53 ou des gènes BRCA impliqués dans la prédisposition génétique aux cancers du sein et de l’ovaire. Sur la base des données de transcriptome, un modèle mathématique a été développé pour détecter les variants correspondant à des mutations délétères. Nous avons ensuite sélectionné les biomarqueurs les plus discriminants pour les intégrer dans un test fonctionnel de RT-MLPA dédié à la voie p53. Nous avons enfin adapté cet essai pour qu’il soit réalisable sur une simple prise de sang, sans immortalisation des lymphocytes du patient
The identification of the constitutional mutation responsible for a genetic predisposition to cancer is essential to the clinical management of the patient and its relatives. With the implementation of high-throughput sequencing to the diagnostic routine of these pathologies, the challenge no longer lies within the detection of alterations but in their biological and clinical interpretation. While specific treatments are emerging, simple functional assays to help with the interpretation of the detected variants are needed. In this context, we used a functional test developed by our team to classify variations in the TP53 gene responsible for Li-Fraumeni syndrome and to understand the genotype-phenotype correlation in LFS patients. On the other hand, we assessed the relevance of a multi-omic approach (RNA-Seq and metabolomics) to discriminate wild-type cells from cells with a deleterious heterozygous mutation in TP53 or in the BRCA genes implicated in genetic predisposition to breast and ovarian cancers. Based on the transcriptomic data, a mathematical model has been developed to detect variants corresponding to deleterious mutations. Then we selected the most discriminating biomarkers and integrated them into a RT-MLPA functional assay dedicated to the p53 pathway. We finally adapted this test to be feasible on a simple blood test, without immortalization of the patient's lymphocytes

L’aspect novateur de ce projet réside dans l’étude des acylations au niveau de la lysine K27 de l’histone H3, classiquement étudiée sous forme méthylée ou acétylée. Nous avons réalisé ce travail sur des cellules germinales méiotiques et post-méiotiques de souris. La spermiogénèse, qui implique un programme d’expression spécifique ainsi qu’une régulation fine de la transcription, est un processus particulièrement adapté à la compréhension du rôle des nouvelles modifications d’histones. Ces travaux regroupent l’utilisation de quatre approches omiques différentes, à savoir la protéomique, la métabolomique, la transcriptomique et le séquençage ChIP-seq afin de décrypter la régulation des acylations sur H3K27.Dans la première partie de ce projet, nous avons exploré la dynamique d’acétylation et de crotonylation sur les lysines d’histones au cours des processus de sporulation de la levure et de spermatogénèse de la souris, ce qui nous a permis de mettre en évidence un site d’intérêt H3K27 crotonylé. Son accumulation sur le variant d’histone H3.3 et sa stoechiométrie importante par rapport à la forme acétylée H3K27ac dans les cellules germinales post-méiotiques de souris nous ont conduits à étudier la distribution génomique de cette marque, par des analyses de ChIP-seq. L’analyse comparative de H3K27ac et H3K27cr a révélé une synergie entre la présence de ces deux marques à la fois au niveau des promoteurs et des enhancers distants, ce qui suggére une possible alternance des deux marques afin de réguler la transcription. Au niveau des promoteurs, nous avons observé une augmentation de ces modifications entre les stades méiotiques et post-méiotiques en amont des gènes caractéristiques de la spermiogénèse. D’autre part, la présence simultanée des deux marques coïncide avec la co-localisation de plusieurs régulateurs de la transcription spécifiques de ce processus (SLY, SOX30) et de protéines de liaison à la chromatine (BRD4, BORIS et CTCF), tandis qu’une sélectivité de fixation est observée lorsque H3K27ac et H3K27cr sont identifiées seules aux promoteurs. De façon intéressante, nous observons des résultats similaires au niveau des enhancers ainsi que des super-enhancers, confirmant que la régulation de la transcription est modulée par la présence alternative de ces deux acylations.La deuxième partie de ma thèse a porté sur l’étude de la propionylation et de la butyrylation de H3K27 au cours de la sporulation de la levure et de la spermatogénèse de la souris. Cependant, cette partie s’est avérée pleine de surprises car les analyses MS/MS en cellules HCD et la comparaison avec les peptides synthétiques correspondants n’ont pas permis de valider une propionylation et une butyrylation sur H3K27. Il s’est avéré qu’il s’agissait de structures strictement isobares avec ces modifications connues, mais de nature différente, puisque plus hydrophile que ces acylations. Plusieurs hypothèses ont été testées afin de déterminer la composition de ces modifications, mais au moment de la finalisation de ce manuscrit, nous n’avons pas encore trouvé le fin mot de l’histoire.Mes travaux de thèse rappellent les obervations de Goudarzi et al., à savoir une dynamique entre acétylation et acylation sur les résidus lysines à l’origine de la fixation différentielle de facteurs de régulation ou de protéines se liant à la chromatine et responsables de la régulation de la transcription. Ils ont également mis en lumière un rôle importante de H3K27cr, au niveau des enhancers en combinaison avec H3K27ac, lesquels ne sont classiquement pas étudiés dans les études fonctionnelles portant sur la compréhension du rôle de nouvelles acylations
The innovative aspect of this project lies in the study of acylations at lysine 27 from histone H3 (H3K27), conventionally studied in a methylated or an acetylated form. We performed this work on meiotic and post-meiotic mouse germ cells. Spermiogenesis, which involves a specific expression program as well as a fine regulation of transcription, is a process that is particularly well suited to understanding the roles of new histone modifications. This work combines the use of four different omics approaches, namely proteomics, metabolomics, transcriptomics and ChIP- sequencing to decipher the regulation of acylations on H3K27.In the first part of this project, we explored the dynamics of acetylation and crotonylation on histone lysines during the processes of yeast sporulation and mouse spermatogenesis, which allowed us to highlight in particular crotonylated H3K27. Its accumulation on the histone variant H3.3 and its important stoichiometry compared to the acetylated form H3K27ac in mouse post-meiotic germ cells led us to study the genomic distribution of this mark by ChIP-seq analysis. The comparative analysis of H3K27ac and H3K27cr revealed a synergy between the presence of these acylations at both promoters and distal enhancers, suggesting a possible alternation of the two marks to regulate transcription. At the promoter level, we observed an increase of these modifications between the meiotic and post-meiotic stages upstream of the genes characteristic of spermiogenesis. In addition, the simultaneous presence of the two marks coincides with the co-localization of several transcriptional regulators specific for this process (SLY, SOX30) and of chromatin-binding proteins (BRD4, BORIS and CTCF), whereas a binding selectivity is observed when H3K27ac and H3K27cr are identified alone at promoters. Interestingly, we observe similar results at enhancers as well as super-enhancers, confirming that the regulation of transcription is modulated by the alternative presence of these two acylations.The second part of my thesis focused on the study of the possible propionylation and butyrylation of H3K27 during yeast sporulation and mouse spermatogenesis. However, this part proved to be full of surprises because the MS/MS analyses and the comparison with the corresponding synthetic peptides did not make it possible to validate a propionylation and a butyrylation on H3K27. It turned out that the modifications observed on H3K27 from mouse histones were strictly isobaric with these known modifications, but of a different nature, since they are more hydrophilic. Several hypotheses were tested in order to determine the structure of these modifications, but at the time of finalizing this manuscript, we have not found out what it is all about.My PhD work contributes further to the idea of a dynamics between acetylation and acylations on lysine residues at the origin of the differential binding of chromatin-binding proteins responsible for regulating transcription. It also highlighted an important role of H3K27crat enhancers which are not classically considered in studies aiming at understanding the roles of new acylations

Cyanobacteria are photosynthetic organisms with the potential to sustainably produce carbon-based end products by fixing carbon dioxide from the atmosphere. Optimizing the growth or biochemical production in cyanobacteria is an ongoing challenge in metabolic engineering. Rational design of metabolic pathways requires a deep understanding of regulatory mechanisms. Hence, a deeper understanding of photosynthetic regulation of the influence of the environment on metabolic fluxes provides exciting possibilities for enhancing the photosynthetic Calvin-Benson-Bassham cycle. One approach to study metabolic processes is to use omic-level techniques, such as proteomics and fluxomics, to characterize varying phenotypes that result from different environmental conditions or different genetic perturbations.

This dissertation examines the influence of light intensity on enzymatic abundances and the resulting Calvin-Benson-Bassham cycle fluxes using a combined proteomic and fluxomic approach in the model cyanobacteria Synechocystis sp. PCC 6803. The correlation between light intensity and enzymatic abundances is evaluated to determine which reactions are more regulated by enzymatic abundance. Additionally, carbon enrichment data from isotopic labelling experiments strongly suggest metabolite channeling as a flexible and light-dependent regulatory mechanism present in cyanobacteria. We propose and substantiate biological mechanisms that explains the formation of metabolite channels under specific redox conditions.

The same multi-omic approach was used to examine genetically modified cyanobacteria. Specifically, genetically engineered and conditionally growth-enhanced Synechocystis strains overexpressing the central Calvin-Benson-Bassham cycle enzymes FBP/SBPase or transketolase were evaluated. We examined the effect of the heterologous expression of each of these enzymes on the Calvin-Benson-Bassham cycle, as well as on adjacent central metabolic pathways. Using both proteomics and fluxomics, we demonstrate distinct increases in Calvin-Benson-Bassham cycle efficiency as a result of lowered oxidative pentose phosphate pathway activity. This work demonstrates the utility of a multi-omic approach in characterizing the differing phenotypes arising from environmental and genetic changes.

Advances in instrumentation and sample preparation techniques enable evermore in-depth molecular profiling to catalyze exciting research into complex biological processes. Current platforms survey biomolecular classes with varying depth. While sequencing is near comprehensive, and even enabled at single cell resolution, challenges remain in global metabolite surveys primarily due to the increased chemical diversity relative to other “omics” data types. At the same time, metabolism and the interaction of diverse biomolecules are increasingly recognized as vitally important components of many disease processes. Presented here is work describing the development and use of molecular interaction subnetworks for the functional interpretation of multi-omic data. Metabolic pathway-centric subnetworks for functional inference with protein or gene derived global profiling data were created from the integration of disparate network models: Protein- protein interaction (PPI) networks and metabolic models. The subnetworks were shown to increase mapping between metabolic pathways and the proteome, and the subnetwork- derived analysis shows dramatic improvement over primary enzymes alone with direct metabolomic experimental measurements for validation of pathway findings. We illustrate the functional utility of integrating PPI data with metabolic models by finding network modules previously but independently implicated in disease. Specifically, the analysis reveals abundance increases in known oncogenes in response to changes in breast cancer metabolism. Additionally, we reveal cellular mechanisms related to metabolic stress observed in patient sera following viral SARS-CoV-2 infection, and metabolic changes in a model of heart disease, where the characteristic muscle fibers make in-depth proteomic profiling difficult. Functional network models were additionally used to compare the response of varying cell lines in response to viral infection, showing significant context- specific differences. All of these findings demonstrate the importance of functional models to help interpret multi-omic data. The implications of revealing the connections between metabolism and protein subnetwork rewiring may be profound; for example, suggesting metabolic pathway activity may be as important a biomarker as mutation status in cancer. This research points to a means of practically inferring metabolic state from proteomic data. We further describe the release of our open-source software to accelerate integrative multi-omic analysis in the broader research community.
2023-06-16T00:00:00Z

Despite decades of effort, the mechanistic underpinnings of many cancers remain unsolved It has increasingly become appreciated that cancers can be more readily classified by their transcriptional identities rather than by genomics alone. A fuller understanding of the mechanistic connections between the aberrant genomics leading to the transcriptional dysregulation of tumors is key to both improving our knowledge of cancer biology as well as developing more precise and effective therapeutics. This thesis explores the development and application of a network based multi-omic master regulator framework designed to elucidate these pathways. In Chapter 2 we apply this analysis across 20 tumor types from the Cancer Genome Atlas and in doing so identify 407 key master regulators responsible for canalizing a high percentage of the driver genetics present across these samples. Further evaluation of these key regulators revealed a highly modular structure, indicating that the regulators work in coordinated groups to implement a variety of key cancer hallmarks. Genetic and pharmacological validation assays confirmed the predicted interactions and biological phenotypes. Chapter 3 focuses on the application of this analytical framework specifically on gastroesophageal tumors. Using a more fine-grained approach we find 15 distinct subtypes across a cohort of these heterogenous tumors. These subtypes align well with previously identified features of these cancers but also reveal novel genomic associations and key master regulators that can serve as potential avenues for therapeutic treatment.

Background: Osteogenesis imperfecta (OI) also known as brittle bone disease, is a genetic pathology in which bones do not form properly and therefore are fragile and break easily. Is a heterogeneous congenital heritable disease that mainly affects connective tissues. Nowadays we number 18 types of OI, distinguished into autosomal dominant, recessive and X-linked inheritance. OI type VI is caused by loss-of-function mutations in SERPINF1, which shows recessive inheritance. Lack of the gene product, PEDF, causes an atypical bone mineralization defect determining a unique clinical phenotype. Aim: the aim of the study is to identify genomic, epigenomic and metabolomic variations that are associated to the disease status in individuals that belong to a nuclear Pakistan family in which is supposed to be segregate type VI osteogenesis imperfecta. Results: exome sequencing confirmed the consanguinity between the parents and shared regions of homozygosity between affected were observed in chr7, chr12 and chr22. In the hypothesis of Autosomal Recessive disease, any compatible mutation was found, and no clear pathogenic variant have been detected. Thus, we explore compound heterozygosis model, identifying and suggesting as potential candidate CERCAM gene, but it is role in bone homeostasis it is still unknown. Epigenetic investigation highlights some interesting genes, known to be involved in bone metabolism, such as RXRA (Retinoid X Receptor Alpha), ELK3 (ETS Transcription Factor) and GLI2 (GLI Family Zinc Finger 2). Metabolomic profiling found 4 modulated pathways: phenylalanine, tyrosine and tryptophan biosynthesis, tryptophan metabolism, pyrimidine metabolism, and Vitamin B6 metabolism. Conclusions: the future investigation will try to enhance and integrate the results from the present omics (transcriptomic analysis is ongoing) into a context of system biology aimed to depict and clarify the defects and biological processes associated to the disease.

Lung cancer (LC) is the leading cause of cancer death in the US, claiming over 160,000 lives annually. Although CT screening has been shown to be efficacious in reducing mortality, the limited access to screening programs among high-risk individuals and the high number of false positives contribute to low survival rates and increased healthcare costs. As a result, there is an urgent need for preventative therapeutics and novel interception biomarkers that would enhance current methods for detection of early-stage LC. This thesis addresses this challenge by examining the hypothesis that transcriptomic changes preceding the onset of LC can be identified by studying bronchial premalignant lesions (PMLs) and the normal-appearing airway epithelial cells altered in their presence (i.e., the PML-associated airway field of injury). PMLs are the presumed precursors of lung squamous cell carcinoma (SCC) whose presence indicates an increased risk of developing SCC and other subtypes of LC. Here, I leverage high-throughput mRNA and miRNA sequencing data from bronchial brushings and lesion biopsies to develop biomarkers of PML presence and progression, and to understand regulatory mechanisms driving early carcinogenesis. First, I utilized mRNA sequencing data from normal-appearing airway brushings to build a biomarker predictive of PML presence. After verifying the power of the 200-gene biomarker to detect the presence of PMLs, I evaluated its capacity to predict PML progression and detect presence of LC (Aim 1). Next, I identified likely regulatory mechanisms associated with PML severity and progression, by evaluating miRNA expression and gene coexpression modules containing their targets in bronchial lesion biopsies (Aim2). Lastly, I investigated the preservation of the PML-associated miRNAs and gene modules in the airway field of injury, highlighting an emergent link between the airway field and the PMLs (Aim 3). Overall, this thesis suggests a multi-faceted utility of PML-associated genomic signatures as markers for stratification of high-risk smokers in chemoprevention trials, markers for early detection of lung cancer, and novel chemopreventive targets, and yields valuable insights into early lung carcinogenesis by characterizing mRNA and miRNA expression alterations that contribute to premalignant disease progression towards LC.
2020-01-25

Head and neck squamous cell carcinoma (HNSCC) is an aggressive malignancy associated with molecular heterogeneity, locoregional spread, resistance to therapy and relapse after initial treatment. Increasing evidence suggests that master developmental pathways with key roles in adult tissue homeostasis, including Hippo and Wnt/β-catenin signaling, are dysregulated in the initiation and progression of HNSCC. However, a comprehensive investigation into the crosstalk between these pathways is currently lacking, and may prove crucial to the discovery of novel targets for HNSCC therapy. More recent evidence points to the tumor microenvironment, mainly comprising cancer-associated fibroblasts (CAFs), as capable of influencing tumor cell behavior and promoting invasive HNSCC phenotypes. Nonetheless, current methods to screen for CAF markers in tumors are restricted to targeted immunostaining experiments with limited success and robustness across tissue types. The Cancer Genome Atlas network has generated multi-tiered molecular profiles for over 10,000 tumors spanning more than two dozen different cancer types, providing an unprecedented opportunity for the application and development of integrative methods aimed at the in silico interrogation of experimentally-derived signatures. These multi-omic profiles further enable one to link genomic anomalies, including somatic mutations and DNA copy number alterations, with phenotypic effects driven by pathogenic pathway activity. Effectively querying this vast amount of information to help elucidate subsets of functionally and clinically-relevant oncogenic drivers, however, remains an ongoing challenge. To address these issues, I first investigate the effects of oncogenic pathway perturbation in HNSCC using experimental models coupled with in vitro genome-wide transcriptional profiling. Next, I describe a new computational approach for the unbiased identification of CAF markers in HNSCC solely using bulk tumor RNA-sequencing information. Lastly, I have developed Candidate Driver Analysis or CaDrA - a statistical framework that allows one to query genetic and epigenetic alterations for candidate drivers of signature activity within a given disease context. Collectively, this work offers new perspectives on the molecular cues underlying HNSCC development, while simultaneously highlighting the power of integrative genomics methods capable of accelerating the discovery of novel targets for cancer diagnosis and therapy.

Modern healthcare research demands collaboration across disciplines to build preventive measures and innovate predictive capabilities for curing diseases. Along with the emergence of cutting-edge computational and statistical methodologies, data generation and analysis has become cheaper in the last ten years. However, the complexity of big data due to its variety, volume, and velocity creates new challenges for biologists, physicians, bioinformaticians, statisticians, and computer scientists. Combining data from complex multiple profiles is useful to better understand cellular functions and pathways that regulates cell function to provide insights that could not have been obtained using the individual profiles alone. However, current normalization and artifact correction methods are platform and data type specific, and may require both the training and test sets for any application (e.g. biomarker development). This often leads to over-fitting and reduces the reproducibility of genomic findings across studies. In addition, many bias correction and integration approaches require renormalization or reanalysis if additional samples are later introduced. The motivation behind this research was to develop and evaluate strategies for addressing data integration issues across data types and profiling platforms, which should improve healthcare-informatics research and its application in personalized medicine. We have demonstrated a comprehensive and coordinated framework for data standardization across tissue types and profiling platforms. This allows easy integration of data from multiple data generating consortiums. The main goal of this research was to identify regions of genetic-epigenetic co-ordination that are independent of tissue type and consistent across epigenomics profiling data platforms. We developed multi-‘omic’ therapeutic biomarkers for epigenetic drug efficacy by combining our biomarker regions with drug perturbation data generated in our previous studies. We used an adaptive Bayesian factor analysis approach to develop biomarkers for multiple HDACs simultaneously, allowing for predictions of comparative efficacy between the drugs. We showed that this approach leads to different predictions across breast cancer subtypes compared to profiling the drugs separately. We extended this approach on patient samples from multiple public data resources containing epigenetic profiling data from cancer and normal tissues (The Cancer Genome Atlas, TCGA; NIH Roadmap epigenomics data).