Heterozygous mutations in the transcription factor, hepatocyte nuclear factor 1B (HNF1B), result in multisystem disease including diabetes due to beta-cell dysfunction and pancreatic hypoplasia. However, the mechanisms that underlie development of diabetes in HNF1B mutation carriers are still not fully understood due to lack of an appropriate model system. Human induced pluripotent stem cells (hiPSCs), which are capable of self-renewal and can differentiate into any cell type, provide an advantageous alternative to model human developmental diseases. The aim of this project was to develop a hiPSC based model system to determine the molecular mechanisms by which HNF1B mutations cause pancreatic hypoplasia and diabetes. HNF1B mutant hiPSC lines were produced using CRISPR-Cas9 genome editing. Isogenic HNF1B wild-type, homozygous and heterozygous mutant hiPSC lines were directed to differentiate along the pancreatic lineage and cells were phenotyped at each stage of the differentiation process to check for appropriate expression of lineage markers. The normal expression pattern of HNF1B in human pancreas development was analysed and showed up-regulation of HNF1B at the foregut stage, and during pancreas specification. Homozygous knockout of HNF1B resulted in failure of foregut and pancreatic progenitor development, while heterozygous knockout of HNF1B resulted in impairment of pancreatic progenitor and endocrine cell differentiation as well as impaired insulin secretion upon glucose stimulation. Cell proliferation analyses showed a significant decrease in the proliferation rate in HNF1B heterozygous and homozygous mutant cells compared with wild-type cells at the foregut stage while no change in the apoptosis rate could be detected. RNA-sequencing and ATAC-sequencing, were used to further define the molecular mechanisms controlled by HNF1B and the effect HNF1B on modulation of chromatin accessibility during pancreas development. These results provide further insights into the molecular mechanisms by which HNF1B regulates human pancreas development and function, revealing that HNF1B haploinsufficiency impairs the expansion and maintenance of pancreatic progenitor cells in vitro. In vivo, this would likely result in reduced beta cell numbers at birth and diabetes later in life in patients with HNF1B-associated disease. These mechanisms suggest that the capacity to produce pancreatic progenitor cells during embryonic life could determine individual susceptibility to diabetes.

Monogenic autoimmune diseases are highly variable syndromes that usually have onset in the first year of life and are often fatal in early childhood. Identifying monogenic autoimmune diabetes is important as it can have implications for medical management of patients, informs families and clinicians of prognosis and recurrence risk, and gives insights into beta-cell autoimmunity and immune tolerance. The first section of this thesis introduces monogenic autoimmune disease, with focus on the conditions that have autoimmune endocrine disorders as part of their clinical phenotype. The following section details the methodologies used throughout this thesis. In chapter 1, we used a type 1 diabetes genetic risk score (T1D-GRS) based on the top 10 risk alleles for T1D to identify patients with monogenic autoimmunity from patients with early-onset polygenic diabetes and additional autoimmunity. We showed that the T1D-GRS was highly discriminatory of monogenic autoimmunity, especially when combined with age of onset (ROC-AUC 0.88). We also identified 16 families for gene discovery studies. Furthermore, this work shows that polygenic risk for the development of T1D does not affect the development of diabetes in monogenic autoimmunity. Chapter 2 describes the genetic and phenotypic information for the largest cohort of patients with IPEX syndrome, caused by hemizygous mutations in FOXP3, reported to date (n=48). We analysed this data to determine if there were any genotypic or clinical characteristics of IPEX syndrome that could predict prognosis. We did not find evidence of phenotype-genotype relationships and showed that presenting feature did not predict prognosis. Medical management of IPEX syndrome cannot, therefore, be based on genotype or presentation. In chapter 3 we employed whole exome sequencing to look for causal variant(s) in a patient with diabetes (diagnosed aged 7 weeks) and autoimmune lymphoproliferative disease. This identified recessively inherited causative variants in LRBA. We then used targeted next generation sequencing (NGS) to screen a large cohort of patients (n=169) and identified an additional 8 probands and an affected family member. This confirms the role of LRBA as a neonatal diabetes gene, bringing the total number of genes to 25. In chapter 4, we assessed if immunoglobulin E (IgE) could be useful to identify patients with early-onset multisystem autoimmune disease caused by gain of function (GOF) STAT3 mutations. We showed that serum IgE was below the lower limit of the normal reference range (2KU/L) in all patients with STAT3 GOF (n=6), giving this threshold a sensitivity of 100% (95% CI: 54.1 – 100) and specificity 97.2% (95% CI: 96.2-97.9). We also found that IgE in patients with IPEX (n=16) was significantly higher than those with STAT3 GOF (p=0.002) suggesting it could be useful to identify IPEX from STAT3 GOF in non-consanguineous males with early-onset autoimmunity. The final concluding section summarises the key findings of each chapter, the impact of these findings and suggests future avenues for research. Identifying monogenic autoimmunity has enabled prenatal diagnoses, given families and clinicians knowledge on recurrence risk, and could enable targeted therapies to be employed. This body of work will enable better discrimination of monogenic autoimmunity from polygenic clustering of early-onset autoimmunity, and gives insights into the factors that determine disease phenotype and clinical course in monogenic autoimmunity. Gene discovery on the remaining patients will give new insights into the mechanisms of beta-cell autoimmunity and the regulation of the adaptive immune system and maintenance of immune tolerance.

The aim of this thesis is to explore the role of common genetic variation, identified through genome-wide association (GWA) studies, in human traits and diseases, using height as a model polygenic trait, type 2 diabetes as a model common polygenic disease, and maturity onset diabetes of the young (MODY) as a model monogenic disease. The wave of the initial GWA studies, such as the Wellcome Trust Case-Control Consortium (WTCCC) study of seven common diseases, substantially increased the number of common variants associated with a range of different multifactorial traits and diseases. The initial excitement, however, seems to have been followed by some disappointment that the identified variants explain a relatively small proportion of the genetic variance of the studied trait, and that only few large effect or causal variants have been identified. Inevitably, this has led to criticism of the GWA studies, mainly that the findings are of limited clinical, or indeed scientific, benefit. Using height as a model, Chapter 2 explores the utility of GWA studies in terms of identifying regions that contain relevant genes, and in answering some general questions about the genetic architecture of highly polygenic traits. Chapter 3 takes this further into a large collaborative study and the largest sample size in a GWA study to date, mainly focusing on demonstrating the biological relevance of the identified variants, even when a large number of associated regions throughout the genome is implicated by these associations. Furthermore, it shows examples of different features of the genetic architecture, such as allelic heterogeneity and pleiotropy. Chapter 4 looks at the predictive value and, therefore, clinical utility, of variants found to associate with type 2 diabetes, a common multifactorial disease that is increasing in prevalence despite known environmental risk factors. This is a disease where knowledge of the genetic risk has potentially substantial clinical relevance. Finally, Chapter 5 approaches the monogenic-polygenic disease bridge in the direction opposite to that approached in the past: most studies have investigated genes mutated in monogenic diseases as candidates for harboring common variants predisposing to related polygenic diseases. This chapter looks at the common type 2 diabetes variants as modifiers of disease onset in patients with a monogenic but clinically heterogeneous disease, maturity onset diabetes of the young (MODY).

Quatre études de liaison génome entier ont mis en évidence une région commune de5,6 Mb dans la région du chromosome 5q15 liée à des traits associés à l’obésité, cette région incluant le gène de la prohormone convertase 1 (PCSK1). Une mutation Pc1 chez la souris a été associée à l’obésité, l’hyperphagie et à une augmentation de l’efficacité du métabolisme. La déficience complète en PCSK1 a été associée à une forme récessive rare d’obésité chezl’homme, et depuis 1997 seuls trois patients présentant cette déficience ont été décrits dans la littérature. Les porteurs de mutations délétères PCSK1 présentent des phénotypes sévères,incluant l’obésité, des hypoglycémies post-prandiales et des problèmes intestinaux ethormonaux. Contrairement aux observations faites chez la souris, les membres des famillesporteurs hétérozygotes ont été considérés comme cliniquement sains. Toutes ces études ontdésigné PCSK1 comme un gène candidat important pour l’obésité.Dans un premier temps, la contribution du gène PCSK1 au risque d’obésitépolygénique a été évaluée chez 13,659 individus d’origine européenne issus de huit cohortescas contrôles ou familiales indépendantes. Neuf variants fréquents couvrant 92% de lavariabilité génétique du locus ont été génotypés. Les méta-analyses des huit études pour levariant commun rs6232 et pour le cluster rs6234-rs6235 ont montré une associationreproductible avec l’obésité chez l’adulte et chez l’enfant (P=7.27x10-8 et P=2.31x10-12respectivement). Le rs6232 était associé à une augmentation du risque d’obésité de 34%, alorsque le cluster rs6234-rs6235 augmentait le risque d’obésité de 22%. Les analysesfonctionnelles ont montré une diminution significative de 10,4% de l’activité catalytique de laprotéine PC1/3 pour le N221D, et une diminution non significative de l’activité catalytique dela protéine PC1/3 pour le cluster Q665E/S690T.L’implication du gène PCSK1 dans l’obésité monogénique a ensuite été entreprise parle séquençage des exons de PCSK1 chez 845 sujets obèses non-consanguins d’origineeuropéenne,. Huit nouvelles mutations non-synonymes ont été identifiées. L’étude des conséquences fonctionnelles des mutations détectées sur la protéine PC1/3 a montré que62.5% de ces mutations détectées étaient prédites délétères par les analyses in silico et 87.5%de ces mutations avaient un effet sur l’auto-activation ou sur l’activité enzymatique de PC1/3in vitro. Dans le but d’estimer le degré de pénétrance pour ces sept mutations pathogéniques,6,060 obèses et 6,274 sujets minces ont été génotypés, démontrant un enrichissement par sixde ces mutations PCSK1 chez les sujets obèses (P=0.007). Cette étude a mis en évidence pourla première fois une augmentation du risque d’obésité chez les porteurs hétérozygotes de mutations perte de fonction du gène PCSK1, confirmant un mode de transmission codominantde l’obésité avec une pénétrance incomplète. La pénétrance de l’obésité a été105estimée à 54.5% pour les porteurs hétérozygotes de mutations délétères PCSK1. Unedéficience partielle en PCSK1 pourrait expliquer environ 0.83% des formes extrêmesd’obésité et représenter la seconde forme la plus fréquente d’obésité monogénique après ladficience en MC4R.Pour conclure, en plus des formes syndromiques très rares d’obésité dues à unedéficience complète en PCSK1, ce travail a permis de démontrer le rôle des variants codantsfréquents non-synonymes dans le risque d’obésité, ainsi que l’importance longtempsinsoupçonné d’une déficience partielle en PCSK1 dans les formes monogéniques d’obésité. / Four whole genome studies basing on positional cloning approach revealed a region ofchromosome 5q linked to traits related to obesity, this region contained the gene coding forthe prohormone convertase 1 named PCSK1. Pc1 mutation in mice has been associated withobesity, hyperphagia and increased metabolic efficiency. In human, PCSK1 deficiency is amonogenic form of obesity. The first case of complete PCSK1 deficiency has been identifiedin 1997 and since two other cases were discovered. Deleterious PCSK1 mutations carrierswere either homozygous or compound heterozygous and presented severe phenotypes, such asobesity, intestinal troubles and endocrine disorders. Surprisingly, the family members whowere heterozygous for these mutations appeared clinically unaffected. Overall of these studieshighlighted PCSK1 as a candidate gene for obesity.We have therefore decided to assess the contribution of PCSK1 gene to polygenicobesity risk. To assess the contribution of PCSK1 to polygenic obesity risk, we genotyped tagsingle nucleotide polymorphisms in a total of 13,659 European individuals from eightindependent case-control or family-based cohorts. The non-synonymous variants rs6232,encoding N221D, and cluster rs6234-rs6235, encoding the Q665E-S690T pair, wereconsistently associated with obesity in adults and children (P=7.27 x 10-8 and P=2.31 x 10-12,respectively). Functional analysis revealed a significant impairment of the N221D mutant onPC1/3 protein catalytic activity.In continuity of this study we decided to assess the involvement of PCSK1 gene inmonogenic obesity, knowing that only three cases of complete PCSK1 deficiency have beenreported up to now. The objectives of this study were to evaluate the prevalence of rarePCSK1 mutations contributing to human obesity and to investigate the mode of inheritance ofobesity in the context of PCSK1 deficiency. We sequenced exons of the PCSK1 gene in 845non-consanguineous extremely obese subjects of European origin and we identified eightnovel PCSK1 non-synonymous mutations in eight carriers, all heterozygous. Wecharacterized the functional consequences of the detected mutations on PC1/3 protein and wefound that 62.5% of mutations detected were predicted to be deleterious in silico and werevealed that 87.5% of mutations had an effect on the autoactivation or on the enzymaticactivity of PC1/3 in vitro. In order to estimate the degree of penetrance for the sevenpathogenic mutations, we genotyped 6,060 obese and 6,274 lean subjects. We assessed a 6-fold enrichment of these PCSK1 mutations in obese subjects (P = 0.007). We provided thefirst evidence of an increased obesity risk in heterozygous carriers of loss of functionmutations in PCSK1 gene, confirming a co-dominant mode of transmission of obesity withincomplete penetrance for this gene. The penetrance of obesity was estimated to 54.5% for108heterozygous carriers of deleterious PCSK1 mutations. Partial PCSK1 deficiency mightexplain ~ 0.83% of extreme obesity.To conclude, in addition of the syndromic forms of obesity due to a complete PCSK1deficiency, we provided the strong evidence of the contribution of common non-synonymousvariants in obesity risk and we highlighted that a partial PCSK1 deficiency is associated withan increased risk of obesity.

Monogenic diseases make excellent models for the study of gene functions and basal cellular mechanisms in humans. The aim of this thesis was to elucidate how genetic mutations affect the basal cellular mechanisms in the monogenic diseases Nerve growth factor (NGF) dependent pain insensitivity and Iron-Sulphur cluster assembly protein U (ISCU) myopathy. NGF dependent pain insensitivity is a rare genetic disorder with clinical manifestations that include insensitivity to deep pain, development of Charcot joints, and impaired temperature sensation but with no effect on mental abilities. The disease is caused by a missense mutation in the NGFβ gene causing a drastic amino acid substitution (R221W) in a well-conserved region of the protein. NGF is secreted in limited amounts by its target tissues and is important for the development and maintenance of the cholinergic forebrain neurons as well as the sensory and sympathetic neurons. To reveal the underlying mechanisms of disease we performed functional studies of the mutant NGF protein. We could show that mutant NGF was unable to induce differentiation of PC12 cells as a consequence of impaired secretion. Furthermore, mutant NGF had different intracellular localisation compared to normal NGF and resided mostly in its unprocessed form proNGF. Mature NGF and proNGF have different binding properties to the receptors TrkA and p75. Individuals with mutations in TRKA are, aside from pain insensitive mentally affected; therefore it has been proposed that the R221W mutation mainly affects the interaction with p75. In agreement with this, we could show that R221W NGF was able to bind and activate TrkA whereas the interaction with p75 was impaired as compared to normal NGF. ISCU myopathy is a monogenic disease where the affected patients suffer from severe exercise intolerance resulting in muscle cramps and sometimes severe lactic acidosis. The disease is caused by a point mutation in the last intron of the Iron sulphur cluster assembly gene, ISCU, resulting in the inclusion of a part of the intron in the mRNA. ISCU functions as a scaffold protein in the assembly of iron-sulphur (Fe-S) clusters important for electron transport in Kreb’s cycle and the respiratory chain. We have shown that ISCU is vital in mammals since complete knock-down of Iscu in mice results in early embryonic death. The deletion of ISCU homologous in lower organisms has also been shown fatal. In spite this central role in energy metabolism the disease is restricted to the patient’s skeletal muscles while other energy demanding organs seem unaffected. To address this contradiction we examined if tissue-specific differences in the splicing of mutant ISCU could explain the muscle-specific phenotype. We could show that the splicing pattern did, indeed, differ with more incorrectly spliced ISCU in muscle compared to other tissues. This was accompanied by a decrease in Fe-S containing proteins in muscle, while no decrease was observed in other tissues. Alternative splicing is more common then previously thought and may depend upon interacting factors and/or differences in the surrounding milieu. To reveal plausible mechanisms involved in the tissue-specific splicing we identified nuclear factors that interacted with the region where the mutation was located. Five interacting factors were identified, out of which three affected the splicing of ISCU. PTBP1 was shown to repress the incorrect splicing while IGF2BP1 and RBM39 repressed the formation of normal transcript and could also counteract the effect of PTBP1. IGF2BP1 was the only factor that showed higher affinity to the mutant sequence making it a possible key factor in the incorrect splicing of the mutant ISCU gene. Together, these results offer important insights into the cellular mechanisms causing these diseases. We found impaired secretion and inaccurate sorting of NGF to be cellular mechanisms contributing to NGF dependent pain insensitivity while tissue-specific splicing of ISCU was found to be the event contributing to the phenotype of ISCU myopathy.

Type 2 diabetes (T2D) and fasting plasma glucose (fpg) levels have distinct genetic components which are as yet only modestly understood. Understanding the genetics of this complex disorder and its related traits is likely to be of significant benefit to the field. Not only will it shed light on critical genes, pathways and mechanisms of regulation, but it may also contribute to the development of pharmaceutical anti-hyperglycaemic agents via the identification of key therapeutic targets. Therefore the aim of this thesis was to utilise a broad, multidisciplinary approach to study the genetics of insulin secretion and signalling. Traditionally genes which harbour rare variants causing monogenic beta-cell dysfunction have also been found to harbour common variants associated with T2D and fpg. As genome-wide association studies (GWAS) identify an increasing number of common variants and genes, they also increase the number of genes which act as monogenic candidates. I screened G6PC2, a novel fpg associated gene, in patients with monogenic forms of beta-cell dysfunction and demonstrated that rare variants in this gene are unlikely to be a common cause of monogenic beta-cell dysfunction. Although GWAS have been of considerable benefit to our understanding of complex disease genetics, they are not without their own limitations, primarily concerning signal refinement. To try to overcome this barrier for T2D and fpg signals I established a pipeline for fluorescence activated cell sorting of human islets to obtain pure beta-cells. In these cells, I performed transcript profiling of genes falling within T2D and fpg associated loci, demonstrating how this approach, alongside physiological analysis, can be of benefit for GWAS researchers and provide a starting point for fine mapping. Access to human beta-cells also enabled me to follow up one novel fpg association signal, SLC2A2. Through analysis within this metabolically relevant tissue I was able to establish that the mechanism for increased fpg levels is unlikely to be mediated via a beta-cell pathway. Although GWAS have highlighted a number of key genes associated with beta-cell dysfunction; they have been far less successful at identifying genes associated with insulin resistance, another key component of T2D pathogenesis. Additional approaches, including the study of rodent models, may be required to study this aspect of T2D. PTEN is known to negatively regulate the insulin signalling pathway and adipose tissue specific Pten-/- animals were shown to be markedly insulin sensitive. To assess the role of PTEN in human insulin sensitivity I performed mRNA expression profiling of PTEN in human adipose tissue biopsies from subjects with T2D and matched controls, demonstrating that PTEN is significantly reduced in the subcutaneous adipose tissue of the former. This response is likely to be a compensatory mechanism to counteract muscular insulin resistance although further investigation needs to be performed to determine the mechanism of compensatory downregulation. These data provide insights into a number of aspects of T2D genetics, and demonstrate how a multidisciplinary approach is of benefit to T2D genetic research.

Le Lupus est une maladie systémique auto-immune complexe, caractérisée notamment par la perte de tolérance vis-à-vis d’antigènes nucléaires, qui affecte majoritairement les femmes à l’âge adulte. Cette maladie est hétérogène tant du point de vue clinique que génétique, ce qui a considérablement limité les progrès dans la compréhension de sa pathogénèse. Dans ce projet, nous avons mis à profit la technologie de séquençage de nouvelle génération de type « Whole Exome Sequencing (WES)» pour explorer la génétique du Lupus à l’échelle de familles individuelles. De plus, nous avons centré notre étude sur des formes à début pédiatrique qui sont cliniquement plus sévères que les formes adultes. Nous avons fait l’hypothèse que ce sous-groupe de patients devrait être enrichi en formes monogéniques. Nous avons ainsi inclus dans cette étude des cas familiaux de lupus ou des familles dans lesquelles il existait un patient lupique et des apparentés atteints de pathologies auto-immunes. Un séquençage WES a été réalisé dans ces familles et la validation des gènes identifiés par cette stratégie a fait l’objet de ce travail. Nous avons découvert une mutation hétérozygote gain de fonction du gène TMEM173 qui code STING, un adaptateur clé dans la signalisation de la détection des ADNs cytosoliques et la production d’interféron de type-I (IFN-I). Quatre membres de cette famille porteurs de la mutation présentent une pathologie auto-inflammatoire et auto-immune, dont un cas de Lupus, avec une expression clinique hautement variable. Une modélisation structurale in silico de la forme mutée prédisait initialement une stabilisation de la forme dimérique de STING avec pour conséquence possible une activation constitutive. En accord avec ces données, nous avons montré, en absence du ligand naturel, une activité constitutive de la forme mutée de STING exprimée dans une lignée cellulaire, ainsi qu’une localisation spontanée dans l’appareil de Golgi des fibroblastes de patients indiquant un état activé. Corrélativement, nous avons observé une concentration sérique élevée d’IFNa chez les patients ainsi qu’une signature transcriptionnelle fortement augmentée d’un panel de gènes induits par les IFN-I. A la suite de cette découverte, nous avons identifié de nouvelles mutations activatrices de STING chez des patients atteints de pathologies vasculaires et de fibroses pulmonaires sévères en relation avec un syndrome inflammatoire, et deux patients ont pu recevoir un traitement ciblant l’inhibition de la voie des IFN-I. Ce travail a permis de mettre en lumière le rôle clé de STING dans l’homéostasie du système immunitaire chez l’homme et de mieux comprendre la physiopathologie de maladies inflammatoires et auto-immunes, motivant la mise en place d’un traitement spécifique chez les patients. Au-delà de cette découverte importante pour ces malades, ce travail suggère que des variants rares et délétères permettent de décrire les causes génétiques de maladies auto-immunes dites complexes. / Lupus is a complex systemic autoimmune disease characterized by a loss of tolerance to nuclear antigens predominantly afflicting women of childbearing age. The disease is both clinically and genetically heterogeneous and this has greatly limited progress in understanding disease pathogenesis. In this project, we utilize next generation sequencing technology such as “Whole Exome Sequencing (WES)” to explore the genetics of lupus at the level of individual families. Furthermore, we focused on a rare subgroup of lupus patients, which develop the disease in childhood and usually present with a more severe clinical phenotype compared to the adult-onset form. We hypothesized that this subgroup would be enriched for monogenic forms of the disease. Thus, we recruited several families with early-onset in the proband and at least one other familial member presenting with lupus or associated autoimmune disease. WES was performed and the validation of the gene identified by this strategy is described. We identified a familial gain of function dominant mutation in TMEM173, encoding STING, a key adaptor molecule in the cytosolic DNA sensing pathway and Type-I interferon (IFN-I) production. The four affected family members demonstrate a systemic inflammatory and autoimmune condition, including lupus, with variable clinical expression. Structural modeling initially predicted a stabilized dimerization of the mutant STING and thereby a constitutive activation. In agreement with this, we found that the STING mutant spontaneously localizes in the Golgi of patient fibroblasts, and is constitutively active in the absence of its ligand in vitro. Accordingly, we observed elevated serum interferon activity and a IFN-I signature in peripheral blood. We have also identified several other patients with activating STING mutations presenting with vascular inflammation and or pulmonary fibrosis, and two patients received a treatment targeting the IFN-I signaling pathway. This discovery highlights the key role of STING in human immune homeostasis and its implication in severe inflammatory and autoimmune diseases, leading to new and more specific therapeutical approaches. Beyond these important findings with regards to patients, this work suggests that rare deleterious variants can describe the genetic cause of autoimmune pathologies previously defined as complex diseases.

La génétique, et par extension le séquençage de l’ADN, sont des outils qui ont transformé la compréhension des mécanismes impliqués dans la survenue de nombreuses pathologies, dont l’obésité. Les technologies aujourd’hui à notre disposition nous permettent de déterminer rapidement si un patient est ou non porteur d’un évènement génétique pouvant expliquer sa pathologie. L’une des techniques les plus utilisées en diagnostic aujourd’hui est le séquençage d’exome, ou WES, qui permet une excellente détection des mutations ponctuelles dans les régions codantes du génome. Mais d’autres évènements comme les copy number variants, ou CNV, peuvent également expliquer certaines pathologies, dont l’obésité, via entre autres les CNV de la région 16p11.2. Actuellement, la technique de référence pour la détection de ces copy number variants est l’analyse de puces CGH (Comparative Genomic Hybridization), mais celles-ci ne permettent pas de détecter des mutations non répertoriées au préalable lors de la création de la puce. Sur le principe, le séquençage d’exome peut lui aussi être utilisé pour détecter les CNV, mais son absence de couverture des régions non codantes du génome ne permet pas une détection efficace de ces CNV, car ceux-ci peuvent survenir sur l’ensemble du génome, en englobant des régions codantes et non codantes au sein d’un seul évènement. Le séquençage génome complet peut détecter ces deux types d’évènement, mais son cout est encore élevé ce qui freine sa démocratisation, et l’analyse de données associées nécessite d’importantes ressources informatiques, et le rend difficilement utilisable en diagnostic de routine en l’état actuel des choses. Il est donc pour l’instant nécessaire d’avoir recours à deux techniques différentes pour couvrir ces deux types d’évènements génétiques. Cela implique d’utiliser des échantillons parfois très précieux à deux reprises, de supporter les couts liés à deux techniques diagnostiques (d’environ 450 euros pour le séquençage d’exome au laboratoire et un cout un peu plus élevé pour une puce à ADN dans un laboratoire clinique), et d’allonger les temps de rendu de résultats et donc la durée d’établissement du diagnostic du patient. Cet état de fait nous a conduit à développer une technique de séquençage, que nous avons nommé CoDE-seq (Copy number variation Detection and Exome sequencing), et qui permettra la détection simultanée de ces deux types d’évènements, pour diminuer les temps d’établissement de diagnostics, leurs couts, et la quantité d’échantillon nécessaire. Ce travail a nécessité deux aspects : la mise au point technique et la mise au point analytique. La mise au point technique est passée par la création d’une nouvelle « capture », permettant une détection correcte des mutations ponctuelles de l’exome et des CNV de tout le génome. La mise au point analytique a consisté à définir la méthode à employer, et à permettre d’arriver à une détection fiable, à la fois sensible et spécifique, des CNV sur l’ensemble du génome. Une fois ces CNV identifiés, la question de leur signification fonctionnelle se pose également, et une seconde partie de ma thèse porte sur l’étude de cette signification fonctionnelle, via l’étude de la conformation spaciale de la chromatine et de l’influence des CNV sur celle-ci. / Genetics, and by extention DNA sequencing, are tools that have modified the understanding of the mechanisms involved in genetic diseases, like obesity. Today’s technology has allowed us to rapidly find if a patient carries a genetic event that may explain his/her pathology. One of the most used technology for diagnostic is exome sequencing, or WES, which enables an excellent detection of point mutations in coding regions of the genome. However other events, such as copy number variations, or CNV, can also explain some pathologies, like a severe form of obesity due to CNV in the chr16p11.2 region. Actually, the gold standard method for an accurate detection of CNV is array CGH, but this technology cannot detect new point mutations. Exome sequencing can be used to detect CNV, but the lack of coverage in non-coding regions limits CNV detection sensitivity. Of note, whole genome sequencing can detect both CNVs and point mutations, but it is still very expensive and needs huge informatics capacities, which is an obvious limitation for a routine diagnostic use.For now, we have had to use two different methods in order to accurately detect both CNVs and point mutations. In other words, we have had to use precious samples two times, to assume the cost of two different methods (which is nearly 450 euros in the laboratory for exome sequencing, and a bit more for array CGH in a clinical laboratory), and to consider the time of the realization of two different methods in order to achieve a complete diagnostic.In this context, we aimed to develop an innovative sequencing method, named CoDE-seq (Copy number variation Detection and Exome sequencing), which would allow us to simultaneously detect both CNVs and point mutations, in order to reduce the time of diagnostic, the cost, and the needed quantity of sample.This work included the method conception, and the data analysis steps. The method conception has been done through the creation of a new capture enabling the detection of point mutations in the exome, and CNVs all along the genome. Furthermore, the data analysis step included the choice of the bioinformatics methods to be used, in order to get a specific and sensitive CNV detection, all along the genome.We were also interested in the fonctional significance of identified CNV, and tried to decipher it by the study of chromatine spacial conformation and the influence of these CNV.

Les souris invalidées pour Hnf1a présentent un diabète sévère avec défaut d’insulino-sécrétion proche du phénotype observé chez les patients MODY3 (Maturity Onset Diabetes of te Young). Les mécanismes moléculaires responsables du diabète lié à la déficience en Hnf1a ne sont pas clairement compris. Dans ce travail, nous avons montré que des souris congéniques présentant la même délétion de Hnf1a dans des fonds génétiques différents présentent des phénotypes radicalement opposés. En effet, la déficience en Hnf1a conduit à un diabète sévère dans la plupart des lignées telles que 129, B6, BALB/c ou AJ (lignées sensibles). Mais à l’inverse, aucun diabète n’est observé chez les souris congéniques des lignées CBA et C3H malgré le défaut d’expression de Hnf1a (lignées résistantes). Les souris sensibles présentent une altération de leurs îlots de Langerhans avec notamment une diminution de taille de ces îlots. A l’inverse les souris résistantes présentent des îlots beta de taille normale malgré la déficience en Hnf1a. Nous avons pu montrer que le ou les variants génétiques présents dans les deux lignées résistantes agissent de façon dominante et, par un balayage du génome, nous avons identifié un locus majeur suppresseur du phénotype diabétique au niveau du chromosome 3. Ce locus majeur contient 11 gènes présentant des variations de SNP non synonymes et il interagit avec cinq autres loci ancillaires au niveau des chromosomes 4, 11 et 18. Notre travail montre finalement que les variations génétiques naturellement présentes dans les lignées de souris de laboratoire peuvent supprimer le phénotype diabétique lié à la déficience en Hnf1a. / Hnf1a-/- mice exhibit a severe diabetes mellitus due to a drastic defect in insulin secretion that closely resembles to the phenotype presented by MODY3 (Maturity Onset Diabetes of the Young type 3) patients. The molecular mechanisms responsible for the diabetes are still poorly understood. Here we show that congenic mice of different genetic backgrounds carrying the same Hnf1a deletion presented with drastically different phenotypes. Hnf1a-deficiency led to severe diabetes when introgressed into 129, B6, BALB/C or A/J genetic backgrounds (sensitive strains). Conversely, when the same null mutation was introgressed into CBA or C3H genetic backgrounds (resistant strains), the diabetic phenotype was suppressed. In sensitive strains, pancreatic islets did not increase in size compared to control animals and on the other hand average islet-size growth was normal in resistant strains. The genetic variations naturally present in these two resistant strains acted in a dominant way and a genome scan analysis led to the identification of a major suppressor locus on chromosome 3 that accounted for more than 60% of the variance of glycemia. The major locus contained 11 genes with non-synonymous SNPs changes and it interacted with 5 additional ancillary loci on chromosomes 4, 11 and 18. Our study demonstrated that the naturally occurring genetic variation present in distinct mouse laboratory strains is able to suppress the phenotype of a monogenic disorder.

Le diabète et l’obésité ont atteint de telles proportions dans le monde qu’on parle de pandémie. Les enjeux médicaux et financiers font de ces deux maladies un problème majeur de santé publique. Deux groupes de facteurs contribuent à ces deux maladies : l’environnement, et la génétique sur laquelle cette thèse s’appuie. Ce travail s’est focalisé sur les formes rares et monogéniques qui constituent les formes extrêmes de diabète de type 2 et d’obésité.Ces formes sont loin d’être totalement élucidées. Mon projet s’est concentré sur l’utilisation du séquençage de nouvelle génération (NGS) pour identifier de manière plus optimale, par rapport au séquençage classique de type Sanger, des mutations dans des gènes déjà connus chez de nouveaux patients introduits dans notre cohorte dans une optique de diagnostic. Le deuxième objectif était d’utiliser les techniques de NGS pour découvrir de nouveaux loci, liés à de nouvelles voies de signalisation impliquées dans la physiopathologie du diabète et de l’obésité.La première approche utilise une technique d’enrichissement par hybridation en phase liquide et se focalise sur 34 gènes associés à des formes monogéniques et polygéniques d’obésité. Le criblage a été réalisé sur 201 individus dans 13 familles dont la cause d’obésité est inconnue. Cette approche a mené à l’identification d’une mutation dans un gène connu de l’obésité : PCSK1. Cette mutation est causale, car elle conduit à un codon-stop au début de la protéine et n’est présente que chez des individus obèses. De plus, l’étude fonctionnelle a démontré une inhibition partielle de PC1/3 par la protéine tronquée et un possible impact sur la maturation et la sécrétion de l’enzyme.La deuxième approche se base sur une technique d’amplification par PCR dans des microgouttelettes lipidiques développée par la société Raindance, dont le premier test vise à réidentifier les mutations causales du diabète et/ou de l’obésité chez 40 patients. Cette approche a donné des résultats satisfaisants, car pour une large majorité de patients, les mutations causales ont pu être à nouveau identifiées. Seul un patient n’a pu être reconfirmé à cause des outils bioinformatiques actuels qui restent limités dans la détection des indels complexes. Parmi les 39 patients identifiés, 3 d’entre eux sont potentiellement porteurs de plusieurs mutations causales. Cette technique pourrait être envisagée dans le domaine clinique, car elle permet une approche multigénique en fournissant un diagnostic rapide, moins couteux et qualitativement aussi bon que le séquençage Sanger.La troisième approche met en jeu le séquençage de l’exome entier (WES) chez 4 individus où la famille entière s’est précédemment révélée négative pour tous les gènes connus du diabète. Cette approche a permis la découverte d’un 13e gène du MODY, KCNJ11, et confirme le large spectre phénotypique qui va du diabète néonatal au MODY selon les mutations. La difficulté majeure de cette technique est le filtrage des variants en vue d’aboutir à une seule mutation causale (ou éventuellement plusieurs sur un même gène) pour identifier de nouveaux gènes du MODY. La stratégie utilisée combinait à la fois un filtrage bioinformatique, avec par exemple des filtres sur la coségrégation familiale et sur des bases de SNPs référencés, et un filtrage biologique, avec l’utilisation d’une technique de génotypage haut débit. En conclusion, ce travail a permis de tirer parti des avancées technologiques comme la capture, le séquençage ciblé de masse et le NGS pour élucider et améliorer le criblage des formes monogéniques de diabète et d’obésité. Cette amélioration de la compréhension des mécanismes moléculaires conduira peut-être au développement de meilleurs traitements comme la médecine personnalisée. On espère voir des améliorations directes pour le patient dans un futur proche, par exemple un diagnostic moléculaire plus rapide, plus sûr et plus exhaustif. / Diabetes and obesity have reached such proportions worldwide we are talking about pandemic. Both diseases are a major cause of mortality and multiple complications. Medical and financial issues are for both diseases a major public health problem. Two groups of factors contribute to these two diseases: environment, and genetics on which this thesis is based. This work focused on rare and monogenic forms which are extreme forms of type 2 diabetes and obesity.These forms are far from being fully understood. My project focused on the use of next generation sequencing (NGS) to identify more optimally, compared to conventional Sanger sequencing, mutations in already known genes among new patients in our cohort for diagnostic purposes. The second objective was to use NGS to discover new loci associated with new signaling pathways involved in the pathophysiology of diabetes and obesity.The first approach uses a liquid-phase hybridization technique and focuses on 34 genes associated with monogenic and/or polygenic obesity. The screening was carried out on 201 people in 13 families for which the cause of obesity is unknown. This approach led to the identification of a mutation in a known gene of obesity: PCSK1. This mutation is causal because it leads to a stop codon at the beginning of the protein and is present only in obese individuals. Additionally, functional studies have demonstrated partial inhibition of PC1/3 by the truncated protein and the possible impact on the processing and secretion of this enzyme. This study has been published published in the "International Journal of Obesity" newspaper.The second approach is based on a PCR amplification technique in lipid microdroplets developed by Raindance. The first test is to re-identify the causal mutations of diabetes and/or obesity in 40 patients. This approach has yielded satisfactory results because for a large majority of patients, the causative mutations have been identified again. Only one patient was unable to be reconfirmed because current bioinformatics tools are limited in the detection of complex indels. Of the 39 patients identified, 3 of them are potential carriers of several causative mutations. This technique could be considered in the clinical field because it allows a multigene approach by providing a rapid diagnosis, cheaper and with a quality similar to the gold standard Sanger sequencing. For us, the purpose of this technique is a fast and optimal clinical diagnosis in order to identify unsolved cases, which are candidates for exome sequencing. This second study was published in "Diabetes Care" journal.The third approach involves whole exome sequencing (WES) in 4 individuals where the whole family was previously tested negative for all known genes of diabetes. This approach led to the discovery of a thirteen MODY gene, KCNJ11, and confirms the broad phenotypic spectrum that goes from neonatal diabetes to MODY depending on the mutations. The major difficulty with this technique is filtering variants in order to get a single causal mutation (or possibly several on the same gene) to identify new MODY genes. The strategy we used combined both a bioinformatics filter for example with filters on family cosegregation and on SNP databases and a biological filter, with the use of a technique for high-throughput genotyping. This pioneering study in the use of NGS to identify new genes of MODY has been published in "PLoS ONE".In conclusion, this work took advantage of technological advances such as capture, targeted sequencing and NGS to elucidate and to improve the screening of monogenic forms of diabetes and obesity. This improved understanding of the molecular mechanisms may lead to the development of better treatments like personalized medicine. We hope to see direct improvements for patients in the near future, such as a more accurate, faster and more comprehensive molecular