L’acrodysostose est une chondrodysplasie associant une petite taille et des anomalies des extrémités et de la face. Les mécanismes moléculaires sous-jacents ont récemment été identifiés, et deux types d’acrodysostose sont individualisés : 1) l’acrodysostose avec résistance hormonale (acroR1a), en lien avec des mutations du gène PRKAR1A, codant pour une sous-unité régulatrice de la PKA et 2) l’acrodysostose « sans » résistance hormonale (acroPDE), en lien avec des mutations du gène PDE4D, codant pour la phosphodiestérase 4D. Dans les deux cas, il existe une altération de la voie de signalisation de l’AMPc entraînant une résistance au PTHrp à l’origine du phénotype osseux des patients, similaire à celui observé dans la PHP 1a due à des mutations inactivatrices du gène GNAS, codant pour la protéine Gsα. Je me suis intéressée dans ce travail aux mécanismes cellulaires à l’origine du phénotype de l’acroPDE et pouvant expliquer ses différences phénotypiques avec l’acroR1a et la PHP1a. Ces trois pathologies offrent un outil précieux pour l’étude de la voie de signalisation des RCPG puisqu’elles sont toutes dues à un défaut génétique altérant cette voie, mais s’exprimant cliniquement de façon différente bien que proche, notamment en ce qui concerne les résistances hormonales autres que celle du PTHrp dans les chondrocytes. Dans la première partie, afin de mettre en évidence une spécificité tissulaire pouvant expliquer les différences d’expression phénotypique de ces mutations, j’ai utilisé deux modèles cellulaires humains, des fibroblastes et la lignée HEK293, et observé les effets de l’inhibition sélective de la PDE4 par du rolipram à différentes étapes de la voie de signalisation, après stimulation par deux agonistes, la PTH et la PGE2. Les résultats indiquent que l’impact du rolipram est d’autant plus important que la stimulation par l’agoniste est faible. La modulation du signal par la PDE4 dépend donc à la fois du type cellulaire et de l’intensité initiale du stimulus par l’agoniste. Dans la seconde partie, pour comparer plus directement l’effet des mutations de ces trois gènes au niveau cellulaire, j’ai réalisé une étude détaillée du phénotype des fibroblastes de témoins et de patients atteints de PHP1a, d’acroR1a et d’acroPDE, à différentes étapes de la voie de signalisation, après stimulation par la PTH et la PGE2, et en présence ou non d’inhibiteur des PDE. Il n’existe pas de différence significative dans l’expression des transcrits ou la traduction des protéines impliquées dans la voie de signalisation dans les différents groupes de fibroblastes. En revanche, l’effet de l’inhibition des PDE4 sur la quantité d’AMPc intracellulaire est plus important dans les fibroblastes mutés PDE4D, allant dans le sens de mutations activatrices. Nous n’avons pas pu mettre en évidence de différence significative dans la phosphorylation de CREB dans ce modèle. Et enfin, dans la 3ème partie, je présente la mise au point d’une technique de BRET permettant l’étude fonctionnelle des différentes mutations de PDE4D connues pour être responsables d’acroPDE. Les résultats sont en faveur d’un caractère activateur des mutations de PDE4D à l’origine d’une dégradation plus rapide d’AMPc aboutissant à une diminution de l’activité PKA. En conclusion, ces résultats concordent à montrer que les mutations de PDE4D sont activatrices, augmentant ainsi la dégradation de l’AMPc ce qui serait à l’origine des résistances hormonales. Ce phénomène est d’autant plus important dans les situations où l’augmentation d’AMPc est modeste, expliquant probablement l’absence de résistance hormonale dans certains tissus. Bien sûr, ces résultats sont à pondérer par le fait qu’il existe de nombreuses autres phosphodiestérases pouvant compenser ces phénomènes, qu’il existe une troisième voie d’utilisation de l’AMPc (Epac), et qu’il faudrait poursuivre les investigations notamment en étudiant les phénomènes de compartimentalisation de la cellule pour élucider ces mécanismes. / No abstract

Stroke is a common disorder of later life with a complex etiology, including both environmental and genetic risk factors. The inherited predisposition is challenging to study due to the complexity of the stroke phenotype. Genetic studies in an isolated population have successfully identified a positional candidate gene for stroke, phosphodiesterase 4D (PDE4D). The aim of this thesis was to identify stroke susceptibility loci and positional candidate genes, taking advantage of low genetic variation in the northern Sweden population. All stroke cases were identified in a population-based stroke registry at the northern Sweden MONICA Centre. 56 families containing multiple cases of stroke and a follow up set of an additional 53 families were used for linkage studies. For association studies, 275 cases of first ever stroke together with 550 matched community controls were included. In paper I, we used a candidate region approach to investigate the PDE4D region on chromosome 5q. Linkage was obtained with a maximum allele-sharing LOD score of 2.06; P = 0.001. However, no significant association of ischemic stroke to the previously defined at-risk allele in PDE4D was observed. We next performed a genome wide linkage scan to explore new susceptibility loci for common forms of stroke (paper II). Non-parametric multipoint linkage analysis yielded allele-sharing LOD scores > 1.2 at nine locations; 1p34, 5q13, 7q35, 9q22, 9q34, 13q32, 14q32, 18p11, 20q13. The highest allele-sharing LOD score was obtained on chromosome 18p (LOD = 2.14). Fine mapping resulted in increased allele-sharing LOD scores for chromosome 5q13 and 9q22. In the follow up analysis of the nine regions, including all 109 families, the highest allele-sharing LOD scores were obtained on chromosomes 5q, 13q and 18p although none reached the initial genome wide values. In paper III, we focused on the chromosome 5q region, and further mapping and haplotype analysis in the families was performed. A common 1 cM haplotype was found to be shared among affected members of five families. In this region only the regulatory subunit 1 of phosphatidylinositol 3-kinase (PIK3R1) gene was located. Association of three single nucleotide polymorphisms in the PIK3R1 gene to common stroke was obtained in the case-control material. Finally, in paper IV, an extended pedigree containing seven families connected to common founders eight generations back was identified by genealogical analysis, and submitted to a separate genome wide scan analysis. A significant allele-sharing LOD score of 4.66 (genome wide P < 0.001) at chromosome 9q31-33 was obtained. Haplotype analysis identified a minimal common region of 3.2 cM, which was shared by four of the seven families. These four families contained all of the primary intracerebral hemorrhagic cases present in the extended pedigree. In conclusion we have replicated linkage of stroke susceptibility to the PDE4D region on chromosome 5q, but no significant association of ischemic stroke to PDE4D was observed. Linkage analysis of stroke did not identify any new major stroke loci, indicating that multiple minor susceptibility loci in addition to the previously known locus on chromosome 5q could contribute to the disease. In the chromosome 5q region a novel positional candidate gene for stroke was identified, the PIK3R1 gene. The PIK3R1 protein has several biological actions with potential roles in stroke susceptibility. Also a novel susceptibility locus for common forms of stroke at chromosome 9q was identified in a large pedigree, which may be of special importance for susceptibility to hemorrhagic stroke.

A systems biology approach was adopted in order to assess various aspects of the disease oral squamous cell carcinoma. Three main aims were addressed; assess the ability of CoCl2 to mimic the hypoxic response in a eukaryotic cell line, assess the role of PDE4D in oral squamous cell carcinoma (OSCC) and the construction of a normoxic/hypoxic mathematical model to identify therapeutic targets.Cancer cells often acquire a revised metabolism which aids in initiation, survival and progression of the tumour. This is predominantly due to the transcription factor HIF-1 which is activated under hypoxic conditions. Certain compounds such as cobalt chloride (CoCl2) have been used extensively to inhibit the degradation of HIF-1α and simulate hypoxia. CoCl2 is likely to have off-target effects on metabolism; these effects were examined when exposing human telomerase reverse transcriptase (hTERT) cells to 100μM CoCl2. Gas chromatography-mass spectrometry (GC-MS), liquid chromatography-mass spectrometry (LC-MS) based metabolomics were utilised in combination with ELISA assays for HIF-1α and ATP. Central metabolism was accurately mimicked when hTERT cells were subjected to 100μM CoCl2, however; it was apparent that this concentration of CoCl2 does not induce an equal extent of hypoxia as 1% oxygen. A number of off-target effects of CoCl2 were observed in secondary metabolism, specifically in lipids and fatty acids. In conclusion, CoCl2 should be used with caution as a hypoxic mimicker with the caveat that interpretation of results should be restricted to its effects on central metabolism.The transcription factor CREB has the ability to regulate approximately 4000 genes, a number of which are associated with cancer initiation and progression. Cyclic adenosine monophosphate (cAMP) is required to activate CREB and is partially regulated through its degradation via the enzyme phosphodiesterase type 4D (PDE4D). A homozygous deletion of PDE4D has been associated with OSCC; however; the exact consequence of this deletion has not been fully elucidated. PDE4D was knocked down in the OSCC cell line BicR16 and cellular proliferation, migration, resistance to ionising radiation and central metabolism was investigated using MTT, scratch, clonogenic and GC-MS, respectively. The knockdown resulted in an increase in proliferation, migration and radiation resistance suggesting the role of PDE4D as a TSG. Amino acids, cholesterol, fatty acids, carbohydrates and TCA intermediates were found to be altered in concentration.A mathematical model of glycolysis, TCA and glutaminolysis under normoxia and hypoxia was constructed through the amalgamation of two established models from the literature. New reactions, parameters and metabolite concentrations were added and unnecessary entities were deleted. COmplex PAthway SImulator (COPASI) was utilised to construct the model before validating the model using experimental data from the literature and steady state and flux analyses. Sensitivity analysis and a reduction in external glucose and glutamine were mimicked and the alterations in hypoxic and normoxic metabolism analysed. The reactions vCSII, vGS, vPGK and vGII were identified as potential therapeutic targets which may affect metabolism in hypoxia only. However, certain validation methods proved unsuccessful and hence the model requires further work before attempting the analyses again.