This thesis is based on 9 papers dealing with the synthesis, physicochemical and biochemical properties of two types of chemically modified oligonucleotides which have the potential to down-regulate gene expression: (i) The first set is comprised of antisense oligonucleotides (AONs) conjugated with different chromophores of varying size, charge and π-electron density. Conjugation of the chromophores at the 3'- or 5'-end enhanced the target RNA binding affinity and RNase H recruitment capabilities compared to the native counterpart without changing the global helical conformation of their AON/RNA hybrid duplexes. The 3'-dipyridophenazine (DPPZ) has emerged as the most promising non-toxic chromophore in this series. (ii) The second set encompasses a new class of AONs containing North-East conformationally constrained 1',2'-oxetane-nucleosides. The introduction of oxetane-<b>T</b> and -<b>C</b> units imparts lowering of the Tm by ~ 6º and ~ 3 ºC/modification, respectively, of the AON/RNA hybrids, whereas the incorporation of the corresponding oxetane-<b>A</b> and-<b>G</b> units into AONs did not alter the thermostability in comparison with that of the native hybrid duplex. The oxetane-modified AONs have been found to possess enhanced serum stability compared to that of the native, whereas oxetane-<b>T</b> and -<b>C</b> containing AONs were more endonuclease-resistant than oxetane-<b>A</b> and-<b>G</b> modified AONs. All oxetane-modified mixmer AON/ RNA hybrid duplexes were, however, found to be excellent substrates for RNase H cleavage, which has been analyzed by Michaelis-Menten kinetics. The oxetane-modified mixmer AONs have shown effective down-regulation of the proto-oncogene c-myb mRNA in the K562 human leukemia cells, which was analyzed by QRT-PCR and Western Blot. Based on the amount of AON uptake after delivery, determined by slot blot, it was apparent that the oxetane-modified AONs are 5-6 times more effective antisense agents than the corresponding isosequential phosphorothioate analogues. The electrochemical assay based on sensitive nucleic acid mediated charge transport (CT) has revealed that the presence of oxetane-<b>T</b> unit causes more stacking perturbations in a DNA/DNA duplex than in a DNA/RNA duplex.

Non-coding RNAs are highly abundant regulators of gene expression in all kingdoms of life that often play important roles in vital cellular functions. In bacteria, small regulatory RNAs (sRNAs) usually act post-transcriptionally by regulating mRNAs through base pairing within ribosome binding sites (RBS), thereby inhibiting translation initiation. tisB encodes a toxin, TisB, whose synthesis is controlled by the sRNA IstR-1. Intriguingly, IstR-1 base pairs far upstream of the RBS but nevertheless inhibits translation initiation. The tisB mRNA is unusual in that ribosomes cannot access the RBS directly, but instead need an unstructured upstream region. This is precisely where IstR-1 exerts its inhibitory effect. We propose this region to serve as a ribosome loading site (standby site) which permits ribosomes to overcome the obstacle of inhibitory RBS-containing structures. Sequence-independent ribosome binding to the standby site allows for efficient relocation to the RBS structure when it is transiently open. Thus, standby sites are translation enhancer elements. I also characterized TisB-mediated toxicity. The hydrophobic protein TisB is targeted to the inner membrane and causes damage. This decreases the intracellular ATP concentration and entails decreased  replication, transcription and translation rates. It is likely that this toxin is involved in multidrug tolerance under certain conditions. We identified the sRNA MicF as a negative regulator of lrp expression. Lrp is a global transcription factor that controls genes involved in amino acid metabolism and transport of small molecules. Interestingly, Lrp also downregulates MicF. Thus, this study established that the mutual downregulation of MicF/Lrp creates a positive feedback loop which gives a switch-like behavior important for fast adaptations.

Reducing the production cost of industrial biofuels will greatly facilitate their proliferation and co-integration with fossil fuels. The cost of feedstock is the largest cost in most fermentation bioprocesses and therefore represents an important target for cost reduction. Meanwhile, the biorefinery concept advocates revenue growth through complete utilization of by-products generated during biofuel production. Taken together, the production of biofuels from low-cost crude glycerol, available in oversupply as a by-product of bioethanol production, in the form of thin stillage, and biodiesel production, embodies a remarkable opportunity to advance affordable biofuel development. However, few bacterial species possess the natural capacity to convert glycerol as a sole source of carbon and energy into value-added bioproducts. Of particular interest is the anaerobe Clostridium pasteurianum, the only microorganism known to convert glycerol alone directly into butanol, which currently holds immense promise as a high-energy biofuel and bulk chemical. Unfortunately, genetic and metabolic engineering of C. pasteurianum has been fundamentally impeded due to a complete lack of genetic tools and techniques available for the manipulation of this promising bacterium. This thesis encompasses the development of fundamental genetic tools and techniques that will permit extensive genetic and metabolic engineering of C. pasteurianum. We initiated our genetic work with the development of an electrotransformation protocol permitting high-level DNA transfer to C. pasteurianum together with accompanying selection markers and vector components. The CpaAI restriction-modification system was found to be a major barrier to DNA delivery into C. pasteurianum which we overcame by in vivo methylation of the recognition site (5’-CGCG-3’) using the M.FnuDII methyltransferase. Systematic investigation of various parameters involved in the cell growth, washing and pulse delivery, and outgrowth phases of the electrotransformation procedure significantly elevated the electrotransformation efficiency up to 7.5 × 104 transformants µg-1 DNA, an increase of approximately three orders of magnitude. Key factors affecting the electrotransformation efficiency include cell-wall-weakening using glycine, ethanol-mediated membrane solubilization, field strength of the electric pulse, and sucrose osmoprotection. Following development of a gene transfer methodology, we next aimed to sequence the entire genome of C. pasteurianum. Using a hybrid approach involving 454 pyrosequencing, Illumina dye sequencing, and single molecule real-time sequencing platforms, we obtained a near-complete genome sequence comprised of 12 contigs, 4,420,100 bp, and 4,056 candidate protein-coding genes with a GC content of 30.0%. No extrachromosomal elements were detected. We provide an overview of the genes and pathways involved in the organism’s central fermentative metabolism. We used our developed electrotransformation procedure to investigate the use of established clostridial group II intron biology for constructing chromosomal gene knockout mutants of C. pasteurianum. Through methylome analysis of C. pasteurianum genome sequencing data and transformation assays of various vector deletion constructs, we identified a new Type I restriction-modification system that inhibits transfer of vectors harboring group II intron gene knockout machinery. We designated the new restriction system CpaAII and proposed a recognition sequence of 5’-AAGNNNNNCTCC-3’. Overcoming restriction by CpaAII, in addition to low intron retrohoming efficiency, allowed the isolation of a gene knockout mutant of C. pasteurianum with a disrupted CpaAI Type II restriction system. The resulting mutant strain should be efficienty transformed with plasmid DNA lacking M.FnuDII methylation. Lastly, we investigated the use of plasmid-based gene overexpression and chromosomal gene downregulation to alter gene expression in C. pasteurianum. Using a β-galactosidase reporter gene, we characterized promoters corresponding to the ferredoxin and thiolase genes of C. pasteurianum and show that both promoters permitted high-level, constitutive gene expression. The thiolase promoter was then utilized to drive transcription of an antisense RNA molecule possessing complementarity to mRNA of our β-galactosidase reporter gene. Our antisense RNA system demonstrated 52-58% downregulation of plasmid encoded β-galactosidase activity throughout the duration of growth. In an attempt to perturb the central fermentative metabolism of C. pasteurianum and enhance butanol titers, we prepared several antisense RNA constructs for downregulation of 1,3-propanediol, butyrate, and hydrogen production pathways. The resulting downregulation strains are expected to exhibit drastically altered central fermentative metabolism and product distribution. Taken together, we have demonstrated that C. pasteurianum is amendable to genetic manipulation through the development of methods for plasmid DNA transfer and gene overexpression, knockdown, and knockout. Further, our genome sequence should provide valuable nucleotide sequence information for the application of our genetic tools. Thus, the genome sequence, electrotransformation method, and associated genetic tools and techniques reported here should promote extensive genetic manipulation and metabolic engineering of this biotechnologically important bacterium.

The last few years have seen dramatic advances in genomics, including the discovery of a large number of non-coding and antisense transcripts. This has revolutionised our understanding of multifaceted transcript structures found within gene loci and their roles in the regulation of development, neurogenesis and other complex processes. The recent and continuing surge of knowledge has prompted researchers to reassess and further dissect gene loci. The ghrelin gene (GHRL) gives rise to preproghrelin, which in turn produces ghrelin, a 28 amino acid peptide hormone that acts via the ghrelin receptor (growth hormone secretagogue receptor/GHSR 1a). Ghrelin has many important physiological and pathophysiological roles, including the stimulation of growth hormone (GH) release, appetite regulation, and cancer development. A truncated receptor splice variant, GHSR 1b, does not bind ghrelin, but dimerises with GHSR 1a, and may act as a dominant negative receptor. The gene products of ghrelin and its receptor are frequently overexpressed in human cancer While it is well known that the ghrelin axis (ghrelin and its receptor) plays a range of important functional roles, little is known about the molecular structure and regulation of the ghrelin gene (GHRL) and ghrelin receptor gene (GHSR). This thesis reports the re-annotation of the ghrelin gene, discovery of alternative 5’ exons and transcription start sites, as well as the description of a number of novel splice variants, including isoforms with a putative signal peptide. We also describe the discovery and characterisation of a ghrelin antisense gene (GHRLOS), and the discovery and expression of a ghrelin receptor (growth hormone secretagogue receptor/GHSR) antisense gene (GHSR-OS). We have identified numerous ghrelin-derived transcripts, including variants with extended 5' untranslated regions and putative secreted obestatin and C-ghrelin transcripts. These transcripts initiate from novel first exons, exon -1, exon 0 and a 5' extended 1, with multiple transcription start sites. We used comparative genomics to identify, and RT-PCR to experimentally verify, that the proximal exon 0 and 5' extended exon 1 are transcribed in the mouse ghrelin gene, which suggests the mouse and human proximal first exon architecture is conserved. We have identified numerous novel antisense transcripts in the ghrelin locus. A candidate non-coding endogenous natural antisense gene (GHRLOS) was cloned and demonstrates very low expression levels in the stomach and high levels in the thymus, testis and brain - all major tissues of non-coding RNA expression. Next, we examined if transcription occurs in the antisense orientation to the ghrelin receptor gene, GHSR. A novel gene (GHSR-OS) on the opposite strand of intron 1 of the GHSR gene was identified and characterised using strand-specific RT-PCR and rapid amplification of cDNA ends (RACE). GHSR-OS is differentially expressed and a candidate non-coding RNA gene. In summary, this study has characterised the ghrelin and ghrelin receptor loci and demonstrated natural antisense transcripts to ghrelin and its receptor. Our preliminary work shows that the ghrelin axis generates a broad and complex transcriptional repertoire. This study provides the basis for detailed functional studies of the the ghrelin and GHSR loci and future studies will be needed to further unravel the function, diagnostic and therapeutic potential of the ghrelin axis.

Diss. (sammanfattning) Stockholm : Karolinska institutet, 2007. / Härtill 4 uppsatser.

There is a growing appreciation that eukaryotic genomes are transcribed into numerous, previously undetected - and thus uncharacterized regulatory long non-coding RNAs (lncRNAs). Recent studies are primarily focused on lncRNAs transcribed from intergenic regions and enhancers, leaving antisense lncRNAs the least studied group of lncRNAs. At the same time, antisense transcription occurs in up to 74 % of human gene loci, frequently - from the opposite strand of genes encoding proteins involved in regulation of transcription. Here, we identified HIPSTAR (Heterogeneously expressed from the Intronic Plus Strand of the TFAP2A-locus RNA), a novel conserved lncRNA that is transcribed antisense to the TFAP2A gene. Unlike previously reported antisense lncRNAs, HIPSTR expression does not correlate with the expression of its antisense counterpart. Although HIPSTAR and TFAP2A are co-expressed in in vitro derived neural crest and trophoblast cells, only HIPSTAR and not TFAP2A is specifically expressed in a subset of cells within 8-cell- and morula-stage human embryos. We show that, similar to HIPSTAR, in the individual cells of developing human embryos or of stable cell lines the expression of lncRNAs is more highly heterogeneous than the expression of mRNAs. Finally, we demonstrate that HIPSTAR depletion in HEK293 and H1BP, a human embryonic stem cell line, predominantly affects the expression levels of genes involved in early organismal development and cell differentiation. Together, we show that expression of HIPSTAR and hundreds other lncRNAs is highly heterogeneous in human embryos and cell lines. We use HIPSTAR to exemplify the functional relevance of lncRNAs with heterogeneous and developmental stage-specific expression patterns. / Tem sido cada vez mais reconhecido que a transcrição dos genomas eucarióticos produz múltiplos transcritos novos, anteriormente não detectados e ainda não caracterizados, sendo que a maioria é constituida de RNAs não-codificantes longos (lncRNAs) regulatórios. Estudos recentes estão focados principalmente nos lncRNAs transcritos de regiões intergênicas e enhancers; assim, o grupo dos lncRNAs antisenso permanece o menos estudado de todos. Ao mesmo tempo, a transcrição antisenso ocorre em até 74% dos loci de genes humanos, frequentemente - a partir da fita oposta de genes que codificam proteínas envolvidas na regulação da transcrição. No presente trabalho, nós identificamos HIPSTR (Heterogeneously expressed from the Intronic Plus Strand of the TFAP2A-locus RNA), um lncRNA novo conservado que é transcrito a partir da fita antisenso do gene TFAP2A. Ao contrário do anteriormente relatado para os lncRNAs antisenso, a expressão de HIPSTR não está correlacionada com a expressão do gene da fita oposta. HIPSTR e TFAP2A são co-expressos em células da crista neural e em trofoblastos derivadas in vitro, mas somente HIPSTR e não TFAP2A está especificamente expresso num subconjunto de células de embriões humanos nos estágios de 8-células e mórula. Mostramos que, semelhante a HIPSTR, a expressão de lncRNAs é mais altamente heterogênea que a expressão de mRNAs em células individuais de embriões humanos em desenvolvimento ou em linhagens estáveis de células. Finalmente, nós demonstramos que a depleção de HIPSTAR em células HEK293 e H1BP, uma linhagem de células tronco embrionárias humanas, afeta predominantemente os níveis de genes envolvidos no início do desenvolvimento do organismo e na diferenciação de células. No conjunto, nós mostramos que a expressão de HIPSTR e de centenas de outros lncRNAs é altamente heterogênea em embriões humanos e linhagens celulares. Usamos HIPSTR para exemplificar a relevância funcional de lncRNAs com padrões de expressão heterogêneos e estágio-de-desenvolvimento específicos.

Les systèmes naturels ont montré l'intérêt de la multiplication des interactions pour une cible, permettant d'améliorer l'affinité et de moduler la spécificité de reconnaissance. Il est ainsi important pour des applications biologiques de concevoir des systèmes multivalents et biocompatibles. Le travail entreprit au cours de ce doctorat porte sur le développement de nouvelles méthodologies pour accéder à des systèmes multivalents originaux.Ainsi, nous avons conçu, par synthèse multi-étapes, une nouvelle plate-forme fonctionnalisée, basée sur un châssis α-PNA pour la reconnaissance multivalente d'oligonucléotides. Ce nouveau système peut potentiellement être impliqué dans la reconnaissance sélective multipoint d'ADN.En parallèle, nous avons préparé des clusters multivalents d'iminosucres sur des châssis peptidiques, construits à partir de ligations click sans métaux, pour l'inhibition enzymatique de glycosidases. En effet, des systèmes multivalents ont été récemment développés en tant qu'inhibiteurs de glycosidases. Cependant, leur méthodologie de synthèse repose quasiment exclusivement sur la ligation azoture-alcyne catalysée au cuivre, ce qui limite son application biologique en raison de sa toxicité. Nos travaux ont ainsi conduit à l'identification d'inhibiteurs efficaces d'α-mannosidases par une approche synthétique sans métaux.Dans le contexte de la vectorisation d'oligonucléotides, il existe un besoin de concevoir des systèmes dynamiques qui permettent un relargage contrôlé. Nous avons appliqué une stratégie d'auto-assemblage, par ligation click de type acylhydrazone, pour la génération in situ de clusters biomoléculaires à partir de châssis peptidiques et de ligands d'acides aminés modifiés. Etant donné le caractère dynamique de la ligation qui confère une adaptabilité au système, nous avons démontré que a) la présence d'une cible permet d'assister la formation des clusters par sélection de certains composants et b) l'ADN peut être relargué par échange de ligands. Cette technique efficace et rapide d'auto-assemblage de fragments a ensuite permis de réaliser un criblage pour sonder l'effet de l'architecture et de la valence sur la complexation. Ce projet a finalement conduit à l'identification de vecteurs efficace pour la transfection de siARN sur cellules.Enfin, dans un dernier projet, nous avons exploité diverses techniques orthogonales et chimiosélectives de ligations click dans le but de générer des nanostructures peptidiques. Deux cages ont ainsi été obtenues par la formation de ligations acylhydrazones et thiol-maléimides selon une approche one-pot.En résumé, ces travaux d'ingénierie et d'auto-assemblage de systèmes biomoléculaires multivalent ont permis le développement de méthodes innovantes pour répondre à des besoins d'actualité et permettre la construction de systèmes multivalents destinés à la reconnaissance d'oligonucléotides, la vectorisation et l'inhibition enzymatique. / Natural systems are inspiring in showing that the combination of multiple interactions enables improvement in binding affinity and selectivity for a target. Thus, the design of synthetic and biocompatible multivalent systems is of great importance for biological applications. The work described in this PhD thesis aims at developing novel methodologies for generating functional multivalent systems.In order to engineer multivalent systems for the recognition of oligonucleotides, we elaborated a multi-step synthesis of functionalized α-PNA scaffolds bearing side-groups. This new scaffold can potentially serve for the multi-point sequence-selective recognition of DNA.Multivalent nanoconstructs are emerging tools for enzyme inhibition. In this context, we prepared multivalent clusters of iminosugars – by metal-free click ligations on peptide scaffolds – as candidates for glycosidases inhibition. Although such enzyme inhibitors based on iminosugar clusters were recently reported, their synthesis relies almost exclusively on copper-catalyzed azide-alkyne cycloaddition, which notorious toxicity represents a serious limitation for biological applications. Our approach demonstrates that iminosugar clusters can be prepared in a metal-free fashion and exhibit strong multivalent effects for the inhibition of α-mannosidases. Multivalent biomolecular systems are also candidates for gene delivery application. In this context, the design of dynamic systems is of interest for achieving controlled release. We implemented a self-assembly strategy, using the acylhydrazone click ligation, for the in situ generation of biomolecular clusters starting from peptide scaffolds and modified amino acids building blocks. We showed that, whereas both compounds are ineffective for DNA complexation, the mixed system spontaneously expresses cationic clusters that effectively complex DNA. We further demonstrated that, given the dynamic character of the acylhydrazone ligation, the system is able to a) adapt to the presence of the DNA target by selecting the optimal building blocks for the cluster self-assembly, and b) trigger DNA release by component exchange. This modular and versatile self-assembly approach was further exploited to perform a fragments screening varying molecular structure and valency. Thereby, we identified new and effective vectors for the transfection of siRNA in living cells.The last project described in this manuscript deals with the generation of cage-type peptide nanoconstructs by using a set of orthogonal and chemoselective click ligations. Two cages, based on acylhydrazone ligation on one side and thiol-maleimide on the other, were obtained successfully in one-pot.In summary, this work has led to the development of novel methodologies for the engineering and self-assembly of multivalent biomolecular nanoconstructs for diverse biological applications such as oligonucleotide recognition, delivery and enzyme inhibition.

La myopathie myotubulaire est une maladie musculaire congénitale très sévère. Le laboratoire d’accueil a démontré que les échantillons de muscle de patients atteints de cette maladie ainsi que le modèle murin présentent une surexpression de DNM2, alors que sa réduction par croisement génétique améliore les signes cliniques et histologiques de la maladie. Le but de ce travail consistait à développer, tester et valider des composés injectables qui ciblent DNM2 et diminuent son niveau. Deux approches thérapeutiques ont été développées l’une basée sur l’utilisation de virus adéno-associés (AAV) exprimant des shRNA, l’autre sur les oligonucleotides antisens (ASO). L’injection des vecteurs AAV-shDnm2 ou bien les ASO-Dnm2 pouvait corriger les défauts histologiques et fonctionnels des muscles des souris myopathes.Les résultats obtenus montrent le potentiel thérapeutique de la réduction de DNM2, et présente une nouvelle approche pour le traitement de la myopathie myotubulaire. / Myotubular myopathy is a severe muscle disease. We previously have shown that muscle specimens of both patients and the mouse model presented an overexpression of DNM2, while its genetic reduction prevents the development of the muscle phenotypes. The aim of this work was to develop, test and validate deliverable compounds. Two therapeutic approaches were used. Injection of antisense oligonucleotide or adeno-associated virus expressing shRNA restores a normal lifespan with improved muscle structure and function of the myopathic mice. These results demonstrate that therapeutic potential of reduction of DNM2 level and provides an attractive therapeutic strategy that could be applied to treat myotubular myopathy.

Récemment, des analyses dans divers organismes eucaryotes ont révélé que l'ensemble du génome est transcrit et produit en plus des ARNs messagers, une grande variété d’ARNs non codants de différentes longueurs. Les ARNs non codants de plus de 200 nucleotides, classés comme longs ARNs non codants (LARNnc), représentent la classe la plus abondante de transcripts non codants. Les études des fonctions des LARNnc suggèrent que beaucoup d'entre eux seraient impliqués dans la régulation de la transcription. L'objectif de ma thèse de doctorat était d'élucider les mécanismes de la régulation transcriptionnelle médiée par des LARNnc dans différents systèmes eucaryotes. Dans mon premier projet, j'ai étudié le rôle d'un long ARN non codant antisens dans la régulation transcriptionnelle du gène PHO84, codant un transporteur de phosphate à haute affinité, chez S. cerevisiae. Des études antérieures ont montré que la suppression d’une proteine de l’exosome Rrp6 entraîne une augmentation de l'expression antisens et la répression de PHO84. Il a été suggéré que la perte de Rrp6 entraîne une stabilisation antisens au locus PHO84, entraînant le recrutement de l'histone de-acétylase Hda1 et la répression de PHO84. Cependant, le mécanisme par lequel Rrp6p régule la transcription de PHO84 n’était pas connu. En combinant des méthodes à l’échelle de cellule unique, des approches biochimiques et génétiques, nous avons montré que les niveaux d'ARN antisens sont régulés principalement lors de l'élongation par le complexe Nrd1-Nab3-Sen1, qui nécessite Rrp6 pour un recrutement efficace à l`extrémité 3`de PHO84. De plus, nous révélons l'expression anticorrelé du sens et de l'antisens, En résumé, nos données suggèrent que la transcription antisens régule le seuil d'activation du promoteur PHO84. Dans mon second projet, j'ai étudié les rôles des ARNs dérivés des amplificateurs (ARNa) dans la regulation de la transcription. En utilisant les cellules de cancer du sein MCF7 comme système modèle, nous avons cherché à déterminer comment les ARNa induits par l'oestrogène (E2) participent à la régulation de la transcription médiée par le recepteur d’oestrogène (ERα) au niveau de l'allèle unique. À l'aide de l’hybridation fluorescente à l’échelle de molécule unique (smFISH), nous avons révélé qu`après induction d'E2, les ARNa sont induits avec une cinétique similaire à celle des ARNm cibles, sont localisés exclusivement dans le noyau, principalement associés à la chromatine, et sont moins abondants que les ARNm. De manière surprenante, nous avons constaté que les ARNa sont rarement co-transcrits avec leurs loci cibles, indiquant que la transcription active des gènes ne nécessite pas la synthèse continue ou l'accumulation d'ARNa sur l'amplificateur. En outre, en utilisant des mesures de la distance à sous-diffraction, nous avons démontré que la cotranscription des ARNa et des ARNm se produit rarement dans une boucle amplificateurpromoteur. De plus, nous avons révélé que la transcription basale d'ARNa n'exige pas ERα ou l'histone méthyltransférase MLL1 qui active l'amplificateur par la mono-méthylation H3K4. Dans l'ensemble, nos résultats ont montré que les ARNa peuvent jouer un rôle lors de l'activation du promoteur, mais ne sont pas nécessaires pour maintenir la transcription de l'ARNm ou pour stabiliser les interactions amplificateur-promoteur. / Transcription is the initial step in gene expression and is subject to extensive regulation. Recently, analyses in diverse eukaryotes have revealed that in addition to protein coding genes, transcription occurs throughout the noncoding genome, producing non-coding RNAs of various lengths. Non-coding RNAs longer than 200 nucleotides, classified as long non-coding RNAs (lncRNAs), represent the most abundant class of non-coding transcripts, whose functions however are poorly understood. Recent studies suggest that many lncRNAs might have roles in transcription regulation. The goal of my PhD thesis was to elucidate the mechanisms of lncRNA mediated transcription regulation in different eukaryotic systems. For my first project, I investigated the role of an antisense long noncoding RNA in transcription regulation of the high-affinity phosphate transporter gene PHO84 in the unicellular eukaryote S. cerevisiae. Previous studies showed that deletion of the nuclear exosome component Rrp6 results in increased antisense expression and repression of PHO84. It was suggested that the loss of Rrp6 results in antisense stabilization at the PHO84 locus, leading to recruitment of the histone de-acetylase Hda1 and repression of PHO84. However, most of the mechanistic details of how Rrp6p functions in regulating PHO84 transcription were not understood. Combining single cell methods with biochemical and genetic approaches, we showed that antisense RNA levels are regulated primarily during transcriptional elongation by the Nrd1-Nab3-Sen1 complex, which requires Rrp6 for efficient recruitment to the 3’end of PHO84. Furthermore, we reveal anti-correlated expression of sense and antisense, which have distinct modes of transcription. In summary, our data suggest a model whereby antisense transcriptional read-through into the PHO84 promoter regulates the activation threshold of the gene. For my second project, I investigated the roles of enhancer derived RNAs (eRNAs). eRNAs are lncRNAs transcribed from enhancers that have been suggested to regulate transcription through different mechanisms, including enhancer-promoter looping, RNA polymerase elongation, and chromatin remodeling. However, no coherent model of eRNA function has yet emerged. Using MCF7 breast cancer cells as a model system, we sought to determine how estrogen (E2) induced eRNAs participate in estrogen receptor alpha (ERα) mediated transcription regulation at the single allele level. Using single molecule fluorescent in situ hybridization (smFISH), we revealed that upon E2 induction eRNAs are induced with similar kinetics as target mRNAs, but are localized exclusively in the nucleus, mostly chromatin associated, and are less abundant than mRNAs. Surprisingly, we found that eRNAs are rarely co-transcribed with their target loci, indicating that active gene transcription does not require the continuous synthesis or accumulation of eRNAs at the enhancer. Furthermore, using sub-diffraction-limit distance measurements, we demonstrated that co-transcription of eRNAs and mRNAs rarely occurs within a closed enhancer-promoter loop. Moreover, we revealed that basal eRNA transcription does not require ERα or the histone methyltransferase MLL1, which activates the enhancer through H3K4 mono-methylation. Altogether, our findings showed that eRNAs may play a role during promoter activation, but are not required to sustain mRNA transcription or stabilize enhancer-promoter looping interactions.

Les récentes études transcriptomiques chez divers organismes ont montré que la transcription des gènes convergents peut produire des ARN messagers (ARNm) chevauchants. Ce phénomène a été analysé dans le contexte de l’interférence par ARN (ARNi) nucléaire, et peu d’information existe quant au destin cytoplasmique des messagers 3’ chevauchants ou leur impact sur l’expression des gènes. Dans ce travail, nous avons abordé les conséquences potentielles de l’interaction entre des paires d’ARNm sens-antisens chez Saccharomyces cerevisiae, un organisme modèle naturellement dépourvu de l’ARNi. Nous avons démontré que les extrémités 3’ complémentaires des ARNm peuvent interagir dans le cytoplasme et moduler la stabilité ainsi que la traduction d’ARNm. Nos résultats sont issus d’une étude détaillée d’une paire d’ARNm convergents, POR1 et OCA2, ensuite généralisée par l’approche de l’ARNi reconstituée chez S. cerevisiae. L’analyse globale a confirmé que dans les cellules sauvages, les paires d’ARNm sens-antisens forment des duplexes d’ARN in vivo et ont un rôle potentiel à moduler l’expression d’ARNm ou de protéines respectifs, dans des différentes conditions de croissance. Nous avons montré que le destin de centaines des messagers convergents est contrôlé par Xrn1, révélant l’importance de cette exoribonucléase 5’-3’ cytoplasmique très conservée dans la régulation post-transcriptionnelle des gènes convergents. Notre travail ouvre donc la perspective de considérer un nouveau mécanisme de l’interaction entre les paires d’ARNm sens-antisens dans le cytoplasme, chez les organismes contenant ou non la voie de l’interférence par ARN. / Recent transcriptome analyses have revealed that convergent gene transcription can produce many 3’ overlapping mRNAs in diverse organisms. This phenomenon has been studied in the context of nuclear RNA interference (RNAi) pathway, however little is known about the cytoplasmic fate of 3’ overlapping messengers or their impact on gene expression. In this work, we address the outcomes of interaction between sense-antisense mRNA pairs in Saccharomyces cerevisiae, a model organism naturally devoid of RNAi. We demonstrate that the complementary tails of 3’ overlapping mRNAs can interact in the cytoplasm in a sequence-specific manner and promote post-transcriptional remodeling of mRNA stability and translation. Our findings are based on the detailed analysis of a convergent mRNA pair, POR1 and OCA2, subsequently generalized using the reconstituted RNAi approach in S. cerevisiae. Genome-wide experiments confirm that in wild-type cells, sense-antisense mRNA pairs form RNA duplexes in vivo and thus have potential roles in modulating the respective mRNA or protein levels under different growth conditions. We show that the fate of hundreds of messenger-interacting messengers is controlled by Xrn1, revealing the extent to which this conserved 5’-3’ cytoplasmic exoribonuclease plays an unexpected but key role in the post-transcriptional control of convergent gene expression. In sum, our work opens a perspective to consider an additional, cytoplasmic mechanism of interaction between sense-antisense mRNA pairs, in both RNAi-positive and negative organisms.