Дисертації з теми "Variante somatica"

It is important to understand the entire spectrum of somatic variants to gain more insight into mutations that occur in different cancers for development of better diagnostic, prognostic and therapeutic tools. This thesis outlines our work in understanding somatic variant calling, improving the identification of somatic variants from whole genome and whole exome platforms and identification of biomarkers for lung cancer. Integrating somatic variants from whole genome and whole exome platforms poses a challenge as variants identified in the exonic regions of the whole genome platform may not be identified on the whole exome platform and vice-versa. Taking a simple union or intersection of the somatic variants from both platforms would lead to inclusion of many false positives (through union) and exclusion of many true variants (through intersection). We develop the first framework to improve the identification of somatic variants on whole genome and exome platforms using a machine learning approach by combining the results from two popular somatic variant callers. Testing on simulated and real data sets shows that our framework identifies variants more accurately than using only one somatic variant caller or using variants from only one platform. Short tandem repeats (STRs) are repetitive units of 2-6 nucleotides. STRs make up approximately 1% of the human genome and have been traditionally used as genetic markers in population studies. We conduct a series of in silico analyses using the exome data of 32 individuals with lung cancer to identify 103 STRs that could potentially serve as cancer diagnostic markers and 624 STRs that could potentially serve as cancer predisposition markers. Overall these studies improve the accuracy in identification of somatic variants and highlight the association of STRs to lung cancer.
Ph. D.

The rapid development of high-throughput sequencing technologies pushed forward the fields of medical genomics and precision medicine, creating many new applications for diagnostics and clinical studies that require high quality data and highly accurate analysis methods. Distinguishing errors from real variants in Next Generation Sequencing data is a challenge when systematic errors, random sequencing errors, germline variants or somatic variants at very low allele frequency are present in the same data. In the first part of this thesis, we developed a genotype callability filter (ABB) able to identify systematic variant calling errors that were not found by state-of-the art methods. This tool cleans false positive calls from somatic and germline variant callsets, as well as detects false gene-disease associations in case-control studies. Secondly, we developed a set of novel methods able to distinguish and correct sequencing and PCR errors with the use of molecular barcodes, permitting us to build error rate models for the detection of somatic mutations at extremely low allele frequencies in liquid biopsies. As final part of this thesis, we characterized mosaic mutations in a multi-tissue, multi-individual study using a cohort of thousands of samples from hundreds of healthy individuals.
El ràpid desenvolupament de les tecnologies de seqüenciació d’alt rendiment ha impulsat els camps de la genòmica mèdica i la medicina d’alta precisió, creant una gran varietat de noves aplicacions, les quals requereixen dades d’una qualitat excel·lent i mètodes d’anàlisi altament precisos. La distinció entre errors i variants reals en dades de seqüenciació de propera generació (NGS) és un repte quan hi ha errors sistemàtics o aleatoris mesclats amb variants germinals o somàtiques a freqüències al·lèliques molt baixes. En la primera part d'aquesta tesi, hem desenvolupat un filtre per al genotipatge de variants (ABB) capaç d'identificar errors sistemàtics durant el procés de detecció de variants que altres mètodes convencionals no poden trobar. Aquesta eina filtra falsos positius del conjunt de variants finals en estudis de variacions somàtiques i germinals, així com també detecta falses associacions de malalties gèniques en estudis de casos-controls. En segon lloc, hem desenvolupat un conjunt de nous mètodes capaços de distingir i corregir els errors de seqüenciació i PCR amb l’ús d’identificadors moleculars. Aquests ens permeten modelar les taxes d’error i conseqüentment detectar mutacions somàtiques a freqüències al·lèliques extremadament baixes en l’anàlisi de biòpsies líquides. Per finalitzar aquesta tesi, hem caracteritzat les mutacions mosaiques en un estudi multi-teixit multi-individu utilitzant una cohort de centenars d'individus sans amb milers de mostres.

Germline copy-number variants (CNVs), as well as somatic copy-number alterations (CNAs), play an important role in many phenotypic traits, including genetic diseases and cancer. Next Generation Sequencing (NGS) allows accurate detection of short variants, but reliable detection of large-scale CNVs in NGS data remains challenging. In this work, I address this issue and describe a novel statistical method for detection of CNVs and CNAs implemented in the tool called ClinCNV. I present analytical performance measures of “ClinCNV” in different datasets, compare it with the performance of other existing methods, and show the advantages of ClinCNV. ClinCNV is already implemented as a part of the diagnostics pipeline at the Institute of Medical Genetics and Applied Genomics (IMGAG), Tuebingen, Germany. ClinCNV has the potential to facilitate molecular diagnostic of genetic-based diseases as well as cancer through accurate detection of copy-number variants.
Las variantes en el número de copias genéticas, tanto en estado germinal (CNV) como en somático (CNA), juegan un papel muy importante en muchos rasgos fenotípicos y están frecuentemente relacionadas con una gran variedad enfermedades genéticas y cáncer. Aunque la secuenciación de próxima generación (NGS) permite detectar variantes cortas con una gran precisión, la correcta detección de CNVs a gran escala con datos de secuenciación sigue siendo un gran desafío. En esta tesis, me centro en abordar este problema y describo un nuevo método estadístico para la detección de CNV y CNA englobado en una nueva herramienta llamada ClinCNV. Para el análisis del rendimiento de ClinCNV y demostrar las ventajas de este nuevo algoritmo, comparamos nuestra herramienta con otras existentes en distintos conjuntos de datos. Por otra parte, ClinCNV ya está implementado como parte del sistema de trabajo de diagnóstico en el Instituto de Genética Médica y Genómica Aplicada (IMGAG) en Tuebingen (Alemania). En resumen, ClinCNV tiene el potencial de facilitar el diagnóstico molecular de enfermedades genéticas y cáncer mediante la precisa detección de variantes en el número de copias genéticas.

Somatic mutations are those that arise after the zygote is formed and are therefore inherited by a fraction of the cells of an individual. Their relevance in certain skin diseases has been known for almost half a decade and cancer, the most common disease caused by somatic mutations, has been extensively studied. Yet, their prevalence in healthy individuals as well as their putative role in other human disorders such as neurodegenerative diseases are still unanswered questions. Furthermore, accurate detection of somatic variants from bulk sequencing data still poses a technical challenge. This work focuses on detecting and circumventing the biases that hinder their identification. Using this knowledge, we identified somatic point mutations in the exomes of five different tissues from sporadic Parkinson disease patients. We also assessed the detection of somatic copy number variants from array CGH data using two tissues from Alzheimer disease patients. Finally, we participated in the identification of somatic variants in an extensive genomic dataset from a neurotypical individual.
Las mutaciones somáticas son aquellas que surgen tras la formación del cigoto y son, por tanto, heredadas por una fracción de las células de un individuo. Su importancia en algunas enfermedades cutáneas se conoce desde hace casi medio siglo. El cáncer, la enfermedad más común causada por mutaciones somáticas, se ha estudiado extensamente. Sin embargo, su prevalencia en individuos sanos, así como su potencial relevancia en otras afecciones humanas, como las enfermedades neurodegenerativas, son cuestiones todavía por resolver. Asimismo, detectar variantes somáticas con precisión en datos de secuenciación de muestras homogeneizadas sigue siendo complejo técnicamente. Este trabajo se centra en la detección y resolución de los sesgos que dificultan su identificación. Aplicando este conocimiento, identificamos mutaciones somáticas de una sola base en datos de secuenciación del exoma de cinco tejidos diferentes de pacientes de la enfermedad de Parkinson. También evaluamos la detección de variantes de número de copia somáticas en datos de array CGH de dos tejidos de pacientes de Alzheimer. Finalmente, participamos en la identificación de variantes somáticas en un amplio conjunto de datos genómicos de un individuo neurotípico.

Insertion and deletion (indel), a common form of genetic variation, has been shown to cause or contribute to human genetic diseases and cancer. Despite this importance and being the second most abundant variant type in the human genome, indels have not been studied as much as the single nucleotide polymorphism (SNP). With the advance of next-generation sequencing technology, many indel calling tools have been developed. However, performance comparison of commonly used tools has shown that (1) the tools have limited power in identifying indels and there are significant number of indels undetected, and (2) there is significant disagreement among the indel sets produced by the tools. These findings indicate the necessity of improving the existing tools or developing new algorithms to achieve reliable and consistent indel calling results. Two indels are biologically equivalent if the resulting sequences are the same. Storing biologically equivalent indels as distinct entries in databases causes data redundancy and misleads downstream analysis. It is thus desirable to have a unified system for identifying and representing equivalent indels. This dissertation describes UPS-indel, a utility tool that creates a universal positioning system for indels so that equivalent indels can be uniquely determined by their coordinates in the new system. Results show that UPS-indel identifies more redundant indels than existing algorithms. While mapping short reads to the reference genome, a significant number of short reads are unmapped and excluded from downstream analyses, thereby causing information loss in the subsequent variant calling. This dissertation describes Genesis-indel, a computational pipeline that explores the unmapped reads to identify missing novel indels. Results analyzing sequence alignment of 30 breast cancer patients show that Genesis-indel identifies many novel indels that also show significant enrichment in oncogenes and tumor suppressor genes, demonstrating the importance of rescuing indels hidden in the unmapped reads in cancer and disease studies. Somatic mutations play a vital role in transforming healthy cells into cancer cells. Therefore, accurate identification of somatic mutations is essential. Many somatic mutations callers are available with different strengths and weaknesses. An ensemble approach integrating the power of the callers is warranted. This dissertation describes SomaticHunter, an ensemble of two callers, namely Platypus and VarDict. Results on synthetic tumor data show that for both SNPs and indels, SomaticHunter achieves recall comparable to the state-of-the-art somatic mutation callers and the highest precision, resulting in the highest F1 score.
Doctor of Philosophy
Insertion and deletion (indel), a common form of genetic variation in the human genome, is associated with genetic diseases and cancer. However, indels are heavily understudied due to experimental and computational challenges. This dissertation addresses the computational challenges in three aspects. First, the current approach of representing indels is ambiguous and causes significant database redundancy. A universal positioning system, UPS-indel, is proposed to represent equivalent indels unambiguously and the UPS-indel algorithm is theoretically proven to find all equivalent indels and is thus exhaustive. Second, a significant number of indels are hidden in DNA reads not mapped to the reference genome. Genesis-indel, a computational pipeline that explores the unmapped reads to identify novel indels that are initially missed, is developed. Genesis-indel has been shown to uncover indels that can be important genetic markers for breast cancer. Finally, mutations occurring in somatic cells play a vital role in transforming healthy cells into cancer cells. Therefore, accurate identification of somatic mutation is essential for a better understanding of cancer genomes. SomaticHunter, an ensemble of two sensitive variant callers, is developed. Simulated studies using whole genome and whole exome sequences have shown that SomaticHunter achieves recall comparable to state-of-the-art somatic mutation callers while delivering the highest precision and therefore resulting in the highest F1 score among all the callers compared.

Somatic single nucleotide variants (SNVs) are mutations resulting from the substitution of a single nucleotide in the genome of cancer cells. Somatic SNVs are numerous in the genomes of most types of cancers. SNVs can contribute to the malignant phenotype of cancer cells, though many SNVs likely have negligible selective value. Because many SNVs are selectively neutral, their presence in a measurable proportion of cells is likely due to drift or genetic hitchhiking. This makes SNVs an appealing class of genomic aberrations to use as markers of clonal populations and ultimately tumour evolution. Advances in sequencing technology, in particular the development of high throughput sequencing (HTS) technologies, have made it possible to systematically profile SNVs in tumour genomes. We introduce three probabilistic models to solve analytical problems raised by experimental designs that leverage HTS to study cancer biology. The first experimental design we address is paired sequencing of normal and tumour tissue samples to identify somatic SNVs. We develop a probabilistic model to jointly analyse data from both samples, and reduce the number of false positive somatic SNV predictions. The second experimental design we address is the deep sequencing of SNVs to quantify the cellular prevalence of clones harbouring the SNVs. The key challenge we resolve is that allele abundance measured by HTS is not equivalent to cellular prevalence due to the confounding issues of mutational genotype, normal cell contamination and technical noise. We develop a probabilistic model which solves these problems while simultaneously inferring the number of clonal populations in the tissue. The final experimental design we consider is single cell sequencing. Single cell sequencing provides a direct means to measure the genotypes of clonal populations. However, sequence data from a single cell is inherently noisy which confounds accurate measurement of genotypes. To overcome this problem we develop a model to aggregate cells by clonal population in order to pool statistical strength and reduce error. The model jointly infers the assignment of cells to clonal populations, the genotype of the clonal populations, and the number of populations present.
Science, Faculty of
Graduate

The human brain is one of the most exquisite structures in nature, featuring extreme functional complexity and capacities that allow for advanced cognitive abilities. During the development of the human brain, neural progenitors undergo massive proliferation, which is known to inevitably result in spontaneous mutations; yet the degree of somatic mosaicism within the human brain is unexplored. Several hypotheses have been proposed that various types of somatic mosaicism may serve as an adaptive mechanism to diversify neurons and thereby promote the functional complexity of human brains. Previously proposed mechanisms to increase somatic mosaicism within the brain include elevated somatic LINE-1 element retrotransposition, and the creation of somatic aneuploidy during neurogenesis. On the other hand, genomic diversity needs to be balanced by genomic stability, in order to protect against deleterious mutations that reduce the fitness of the cells, or oncogenic mutations that might promote cancers. In fact, brain-specific somatic mutations have also been proposed to contribute to the unexplained burden of neurological diseases. To directly study genomic variability from cell-to-cell within the human brain, we developed a method to isolate and amplify single neuronal genomes from postmortem and surgically resected human brain tissues. We quantified the frequency of somatic LINE-1 retrotransposition events and aneuploidy in human cortical neurons, and found that the frequencies of both are low, with no sign of brain-specific elevation, arguing against the hypotheses that these two mutational sources are obligate generators of neuronal diversity. Additionally, aneuploidy analysis was performed on bulk and single cortical cells from a hemimegalencephaly brain. Hemimegalencephaly is an asymmetrical brain overgrowth syndrome caused by somatic mutations in brain. Single-cell analysis identified an unexpected mosaic tetrasomy of chromosome 1q, affecting both neuronal and glial populations, as a genetic cause of hemimegalencephaly. These results demonstrate that single-neuron sequencing allows systematic assessment of genomic diversity in the human brain and the identification and characterization of pathogenic somatic mutations underlying neurological disorders.

The identification and analysis of genomic variation across individuals has been central in biology, first through comparative genomics to answer evolutionary questions, and then in the context of biomedicine, where it is actually becoming central to the study of most diseases. Next generation sequence technologies are allowing the systematic analysis of thousands of different types of genetic variation, enhancing the identification of disease markers and the understanding of the molecular basis of disease. For the past years, there has been a burst of new methodology for genome analysis around diseases coming from hundreds of groups around the world. Specific computational methods and strategies are being designed and improved around the identification and interpretation of genomic variation. The identification and classification of different types of genomic variants in the context of biomedicine is a key and foundational step for the development of a personalized medicine. This has been particularly central in the field of cancer genomics, which has based the research of the past ten to fifteen years in the sequencing of genomic DNA, and the identification and interpretation of (mostly) somatic and germline variation. Throughout these years, a large number of methods for variant detection have been developed with different action ranges. Despite all these developments, the identification of genomic variants has still room for improvement, not only at the level of sensitivity and specificity, but also at the computational level. Given the emergence of many initiatives for personalized medicine around the world, and the expected number of genomes that will have to be analyzed within health care systems, we require robust algorithms, designed together with a matching implementation that will minimize the computational costs of the analysis. With this aim, during this thesis, I have pushed and designed and implemented an algorithm for the efficient processing of genomic data, in close collaboration with computer scientists of our center that defined the implementation, focusing on lowering the energy and the time of the analysis. This methodology, which relies on a reference free approach of read classification, has been protected with a patent, and is being used as the foundation for the development of SMuFin2, a more accurate and computationally efficient version of the initial SMuFin from 2014. We here show that our method is able to process whole genome sequences very fast and with a minimal energy consumption, compared with existing methods, and that has great potential for the identification of all ranges of variants, including insertions of non-human DNA. Further developments on SMuFin2 are needed to finally assess its full variant calling capabilities. Despite their great importance and their clear role in the biology of the cell, somatic variation that occurs in healthy tissues has remained diffuse in their roles. In the case of development, some hypotheses have been proposed to explain the observed somatic DNA damage that occurs during brain development (e.g., replication stress). But the real impact and the underlying mechanisms of this somatic variation are not yet understood. In order to seed light on the type and potential functional impact of somatic variation in brain development, we established a new collaboration to identify, and describe somatic DNA rearrangements induced by Pgbd5 during brain development and adult state in 36 mice neural tissue samples. The detection of somatic variants in healthy tissues presents more challenges than in the cancer scenario, where a variant is present in a significant number of cells and is easier to detect. We have identified, classified and interpreted the landscape of somatic variation in neural development and identified interesting differences between adult and embryonic variation load, and specific types of variants, as the potential result of the activity of these transposase-like genes.

The most recent cancer classification from NIH includes ~200 types of tumor that originates from several tissue types (http://www.cancer.gov/types). Although macroscopic and microscopic characteristics varies significantly across subtypes, the starting point of every cancer is believed to be a single cell that acquires DNA somatic alterations that increases its fitness over the surrounding cells and makes it behave abnormally and proliferate uncontrollably. Somatic mutations are the consequence of many possible defective processes such as replication deficiencies, exposure to carcinogens, or DNA repair machinery faults. Mutation development is a random and mostly natural process that frequently happens in every cell of an individual. Only the acquisition of a series of subtype-specific alterations, including also larger aberrations such as translocations or deletions, can lead to the development of the disease and this is a long process for the majority of adult tumor types. However, genetic predisposition for certain cancer types is epidemiologically well established. In fact, several cancer predisposing genes where identified in the last 30 years with various technologies but they characterize only a small fraction of familial cases. This work will therefore cover two main steps of cancer genetics and genomics: the identification of the genes that somatically changes the behavior of a normal human cell to a cancer cell and the genetic variants that increase risk of cancer development. The use of publicly available datasets is common to all the three results sections that compose this work. In particular, we took advantage of several whole exome sequencing databases (WES) for the identification of both driver mutations and driver variants. In particular, the use of WES in cancer predisposition analysis represents one of the few attempts of performing such analysis on genome-wide sequencing germline data.

Although the understanding of genetic predisposition to prostate cancer (PCa) has been improved through genome-wide association studies (GWAS), little is known about the biological implication of germline variants residing in coding or non-coding regions in cancer development and progression. Our hypothesis is that inherited variants may predispose to specific early recurrent genomic events observed in PCa adenocarcinomas, possibly in the context of variable androgen receptor (AR) signaling that changes during a man’s lifetime. Recent in silico analysis by our group on potential association between germline variants and PCa specific somatic lesions identified a non-coding polymorphic regulatory element at the 7p14.3 locus associated with DNA repair and hormone regulated transcript levels and with an early recurrent prostate cancer specific somatic mutation in the Speckle-Type POZ protein (SPOP) gene (OR=5.54, P=1.22e-08) in human prostate tissue data. In order to functionally characterize the polymorphic 7p14.3 locus (rs1376350, single nucleotide polymorphism, G>A), we set up to establish isogenic cell lines harboring the minor allele by using the CRISPR/Cas9 system. In parallel, CRISPR/Cas9 system was used to knock out different portion of the region encompassing the 7p14.3 variant and to eliminate transcription factors (TFs) binding sites that were identified from previous in silico analysis (i.e. AR and CCAAT/Enhancer Binding Protein (C/EBP) beta (CEBPβ)). The transcriptomes of edited pools and edited single clones from macrodeletion (731 bp), microdeletion (50 bp) and alterations of TFs binding sites were analyzed and compared to the transcriptomes of isogenic cells heterozygous (A/G) and homozygous (A/A) for the minor allele A of the risk variant rs1376350 (with or without AR overexpression). These data identified a set of genes scattered throughout the genome with the same pattern of deregulation suggesting the implication of the variant on the regulation of genes residing in different chromosomes. Additionally, ChIP-qPCR experiments for histone modification supported the identification of the 7p14.3 locus with enhancer activity. Furthermore, ChIP-qPCR of histone mark associated with transcriptional activation or repression in isogenic cells harboring the minor allele A upon AR overexpression showed that the activity of the locus is higher for the minor allele A compared to G, independently from AR activation. Despite the limitations of our model and the current lack of validation in other cells, we confirmed that some of the differentially expressed genes that emerged from the comparative analysis of edited cells are deregulated in human normal and tumor prostate samples as well. This work is a proof of concept of germline predisposition to molecularly distinct cancer subclasses and has the potential to nominate new mechanisms of cancer development. Future work aims to elucidate the mechanisms implicated in the deregulation of the transcriptome by combining the information obtained until now with potential new players that we expect to identify by Mass Spectrometry experiments. To clarify the link between the 7p14.3 variant and the somatic mutations in SPOP, we plan to express mutant SPOP in isogenic cells harboring the minor allele and to asses DNA damage response upon overexpression or silencing of TFs binding at and around the rs1376350 variant. My work is an example of how the CRISPR/Cas9 system can be used to develop a technical framework with convergent approaches to functionally characterize polymorphic regulatory regions including but not limited to the establishment of isogenic cells upon single nucleotide editing.

Although the understanding of genetic predisposition to prostate cancer (PCa) has been improved through genome-wide association studies (GWAS), little is known about the biological implication of germline variants residing in coding or non-coding regions in cancer development and progression. Our hypothesis is that inherited variants may predispose to specific early recurrent genomic events observed in PCa adenocarcinomas, possibly in the context of variable androgen receptor (AR) signaling that changes during a man’s lifetime. Recent in silico analysis by our group on potential association between germline variants and PCa specific somatic lesions identified a non-coding polymorphic regulatory element at the 7p14.3 locus associated with DNA repair and hormone regulated transcript levels and with an early recurrent prostate cancer specific somatic mutation in the Speckle-Type POZ protein (SPOP) gene (OR=5.54, P=1.22e-08) in human prostate tissue data. In order to functionally characterize the polymorphic 7p14.3 locus (rs1376350, single nucleotide polymorphism, G>A), we set up to establish isogenic cell lines harboring the minor allele by using the CRISPR/Cas9 system. In parallel, CRISPR/Cas9 system was used to knock out different portion of the region encompassing the 7p14.3 variant and to eliminate transcription factors (TFs) binding sites that were identified from previous in silico analysis (i.e. AR and CCAAT/Enhancer Binding Protein (C/EBP) beta (CEBPβ)). The transcriptomes of edited pools and edited single clones from macrodeletion (731 bp), microdeletion (50 bp) and alterations of TFs binding sites were analyzed and compared to the transcriptomes of isogenic cells heterozygous (A/G) and homozygous (A/A) for the minor allele A of the risk variant rs1376350 (with or without AR overexpression). These data identified a set of genes scattered throughout the genome with the same pattern of deregulation suggesting the implication of the variant on the regulation of genes residing in different chromosomes. Additionally, ChIP-qPCR experiments for histone modification supported the identification of the 7p14.3 locus with enhancer activity. Furthermore, ChIP-qPCR of histone mark associated with transcriptional activation or repression in isogenic cells harboring the minor allele A upon AR overexpression showed that the activity of the locus is higher for the minor allele A compared to G, independently from AR activation. Despite the limitations of our model and the current lack of validation in other cells, we confirmed that some of the differentially expressed genes that emerged from the comparative analysis of edited cells are deregulated in human normal and tumor prostate samples as well. This work is a proof of concept of germline predisposition to molecularly distinct cancer subclasses and has the potential to nominate new mechanisms of cancer development. Future work aims to elucidate the mechanisms implicated in the deregulation of the transcriptome by combining the information obtained until now with potential new players that we expect to identify by Mass Spectrometry experiments. To clarify the link between the 7p14.3 variant and the somatic mutations in SPOP, we plan to express mutant SPOP in isogenic cells harboring the minor allele and to asses DNA damage response upon overexpression or silencing of TFs binding at and around the rs1376350 variant. My work is an example of how the CRISPR/Cas9 system can be used to develop a technical framework with convergent approaches to functionally characterize polymorphic regulatory regions including but not limited to the establishment of isogenic cells upon single nucleotide editing.

This thesis reports the cell culture establishment and a somatic cell selection system optimized for the isolation of chlorsulfuron-resistant variants in asparagus (Asparagus officinalis L.). The development of this cell selection system benefited the isolation of chlorsulfuron-resistant variants from an elite asparagus genotype. A cell culture system, suitable for somatic cell selection, was established for asparagus genotype CRD 168. Friable callus was initiated from etiolated shoots in darkness and used to produce a high density of single cells in suspension. Cell density was estimated based on a linear relationship with settled cell volume. A mean plating efficiency of 0.19 % was recorded between 1-4x10⁵ cells/Petri dish. In vitro cell selection techniques were developed to identify mutant asparagus cells with resistance to a sulfonylurea herbicide, chlorsulfuron. A few key aspects were important to achieve this: a cell culture system for cell selection was initially established; a toxic concentration for the complete growth inhibition of the wild type asparagus cells was defined; rare, resistant cell colonies were isolated and characterized; and chlorsulfuron-resistant plants were regenerated. From about 50 million cells, 165 cell colonies were isolated in the presence of 8 nM chlorsulfuron. Characterization of these selected cell colonies yielded 24 escapes, 98 unstable variants, and 43 stable-resistant variants. Callus cultures from 34 of these stable variants retained resistance following 11 months growth in the absence of the selection agent. Plants were regenerated from 36 of these stable herbicide-resistant variants. Six of these chlorsulfuron-resistant variants were screened for their degree of resistance to chlorsulfuron, cross resistance to other acetohydroxyacid synthase (AHAS) inhibiting herbicides and AHAS enzyme activity. Cross resistance to imazamox was evident in four of the resistant variants, while lack of cross resistance to metsulfuron methyl was observed in all six resistant variants. A varying degree of resistance to chlorsulfuron was observed among the resistant variants. Both in the original and secondary callus, an uninhibited AHAS enzyme activity in all six resistant variants was recorded in the presence of high chlorsulfuron concentration (70-140 nM), compared to the total inhibition in the wild type. One chlorsulfuron-resistant variant, R-45, was used to compare the biochemical and physiological basis of resistance with the wild type. The AHAS enzyme activity in the tissue culture and greenhouse foliage of R-45 was significantly higher in the presence of up to 280 nM chlorsulfuron compared with the wild type. Chlorsulfuron retention was considerably higher due to the reduction of epicuticular wax deposits on the foliage of R-45, in comparison with the wild type. Consequently, the resistant line absorbed at least 1.6 fold more chlorsulfuron than the wild type plants. Therefore, foliar application of 15 g a.i./ha Glean (commercial formulation of chlorsulfuron) produced typical symptoms of chlorosis in R-45, similar to the wild type, in the greenhouse plants. Somatic cell selection was carried out using two elite asparagus genotypes, CRD 74 and Clone X. Of the 33 rare cell colonies isolated from Clone X, 22 unstable variants and 6 escapes were discarded. All five remaining resistant variants produced plants. One of the stable-resistant variants (Clone X-24) was evaluated for resistance to chlorsulfuron. Both in vitro shoot cultures and greenhouse-grown plants of Clone X-24 showed increased resistance to chlorsulfuron compared with the wild type. The AHAS enzyme activity in the foliar extracts also showed the presence of higher enzyme activity in Clone X-24.

Les Lymphomes Primitifs Vitréo-Rétiniens (LPVR) représentent un sous-type de Lymphome Primitif du Système Nerveux Central (LPSNC). Ces hémopathies très rares sont caractérisées par leur localisation anatomique atypique, dans des sites physiologiquement dépourvus de lymphocytes B. Les lymphomes du SNC sont rattachés histologiquement aux Lymphomes B Diffus à Grandes Cellules (LBDGC) de type post-germinatif (ABC). L’objectif de notre étude était de définir le répertoire immunologique (chaînes lourdes et légères) des LPVR et des LPSNC, et de les comparer aux LBDGC. Nous avons mené une étude immunologique détaillée de ces tumeurs afin de rechercher des éléments de réponse expliquant ces localisations ectopiques. Notre projet, réalisé sur la plus grande série de LPVR à ce jour, a mis en évidence un biais de répertoire majeur, avec une sur-représentation massive du gène IGHV4-34 (63,6% des cas), significativement plus utilisé dans les LPVR comparativement aux LPSNC et aux LBDGC systémiques. Bien que la proportion de ce gène soit élevée dans d’autres SLP, cette fréquence n’a jamais été atteinte. Un subset a été décrit pour 50% des LPVR utilisant le gène IGHV3-7. Ces données suggèrent fortement l’implication d’un antigène dans leur développement. En conclusion, le LPVR représente un modèle surprenant et singulier de lymphome dirigé par l’antigène, dont l’identification offrirait des perspectives physiopathologiques et thérapeutiques prometteuses
Primary vitroretinal lymphoma (PVRL) is a high-grade lymphoma considered as a subtype of primary central nervous system lymphoma (PCNSL). Unusual localization is a feature of these rare entities. The vast majority of cases are diffuse large B cell lymphoma (DLBCL), mostly of activated B-cell (ABC). To investigate whether PVRLs display a specific IG repertoire contributing to explain their unusual localization, we analysed in detail the IG heavy and light chain sequences from PVRL and PCNSL cases, and we compared their repertoire to that of a publicly available IG heavy chain sequences dataset from systemic ABC-type DLBCLs. Our study was carried out on the largest PVRL series reported to date and showed that PVRL displayed a strikingly biased repertoire as the IGHV4-34 gene was used in 63.6% of cases. The frequency was significantly higher in PVRL compared to PCNSL and DLBCL. This gene has been repeatedly found to be preferentially used in various B-cell malignancies, but never to such an extent. Half of PVRL cases expressing the IGHV3-7 gene had stereotyped VH CDR3 features (subset). Altogether our data showed that PVRLs display a very biased IG repertoire strongly suggesting that antigen selection plays a major role in their development. Thus, PVRL display a highly restricted IG repertoire indicative of antigen selection, and distinct from that of PCNSL. Antigen(s) identification may provide promising perspectives in physiopathology concepts and therapeutic approaches

Malgré les bénéficies cliniques considérables de la prise en considération du contexte moléculaire tumoral, les thérapies ciblées présentent certaines limitations majeures. La première partie de ces travaux se concentre sur l’étude des inhibiteurs de tyrosine kinase ciblant l’EGFR dans les cancers bronchiques non à petites cellules. L’amélioration des méthodes de détection des biomarqueurs a été complétée par la caractérisation in vitro d’un mécanisme de résistance acquise non précédemment identifié. L’exposition de cellules pulmonaires à un agent mutagène puis une pression de sélection a ainsi mis en évidence la signature TBK1 et l’effet synergique de la co-inhibition s’est révélé intéressant. Le deuxième aspect concernent les inhibiteurs de PARP, incontournables dans la prise en charge des cancers de la sphère gynécologique. Ces thérapies reposant sur le principe de létalité synthétique, la présence de mutations pathogènes de BRCA1/2 est un pré-requis, illustrant la problématique des variants de signification inconnue. Face à la nécessité de leur caractérisation fonctionnelle, une étude transcriptionnelle des variants d’épissage par extraction de l’ARNm depuis les prélèvements FFPE a d’abord été réalisée. Puis, afin d’évaluer tous les variants, l’édition génomique a été développée comparant les efficacités d’édition d’un variant d’intérêt et d’un variant silencieux dans un contexte haploïde où ces gènes sont essentiels. La signification biologique de variants de BRCA1/2, et la généralisation de notre approche à l’ensemble des gènes suppresseurs de tumeur essentiels de nos cellules, peut ainsi être évaluée en 3 semaines, délai compatible avec les impératifs cliniques
Despite significant clinical benefit from the consideration of molecular context, targeted therapies are still challenging. First part of this work focused on tyrosine kinase inhibitors targeting EGFR in non small cell lung cancers. Thus, improvement of biomarkers detection methods was completed by in vitro characterization of an unreported mechanism of acquired resistance. Briefly, pulmonary cells were exposed to a mutagen agent and a selection pressure was applied with EGFR inhibitors allowing the detection of TBK1 signature. Finally, synergic effect of that co-inhibition was highlighted. Now essentials in gynaecological cancers management, PARP inhibitors represent the second part of that work. Those targeted therapies are based on synthetic lethality. Consequently, BRCA1/2 pathogenic mutations are required for their administration, illustrating the issue of variants of uncertain significance. Toward their functional characterization necessity, a transcriptional analysis of splicing variant was first conducted on mRNA extracted from FFPE samples. Then, to evaluate functional signification of all types of variants, genomic edition was developed. Editing efficiencies of the unknown variant and a silent control one were compared in a haploid model where those genes are essentials. Functional signification of BRCA1/2 variants, and thereby mutations from all essential tumor suppressor genes in our model, can be evaluated in three weeks which is compatible with clinical management

Animal traits differ not only in mean, but also in variation around the mean. For instance, one sire’s daughter group may be very homogeneous, while another sire’s daughters are much more heterogeneous in performance. The difference in residual variance can partially be explained by genetic differences. Models for such genetic heterogeneity of environmental variance include genetic effects for the mean and residual variance, and a correlation between the genetic effects for the mean and residual variance to measure how the residual variance might vary with the mean. The aim of this thesis was to develop a method based on double hierarchical generalized linear models for estimating genetic heteroscedasticity, and to apply it on four traits in two domestic animal species; teat count and litter size in pigs, and milk production and somatic cell count in dairy cows. The method developed is fast and has been implemented in software that is widely used in animal breeding, which makes it convenient to use. It is based on an approximation of double hierarchical generalized linear models by normal distributions. When having repeated observations on individuals or genetic groups, the estimates were found to be unbiased. For the traits studied, the estimated heritability values for the mean and the residual variance, and the genetic coefficients of variation, were found in the usual ranges reported. The genetic correlation between mean and residual variance was estimated for the pig traits only, and was found to be favorable for litter size, but unfavorable for teat count.

In a comprehensive attempt to understand the molecular and cellular processes driving seedlessness in grapevine, a seeded variety (wild-type) and its seedless somatic variant (mutant) were characterized at the morphological, genomic and transcriptomic levels in relation to berry development and seed content. The overall importance of clonal variability and the application of Next Generation Sequencing technology in highlighting the molecular events during seed formation within a developing berry have been clearly demonstrated. In this thesis three hypothesis were formulated, tested and confirmed. First it was hypothesized that the mutant has a gross morphology identical to the wild-type except for berry size and seed content. In testing this hypothesis quantitative and qualitative traits that relate to berry development and seed content were compared in the two clones. Here traits that were significantly different in the two lines are those that relate only to berry size and seed content. This evaluation was performed both in control conditions (self-pollination) and after anther/stigma removal which allowed the investigation of a possible role for parthenocarpy, stenospermocarpy or other mechanisms in promoting the phenotype of the seedless somatic variant. The second hypothesis states that the mutant is sterile or partly sterile hence cannot produce viable seeds. In order to verify this hypothesis pollen germination and viability assays were carried out in both clones. The tests confirmed pollen germination and vitality percentage of the mutant was significantly lower than that of the wild-type. The third hypothesis concerned the existence of genomic/transcriptomic differences between the two lines and could be tested through the power of the Next generation Sequencing technology. In particular, we raised the following questions. Are there somatic mutations that can allow the wild-type and mutant to be distinguished? What are the temporal and spatial changes that could occur in their respective transcriptomes? Especially how does expression levels of key regulatory genes change before, during and after fertilization in the two clones? These key questions were addressed with the aid of Molecular marker analysis, Array based SNP genotyping method and RNA-Seq approach. Using 58 microsatellites, the analyzed loci showed identical profile in the wild-type and the mutant. The 20K grapevine Illumina Chip revealed 16333 identical SNP loci in the two clones, thus a further confirmation of the true identity of the seedless line. Conversely variant calling from RNA-Seq enabled the identification of several somatic mutations at the whole-genome level in the two lines. At the same time, RNA-Seq allowed the creation of inventories of gene expression at successive stages of seed formation. i.e. stages E-L 15 (single flowers in compact groups), E-L 27 (young berries enlarging) and E-L 38 (berries harvest-ripe). Here the transcriptomes revealed by Illumina mRNA-Seq technology had approximately 98% of grapevine annotated transcripts and about 80% of them were commonly expressed in the two lines. Differential gene expression analysis revealed a total of 1075 differentially expressed genes (DE) in the pairwise comparison of developmental stages, which included DE genes specific to the wild-type background, DE genes specific to the mutant background and DE genes commonly shared in both backgrounds. The analysis of differential expression patterns and functional category enrichment of wild-type and mutant DE genes highlighted significant coordination and enrichment of pollen and ovule developmental pathways. The expression of some selected DE genes was further confirmed by real-time RT-PCR analysis. To the best of our knowledge the work presented in this thesis represents the most comprehensive attempt to characterize the genetic bases of seed formation in grapevine. We have shown that a seeded wine grape and its seedless somatic variant are similar in several biological processes except for berry size and seed content. With a high throughput method we could identify an inventory of genes with altered expression in the mutant compared to the wild-type, which may be responsible for the seedless phenotype. The genes located within known genomic regions regulating seed content may be used for the development of molecular tools to assist table grape breeding. Therefore the data reported here have provided a rich genomic resource for practical use and functional characterization of the genes that potentially underpin seedlessness in grapevine.

Increasing evidence supports the existence of intratumor heterogeneity (ITH) in many cancer types, with potential clinical implications for both cancer diagnosis and treatment. Expansion of genetically distinct cell populations within a tumor creates a subclonal architecture that varies dynamically throughout cancer progression. This acquired cancer trait, ITH is the substrate for Darwinian evolution to act upon, selecting the subclones that carry beneficial phenotypes. The outgrowth of such subclones impacts cancer development, drug resistance and tumor relapse. Nevertheless, despite the key role ITH plays in cancer, important questions regarding its magnitude, origin and genetic drivers across different cancer types remain largely unanswered. By facilitating the emergence of nucleotide sequence mutations, copy-number alterations (CNA), chromosomal translocations or aneuploidies, genomic instability has been regarded as the major source of ITH. However, discrepancies in the rates of genomic instability and ITH observed in previous comprehensive studies suggest that additional events congregate to increase genetic heterogeneity in tumors. It is widely accepted that genetic alterations and disruption of epigenetic regulatory mechanisms are hallmarks of cancer. Nonetheless, while the genetic contribution for cancer development is easily illustrated by mutations in tumor suppressors or oncogenes, the epigenetic involvement and its functional relevance is far more complex and has only recently become a major focus of cancer research. Besides genetic mutations, cancer cells invariably present with some degree of epigenetic alterations that contribute to the acquisition of the cancer hallmarks. Indeed, there is evidence that epigenomic reprogramming plays a seminal role in tumorigenesis by creating a progenitor-like cell state that facilitates expression of driver mutations and tumor initiation. High throughput genome sequencing (HTS) efforts have identified driver mutations in genes that regulate the epigenome, namely genome-wide chromatin and deoxyribonucleic acid (DNA) methylation. While epigenetic deregulation in cancer arises primarily as a consequence of DNA mutations, the view that altered epigenomes may also change DNA mutations rates highlights reciprocal interactions that contribute to cancer development. Accordingly, epigenomic disruption should favour the development of genetically diverse tumor cell populations, fuelling ITH. In fact, a possible relationship between genomic and epigenomic alterations during clonal evolution of tumors has been suggested in esophageal squamous cell carcinoma and glioma where high concordance was observed between the evolution of genetic and epigenetic diversification. In this PhD thesis, I conducted an exhaustive characterization of ITH across 2,807 tumor samples from 16 different carcinomas using whole-exome sequencing (WES) datasets from The Cancer Genome Atlas (TCGA). Integration of ITH scores and somatic variants detected in each tumor sample revealed that mutations in epigenetics modifier genes are the stronger determinants of ITH and display an association with increased clonal evolution across several cancer types. To investigate whether epigenomic deregulation drives the development of tumors with high levels of ITH, we focused our analysis on kidney renal clear cell carcinoma (KIRC), the cancer type with the highest frequency of mutations in epigenetic modifiers. The important role of epigenomic deregulation in the development and progression of KIRC is illustrated by the finding that patients with mutations in epigenetic modifiers have worse overall survival than those without mutations in these genes. In particular, genes that regulate genome-wide histone and DNA methylation emerged as candidate drivers of ITH. Knockout of histone methyltransferase SETD2 or DNA methyltransferase DNMT3A using the CRISPR/Cas9 system on renal carcinoma cells led to significant expansion of genetically distinct clones and culminated in highly heterogeneous cell populations which recapitulate the heterogeneity levels observed in the TCGA patients. We thus reasoned that the increased ITH observed after DNMT3A or SETD2 knockouts likely underpins variations in mitochondrial metabolism upon which natural selection can act. Indeed, changes in mitochondrial metabolism constitute an important source of variability for natural selection during cancer evolution. Upon DNMT3A or SETD2 knockout we observed that positively selected clones displayed similar mutational spectra and increased mitochondrial bioenergetic performance under stress conditions, suggesting that specific patterns of genotypic and phenotypic variation emerge upon epigenomic deregulation. Our work provides new insights on tumor development and unravels new drivers of ITH, showing an unprecedented pan-cancer portrait of the major determinants of ITH. This work validates experimentally the role of specific epigenetic modifier genes and lays a foundation for more effective cancer prognosis and treatment.

Indiana University-Purdue University Indianapolis (IUPUI)
Reliable detection of low-frequency single nucleotide variants (SNVs) carries great significance in many applications. In cancer genetics, the frequencies of somatic variants from tumor biopsies tend to be low due to contamination with normal tissue and tumor heterogeneity. Circulating tumor DNA monitoring also faces the challenge of detecting low-frequency variants due to the small percentage of tumor DNA in blood. Moreover, in population genetics, although pooled sequencing is cost-effective compared with individual sequencing, pooling dilutes the signals of variants from any individual. Detection of low frequency variants is difficult and can be cofounded by multiple sources of errors, especially next-generation sequencing artifacts. Existing methods are limited in sensitivity and mainly focus on frequencies around 5%; most fail to consider differential, context-specific sequencing artifacts. To face this challenge, we developed a computational and experimental framework, RareVar, to reliably identify low-frequency SNVs from high-throughput sequencing data. For optimized performance, RareVar utilized a supervised learning framework to model artifacts originated from different components of a specific sequencing pipeline. This is enabled by a customized, comprehensive benchmark data enriched with known low-frequency SNVs from the sequencing pipeline of interest. Genomic-context-specific sequencing error model was trained on the benchmark data to characterize the systematic sequencing artifacts, to derive the position-specific detection limit for sensitive low-frequency SNV detection. Further, a machine-learning algorithm utilized sequencing quality features to refine SNV candidates for higher specificity. RareVar outperformed existing approaches, especially at 0.5% to 5% frequency. We further explored the influence of statistical modeling on position specific error modeling and showed zero-inflated negative binomial as the best-performed statistical distribution. When replicating analyses on an Illumina MiSeq benchmark dataset, our method seamlessly adapted to technologies with different biochemistries. RareVar enables sensitive detection of low-frequency SNVs across different sequencing platforms and will facilitate research and clinical applications such as pooled sequencing, cancer early detection, prognostic assessment, metastatic monitoring, and relapses or acquired resistance identification.

VASA ist ein DEAD-Box-Protein, das in der Keimbahn vorkommt und dem eine essentielle Rolle in der Entstehung und Erhaltung von Keimzellen zugeschrieben wird (Yajima und Wessel 2011a). VASA wurde in jedem bisher untersuchten Organismus in der Keimbahn gefunden (Raz 2000) und ist, da es in Säugetieren als keimzellspezifisch gilt (Castrillon et al. 2000), einer der derzeit meist genutzten Keimzellmarker. Allerdings zeigen neuere Daten aus unkonventionellen Modellorganismen sowie aus Drosophila, dass VASA in diesen Organismen, zusätzlich zu seinen Funktionen in der Keimbahn, noch weitere Funktionen außerhalb der Keimbahn übernimmt. Bisher existieren allerdings keine vergleichbaren Daten, die die Keimzellspezifität von VASA in Säugetieren widerlegen, weshalb VASA hier weiter-hin als spezifischer Keimzellmarker gilt (Alié et al. 2011). Vorarbeiten von Selma Drallé, die in der Arbeitsgruppe von Prof. Dr. rer. nat. R. Behr durchgeführt wurden, ergaben jedoch Hinweise auf eine VASA-Expression in somatischen Zellen des Weißbüschelaffen. In Weiterführung dieser Untersuchungen wurde in der vorliegenden Arbeit eine systematische Expressionsanalyse von VASA in unterschiedlichen Organen des adulten und neugeborenen Weißbüschelaffen durchgeführt. Die Arbeitshypothese hierbei war, dass VASA im Weischbüschelaffen nicht keimzellspezifisch exprimiert wird. Die Untersuchungen erfolgten mittels immunhistochemischer Färbungen, Western Blotting sowie PCR, wobei aufgrund präliminarer Hinweise jeweils Niere, Magen, Leber, Haut und Pankreas des neugeborenen und des adulten Weißbüschelaffen detailliert auf eine VASA-Expression untersucht wurden. Ovar und Hoden adulter Weißbüschelaffen dienten als Positivkontrollen. Die Ergebnisse dieser Arbeit deuten stark auf eine Expression von VASA in der Haut des neugeborenen Weißbüschelaffen hin. Dieser Befund legt nahe, das VASA im neugeborenen Weißbüschel-affen nicht keimzellspezifisch ist und über seine Funktionen in der Keimbahn hinaus noch weitere extragonadale Funktionen innehaben könnte. In den weiteren untersuchten Organen wurde keine VASA-Expression festgestellt; einzelne Befunde, die auf eine VASA-Expression hinwiesen, erwiesen sich als falsch-positiv. In dieser Arbeit wurde zudem eine für C. jacchus noch nicht in der Literatur beschriebene Splice-Variante von VASA im Ovar und im Hoden vom adulten Weißbüschelaffen gefunden, der das Exon 7 fehlt und die zudem modifizierte N- und/oder C-Termini aufweisen könnte. Die vorliegenden Ergebnisse weisen darauf hin, dass VASA im Weißbüschelaffen auch außerhalb der Keimbahn exprimiert wird und somit in dieser Spezies kein spezifischer Keim-zellmarker ist. Um diese Frage jedoch abschließend zu klären, sollten sich weitere Untersuchungen anschließen, um die in dieser Arbeit gezeigten Befunde zu untermauern.

L’hypothyroïdie congénitale par dysgénésie thyroïdienne (HCDT) est la condition endocrinienne néonatale la plus fréquemment rencontrée, avec une incidence d’un cas sur 4000 naissances vivantes. L’HCDT comprend toutes les anomalies du développement de la thyroïde. Parmi ces anomalies, le diagnostic le plus fréquent est l’ectopie thyroïdienne (~ 50% des cas). L’HCDT est fréquemment associée à un déficit sévère en hormones thyroïdiennes (hypothyroïdisme) pouvant conduire à un retard mental sévère si non traitée. Le programme de dépistage néonatal assure un diagnostic et un traitement précoce par hormones thyroïdiennes. Cependant, même avec un traitement précoce (en moyenne à 9 jours de vie), un retard de développement est toujours observé, surtout dans les cas les plus sévères (c.-à-d., perte de 10 points de QI). Bien que des cas familiaux soient rapportés (2% des cas), l’HCTD est essentiellement considérée comme une entité sporadique. De plus, plus de 92% des jumeaux monozygotiques sont discordants pour les dysgénésies thyroïdiennes et une prédominance féminine est rapportée (spécialement dans le cas d’ectopies thyroïdiennes), ces deux observations étant clairement incompatible avec un mode de transmission héréditaire mendélien. Il est donc cohérent de constater que des mutations germinales dans les facteurs de transcription thyroïdiens connus (NKX2.1, PAX8, FOXE1, and NKX2.5) ont été identifiées dans seulement 3% des cas sporadiques testés et furent, de plus, exclues lors d’analyse d’association dans certaines familles multiplex. Collectivement, ces données suggèrent que des mécanismes non mendéliens sont à l’origine de la majorité des cas de dysgénésie thyroïdienne. Parmi ces mécanismes, nous devons considérer des modifications épigénétiques, des mutations somatiques précoces (au stade du bourgeon thyroïdien lors des premiers stades de l’embryogenèse) ou des défauts développementaux stochastiques (c.-à-d., accumulation aléatoire de mutations germinales ou somatiques). Voilà pourquoi nous proposons un modèle «2 hits » combinant des mutations (épi)génétiques germinales et somatiques; ce modèle étant compatible avec le manque de transmission familial observé dans la majorité des cas d’HCDT. Dans cette thèse, nous avons déterminé si des variations somatiques (épi)génétiques sont associées à l’HCTD via une approche génomique et une approche gène candidat. Notre approche génomique a révélé que les thyroïdes ectopiques ont un profil d’expression différent des thyroïdes eutopiques (contrôles) et que ce profil d’expression est enrichi en gènes de la voie de signalisation Wnt. La voie des Wnt est cruciale pour la migration cellulaire et pour le développement de plusieurs organes dérivés de l’endoderme (p.ex. le pancréas). De plus, le rôle de la voie des Wnt dans la morphogénèse thyroïdienne est supporté par de récentes études sur le poisson-zèbre qui montrent des anomalies du développement thyroïdien lors de la perturbation de la voie des Wnt durant différentes étapes de l’organogénèse. Par conséquent, l’implication de la voie des Wnt dans l’étiologie de la dysgénésie thyroïdienne est biologiquement plausible. Une trouvaille inattendue de notre approche génomique fut de constater que la calcitonine était exprimée autant dans les thyroïdes ectopiques que dans les thyroïdes eutopiques (contrôles). Cette trouvaille remet en doute un dogme de l’embryologie de la thyroïde voulant que les cellules sécrétant la calcitonine (cellules C) proviennent exclusivement d’une structure extrathyroïdienne (les corps ultimobranchiaux) fusionnant seulement avec la thyroïde en fin de développement, lorsque la thyroïde a atteint son emplacement anatomique définitif. Notre approche gène candidat ne démontra aucune différence épigénétique (c.-à-d. de profil de méthylation) entre thyroïdes ectopiques et eutopiques, mais elle révéla la présence d’une région différentiellement méthylée (RDM) entre thyroïdes et leucocytes dans le promoteur de FOXE1. Le rôle crucial de FOXE1 dans la migration thyroïdienne lors du développement est connu et démontré dans le modèle murin. Nous avons démontré in vivo et in vitro que le statut de méthylation de cette RDM est corrélé avec l’expression de FOXE1 dans les tissus non tumoraux (c.-à-d., thyroïdes et leucocytes). Fort de ces résultats et sachant que les RDMs sont de potentiels points chauds de variations (épi)génétiques, nous avons lancé une étude cas-contrôles afin de déterminer si des variants génétiques rares localisés dans cette RDM sont associés à la dysgénésie thyroïdienne. Tous ces résultats générés lors de mes études doctorales ont dévoilé de nouveaux mécanismes pouvant expliquer la pathogenèse de la dysgénésie thyroïdienne, condition dont l’étiologie reste toujours une énigme. Ces résultats ouvrent aussi plusieurs champs de recherche prometteurs et vont aider à mieux comprendre tant les causes des dysgénésies thyroïdiennes que le développement embryonnaire normal de la thyroïde chez l’homme.
Congenital hypothyroidism from thyroid dysgenesis (CHTD) is the most common congenital endocrine disorder with an incidence of 1 in 4,000 live births. CHTD includes multiple abnormalities in thyroid gland development. Among them, the most common diagnostic category is thyroid ectopy (~ 50 % of cases). CHTD is frequently associated with a severe deficiency in thyroid hormones (hypothyroidism), which can lead to severe mental retardation if left untreated. The newborn biochemical screening program insures the rapid institution of thyroid hormone replacement therapy. Even with early treatment (on average at 9 d), subtle developmental delay is still be observed in severe cases (i.e., IQ loss of 10 points). Although there have been some reports of familial occurrence (in 2% of the cases), CHTD is mainly considered as a sporadic entity. Furthermore, monozygotic (MZ) twins show a high discordance rate (92%) for thyroid dysgenesis and female predominance is observed in thyroid dysgenesis (especially thyroid ectopy), these two observations being incompatible with simple Mendelian inheritance. In addition, germline mutations in the thyroid related transcription factors NKX2.1, PAX8, FOXE1, and NKX2.5 have been identified in only 3% of sporadic cases and linkage analysis has excluded these genes in some multiplex families with CHTD. Collectively, these data point to the involvement of non-Mendelian mechanisms in the etiology of the majority of cases of thyroid dysgenesis. Among the plausible mechanisms are epigenetic modifications, somatic mutations occurring in the thyroid bud early during embryogenesis, or stochastic developmental events. Hence, we proposed a two-hit model combining germline and somatic (epi)genetic variations that can explain the lack of clear familial transmission of CTHD. In this present thesis, we assessed the role of somatic (epi)genetic variations in the pathogenesis of thyroid dysgenesis via a genome-wide as well as a candidate gene approach. Our genome wide approach revealed that ectopic thyroids show a differential gene expression compared to that of normal thyroids, with enrichment for the Wnt signalling pathway. The Wnt signalling pathway is crucial for cell migration and for the development of several endoderm-derived organs (e.g., pancreas). Moreover, a role of Wnt signalling in thyroid organogenesis was further supported by recent zebrafish studies which showed thyroid abnormalities resulting from the disruption of the Wnt pathway during different steps of organogenesis. Thus, Wnt pathway involvement in the etiology of thyroid ectopy is biologically plausible. An unexpected finding of our genome-wide gene expression analysis of ectopic thyroids was that they express calcitonin similar to normally located (orthotopic) thyroids. Such a finding, although in contradiction with our current knowledge of the embryological development of the thyroid attributes C cell origins to extrathyroidal structures (ultimobrachial bodies) upon fusion with a fully-formed, normally situated gland. Using a candidate gene approach, we were unable to demonstrate any differences in the methylation profile between ectopic and eutopic thyroids, but nevertheless we documented the presence of a differentially methylated region (DMR) between thyroids and leukocytes in the promoter of FOXE1, a gene encoding the only thyroid related transcription factor known to play a crucial role in regulating the migration of the thyroid precursors during development as shown by animal studies. We demonstrated by in vivo and in vitro studies that the methylation status of this DMR is correlated with differential expression of FOXE1 in non-tumoral tissues (thyroids and leukocytes). Knowing that DMRs are hotspots for epi(genetic) variations, its screening among CTHD patients is justifiable in our search for a molecular basis of thyroid dysgenesis, currently underway in a case-control study. The results generated during my graduate studies represent unique and novel mechanisms underlying the pathogenesis of CHTD, the etiology of which is still an enigma. They also paved the way for many future studies that will aid in better understanding both the normal and pathogenic development of the thyroid gland.