2. Abstract Histone methylation plays an important role in almost all cellular processes and its homeostasis is maintained by histone methyltransferases and histone demethylases. Misregulation of histone methylation levels is associated with gene expression misregulation and consequently also with various developmental defects and diseases. In this thesis we focus on the lysine demethylases KDM2A and KDM2B and on their demethylation deficient isoforms KDM2A-SF and KDM2B-SF. The lysine specific demethylases KDM2A and KDM2B have been predominantly studied for their demethylation function on CpG island-rich gene promoters. However, KDM2A-SF and KDM2B-SF have not been studied in detail. Therefore, the main goal of this thesis was to characterize KDM2A-SF more in detail and to focus on the role that KDM2A/B-SF might potentially play in canonical Wnt signaling pathway. We found that the KDM2A-SF mRNA arises through the action of an alternative intronic promoter and not by alternative splicing. We showed that the KDM2A-SF start codon is located in the exon that corresponds to KDM2A exon 14 and we thus determined the exact amino acid sequence of the KDM2A-SF protein. Furthermore, using an isoform specific knockdown assay we showed that KDM2A-SF, unlike KDM2A-LF, forms distinct nuclear foci on pericentromeric...

Chromatin diminution is defined as the elimination of DNA during the differentiation of early embryonic cells into pre-somatic cells. While it was first observed in the nematode Parascaris equorum, it also been identified in other parasitic nematodes, hagfish and copepods. In the copepod Mesocyclops edax, up to 90% of genomic DNA is eliminated during chromatin diminution. It was previously shown that the eliminated DNA contained highly repetitive heterochromatic sequences. Here, we digested pre- and post-diminution DNA with BamHI and produced small libraries of clones from each. Analyses revealed no decrease in low copy numbered sequences, such as transposable elements. Rather, both libraries are found to be surprisingly similar in all aspects analysed. Further comparison also demonstrated similarity of our libraries with the DNA sequences eliminated from Cyclops kolensis. Consequently, we suggest that M. edax eliminates portions of euchromatic DNA, in addition to the previously characterized satellite sequences.

The Krüppel-associated box (KRAB) zinc finger protein (ZFP) genes comprise one of the largest gene families in the mammalian genome, encoding transcription factors with an N-terminal KRAB domain and C-terminal zinc fingers. The KRAB domain interacts with a co-repressor protein, KAP-1, which can recruit various factors causing transcriptional repression of genes to which KRAB ZFPs bind. Little is currently known about the gene targets of the ~400 human and mouse KRAB ZFPs. Many KRAB ZFPs interact with factors other than KAP-1. To identify proteins that may interact with one particular KRAB ZFP, Zfp647, I previously carried out a yeast two-hybrid screen using the full-length Zfp647 sequence and a mouse embryonic cDNA library. I have now tested the interactions from this screen for their specificity for Zfp647. I show that Zfp647 can interact with itself and at least 20 other KRAB ZFPs through their zinc finger domains, and have confirmed the Zfp647 self-interaction by in vitro co-immunoprecipitation. In my yeast two-hybrid screen, Zfp647 bound to KAP-1 as well as another related protein, ARD1/Trim23. Zfp647 also interacts with proteins that function in ubiquitylation. I have found evidence to suggest that Zfp647 may also interact with proteins encoding jumonji domains both by yeast two-hybrid assay and by co-immunoprecipitation from NIH/3T3 cell extracts. We have previously found that Zfp647 localises to non-heterochromatic nuclear foci in differentiated ES cells, which also contain KAP-1 and HP1, and which lie adjacent to PML nuclear bodies in a high proportion of cells. I have found that these foci are also visible in pMEFs, but not NIH/3T3 tissue culture cells. Immunofluorescence studies with antibodies against proteins from the yeast twohybrid screen have not shown any significant co-localisation with Zfp647. KAP-1 is sumoylated ex vivo, as are two human KRAB ZFPs. Because Zfp647 lies adjacent to PML nuclear bodies and can associate with proteins involved in posttranslational modification, I tested whether Zfp647 is also modified. I characterised a sheep _-Zfp647 antibody previously created in the lab and have shown that it detects Zfp647 by western blot, but not by immunofluorescence. I show that treatment of NIH/3T3 cells with NEM, which prevents the removal of protein modifications, leads to the appearance of higher molecular weight forms of Zfp647. Modification of Zfp647 is not dependent on KAP-1, which is known to function as a SUMO E3 ligase. Attempts to classify the modification as either ubiquitin, SUMO or NEDD8 have suggested that Zfp647 may be mono-ubquitylated. The larger modified forms of Zfp647 are present in both NIH/3T3 and ES cells. Interestingly, I found that the modification profile of the protein changes over the course of ES cell differentiation, during which time Zfp647 relocalises to punctate nuclear foci; thus Zfp647 modification may be involved in this process.

Leptosphaeria maculans ‘brassicae’ (Lmb) est un ascomycète de la classe des Dothideomycètes faisant partie d’un complexe d’espèces présentant différents niveaux d’adaptation au colza. Lmb est responsable d’une des maladies les plus dommageables sur colza : la nécrose du collet. Lmb présente un cycle de vie complexe au cours duquel il alterne différents modes de vie, traduisant l’existence de mécanismes de régulation fine de l’expression des gènes lui permettant de s’adapter rapidement à de nouvelles conditions. Le séquençage de son génome a révélé une structure originale, avec l’alternance de deux types de régions : les isochores GC et les isochores AT. Alors que les isochores GC sont riches en gènes, les isochores AT sont pauvres en gènes et présentent des caractéristiques de l’hétérochromatine (régions génomiques riches en éléments transposables et présentant un faible taux de recombinaison). Bien que pauvres en gènes, les isochores AT représentent une « niche écologique » pour les gènes codant des effecteurs puisque 20 % des gènes des isochores AT codent des effecteurs putatifs contre seulement 4 % des gènes localisés en isochores GC. Les gènes codant des effecteurs situés en isochores AT présentent un comportement transcriptionnel différent de ceux localisés en isochores GC : une faible expression pendant la croissance mycélienne et une forte induction d’expression pendant l’infection primaire du colza. Sur la base de ces observations, l’objectif de ma thèse était de caractériser le déterminisme de la co-expression des effecteurs situés dans les isochores AT et en particulier d’évaluer si la régulation de l’expression de ces gènes se fait par un contrôle épigénétique lié à leur localisation particulière et/ou par l’intervention de régulateurs communs. Afin de déterminer le rôle de la structure des isochores AT, l’analyse fonctionnelle de protéines impliquées dans le remodelage de la chromatine (i.e. HP1, DIM-5 et DMM-1) a été réalisée et leur implication dans la régulation de l’ensemble des gènes prédits dans le génome de L. maculans a été évaluée. Cette étude a permis de démontrer l’implication de la structure hétérochromatinienne des isochores AT dans la répression de l’expression pendant la croissance mycélienne des gènes situés dans cet environnement génomique, en particulier les gènes codant des effecteurs. Parmi les gènes sous contrôle épigénétique, nous avons pu observer qu’en plus des gènes localisés en isochores AT, des zones en isochores GC étaient aussi affectées et pouvaient constituer des « hot-spots » de contrôle épigénétique. Afin d’identifier des régulateurs candidats pouvant être impliqués dans le contrôle de l’expression des effecteurs pendant l’infection, le répertoire des gènes codant des facteurs de transcription (FTs) chez Lmb a été établi et l’analyse de la conservation de ce répertoire parmi les autres espèces du complexe d’espèces Leptosphaeria a permis d’identifier les FTs spécifiques, ou spécifiquement sur-exprimés pendant l’infection du colza, chez Lmb. Des candidats ont été sélectionnés pour réaliser leur analyse fonctionnelle : des gènes codant des FTs sur-exprimés pendant l’infection (9 FTs) ainsi que les orthologues de FTs qui avaient été décrits chez d’autres espèces comme régulateurs majeurs de la pathogénie (StuA, Sge1 et Fox1). L’analyse fonctionnelle de FTs candidats a permis d’établir que StuA, comme chez d’autres champignons phytopathogènes, joue un rôle important dans la mise en place de l’infection et l’expression des effecteurs chez L. maculans. Le « silencing » d’un FT de type AT-Hook, famille de FTs se fixant préférentiellement au niveau de séquences riches en AT, a un fort effet sur la pathogénie du champignon et entraîne une diminution d’expression de 2 effecteurs. Cette thèse a permis d’apporter de nouveaux éléments concernant la régulation des gènes codant des effecteurs chez un champignon phytopathogène impliquant, pour la première fois, un mécanisme épigénétique. / Leptosphaeria maculans is an ascomycete belonging to the Dothideomycete class and is part of a species complex showing different level of adaptation toward oilseed rape. Within this species complex, Lmb is responsible for the most damaging disease of this crop: “stem canker”. Lmb presents a complex life cycle during which it alternates between different life styles and nutritional strategies underlying the involvement of precise regulatory networks for gene expression to rapidly adapt to new conditions. The sequencing of the Lmb genome has revealed an unusual structure, alternating two types of regions, GC- and AT-isochores. While GC-isochores are gene-rich, AT-isochores are gene-poor and have several characteristics of heterochromatin (they are rich in transposable elements and present a lower rate of recombination compared to GC-isochores). Although gene-poor, AT-isochores are “ecological niches” for effector genes as 20% of the genes in these regions encode for putative effectors against only 4% of the genes in GC-isochores. Effector-encoding genes located in AT-isochores present a different transcriptional behavior compared to those located in GC-isochores: a very low expression in axenic culture and a drastic increase in expression during primary leaf infection. On these bases, the aim of my thesis was to characterize the determinism of the concerted effector gene expression. Are AT-isochores targets of reversible epigenetic modifications that affect the regulation of effector genes? and/or are one or several common regulators involved in the control of the concerted expression of effector genes? To assess the role of the structure of AT-isochores, functional analysis of three key players involved in chromatin remodeling (i.e. HP1, DIM-5 and DMM-1) was performed and their role in global gene expression was assessed. This study validated that heterochromatic structure of AT-isochores represses expression of genes located in such a genomic environment, notably effector genes. Among genes under an epigenetic control, we also identified genes located in GC-isochores that were similarly influenced and may represent “hot spots” for epigenetic control. To identify putative regulators of effector gene expression, we established the complete repertoire of transcription factors (TFs) of Lmb and by analyzing the conservation of this repertoire among species of the Leptosphaeria species complex, we identified TFs specific of Lmb, or specifically induced during infection. Functional analysis of 12 TFs was set up: nine TF-encoding genes induced during infection and three orthologs of TFs described as required for pathogenesis in other phytopathogenic fungi (StuA, Sge1, Fox1). This functional analysis showed that StuA, as in other phytopathogenic fungi, plays a major role in infection and expression of effector genes in Lmb. The silencing of an AT-Hook type TF, family of TFs that specifically interact with AT-rich sequences, was associated with a reduction of the expression of two effector genes during infection and with pathogenicity defects. This study brought new insights into the regulation of effector genes in a phytopathogenic fungus involving, for the first time, an epigenetic mechanism.

Heterochromatin assembly in fission yeast (Schizosaccharomyces pombe) requires conserved components that mediate RNA interference (RNAi) directed methylation of histone H3 on lysine 9 (H3K9). Fission yeast heterochromatin is mainly found at centromeres, telomeres, and the mating-type locus. At centromeres, transcripts from repetitive elements are processed to siRNAs and RNAi promotes chromatin modification by recruiting the Clr4 methyltransferase. RNAi is not required to maintain silent chromatin at the mating-type locus. This RNAi-directed form of centromeric heterochromatin provides an ideal system for in vivo screening to allow the identification of compounds that inhibit the activity of proteins involved in RNA silencing, chromatin modification and heterochromatin assembly in fission yeast and may inhibit conserved proteins in other organisms. A dominant selectable marker gene system at fission yeast centromeres that reports loss of heterochromatin integrity by increased resistance to G418 in 96-well plate format liquid cultures was developed. The resulting strain was used to screen a nontargeted chemically diverse compound library in vivo to identify compounds that disrupt the integrity of RNAi-directed heterochromatin. Two compounds, Emi1 and Emi14, were identified and found to cause a significant decrease in the level of H3K9 methylation on the outer repeats at fission yeast centromeres. Growth in the presence of Emi1 or Emi14 also caused a reduction in H3K9 methylation levels at the mating-type locus, suggesting that they do not act through RNAi. Consistent with this, Emi1 and Emi14 did not cause a decrease in centromeric siRNA levels. Analyses therefore suggest that Emi1 and Emi14 do not disrupt RNAi but that they inhibit downstream events in chromatin modification and heterochromatin assembly. Cells lacking RNAi due to loss of Dicer (dcr1Δ) or cells lacking the histone deacetylase (HDAC) Sir2 (sir2Δ) retain significant but lower levels of H3K9 methylation on the centromeric outer repeats. When dcr1Δ or sir2Δ cells were grown in the presence of Emi1 or Emi14 a further reduction in H3K9 methylation levels was observed on the outer repeats. This mimics the effect of combining clr3Δ with dcr1Δ or sir2Δ and suggests that Emi1 and Emi14 may interfere with SHREC function. SHREC is a chromatin remodelling complex that includes the HDAC Clr3 and the chromatin remodeler Mit1 and is known to contribute to heterochromatin integrity. Expression profiling performed on Emi1 and Emi14 treated cells confirmed the previous results. The changes in gene expression following Emi1 and Emi14 treatment were compared to known mutants defective in heterochromatin integrity. The profile of expression changes following Emi14 treatment was found to correlate with alterations in the expression pattern observed in cells with SHREC components deleted. No correlation with mutants lacking other HDACs or RNAi components was detected. Emi1 had a weaker correlation with defective SHREC function and thus may also partially inhibit the SHREC complex. Murine erythroleukemia (MEL) cells harbouring a silenced eGFP reporter transgene were used to assess whether Emi1 and Emi14 also affect silencing in mammalian cells. Emi1 was found to disrupt silencing at the eGFP reporter and this correlated with a decrease in H3K9 methylation. Structurally related analogues of Emi1 and Emi14 were selected and tested in the fission yeast assay. Interpretation of the obtained structure-activity relationships allowed identification of the chemical moieties key to Emi1 and Emi14 activity. Overall, an approach was developed to identify two novel small molecule inhibitors of a well-characterized chromatin modification pathway. The SHREC complex was identified as the putative target of these two compounds and structurally related active analogues were identified for them. Importantly, one of the compounds was also active in mammalian cells, highlighting the usefulness of this approach in identifying compounds that affect higher organisms.

This thesis describes the characterization of the Arabidopsis thaliana mutant terminal flower2 (tfl2), the cloning of the corresponding gene, and the analysis of TFL2 function in plant development. The tfl2 mutant is pleiotropic, exhibiting early floral induction in both long and short day conditions, a terminating inflorescence and dwarfing. TFL2 was isolated using a positional cloning strategy, and was found to encode a homolog to HETEROCHROMATIN PROTEIN1 (HP1), previously identified in yeast and animals where it is involved in gene regulation at the level of chromatin, as well as in the structural formation of constitutive heterochromatin. Investigating the light response during seedling photomorphogenesis I found that the tfl2 hypocotyl is hypersensitive to red and far-red light and that tfl2 is impaired in phytochrome mediated light responses such as the shade avoidance response. In the tightly regulated transition to flowering, we have shown that tfl2 might contribute to the interpretation of both external signals such as light and temperature as well as endogenous cues, via FCA, in the autonomous pathway. The Arabidopsis inflorescence meristem is indeterminate, and TFL2 possibly acts to maintain this indeterminate fate by repression of the floral meristem genes APETALA1 and AGAMOUS. In yeast two hybrid experiments TFL2 was shown to interact with IAA5, a protein with suggested functions in auxin regulation. Further, in tfl2 mutants the levels of the auxin indole-3-acetic acid decrease with age in aerial tissues, suggesting a function of TFL2 in regulation of auxin homeostasis and response. In summary, TFL2 contributes to regulation of several aspects of plant development, in accordance with the mutant phenotype and the identity of the TFL2 protein.

In Drosophila melanogaster, two chromosome-specific targeting and regulatory systems have been described. The male-specific lethal (MSL) complex supports dosage compensation by stimulating gene expression from the male X-chromosome and the protein Painting of fourth (POF) specifically targets and stimulates expression from the heterochromatic 4(th) chromosome. The targeting sites of both systems are well characterized, but the principles underlying the targeting mechanisms have remained elusive. Here we present an original observation, namely that POF specifically targets two loci on the X-chromosome, PoX1 and PoX2 (POF-on-X). PoX1 and PoX2 are located close to the roX1 and roX2 genes, which encode ncRNAs important for the correct targeting and spreading of the MSL-complex. We also found that the targeting of POF to PoX1 and PoX2 is largely dependent on roX expression and identified a high-affinity target region which ectopically recruits POF. The results presented support a model linking the MSL-complex to POF and dosage compensation to regulation of heterochromatin.

Chromatin diminution is defined as the elimination of DNA during the differentiation of early embryonic cells into pre-somatic cells. While it was first observed in the nematode Parascaris equorum, it also been identified in other parasitic nematodes, hagfish and copepods. In the copepod Mesocyclops edax, up to 90% of genomic DNA is eliminated during chromatin diminution. It was previously shown that the eliminated DNA contained highly repetitive heterochromatic sequences. Here, we digested pre- and post-diminution DNA with BamHI and produced small libraries of clones from each. Analyses revealed no decrease in low copy numbered sequences, such as transposable elements. Rather, both libraries are found to be surprisingly similar in all aspects analysed. Further comparison also demonstrated similarity of our libraries with the DNA sequences eliminated from Cyclops kolensis. Consequently, we suggest that M. edax eliminates portions of euchromatic DNA, in addition to the previously characterized satellite sequences.

Spt6 is a conserved eukaryotic transcription factor, known to interact with both nucleosomes and RNA polymerase II (RNAPII) to control transcription. We have initiated study of Spt6 in S. pombe in order to identify both novel and conserved roles in regulation of transcription and chromatin. We first constructed and analyzed spt6 mutants by several approaches. As Spt6 is known to be required for histone H3K36 methylation in both Saccharomyces cerevisiae and human cells, we examined the global levels of several histone modifications; we found that in S. pombe, Spt6 is required for both H3K4 and H3K36 trimethylation. We examined the chromatin state at two highly expressed genes, \(act1^+\) and \(pma1^+\), and found that there is a defect in recruitment of the methyltransferases responsible for those marks, Set1 and Set2, respectively. We also observed loss of nucleosomes, as well as a decrease in histone H2B monoubiquitylation. These results suggest that Spt6 plays an important role in chromatin regulation during transcription. We also conducted transcriptional analysis of an spt6 mutant by both microarray and high-throughput sequencing (RNA-seq) and discovered that Spt6 plays a critical role in maintaining the integrity of transcription genome-wide. We found that Spt6 is required to repress antisense transcription, with nearly 70% of genes having antisense transcripts increased by at least two-fold in an spt6 mutant. We also found that transcription of most long terminal repeats (LTRs) is derepressed. Finally, we found that a major class of transcripts elevated in the spt6 mutant is derived from heterochromatin, which is normally silenced. To study the heterochromatic silencing defect in greater detail, we analyzed the chromatin state of the pericentric repeats and found a decrease in H3K9 trimethylation, elevated levels of H3K14 acetylation, reduced recruitment of several known silencing factors and a loss of siRNA production. We also see a very modest increase in RNAPII recruitment. Based on this combination of phenotypes, Spt6 is likely to contribute to both transcriptional and post-transcriptional silencing mechanisms. Taken together, we have found that Spt6 plays several important roles to control transcription in both euchromatin and heterochromatin in S. pombe.

Within the cell nucleus, there are regions of highly condensed, transcriptionally silent chromatin called heterochromatin. Heterochromatin plays an important role in both chromosomal stability and gene regulation within the cell. Heterochromatin assembly is mediated by Heterochromatin Protein 1 (HP1) binding to epigenetically marked histone tails, most notably methylated H3K9. HP1 is post-translationally phosphorylated at serine and threonine residues, and this phosphorylation has been shown to increase HP1’s binding affinity for methylated H3K9 and heterochromatin formation. To study the effect of phosphorylation on heterochromatin assembly and HP1 localization within the nucleus, the unicellular protozoan Tetrahymena thermophila was used. Tetrahymena is an ideal model for this work because cells have a dynamic chromatin environment. Tetrahymena have an HP1-like protein, tHP1, which localizes to transcriptionally silent chromatin bodies within the otherwise transcriptionally active macronucleus. tHP1 is known to be phosphorylated at threonine-64 (site one) and at either serine-102 or threonine-103 (site two). Previous work shows that when phosphorylation at both sites is prevented, tHP1 exhibits decreased localization to chromatin bodies. In order to determine which site of phosphorylation accounts for tHP1’s localization to regions of heterochromatin, mutant proteins that allow phosphorylation at only one of the two sites were generated. The efforts to engineer a mutant protein that cannot be phosphorylated at site two and to visualize the protein’s localization throughout cell development are discussed. When phosphorylation is prevented at site two, tHP1 localization to regions of heterochromatin remains intact. These results suggest that phosphorylation at site one, not site two, may be responsible for tHP1 localization to macronuclear chromatin bodies. A mechanism by which site one phosphorylation influences tHP1 targeting to regions of heterochromatin is proposed. Furthermore, bioinformatics techniques are employed to identify other tHP1-like proteins within Tetrahymena. Characterization of these proteins will likely contribute to a more complete model of how heterochromatin is assembled in Tetrahymena.