Les protéines du groupe Polycomb (PcG) sont des facteurs épigénétiques dont le rôle est de maintenir la répression de leurs gènes cibles au niveau de la chromatine via la modification des protéines histones. EZH2 est une protéine clé dans les mécanismes de régulation puisqu’elle catalyse la mise en place de la marque répressive H3K27me3. Dans le cadre de ma thèse, je me suis intéressée au modèle des leucémies aiguës myéloïdes (LAM) dans lesquelles, contrairement à d’autres pathologies myéloïdes, des mutations affectant EZH2 ou des membres Polycomb ne sont retrouvées que très rarement (˂1%). Des études ont montré que dans ce type de leucémies, de nombreux gènes cibles d’EZH2 sont dérégulés bien que son activité répressive soit toujours présente, mettant en évidence d’éventuels défauts de recrutement de cette protéine. Parmi les facteurs de transcription susceptibles de réguler l’association des protéines PcG à la chromatine, se trouve PLZF qui est un candidat intéressant. En effet, le laboratoire a mis en évidence une interaction entre PLZF et la protéine Polycomb BMI-1 et a montré que la distribution génomique de PLZF concorde avec celle de certains composants des Polycomb. L’objectif de mon travail de thèse a donc été de déterminer dans quelle mesure PLZF intervient dans le recrutement ou l’activité d’EZH2. / Polycomb group (PcG) proteins are epigenetic factors which play a major role in maintaining epigenetic silencing via histone modifications at the chromatin level. EZH2 is a key regulator that catalyzes the trimethylation of H3K27, which is a repressive mark. During my PhD, I was interested in the acute myeloid leukemia (AML) model in which, unlike other myeloid malignancies, EZH2 or other PcG protein mutations are very rare (˂1%). Studies have shown that in this type of leukemia, many of EZH2 target genes are deregulated although its repressive activity is still present highlighting possible EZH2 recruitment defects. Among the transcription factors that regulate the association of PcG proteins to chromatin, the transcription factor PLZF is an interesting candidate. Indeed, the laboratory has demonstrated an interaction between PLZF and the Polycomb protein BMI -1 and showed that the genomic distribution of PLZF is consistent with that of some Polycomb components. The aim of my thesis was therefore to determine in which extent PLZF is involved in the recruitment or activity of EZH2.

Les protéines du groupe Polycomb (PcG) ont initialement été identifiées chez la Drosophile, en tant que facteurs nécessaires au maintien de l’expression spatio-temporelle de gènes homéotiques le long de l’axe antéro-postérieur. Depuis, leur rôle en tant que régulateurs du développement a été mis en évidence chez la plupart des métazoaires ainsi que chez les plantes, chez lesquelles elles orchestrent les transitions développementales, l’organogenèse et la différenciation cellulaire. Les protéines PcG sont nécessaires au maintien de la répression transcriptionnelle de gènes cibles, par la régulation de leur structure chromatinienne via des modifications post-traductionnelles des histones. Elles forment des complexes multiprotéiques, notamment les Complexes Répressifs Polycomb PRC1 et PRC2. PRC2 est responsable de la tri-méthylation de la lysine 27 de l’histone H3 (H3K27me3) et est constitué de 4 sous-unités principales qui, pour la plupart, sont présentes sous forme de familles multigéniques dans le génome d’Arabidopsis thaliana. Ainsi, il existe plusieurs complexes PRC2 constitués de combinaisons alternatives de ces sous-unités, qui sont potentiellement présents au sein d’une même cellule et dont les rôles sont considérés comme partiellement redondants. En utilisant des approches moléculaires, génétiques et génomiques, nous avons analysé le rôle des sous-unités PRC2 exprimées dans la pointe racinaire d’Arabidopsis. Nous avons montré que l’interaction entre différents PRC2 est nécessaire pour réguler l’activité du méristème, le déroulement temporel de la différenciation cellulaire, ainsi que pour le maintien de l’identité des cellules matures. De plus, notre travail montre que les complexes PRC2 contenant l’une ou l’autre des deux méthyltransférases, CLF et SWN, régulent des groupes de gènes communs ainsi que distincts, à travers des mécanismes différents incluant une fonction non-canonique. Par ailleurs, nos résultats indiquent que les différences fonctionnelles entre CLF-PRC2 et SWN-PRC2 reposent, au moins en partie, sur les sous-unités non-catalytiques avec lesquelles la méthyltransférase interagit. Pour identifier les gènes régulés dynamiquement par PRC2 durant la différenciation cellulaire, nous avons développé des approches permettant d’accéder à la résolution des types cellulaires afin d’analyser les états chromatiniens à l’intérieur de la niche de cellules souches et de la zone de maturation de la racine. Nos données suggèrent que PRC2 participe au maintien de l’identité du Centre Quiescent (QC) en réprimant des voies de signalisations spécifiques. De plus, la différenciation cellulaire en direction de la zone de maturation est accompagnée par un accroissement du répertoire des cibles PRC2, incluant des régulateurs méristématiques ainsi que des gènes spécifiquement exprimés dans différents types cellulaires. Enfin, nos résultats suggèrent qu’une proportion significative des cibles PRC2 sont présentes sous la forme de domaines bivalents H3K27me3-H3K4me3 dans les cellules souches végétales, cette proportion étant moins importante que celle décrite chez les cellules souches embryonnaires de mammifères. Dans l’ensemble, ce travail apporte une vue intégrée de la fonction, la dynamique et la multiplicité de l’activité PRC2 au cours du processus de différenciation cellulaire, dans le contexte d’un organe en développement. Nos résultats mettent en évidence le rôle de PRC2 en tant que régulateur majeur de la différenciation cellulaire, qui apporte à la fois robustesse et plasticité aux programmes transcriptionnels qui sous-tendent l’acquisition spatio-temporelle et le maintien de l’identité cellulaire. / The Polycomb group (PcG) proteins were originally identified in Drosophila as factors required for maintaining the spatio-temporal expression of homeotic genes along the head-to-tail axis. Since then, their role as developmental regulators has been highlighted in most metazoans as well as plants, in which they orchestrate developmental transitions, organogenesis and cell differentiation. PcG proteins are required to maintain the transcriptional repression of target genes by regulating their chromatin structure via post-translational histone modifications. They are found in multiprotein complexes, including Polycomb Repressive Complexes PRC1 and PRC2. PRC2 is responsible for the trimethylation of histone H3 at lysine 27 (H3K27me3) and consists of four core subunits, most of which are represented by multigene families in Arabidopsis thaliana. Thus, distinct PRC2 complexes formed by alternative subunit combinations exist, possibly in the same cell, and are thought to play partly overlapping roles. By combining molecular, genetic and genomic approaches, we have analyzed the role of the PRC2 subunits expressed in the Arabidopsis root tip used as a model. We show that the interplay between distinct PRC2s is necessary to regulate the activity of the meristem and the timing of cell differentiation, as well as the maintenance of cell identity. In addition, our work reveals that PRC2 complexes containing either of the two related methyltransferases CLF or SWN regulate common as well as specific sets of genes through distinct mechanisms, including a non-canonical function. Furthermore, our results indicate that the functional differences between CLF-PRC2 and SWN-PRC2 rely, at least in part, on the non-catalytic subunit they are interacting with. To identify the genes dynamically regulated by PRC2 during cell differentiation, we have developed cell type-specific approaches to analyze chromatin marks in cell populations within the stem cell niche and the maturation zone of the root. Our data suggest that PRC2 participates in the maintenance of the quiescent center (QC) identity by repressing specific signaling pathways. In addition, cell differentiation towards the maturation zone is accompanied by an increase of the repertoire of PRC2 targets including stem cell and meristem regulators, as well as cell type-specific genes. Finally, our findings suggest that bivalent H3K27me3-H3K4me3 domains in the QC represent a significant, though smaller proportion of PRC2 targets in plant stem cells compared to what has been described in mammalian embryonic stem cells. Overall, this work provides an integrated view of the function, dynamics and multiplicity of PRC2 activity during the cell differentiation process, in the context of a developing organ. Our results highlight the role of PRC2s as major regulators of cell differentiation that provide both robustness and plasticity to the transcriptional programs underlying cell fate acquisition and identity maintenance.

Les modifications de la chromatine jouent un rôle fondamental dans le contrôle de l’expression des gènes de stress et la mise en place d’une réponse physiologique appropriée en réponse aux signaux environnementaux. Les complexes répressifs polycomb (PRC pour Polycomb Repressive Complexes), PRC1 et PRC2 permettent la répression des gènes via le dépôt des marques H3K27me3 et H2AUb. Ce projet de thèse vise à explorer le rôle de la protéine LHP1, une sous-unité du complexe PRC1 chez Arabidopsis thaliana, dans la régulation de l’homéostasie de la marque H3K27me3 et des voies de signalisation activées par les stress. Nous avons utilisé une approche intégrative pour identifier les réponses immunitaires dé-régulées dans le mutant lhp1 ; mettant ainsi en évidence le rôle de la protéine LHP1 dans la répression de la branche dépendante de MYC2 de la signalisation par l’Acide Jasmonique et l’éthylène. La perte de la protéine LHP1 induit une baisse du niveau de la marque H3K27 me3 dans le corps des gènes ANAC019 et ANAC055, ce qui augmente leur expression et la dérégulation de leurs cible, conduisant ainsi à une résistance accrue aux pucerons, ainsi qu’à une sensibilité plus importante à l’ABA et à une meilleure tolérance à la sècheresse. D’autre part, l’activation de cette voie de signalisation dans le mutant lhp1 induit une baisse d’accumulation de l’Acide Salicylique (SA), due à la dé-régulation des gènes ICS1 et BSMT1, ainsi qu’à une sensibilité accrue au pathogène hémibiotrophe Pseudomonas syringae pv. tomato DC3000. Afin de mieux comprendre la fonction moléculaire de LHP1, nous avons étudié son interaction génétique avec l’histone déméthylase REF6. Nous avons analysé des données de ChIP-seq menées dans les mutants lhp1 et ref6 ainsi que dans le double mutant à la lumière des phénotypes développementaux et physiologiques de ces lignées, mettant ainsi en évidence le rôle complexe de LHP1 dans l’homéostasie de la marque H3K27me3. D’après ces résultats, nous proposons que d’une part, LHP1 est requise pour le recrutement d’histone méthyltransférases sur certains gènes afin de permettre le dépôt de la marque, et que d’autre part, LHP1 protège d’autre loci de l’activité de déméthylases telles que REF6. Nous avons aussi étudié les mécanismes conduisant à une augmentation des niveaux de H3K27me3 sur certains gènes dans le mutant lhp1, et montrons que LHP1 réprime le gène MEA. L’expression ectopique de ce gène dans les tissus somatiques du mutant lhp1 est probablement responsable de l’hyper-méthylation de nombreux loci. Ce groupe de gènes hyper-méthylés dans le mutant lhp1 est enrichi en gènes annotés comme étant impliqués dans les réponses immunitaires, affectant ainsi la régulation transcriptionnelle des gènes de défense, ce qui contribue à la sensibilité accrue du mutant lhp1 aux pathogènes nécrotrophes. Ainsi, nous proposons que l’homéostasie de la marque H3K27me3 est essentielle au développement harmonieux ainsi qu’aux réponses immunitaires de la plante, et que LHP1 contribue au maintien de l’équilibre entre diverses voies de réponse au stress et les processus de croissance. / Chromatin modifications and regulation play a major role in the expression of stress-responsive genes and the instauration of appropriate physiological responses to environmental cues in plants. Polycomb Repressive Complexes (PRCs), PRC1 and PRC2,are involved in the repression of protein-coding genes through the deposition of H3K27me3 and H2Aub. This thesis project explores the role of LHP1, an Arabidopsis thaliana PRC1 protein, in the regulation of H3K27me3 homeostasis and stress-responsive signaling pathways. We used an integrative approach for the identification of immune related pathways de-regulated in the lhp1 mutant, finding that LHP1 is required for the repression of the MYC2 branch of jasmonic acid (JA)/ethylene (ET) pathway of immunity. Loss of LHP1 induces the reduction in H3K27me3 levels in the gene bodies of ANAC019 and ANAC055, leading to their up-regulation and the mis-regulation of their downstream targets, increasing aphid resistance, ABA sensitivity and drought tolerance in Arabidopsis.On the other hand, the up-regulation of this pathway in lhp1 reduces Salicylic Acid (SA) content caused by a de-regulation of ICS1 and BSMT1, as well as increased susceptibility to the hemibiotrophic pathogen Pseudomonas syringae pv. tomato DC3000.In order to deepen our comprehension of the molecular role of LHP1, we studied the genetic interaction between this histone reader and the REF6 demethylase. We integrated ChIP-seq, developmental and physiological data of the single lhp1 and ref6 mutants, as well as their respective double mutant, finding that LHP1 plays different roles in the regulation of H3K27me3 homeostasis. Based in on our results we propose that on the one hand, LHP1 is necessary for the recruitment of histone methyltransferases to certain genes to promote the deposition of H3K27me3, while in some other targets it protects this histone mark from the demethylase activity of REF6. We also addressed the phenomenon of H3K27me3 gain in lhp1, and we provide evidence for the role of LHP1 in the repression of MEA, a gene that is de-repressed in the lhp1 mutant and may contribute to the ectopic hyper-methylation of several loci. This set of hyper-methylatedloci in lhp1 is enriched in immune-related genes, impacting the transcriptional regulation of immune responses, a process that contributes to the increased lhp1 susceptibility to necrotrophic pathogens. Therefore, we propose that H3K27me3 homeostasis is a key phenomenon behind developmental and innate immune processes in Arabidopsis, and that LHP1 plays a role as a keeper of the trade-off between diverse stress signaling pathways but also between these and developmental programs.

Although all cells in a multicellular organism contain the same set of genes, the spatiotemporal expression of these genes needs to be dynamically regulated for morphogenesis and life cycle transitions to take place. Polycomb-group (PcG) proteins are evolutionarily-conserved epigenetic regulators that function – via epigenetic marks such as H3K27me3 and modifications to chromatin structure – to maintain the repression of developmentally-important genes so that these genes are only expressed in the appropriate cells at the appropriate times. This repressive activity of the PcG is antagonised by the trithorax-group (trxG) of proteins. Although they maintain specific patterns of gene repression, PcG proteins are ubiquitously expressed. How their activity is regulated is largely unknown. To identify such regulatory pathways, a genetic screen for modifiers of PcG activity in Arabidopsis was carried out previously using the PcG mutant curly leaf (clf), which has moderately-severe developmental defects due to the ectopic or untimely expression of developmental regulators such as floral homeotic genes and the important flowering time regulator FLOWERING LOCUS T (FT). I characterised three novel potential regulators identified in this genetic screen: the chromatin-associated protein AT-HOOK MOTIF NUCLEAR LOCALISED PROTEIN 22 (AHL22), the 26S proteasome and the novel trithorax-group members ANTAGONIST OF LHP1 1 and 2 (ALP1 and ALP2). I found that the dominant sop-11D mutation is caused by over expression of AHL22 which suppresses the phenotype of clf by reducing FT expression. However, genetic analysis suggests that AHL22 may act in a parallel pathway to the PcG. I showed that mutations affecting diverse subunits of the 26S proteasome reduce the mis-expression of CLF targets and suppress the phenotypes of clf although they do not restore H3K27me3 levels at these targets. Pharmacological inhibition of the proteasome also alleviated the mis-expression of target genes found in clf mutants. Analysis of PcG protein levels following proteasome inhibition suggests that the 26S proteasome antagonises the PcG by degrading the key PcG member EMBRYONIC FLOWER 1 (EMF1), which is likely to be important for implementing target gene repression. Surprisingly, my proteomic analysis showed that the novel trxG members ALP1 and ALP2 are accessory components of a core PcG complex – the Polycomb Repressive Complex 2 (PRC2) – in vivo, suggesting that that ALP1 and ALP2 may antagonise PcG repression by preventing the association of core PRC2 components with accessory components EMF1, LIKE HETEROCHROMATIN PROTEIN 1 and the PHD finger proteins VERNALISATION5 and VIN3-LIKE 1. My results reveal a previously unknown role for 26S proteasomal degradation in the regulation of PcG activity during vegetative development and identify novel in vivo associators of the core PRC2 and point to their role in modulating PcG activity. These results thereby increase our understanding of how the PcG is regulated and serve as a starting point to discover how specificity is given to the PcG mediated repression, either by targeted degradation of EMF1 by various E3 ligases or by different combinations of PRC2 associators.

The development of animals and plants needs a higher order of regulation of gene expression to maintain proper cell state. The mechanisms that control what, when and where a gene should (or should not) be expressed are essential for correct organism development. The Polycomb group (PcG) is a family of genes responsible for maintaining gene silencing and Suppressor of zeste 12 (Su(z)12) is one of the core components in the PcG. The gene is highly conserved in organisms ranging from plants to humans, however, the specific function is not well known. The main tasks of this thesis was to investigate the function of Su(z)12 and its expression at different stages of Drosophila development. In polytene chromosomes of larval salivary glands, Su(z)12 binds to about 90 specific euchromatic sites. The binding along the chromosome arms is mostly in interbands, which are the most DNA de-condensed regions. The binding sites of Su(z)12 in polytene chromosomes correlate precisely with those of the Enhancer-of-zeste (E(z)) protein, indicating that Su(z)12 mainly exists within the Polycomb Repressive Complex 2 (PRC2). However, the binding pattern does not overlap well with Histone 3 lysine 27 tri-methylations (H3K27me3), the specific chromatin mark created by PRC2. The Su(z)12 binding to chromatin is dynamically regulated during mitotic and meiotic cell division. The two different Su(z)12 isoforms: Su(z)12-A and Su(z)12-B (resulting from alternative RNA splicing), have very different expression patterns during development. Functional analyses indicate that they also have different functions he Su(z)12-B form is the main mediator of silencing. Furthermore, a neuron specific localization pattern in larval brain and a giant larval phenotype in transgenic lines reveal a potential function of Su(z)12-A in neuron development.  In some aspects the isoforms seem to be able to substitute for each other. The histone methyltransferase activity of PRC2 is due to the E(z) protein. However, Su(z)12 is also necessary for H3K27me3 methylation in vivo, and it is thus a core component of PRC2. Clonal over-expression of Su(z)12 in imaginal wing discs results in an increased H3K27me3 activity, indicating that Su(z)12 is a limiting factor for silencing. When PcG function is lost, target genes normally become de-repressed. The segment polarity gene engrailed, encoding a transcription factor, is a target for PRC2 silencing. However, we found that it was not activated when PRC2 function was deleted. We show that the Ultrabithorax protein, encoded by another PcG target gene, also acts as an inhibitor of engrailed and that de-regulation of this gene causes a continued repression of engrailed. The conclusion is that a gene can have several negative regulators working in parallel and that secondary effects have to be taken into consideration, when analyzing effects of mutants. PcG silencing affects very many cellular processes and a large quantity of knowledge is gathered on the overall mechanisms of PcG regulation. However, little is known about how individual genes are silenced and how cells “remember” their fate through cell generations.

In vertebrates, the embryonic hindbrain is transiently subdivided along its anterior-posterior (A-P) axis into 8 well defined segments termed rhombomeres (r1-8). Each rhombomere represents a true cellular compartment in transcriptional profile, lineage restriction and neuronal organization. Thus, the vertebrate hindbrain provides a beautiful model for studying mechanisms of anterior-posterior patterning, signal transduction and interpretation, initiation and maintenance of transcriptional profiles, cell sorting and border formation. The Kreisler/MafB gene, which encodes a basic leucine zipper (bZIP) transcription factor that regulates some Hox genes, is one of the first genes to be expressed segmentally in the hindbrain, and is subject to a dynamic and complex regulatory process. However, unlike the Hox genes, Kreisler/MafB is not located within a large cluster of genes and therefore provides a simple system for dissecting the molecular mechanisms involved in hindbrain compartmentalization. In dissecting the mechanisms that govern Kreisler/MafB regulation, we have identified the S5 regulatory element that directs early MafB expression in the future r5-r6 domain. We have found a binding site within S5 that is specific for the Variant Hepatocyte Nuclear Factor 1 (vHNF1) to be essential, but not sufficient for early induction of r5-r6-specific expression. Thus, early inductive events that initiate MafB expression are clearly distinct from later acting ones that modulate its expression levels. Using mouse mutants, we have shown that MafB is dependent on the M33 polycomb protein and other mechanisms of chromatin remodeling. We then utilized transgenic flies and mice as well as binding assays to identify and validate a PcG/trxG response element (PRE), PRE1 which acts to reorganize the surrounding chromatin, regulating S5-dependent expression. To our knowledge, PRE1 is the first validated vertebrate PcG/trxG response element. Thus, PRE1 provides a springboard for further exploration of the mechanisms governing chromatin remodeling.

In vertebrates, the embryonic hindbrain is transiently subdivided along its anterior-posterior (A-P) axis into 8 well defined segments termed rhombomeres (r1-8). Each rhombomere represents a true cellular compartment in transcriptional profile, lineage restriction and neuronal organization. Thus, the vertebrate hindbrain provides a beautiful model for studying mechanisms of anterior-posterior patterning, signal transduction and interpretation, initiation and maintenance of transcriptional profiles, cell sorting and border formation. The Kreisler/MafB gene, which encodes a basic leucine zipper (bZIP) transcription factor that regulates some Hox genes, is one of the first genes to be expressed segmentally in the hindbrain, and is subject to a dynamic and complex regulatory process. However, unlike the Hox genes, Kreisler/MafB is not located within a large cluster of genes and therefore provides a simple system for dissecting the molecular mechanisms involved in hindbrain compartmentalization. In dissecting the mechanisms that govern Kreisler/MafB regulation, we have identified the S5 regulatory element that directs early MafB expression in the future r5-r6 domain. We have found a binding site within S5 that is specific for the Variant Hepatocyte Nuclear Factor 1 (vHNF1) to be essential, but not sufficient for early induction of r5-r6-specific expression. Thus, early inductive events that initiate MafB expression are clearly distinct from later acting ones that modulate its expression levels. Using mouse mutants, we have shown that MafB is dependent on the M33 polycomb protein and other mechanisms of chromatin remodeling. We then utilized transgenic flies and mice as well as binding assays to identify and validate a PcG/trxG response element (PRE), PRE1 which acts to reorganize the surrounding chromatin, regulating S5-dependent expression. To our knowledge, PRE1 is the first validated vertebrate PcG/trxG response element. Thus, PRE1 provides a springboard for further exploration of the mechanisms governing chromatin remodeling.

Epigenetics continues to be redefined and new discoveries are likely to revolutionise the field still further. This thesis explores different aspects of how epigenetic regulation of gene expression contributes to human disease. Paper I explores the function of the IKKα kinase in regulating gene expression through the nuclear retinoic acid receptor (RAR). We define a set of genes requiring IKKα for their expression and found recruitment of IKKα to the RAR dependent on structural motifs in its protein sequence. This interplay between the NFκB pathway and nuclear receptor regulated transcription is important to consider when designing therapeutic strategies. Papers II and III focus on the plasma cell malignancy multiple myeloma (MM) and define a gene regulatory circuit defining an underexpressed gene profile in MM dependent on the Polycomb proteins. We provide proof-of-principle that the use of small chemical inhibitors may be operational in reactivating genes silenced by H3K27me3 and that this leads to decreased tumour load and increased survival in the 5T33 in vivo model of MM. We explored the genome-wide distribution of H3K27me3 and H3K4me3, and defined their association with gene expression in freshly-isolated malignant plasma cells from MM patients. Importantly, H3K27me3-marked genes in MM associated with more aggressive stages of the disease and less favourable survival. We present evidence that gene targeting by H3K27me3 is likely to not only involve a small population of tumour cells, but rather represent a common MM profile and further provide a rationale for evaluating epigenetic therapeutics in MM. Paper IV shows that pro-inflammatory gene expression in monocytes of psychiatric patients can be induced in vitro by sodium pump inhibitors, as the steroid hormone ouabain. We suggest that the ouabain-induced gene expression is regulated by an intricate network involving microRNAs, Polycomb and the H3K27me3 demethylase JMJD3. Our data indicates that epigenetic regulators play a role in transmitting cues between intrinsic and/extrinsic stimuli and gene expression in psychiatric illness. This thesis provides novel insights on how seemingly unrelated pathways may converge on transcriptional regulation and evidence that epigenetic modifiers contribute to the pathogenesis of human complex diseases such as multiple myeloma and mood disorders.

Inheritance of gene expression patterns through multiple rounds of cell division is crucial for the normal development of multi-cellular organisms and is mediated by epigenetic mechanisms. Many epigenetic mechanisms are believed to involve heritable changes to chromatin structure. This includes maintenance of transcriptional repression by Polycomb Group (PcG) proteins. It is unknown how PcG-dependent repression is maintained during or re-established after mitosis, a process that involves many physical and biochemical changes to chromatin. Understanding the behavior of PcG proteins during mitosis is key to answering this question: if PcG proteins remain bound in mitosis they may constitute the memory themselves, or else transcriptional memory must reside elsewhere, such as in the altered chromatin structures induced by PcG proteins. PcG protein association with chromosomes in mitosis in Drosophila S2 cells was examined by immunofluorescence and cellular fractionation. PcG proteins are associated with mitotic chromosomes, which is consistent with a role in carrying information about transcriptional repression through mitosis. Localization of PcG proteins to specific sites in the genome was assessed by chromatin immunoprecipitation (ChIP) followed by genome-wide sequencing (ChIP- SEQ) on mitotic cells. A method for isolating pure populations of mitotic cells was developed to access PcG protein localization in mitosis unambiguously. PcG proteins were not detected at well-characterized PcG targets including Hox genes on mitotic chromosomes, but a covalent modification of histone H3 associated with PcG- dependent repression, trimethylation of lysine 27 (H3K27me3), is retained at these sites. Two PcG proteins Posterior Sex Combs (PSC) and Polyhomeotic (PH) remain at about 10% of their interphase binding sites in mitosis. PSC and PH are preferentially retained in mitosis at sites that overlap recently described borders of chromatin domains (1), including sites that overlap domain borders flanking Hox gene clusters. These persistent binding sites may serve to nucleate re-establishment of PcG binding at target genes upon mitotic exit, perhaps with assistance of H3K27me3. Thus PcG proteins may form part of the transcriptional memory carried through mitosis, but perhaps not by persistent association at the targets of repression. Retention of elements at chromatin boundaries in mitosis may serve as a general mechanism for epigenetic memory.

Most cells within multicellular organisms contain the same genetic information, yet the appropriate tissue-specific expression of genes is required for the proper formation of adult tissues. Genes can either be “turned on” or “turned off” from the initial zygotic state and maintained during subsequent cell divisions. Maintaining the correct expression profiles during cell divisions is accomplished by a number of different nuclear factors. One of the key families of proteins that maintains the repression of target genes during development is the Polycomb group (PcG) of proteins. PcG proteins form a number of different multi-subunit protein complexes that interact with specific regions of chromatin and direct the repression of nearby genes by reducing transcription. One PcG complex, Polycomb repressive complex 1 (PRC1), inhibits transcription and nucleosome remodeling as well as compacts chromatin, both in vivo and in vitro. The in vitro repressive activities map mainly to one subunit of Drosophila PRC1—the Posterior sex combs (PSC) protein. The PRC1 complex is conserved in many other organisms including mammals. To better understand the mechanisms involved in PcG mediated repression we undertook a biochemical structure/function analysis of mouse PRC1. In chapter one, I review the current understanding of PcG biology and a rationale for the dissertation is provided. In chapter two, data are presented that argues that a mouse PRC1 protein, M33/Cbx2, which is non-homologous to PSC, is responsible for chromatin compaction and repression of nucleosome remodeling. Data are presented that suggests these activities are localized to a basic, natively unfolded region of M33/Cbx2. In chapter three, we extend the findings from chapter two in an attempt to predict whether homologous PcG proteins from other species besides fly and mouse have biochemical activity. In agreement with predictions, a panel of recombinant PcG proteins was generated and data are presented that shows the predicted active PcG proteins are capable of both inhibition of nucleosome remodeling and compaction of chromatin. Finally, in chapter four, the implications of the data presented are discussed, and directions for further inquiry are explored.