identification des fonctions in vivo du facteur de transcription ...

292
THÉRÈSE GAGNON-KUGLER IDENTIFICATION DES FONCTIONS IN VIVO DU FACTEUR DE TRANSCRIPTION UBF ET DE LA MÉTHYLATION DE L'ADN DANS LA TRANSCRIPTION RIBOSOMALE Thèse présentée à la Faculté des études supérieures de l'Université Laval dans le cadre du programme de doctorat en biologie cellulaire et moléculaire pour l'obtention du grade de Philosophiae Doctor (Ph. D.) BIOLOGIE MEDICALE FACULTÉ DE MÉDECINE UNIVERSITÉ LAVAL QUÉBEC 2007 © Thérèse Gagnon-Kugler, 2007

Transcript of identification des fonctions in vivo du facteur de transcription ...

THÉRÈSE GAGNON-KUGLER

IDENTIFICATION DES FONCTIONS IN VIVO DU FACTEUR DE TRANSCRIPTION UBF ET DE LA

MÉTHYLATION DE L'ADN DANS LA TRANSCRIPTION RIBOSOMALE

Thèse présentée à la Faculté des études supérieures de l'Université Laval

dans le cadre du programme de doctorat en biologie cellulaire et moléculaire pour l'obtention du grade de Philosophiae Doctor (Ph. D.)

BIOLOGIE MEDICALE FACULTÉ DE MÉDECINE

UNIVERSITÉ LAVAL QUÉBEC

2007

© Thérèse Gagnon-Kugler, 2007

Il

Abstract Ribosomal transcription produces RNAs essential for the structure and function of

the ribosomes. The up-regulation of this transcription during growth is known to address

the increased needs of the cell for proteins and thus for ribosomes. However, the

mechanisms involved in this régulation are not well characterized. Our laboratory was the

first to demonstrate a direct link between growth and ribosomal transcription. Stimulation

of mammalian cells with sérum or EGF causes an immédiate increase in ribosomal RNA

synthesis. This activation is dépendant on the ERK pathway and on the phosphorylation of

the architectural factor UBF by ERK. However, the increase of transcription resulting from

this stimulation is not correlated with an increase in RNA polymerase I loading on the

ribosomal gènes but is rather explained by a higher elongation rate of the polymerase. We

also showed that UBF blocks elongation when it is not phosphorylated and is permissive to

it when it is phosphorylated by ERK. Thèse results argue against the actual model

suggesting that régulation occurs only at initiation level and demonstrate for the first time

the existence of a régulation at the elongation level.

Theoretically, ribosomal transcription could also be regulated at the level of gène

activation since more than half of the ribosomal gènes are silenced at ail times. According

to the current view, DNA methylation is involved in the silencing process, but we hâve

found that, at least in human cells, it is only important to silence a certain portion of the

ribosomal gènes. Loss of methylation following genetic inactivation of both DNA

methyltransferase 1 and 3b in human colorectal carcinoma cell line results in a decrease in

both ribosomal transcription and prolifération rates and in defects in ribosomal RNA

processing and nucleolar structure. Moreover, loss of DNA methylation leads to an

ingression of RNA polymerase II in the ribosomal locus thus probably causing the

problems observed in thèse cells. We suggest that both DNA methylation and régulation of

the elongation rates are essential mechanisms to maintain a high density of RNA

polymerase I elongating complexes on the ribosomal gènes in order to exclude RNA

polymerase II and ensure an efficient ribosome biogenesis.

Résumé La transcription ribosomale synthétise les ARN qui sont au coeur de la fonction et

de la structure des ribosomes. L'augmentation de cette transcription en condition de

croissance est connue et répond aux besoins accrus de la cellule en protéines et donc en

ribosomes. Toutefois, les mécanismes impliqués dans cette régulation étaient inconnus.

Notre laboratoire a montré l'existence d'un lien direct entre la croissance et la transcription

ribosomale. Nous avons montré que la stimulation de cellules de mammifères par le facteur

EGF ou l'ajout de sérum cause une augmentation rapide de la transcription ribosomale.

Cette activation est dépendante de la voie ERK qui cible le facteur architectural UBF. Nous

avons aussi montré que cette stimulation n'entraîne pas de recrutement massif d'ARN

polymérases 1 aux gènes ribosomaux, et ce malgré l'augmentation de la synthèse d'ARN

ribosomaux. Par contre, la vitesse d'élongation de la polymérase explique quantitativement

cette augmentation et celle-ci est régulée par l'état de phosphorylation d'UBF par ERK.

Ces résultats remettent en question le modèle actuel de régulation unique au niveau de

l'initiation des transcrits et démontrent pour la première fois l'existence d'une régulation au

niveau de l'élongation.

La proportion élevée de gènes ribosomaux silencieux suggère que ceux-ci puissent

être activés et constituer un troisième niveau de régulation de la transcription ribosomale.

La méthylation de l'ADN est suggérée comme étant importante pour l'établissement et le

maintien de ces gènes dans un état silencieux. Or, nous avons montré que la méthylation de

l'ADN explique l'état silencieux d'une partie et non de la totalité de ces gènes dans des

cellules humaines en culture. De plus, nous avons montré que la perte de méthylation, suite

à l'inactivation génique de DNMTl et 3b, entraîne une diminution de la transcription

ribosomale et de la vitesse de prolifération ainsi que des défauts dans la transformation de

TARN précurseur et dans la structure nucléolaire. De plus, la perte de méthylation entraîne

une incursion de TARN polymérase II dans le locus ribosomal ce qui cause

vraisemblablement les problèmes observés. Nous suggérons que la méthylation de l'ADN

et la régulation de l'élongation des transcrits sont des mécanismes essentiels pour le

maintien d'une forte densité de complexes en élongation sur les gènes ribosomaux de façon

à exclure TARN polymérase II pour ainsi assurer une biogenèse efficace des ribosomes.

Avant-propos Je tiens à remercier Tom qui m'a donné ma chance en laboratoire et m'a permis de

continuer jusqu'au doctorat dans son équipe. Sa rigueur scientifique et son honnêteté

intellectuelle en font un directeur de thèse et un chercheur d'une rare qualité. Je veux aussi

remercier les organismes qui m'ont offert des bourses : le Fonds de Recherche en Santé du

Québec pour la maîtrise et les Instituts de Recherche en Santé du Canada pour le doctorat.

Je veux remercier aussi les conférences Keystone qui m'ont offert une bourse d'aide au

congrès.

La force du laboratoire de Tom réside dans le dynamisme, l'intelligence et la

gentillesse des assistants de recherche et des étudiants de son équipe. La complicité qui

s'est développée au cours des années et l'entraide continuelle en ont fait un environnement

de travail exceptionnel. Je tiens à les nommer pour que ma reconnaissance passe à la

postérité : Nicolas Bisson, Steve Jean, Frédéric Langlois, Frédéric Lessard, Alexander

Mikryukov, Victor Stefanovsky et Michel Tremblay. Un merci tout particulier à Victor qui

m'a tout appris au laboratoire.

Ma reconnaissance va aussi à tous ceux et celles qui sont devenus mes amis au

Centre de Recherche et avec qui j 'ai passé de très belles années. Ils m'ont supportée dans

les moments les plus difficiles et accompagnée dans de nombreux moments mémorables. Je

veux aussi remercier le Dr Jean-Yves Masson pour sa gentillesse et sa disponibilité ainsi

que le personnel de soutien du Centre pour leur professionnalisme et leur gentillesse. Un

merci particulier à Sophie Lesage, Lucie Marcoux et Céline Sansfaçon.

A tous ceux qui croient en moi. Tout spécialement à François, à ma famille et à

mes amis.

Table des matières Chapitre 1 : Introduction 1

1.1. Le nucléole 1 1.2. La biogenèse des ribosomes 11

1.2.1 Les gènes ribosomaux et leur transcription 12 1.2.1.1. L'ARN polymérase 1 15 1.2.1.2. SL1 17 1.2.1.3 Rrn3 18 1.2.1.4 UBF 19

1.2.2. De l'ARN ribosomal précurseur aux ribosomes matures 24 1.3. L'établissement et le maintien des gènes ribosomaux silencieux 29

1.3.1. Mécanisme d'inactivation des gènes ribosomaux chez la levure S. cerevisiae.. 31 1.3.2. Mécanisme d'inactivation des gènes ribosomaux : la dominance nucléolaire chez les plantes 34 1.3.3. Mécanisme d'inactivation des gènes ribosomaux chez les mammifères 36

1.3.3.1. Mécanisme d'inactivation des gènes ribosomaux chez la souris 37 1.3.3.2. Les ADN méthyltransférases mammifères 39

1.3.3.2.1 Dnmtl 43 1.3.3.2.2. Dnmt2 48 1.3.3.2.3. Les membres de la famille Dnmt3 48

1.3.3.3. La méthylation de l'ADN et la formation de l'hétérochromatine 52 1.4. Contenu de la thèse 55

Chapitre 2 : Une stimulation par le facteur de croissance EGF entraîne, chez les mammifères, une réponse immédiate de la transcription ribosomalc via la phosphorylation d'UBFparERK; 57 « An Immédiate Response of Ribosomal Transcription to Growth Factor Stimulation in Mammals is Mediated by ERK Phosphorylation of UBF » 57

2.1. Avant-propos 57 2.2. Résumé 58 2.3. Summary 59 2.4. Introduction 59 2.5. Results 60

2.5.1. ERK activation is necessary and sufficient to activate ribosomal transcription in mammalian cells 63 2.5.2. Sustained ERK activity is required to maintain the activated level of ribosomal transcription 64 2.5.3. UBF is a target of ERK phosphorylation both in vitro and in vivo 65 2.5.4. Mutation of the ERK phosphorylation sites on UBF inhibits transcription activation 69 2.5.5. UBF phosphorylation site mutants act as dominant négatives, preventing the activation of Poli transcription by the ERK signalling cascade 69 2.5.6. ERK phosphorylation of the HMG boxes of UBF diminishes their interaction withDNA 73

2.6. Discussion 73 2.7. Acknowledgements 76

2.8. Expérimental Procédures 76

iv

2.8.1. Plasmids and constructs 76 2.8.2. Tissue culture and transfection 77 2.8.3. In vivo puise labclling, RNA extraction and analysis 77 2.8.4. Nuclear Run-On 78 2.8.5. Flow Cytometry 78 2.8.6. Endogenous kinase assays and in vitro UBF phosphorylation reactions 79 2.8.7. Phospho-polypeptide sequencing 80 2.8.8. Pull-down assays 80 2.8.9. In vivo phosphorylation of UBF1 81 2.8.10. Immunocytochemistry 82 2.8.11.DNABindingexperiments 82

2.9. Références 82 Chapitre 3 : Un facteur de croissance régule la vitesse d'élongation de la transcription par l'ARN Polymérase I, chez les mammifères, via la phosphorylation d'UBF et le remodelage de lachromatine ribosomale; 90 « Growth Factor Signaling Régulâtes Elongation of RNA Polymcrase I Transcription in Mammals via UBF Phosphorylation and r-Chromatin Remodeling » 90

3.1. Avant-propos 90 3.2. Résumé 91 3.3. Summary 92 3.4. Introduction 92 3.5. Results 94

3.5.1. Growth Factors Enhance Transcription without Increasing RNAPI Loading on the Mammalian rRNA Gènes 94 3.5.2. ChIP Measurements Confirm that RNAPI Loading Remains Constant over a Broad Range of 45S rRNA Synthesis Rates 94 3.5.3. Direct Measurements of RNAPI Elongation Rates In Vivo Reveals Growth Factor Régulation 96 3.5.4. High DNA template Occupancy by UBF Represses RNAPI Transcription In Vitro 100 3.5.5. ERK Phosphorylation of HMG Boxes 1 and 2 of UBF Derepresses Transcription 101 3.5.6. Unmodified, but Not ERK-Phosphorylated UBF Impedes Transcription Downstream of+34 101 3.5.7. UBF Reversibly Impedes Passage of Elongating RNAPI 102 3.5.8. In Vivo Activation of ERK Does Not Displace UBF from the Mouse rRNA gènes 105

3.6. Discussion 105 3.7. Supplementary Data 111 3.8. Acknowledgments 112 3.9. Expérimental procédures 112

3.9.1. Plasmid constructs and protein expression 112 3.9.2. In vivo puise labeling, RNA extraction and analysis 113 3.9.3. Nuclear Run-On 114 3.9.4. ChIP and Q-PCR assays 114 3.9.5. In vitro Phosphorylation of core UBF 115

V

3.9.6. In vitro transcription 115 3.9.7. Dnasel footprinting 116

3.10. Références 116 Chapitre 4 : La perte de la methylation de F ADN dans une lignée de cellules de carcinome colorectal entraîne des défauts dans la transformation de l'ARNr et une transcription par l'ARN polymérase II des gènes ribosomaux; 123 « Loss of CpG methylation in human colorectal carcinoma cells results in rRNA processing defects and RNA polymérase II transcription of thc rDNA» 123

4.1. Avant-propos 123 4.2. Résumé 124 4.3. Summary 125 4.4. Introduction 125 4.5. Results 126 4.6. Discussion 137 4.7. Acknowledgements 139 4.8 Material and methods 140

4.8.1 Cell Culture 140 4.8.2 Psoralen crosslinking and Southern blot 140 4.8.3. Sodium Bisulfite Genomic Sequencing 141 4.8.4. Prolifération rate assay and FACS analyses 142 4.8.5. In vivo puise labeling, RNA extraction and analysis 142 4.8.6. RNA meaurements 142 4.8.7. Northern blot 143 4.8.8. Western blot 143 4.8.9. Microarrays 144 4.8.10. Immunofluorescence staining 145 4.8.1 l.ChIP and Q-PCR assays 146 4.8.12. Run-On/ChIP 146 4.8.13. SI Mapping 147

4.9. Références 148 4.10. Données supplémentaires 152

4.10.1. Détermination de la production d'ARN ribosomaux po!yA+ 153 4.10.2. Western blot avec des extraits de protéines nucléolaires 155 4.10.3. Inhibition de la methylation de FADN avec 5-aza-2'dexoxycytidine 158 4.10.4 Micropuces à ADN (« microarrays ») 162

Chapitre 5 : Inactivation génique d'UBF, de TAF168 et de TAFi95 dans les souris 180 5.1. Avant-propos 180 5.2. Introduction 180 5.3. L'inactivation génique chez la souris 183 5.4. L'inactivation génique d'UBF avec le vecteur pGKDTAmUBFKO 187

5.4.1. Détermination des clones positifs 188 5.4.1.1. Détermination des clones positifs par PCR 190 5.4.1.2. Confirmation des clones positifs par analyse de type Southern 192

5.4.2. Les souris chimériques 193 5.5. Le vecteur pGKDTAmTAFi68KO 195 5.6. Les clones de cellules ES+/- de l'Institut Sanger 196

vi

5.6.1. Le clone de cellules ES UBF +/- de l'Institut Sanger: AY0043 197 5.6.1.1. Confirmation du clone AY0043 comme étant UBF +/- par RT-PCR 198 5.6.1.2. Confirmation du clone AY0043 comme étant UBF +/- par analyse de type Southern 199 5.6.1.3. Les souris chimériques 199

5.6.2. Les clones de cellules ES TAF,68+/- et TAFi95+/- de l'Institut Sanger 202 5.7. Production de lignées cellulaires ES-/- complémentées par un transgène inductible

202 5.8. Perspectives et Conclusion 204 5.9. Matériel et méthodes 207

5.9.1. Culture des cellules embryonnaires souche 207 5.9.2. Lysc des cellules et extraction de l'ADN génomique 207 5.9.3. Réaction de PCR 208 5.9.4. Analyse de type Southern 209

Chapitre 6 : Conclusion 210 6.1. UBF est un régulateur de l'élongation de l'ARN polymérasc 1 210

6.1.1. La phosphorylation d'UBF, par ERK, modifie son attachement à l'ADN 210 6.1.2. UBF définit une chromatine ribosomale 211 6.1.3. L'élongation de la transcription par l'ARN polymérase I est régulée par UBF.

211 6.1.4. L'étape limitante de l'augmentation de la transcription ribosomale en réponse à un facteur de croissance : initiation, Rrn3, ou élongation, UBF? 212 6.1.5. D'autres protéines pourraient aussi réguler l'élongation de l'ARN polymérase I.

214 6.1.6. Le complexe nucléoprotéique UBF-ADN 215 6.1.7. Pespectives sur les fonctions d'UBF 217

6.2. Le rôle de la méthylation de l'ADN dans l'établissement et le maintien des gènes ribosomaux silencieux 218

6.2. 1. La méthylation de l'ADN est essentielle pour la biogenèse des ribosomes.... 218 6.2.2. La méthylation et la réplication de l'ADN 220 6.2.3. L'hypométhylation du promoteur ribosomal et la tumorogenèse? 221

6.2.3.1. La conception des oligonucléotides favorise l'amplification de l'ADN méthylé 222 6.2.3.2. Sous-clonage préférentiel de l'ADN méthylé 222 6.2.3.3. L'hypométhylation du promoteur ribosomal et son implication suggérée dans la tumorogenèse n'est pas démontrée 223 6.2.3.4. Rôle des Dnmts dans la régulation de la transcription ribosomale 226

6.2.4 La méthylation des gènes ribosomaux et la croissance cellulaire 228 6.2.5. Perspectives 229 6.3. Une haute densité de complexes d'ARN polymérase I engagés sur les gènes ribosomaux est essentielle pour la biogenèse normale des ribosomes 230

Annexe 1 : L'importance des gènes ribosomaux dans la biogenèse des ribosomes; « A housekeeper with power of attorney, the rRNA gènes in ribosome biogenesis» 251 Annexe 2 : Un nouveau paradigme pour la régulation de la transcription ribosomale chez les mammifères; «A new paradigm for the régulation of the mammalian ribosomal RNA gènes» 252

Liste des abréviations 5-aza : 5-aza-2'dexoxycytidine, nucléotide analogue et non méthylable qui est

incorporé dans l'ADN lors de la réplication. ADNr : Gènes ribosomaux ARNr : ARN ribosomaux ARNm : ARN messager

CB : Corps de Cajal; « Cajal Body » Cellules DKO : Cellules HCT 116 dans lesquelles les gènes codant pour DNMTl et DNMT

3b ont été génétiquement inactivés; « Double Knock-out cells » Cellules PS: Cellules parentales HCT 116 ; « Parental Strain cells » ChIP : Technique d'immunoprécipitation de chromatine; « Chromatin

ImmunoPrecipitation » DNMTl : ADN Méthyltransférase 1; « DNA Methyl Transferase 1 » DNMTl-/- : Cellules HCT 116 dans lesquelles le gène codant pour DNMTl a été

génétiquement inactivé DNMT2 : ADN Méthyltransférase 2; « DNA Methyl Transferase 2 » DNMT3A : ADN Méthyltransférase 3a; « DNA Methyl Transferase 3a » DNMT3B: ADN Méthyltransférase 3b; « DNA Methyl Transferase 3b » DNMT3b-/-:Cellules HCT 116 dans lesquelles le gène codant pour DNMT3b a été

génétiquement inactivé DNMT3L : ADN Méthyltransférase 31; « DNA Methyl Transferase 31 » DFC : Compartiment nucléolaire; « Dense Fibrillar Component » EGF : Facteur de croissance épidermal; « Epidermal Growth Factor » ERK : Kinase de la voie des MAPK; « Extracellular signal-Regulated Kinase » EST : Séquence génomique vraisemblablement codante, mais dont le produit n'est

pas encore identifié; « Expressed Séquence TAG » ETS : Séquence de TARN ribosomal précurseur située à l'extérieur des séquences

codant pour les ARN ribosomaux matures; « External Transcribed Spacer » FC : Compartiment nucléolaire; « Fibrillar Center» FISH : Technique d'hybridation par fluorescence in situ; « Fluorescence In Situ

Hybridation » GC : Compartiment nucléolaire; « Granular Component » HDAC1 : Histone Déacétylase 1 HDAC2 : Histone Déacétylase 2

iv

HMG1 : Famille de protéines ayant la capacité de modifier la stucture de l'ADN; « High Mobility Group 1 proteins »

IGS : Séquence ribosomale non transcrite; « Intergenic spacer » ITS : Séquence de TARN ribosomal précurseur située entre les séquences codant

pour les ARN ribosomaux matures; « Internai Transcribed Spacer »

ISW1 : Famille de complexes de remodellage de la chromatine ATP-dépendant KO : Inactivation génétique d'une protéine; « Knock-out » NOR : Région organisatrice nucléolaire; « Nucleolar Organizing Région » NoRC : Complexe nucléolaire de remodelage de la chromatine ATP-dépendant;

« Nucleolar remodelling Complex » PAF53 : Facteur 53 associé à TARN polymérase I; « Polymerase Associated Factor

53 » PNB : Corps prénucléolaires; « Prenucleolar bodies » Poli : ARN polymérase I PolII : ARN polymérase II PolIII : ARN polymérase III RRN3/TIF-IA : Protéine du complexe de pré-initiation; Rrn3 chez la levure et l'humain et

TIF-IA chez la souris; « Transcription Initiation Factor IA » S. cerevisiae : Saccharomyces cerevisiae Sir2 : Protéine de levure essentielle à F inactivation des gènes transcrits par l'ARN

polymérase II dans le locus ribosomal; « Silent Information Regulator 2 » Sir3 : Protéine de levure impliquée dans l'inactivation de la transcription aux

télomères et au locus sexuel ou du « mating type »,« Silent Information Regulator 3»

Sir4 : Protéine de levure impliquée dans l'inactivation de la transcription aux télomères et au locus sexuel ou du « mating type »,« Silent Information Regulator 4»

SLl/TIF-IB : Complexe de sélectivité contenant TBP et ses facteurs associés; SLl pour « Selectivity Factor 1 » chez l'humain et TIF-IB pour « Transcription Initiation Factor IB » chez la souris

snoRNA : Court ARN nucléolaire; « small nucleolar RNA » snoRNP : Complexe ribonucléoprotéique nucléaire; « small nuclear ribonucleoprotein

particule » TAF]48 : Facteur associé à TBP dans le complexe SLl; « TBP Associated Factor 48 » TAFi68/63 : Facteur associé à TBP dans le complexe SLl; « TBP Associated Factor

68/63 »; souris/humain

V

TAFi95/l 10 : Facteur associé à TBP dans le complexe SLl; « TBP Associated Factor 95/110 »; souris/humain

TTF1 : Facteur 1 de terminaison de la transcription; « Termination Transcription Factor 1 ».

UBF : Facteur de transcription de l'ARN polymérase I; « Upstream Binding Factor » UCE : Klément du promoteur ribosomal aussi appelé UPE; « Upstream Control

Elément » ou « Upstream Promoter Elément »

vi

Liste des tableaux Tableau 1.1 Localisation de protéines sélectionnées au cours de la mitose 6

Tableau 4.1: Variation de l'expression de protéines nucléolaires 164

Tableau 5.1 Poids des fragments attendus suite à la digestion du clone 1F2 194 Tableau 5.2 Poids des fragments attendus suite à la digestion du clone AY0043 200

Liste des figures Figure 1. 1 Les domaines nucléolaires 4 Figure 1. 2 Visualisation des locus ribosomaux et des NORs mitotiques chez l'humain 4 Figure 1. 3 Modèle hypothétique tridimentionnel des différents niveaux de compaction des

NOR transcrits 9 Figure 1. 4 Les gènes ribosomaux 12 Figure 1. 5 Principales étapes de la production du ribosome 14 Figure 1. 6 La formation du complexe de pré-initiation chez les mammifères 15 Figure 1. 7 UBF et l'enhancesome 20 Figure 1. 8 Shématisation des étapes de transformations de TARN ribosomal précurseur et

des sous-unités ribosomales chez la levure S. cerevisiae 27 Figure 1. 9 Image de microscopie électronique de la transcription d'un gène ribosomal

mammifère 28 Figure 1. 10 Production des ARN ribosomaux matures 30 Figure 1. 11 La dominance nucléolaire 36 Figure 1. 12 Modèle d'inactivation des gènes ribosomaux chez la souris 39 Figure 1. 13 Structure de l'ADN-B méthylé 40 Figure 1. 14 Les ADN méthyltransférases mammifères et de leurs domaines 42 Figure 1. 15 Transfert du groupement méthyl 44 Figure 1. 16 Variants d'épissage de DNMT1 46 Figure 1. 17 Les deux isoformes de Dnmt3a 51 Figure 1.18: Les cinq isoformes de Dnmt3b 51 Figure 1. 19 Trois modèles expliquant la formation de l'hétérochromatine 53

Figure 2.1 ERK activation leads to an immédiate increase in ribosomal RNA transcription. 62

Figure 2.2 The MAP kinase ERK phosphorylates HMG boxes 1 and 2 of UBF 66 Figure 2.3 In vivo phosphorylation of UBF by ERK 68 Figure 2.4 Transactivation of the mouse and human Poli promoter by UBF1 and ERK site

mutants 71 Figure 2.5 ERK-site mutant UBFs localise correctly to the nucleolus but block MAP-kinase

activation of Poli 72 Figure 2.6 ERK phosphorylation of HMG box 1 diminishes its interaction with DNA 75

Figure 3.1 Growth factor enhancement of rRNA gène expression does not increase in RPI loading as determined by Nuclear Run-On 95

Figure 3.2 ChIP assays of RPI loading also show that growth factor enhancement of ribosomal RNA gène expression in NIH3T3 cells does not resuit from an increase in RPI loading 97

Figure 3.3 Direct détermination of elongation rates in growth-stimulated and inhibited NIH3T3 cells 99

Figure 3.4 High template occupancy by UBF inhibits RPI transcription in vitro 103 Figure 3.5 UBF inhibits RPI transcription downstream of+34 bp dépendent on its

phosphorylation state 104 Figure 3.6 UBF reversibly inhibits the full elongation phase of RPI transcription 106

IV

Figure 3.7 Engagement of UBF throughout the 45S rRNA coding région is not significantly affected by growth factor stimulation 107

Figure 3.8 EGF stimulation of 45S rRNA transcription in human cells does not lead to an increase in the number of RPI transcription complexes on the ribosomal gènes 111

Figure 4.1 : CpG mcthylation status across the human RPI promoter 127 Figure 4.2 : Active gène count in the PS and DNMT inactivated cell lines 129 Figure 4.3 : rRNA gène chromatin composition in the présence and absence of CpG

methylation 130 Figure 4.4 : DKO cells display a reduced prolifération rate and an RNA déficit but a normal

cell cycle distribution 131 Figure 4.5 : rRNA gène transcription rates and RPI loadings in the PS and cell lines and its

derivatives 132 Figure 4.6 : The DKO cells display a distended nucleolar structure 133 Figure 4.7 : RPI transcription factor and nucleolar protein concentrations in PS and DKO

cells 134 Figure 4.8 : 47S rRNA turn-over is affected by loss of DNA methylation 135 Figure 4.9 : Processing of the 47S rRNA is disrupted in the DKO cells 136 Figure 4.10 : RPII silencing is lost in the DKO cells 138 Figure 4.11: Analyse de type Northern faite avec une sonde dirigée contre le 47S sur une

préparation d'ARN total ou d'ARN polyA+ 155 Figure 4.12: Analyse comparative de la présence d'ARN polymérase II nucléolaire dans les

cellules PS et DKO 158 Figure 4.13: Le traitement des cellules parentales avec 5-aza-2'dexoxycytidine entraîne une

perte de la methylation de l'ADN 160 Figure 4.14: Le traitement des cellules parentales avec 5-aza-2'dexoxycytidine entraîne une

augmentation du nombre de gènes actifs 162 Figure 4.15: Classement, dans différents champs d'activité métaboliques, des ARNm dont

l'expression est augmentée 166 Figure 4.16: Classement, dans différents champs d'activité métaboliques, des ARNm dont

l'expression est diminuée 167 Figure 4.17: Cancer, croissance, prolifération cellulaire et maladie du système reproducteur.

168 Figure 4.18: Transport des molécules, biochimie de petites molécules et métabolisme des

glucides 169 Figure 4.19: Cancer, mort cellulaire et maladies respiratoire.. 170 Figure 4.20: Cycle cellulaire, croissance et prolifération cellulaire, développement et

fonctionnement du système sanguin 171 Figure 4.21: Croissance cellulaire et prolifération, mort cellulaire, développement et

fonctionnement du système sanguin 172 Figure 4.22: Maladies génétiques, du système sanguin et morphologie cellulaire 173 Figure 4.23: Métabolisme des acides aminés, transport moléculaire et biochimie des petites

molécules 174 Figure 4.24: Morphologie cellulaire, développement cellulaire et cancer 175 Figure 4.25: Cancer, maladie du système reproducteur et maladies dermatologiques 176

V

Figure 4.26: Développement et fonctionnement du système nerveux, morphologie tissulaire, développement et fonctionnement du tissus conjonctif. 177

Figure 4.27: Développement cellulaire, croissance cellulaire et prolifération, développement et fonctionnement du système sanguin 178

Figure 4. 28: Légende pour les figures 4.17 à 4.27 179

Figure 5. 1 Schématisation des étapes menant à l'inactivation génique chez la souris 186 Figure 5. 2 pGKDTAmUBFKO 189 Figure 5. 3: La protéine UBF de mammifère 190 Figure 5. 4 Criblage par PCR des clones potentiellement positifs 191 Figure 5. 5 Vérification de l'insertion de l'ADN provenant du vecteur pGKDTAmUBFKO

au locus codant pour UBF 193 Figure 5. 6 Southern fait avec l'ADN génomique du clone 1F2 194 Figure 5. 7 Souris chimériques obtenues avec les cellules ES provenant du clone 1F2.... 195 Figure 5. 8: Inactivation génique de TAF168 196 Figure 5. 9 Vérification de l'ARNm de fusion dans le clone AY0043 198 Figure 5. 10 Vérification du clone AY0043 200 Figure 5. 11 Southern fait avec l'ADN génomique du clone AY0043 201 Figure 5. 12: Souris chimériques obtenues suite à l'injection de cellules ES UBF +/-

provenant du clone AY0043 201 Figure 5. 13 Etapes menant à la production de cellules ES -/- complémentées par un

transgène sous le contrôle d'un promoteur inductible 204

Figure 6.1: Séquençage suite à un traitement de l'ADN avec du sodium bisulfite 224 Figure 6.2 : Conception des oligonucléotides pour la réaction d'amplification par PCR de

l'ADN traité au sodium bisulfite 225 Figure 6.3: Vérification de la conversion des cytosines non méthylées par digestion avec

l'enzyme de restriction Taql 226

Chapitre 1 : Introduction La transcription ribosomale est essentielle dans la biogenèse des ribosomes ce qui

lui procure une place centrale dans le fonctionnement cellulaire. Cependant, ses

mécanismes de régulation sont encore peu connus. Mes travaux de doctorat ont porté sur

l'étude de ces mécanismes de régulation et sur les protéines qui en sont responsables. Plus

précisément, je me suis intéressée au rôle de la protéine UBF dans la régulation de la

transcription ribosomale suite à un stimulus de croissance (chapitres 2 et 3). J'ai aussi

fabriqué les outils nécessaires à l'inactivation génique, chez la souris, des protéines UBF,

TAF168 et TAF)95 en vue de déterminer leurs fonctions in vivo et leur importance dans le

développement animal et dans la vie (chapitre 5). Finalement, le cœur de ma thèse concerne

l'étude de la méthylation de l'ADN dans le maintien d'une majorité de gènes ribosomaux

dans un état silencieux dans les cellules humaines (chapitre 4). Les résultats obtenus

permettent non seulement de mieux comprendre le rôle de la méthylation de l'ADN, mais

ils fournissent aussi une explication en ce concerne la conservation nécessaire, au cours de

l'évolution, d'une large proportion de gènes ribosomaux non transcrits.

Cette introduction se veut une présentation du contexte scientifique au

commencement de mon doctorat, soit en septembre 2003. Cependant, pour des raisons

pratiques, l'état des connaissances avant 2004 est présenté. Cette façon de faire permettra

au lecteur de bien comprendre les bases théoriques et logiques qui ont mené aux études

présentées dans cette thèse. Les résultats de ces travaux seront par la suite discutés dans le

contexte actuel des connaissances dans la conclusion.

1.1. Le nucléole Le nucléole fut la première structure intranucléaire identifiée par microscopie et elle

fut décrite en 1802 par Franz Bauer (Dundr et Misteli, 2001). Une des particularités du

nucléole c'est qu'il n'est pas délimité par une membrane. Toutefois, il est considéré comme

étant un compartiment cellulaire puisqu'il contient un ensemble de protéines résidantes

(Leung et al., 2006), qu'il peut être identifié morphologiquement par microscopie et qu'il

peut être isolé biochimiquement sous une forme enrichie (Busch et al., 1963). Le nucléole

est le lieu de la transcription ribosomale et sa formation ainsi que son volume sont

2

dépendants de celle-ci (Moss et Stefanovsky, 2000 et références incluses). Le nucléole est

désassemblé en mitose, ce qui coïncide avec un arrêt de la transcription ribosomale

(Dimario, 2004; Hernandez-Verdun et al., 2000 et références incluses). Par ailleurs, la

grosseur du nucléole fut l'un des premiers indicateurs utilisés pour caractériser les cellules

cancéreuses (White, 2005). En effet, puisque les cellules cancéreuses ont une prolifération

plus rapide que les cellules saines, il est logique de penser qu'elles doivent augmenter leur

transcription ribosomale de façon à produire plus de ribosomes. Conséquemment, leur

nucléole est plus volumineux (Derenzini et al., 2000; White, 2005).

Ce sont les régions organisatrices nucléolaires, ou NOR pour « Nucleolar

Organizing Région », qui sont responsables de la formation du nucléole. Les NORs furent

d'abord décrits morphologiquement comme étant situées aux constrictions secondaires

présentes sur certains chromosomes; la constriction primaire étant le centromère. Chez les

humains, les bras courts des cinq chromosomes acrocentriques, 13, 14, 15, 21 et 22 sont

porteurs de ces constrictions secondaires (Raska et al., 2004). Des observations suggérant

que la formation du nucléole commence aux constrictions secondaires furent publiées dès

1931, mais ce n'est qu'en 1934 que Barbara McClintock suggéra que la formation des

nucléoles était la conséquence d'une activité associée aux NORs. Elle démontra qu'une

translocation réciproque entre deux chromosomes porteurs de NORs résultait en la

formation de deux nucléoles de taille différente et que cette taille était indépendante de la

quantité de matériel (ADN) portée par les chromosomes. Elle suggéra que le chromosome

portant la plus petite quantité de matériel produisait un plus grand nucléole puisqu'il avait

une plus grande capacité fonctionnelle pour une activité non identifiée (Dimario, 2004). Il

est maintenant connu que les NORs sont les locus ribosomaux et que l'activité suggérée par

B. McClintock est la transcription ribosomale (Moss, 2004; Raska et al., 2004).

Le nucléole en interphase est subdivisé en trois domaines (figure 1.1) où ont lieu des

activités distinctes. Il s'agit du centre fibreux, FC ou « Fibrillar Center », du compartiment

fibreux dense, DFC ou «Dense Fibrillar Component », et du compartiment granulaire, GC

ou « Granular Component » (respectivement f, d et g sur la figure 1.1) (Dundr et Misteli,

2001; Moss et al., 2006a et références incluses). Des hybridations in situ et des expériences

3

d'incorporation de Br-UTP ont montré que la transcription ribosomale avait lieu à la

périphérie du FC et non en son centre (Cmarko et al., 1999; Dundr et Raska, 1993; Hozak

et al., 1994; Puvion-Dutilleul et al., 1997). De plus, ces expériences ont montré que les

deux autres compartiments nucléolaires étaient dépourvus de rDNA. Puisque la

transcription a lieu au pourtour du FC, les transcrits en élongation ainsi que les ARNr

complets sont probablement déjà en contact avec le DFC (figure 1.1 B). Cet arrangement

tridimensionnel permet vraisemblablement l'assemblage cotranscriptionel des sous-unités

ribosomalcs. Environ 10 clivages de l'ARNr précurseur ainsi que pas moins de 115

méthylations et près de 95 conversions d'uridines en pseudouridines débutent dans le DFC

pour se terminer dans le GC (Dundr et Misteli, 2001). Environ 200 snoRNP, pour « small

nucleolar ribonucleoprotein particles », contenant des ARN de façon stable sont impliqués

dans ces transformations. Les sous-unités ribosomales en maturation passent ensuite du

DFC au GC ce qui confère au GC son allure granuleuse. Par la suite les sous-unités vont

sortir du nucléole, traverser le noyau et les pores nucléaires pour terminer leur maturation

dans le cytoplasme (Fromont-Racine et al., 2003; Tschochner et Hurt, 2003). Une sous-

unité 40S et une sous-unité 60S seront finalement assemblées pour former un ribosome

mature.

Il est possible de visualiser de façon sélective, par une coloration au nitrate d'argent,

les nucléoles en interphase et les NORs mitotiques (figure 1.2). Cette visualisation est

possible car ces deux structures sont riches en protéines acides qui sont des protéines

argyrophiliques. Les protéines nucléophosmine et nucléoline sont les protéines marquées à

l'argent qui permettent la visualisation du nucléole à l'interphase (Roussel et Ilernandez-

Verdun, 1994; Valdez et al., 1995). Ces protéines sont impliquées dans les étapes de

transformations de l'ARNr précurseur et sont associées au DFC. Le marquage de ces deux

protéines, présentes en abondance dans les cellules tumorales, est utilisé de façon routinière

comme indicateur de pronostic de tumeurs (Derenzini, 2000).

4

Figure 1.1 Les domaines nucléolaires.

A) Microscopie électronique montrant les trois régions du nucléole, f : « Fibrillar Center », d : « Dense Fibrillar Component » et g : «Granular Component». Tiré de Raska et al., 2003 B) Représentation schématique de l'organisation des trois domaines nucléolaires. FC : « Fibrillar Center », DFC : « Dense Fibrillar Component » et GC : «Granular Component». L'ADN est représenté par la ligne noire, les polymérases par des points rouges et les transcrits en élongation par une région verte. Provient de Dundr et Misteli,

FISH Coloration à l'argent DAPI

Figure 1.2 Visualisation des locus ribosomaux et des NORs mitotiques chez l'humain. A) FISH sur les chromosomes humains en utilisant une sonde dirigée contre le 18S. Les lettres A, B et C représentent différents niveaux d'intensité de fluorescence A étant la plus forte intensité et C la plus faible. Les dix chromosomes acrocentriques sont marqués par cette technique. B) Coloration à l'argent sur les mêmes chromosomes qu'en A. Les flèches représentent une faible coloration et la tête de flèche représente une absence de signal. C)

5

Marquage au DAPI sur les mêmes chromosomes qu'en A et en B. Modifié de Guillen et al., 2004.

Par ailleurs, il est suggéré que les protéines responsables de la coloration à l'argent

des NORs mitotiques soient UBF et la plus grosse sous-unité de l'ARN polymérase I (Pol

I) (Roussel et Hernandez-Verdun, 1994) ou la plus grosse sous-unité de Poil seule (Bell et

al., 1997). La comparaison de la localisation de tous les NORs par F1SII, pour

« Fluorescence In Situ Hybridation », à celle des NORs marqués à l'argent montre que

certains NORs ne sont pas associés à des protéines argyrophiliques (figure 1.2A et B)

(Guillen et al., 2004). L'association de protéines impliquées dans la transcription

ribosomale avec une sous-population de NORs mitotiques suggère que ce sont ces NORs

qui sont réactivés à la sortie de la mitose. Une étude faite par Roussel et collègues, dans des

cellules humaines HeLa montrent qu'effectivement ce sont les NORs associés avec des

facteurs de transcription, donc colorés à l'argent, qui sont spécifiquement réactivés

(Roussel et al., 1996). Les auteurs proposent qu'il puisse s'agir d'un mécanisme de

mémoire quant au nombre de NORs actifs d'un cycle cellulaire à un autre.

Un tel mécanisme de mémoire serait important puisque le nucléole se désassemble

au cours de la mitose ou, plus précisément, du début de la prophase à la fin de l'anaphase-

début de la télophase (Dimario, 2004). Ce désassemblement nucléolaire corrèle avec un

arrêt de la transcription ribosomale en mitose suivie d'une réactivation graduelle en Gl et

d'une activité maximale en phases S et G2 (Klein et Grummt, 1999). La mitose commence

par une rapide augmentation de l'activité de la kinase cdkl/cycline B qui phosphoryle un

ensemble de substrats (Basi et Draetta, 1995). Le complexe SL1, plus précisément TBP et

TAFillO, est une des cibles de cette kinase et sa phosphorylation bloque la transcription

Poli in vitro (section 1.2.1.2) (Heix et al., 1998). Cette phosphorylation empêche

l'interaction entre SL1 et UBF ce qui inhibe la formation du complexe de pré-initiation

(PIC) et donc la transcription ribosomale (section 1.2.1) (Heix et al., 1998). L'activité de

SL1, soit sa participation à la formation du PIC, est rétablie au début de la phase Gl, mais

cette activité n'est pas suffisante pour réactiver la transcription (Klein et Grummt, 1999) car

UBF est aussi inactivé par une kinase au cours de la mitose (section 1.2.1.4) (Klein et

Grummt, 1999). La phosphorylation d'UBF sur la serine 484 par cdk4/cycline D et

6

cdk2/cycline E est importante dans la réactivation graduelle de la transcription ribosomale

pendant la phase Gl (Voit et al., 1999). La combinaison de cette phosphorylation d'UBF à

l'activité d'une phosphatase, pour enlever la phosphorylation mitose-spécifique, est

responsable de la réactivation progressive de la transcription ribosomale en phase Gl

(Klein et Grummt, 1999). L'identité de la kinase mitotique et de la phosphatase ainsi que

les effets des phosphorylations par la kinase mitotique et par cdk4/cycline D et

cdk2/cycline E sur la serine 484 ne sont pas caractérisés.

CB Cyto EPC NDF PNB NOR

SnoRNAs(U3,U14) + + + +

Protéines du GFC et du GC :

Fibrillarine

B23

+ + +

+

+

+ +

Protéines du FC (Poli, UBF, topo I) + i

ARNr précurseur (région de) + + +

Tableau 1.1 Localisation de composants nucléolaires sélectionnées au cours de la mitose. CB : corps de Cajal; « Cajal Body », Cyto : cytoplasme, EPC : envelope périchromosomale, NDF : « nucleolar derived foci », PNB : « perinucleolar body », NOR : « nucleolar organizing région ». Tiré de Dimario, 2004.

La localisation mitotique des protéines nucléolaires diffère selon leur compartiment

d'origine (tableau 1.1). Les protéines du FC, impliquées dans la transcription ribosomale,

restent associées aux NORs mitotiques. Il a été montré que des protéines autres qu'UBF et

Poli (Bell et al., 1997; Roussel et Hernandez-Verdun, 1994), aussi importantes pour la

transcription ribosomale, restent attachés aux NORs tout au long de la mitose. Il s'agit de

certaines protéines du complexe SLl (Heix et al., 1998; Roussel et al., 1996), de la protéine

de terminaison TTF1 (Sirri et al., 1999) et de la topoisomérase I (Rose et al., 1988). Des

immunoprécipitations de chromosomes mitotiques ou d'extraits cytoplasmiques suggèrent

que la majorité des protéines associées à la transcription ribosomale, si ce n'est la totalité

d'entres-elles, reste associée aux NORs lors de la mitose (Roussel et al., 1996). Le sort des

7

protéines du DFC et du GC est différent. Certaines d'entres-elles sont relarguées dans le

cytoplasme, alors que d'autres s'associent avec la région périchromosomale, entourant le

chromosome, pour former ce qui est décrit comme une enveloppe (tableau 1.1) (Dimario,

2004 et références incluses). Plusieurs laboratoires ont travaillé à la caractérisation du

contenu nucléolaire de l'enveloppe périchromosomale et ont montré que la majorité des

protéines du GFC et du GC s'y retrouvent (tableau 1.1) (Dimario, 2004 et références

incluses). Les protéines de cette enveloppe vont ségréguer avec les chromosomes clans les

cellules filles (Dimario, 2004).

Plusieurs auteurs ont suggéré des modèles pour expliquer la compaction des NORs

lors de la mitose (Gebranc-Younes et al., 1997; Heliot et al., 1997; Suja et al., 1997). Celui

de Heliot et collègues est présenté à la figure 1.3 et est décrit ici. En interphase, il est

suggéré que les gènes ribosomaux transcrits forment une fibre de 80nm de diamètre qui

émane sous forme de boucle d'une fibre AT riche de 250nm de diamètre (figure 1.3A). Au

cours de la mitose, la fibre de 80nm est de plus en plus condensée, (figure 1.3B et C) et elle

est finalement enroulée autour de la fibre de 250nm qui est considérée comme étant l'axe

de la constriction secondaire (figure 1.3D). Lors de la métaphase, les deux NORs des deux

chromatides sœurs vont partiellement se juxtaposer pour former une structure en croissant

(figure 1.3E, vue frontale et F, vue du profil arrière) (Heliot et al., 1997). L'analyse de la

structure des chromosomes en mitose suggère que la fibre ribosomale, originalement de

80nm, est dix fois moins condensée que le reste de la chromatine (Heliot et al., 1997; Suja

et al., 1997). La figure 1.3 présente le chromosome humain 13 en métaphase (1.3G, vue

frontale, et H, vue arrière). Cette représentation permet de visualiser la constriction

primaire (le centromère) et la constriction secondaire (les NORs) qui résultent d'un

rapprochement des deux chromatides sœurs à ces endroits. Il est à noter que les NORs

silencieux ne forment pas de constriétions secondaires (McStay, 2006).

La nucléologenèse (la formation du nucléole) se fait en deux étapes à la sortie de la

mitose (Dimario, 2004). La première est le recrutement de corps prénucléolaires (ou PNB

pour « Prenucleolar bodies ») par les NORs compétents c'est-à-dire ceux associés avec des

protéines de la transcription (Dousset et al., 2000). Les PNBs contiennent des protéines de

s transformation de l'ARNr précurseur et des ARNr précurseurs qui ont été produits à la fin

de la phase G2 ou au début de la mitose (tableau 1.1). C'est le démantèlement de

l'enveloppe périchromosomale qui mène à la formation séquentielle de foyers nucléolaires,

ou NDF pour « Nucleolar Derived Foci », en anaphase puis de PNBs à la fin de l'anaphase

et au début de la télophase. Cette première étape est indépendante de la transcription

ribosomale (Dimario, 2004). Il est intéressant de noter que les PNBs et les corps de Cajal

ont un contenu protéiniquc en partie semblable (tableau 1.1). En effet, comme les PNB, les

corps de Cajal contiennent le snoRNA U3 et des protéines nucléolaires impliquées dans la

transformation des ARNr comme, par exemple, la fibrillarine, mais pas la nucléoline ou

B23 (Dimario, 2004). Les corps de Cajal contiennent aussi la grande sous-unité de Poli, la

topoisomérase I, la protéine ribosomale S6, le snoRNA U8 et certaines composantes de

transformation des pré-mRNA. Les corps de Cajal ne participent pas à la formation du

nucléole et leur fonction pourrait être de pré-assembler ou de modifier chimiquement des

protéines ou des composantes, par exemple des snoRNA, pour ensuite les diriger vers le

nucléole ou les sites de transformation des ARNm (Dimario, 2004).

La deuxième étape de la nucléologenèse est le transfert du contenu des PNBs au

nucléole en formation. Cette étape est dépendante de la réactivation de la transcription

ribosomale (Dousset et al., 2000). Des études sur le contenu des PNBs ont montré que ces

structures sont dynamiques (Savino et al., 2001) et qu'il en existerait deux types (Dundr et

al., 2000). Le premier contient des protéines impliquées dans les étapes précoces de

transformation de l'ARNr précurseur, comme la fibrillarine, tandis que le deuxième

contient des protéines impliquées dans les étapes plus tardives de transformation. Ainsi, il y

a un transfert séquentiel de matériel des PNBs du premier type au deuxième type puis de

ces PNBs vers le ou les nucléole(s) en formation (Dundr et al., 2000). Ce transfert aurait

comme effet d'accroître le volume du nucléole jusqu'à l'obtention d'un nucléole complet et

fonctionnel. Dans les cellules qui contiennent plusieurs NORs, les nucléoles en formation à

chacun des NORs finiront par s'associer pour former un ou quelques nucléoles. Ce

processus est nommé fusion nucléolaire.

Fibre AT. riche «SOnm?

Fibre ribosomale

Fibre AT riche (250nm:

Fibre ribosomale Constrlctlon secondaire (NOR)

Cenlromère

Chromosome 13 (vue frontale)

Chromosome 13 (vue arrière)

Figure 1.3 Modèle hypothétique tridimensionnel des différents niveaux de compaction des NOR transcrits.

A) Représentation de la chromatine ribosomale en interphase. La fibre ribosomale de 80nm de diamètre, en jaune, émane sous forme de boucle d'une fibre AT riche de 250 nm de diamètre, en rouge. B) à H) Représentations des différents états de compaction de la chromatine ribosomale lors de la mitose. B) et C) Compaction graduelle de la fibre ribosomale par enroulement de celle-ci sur elle-même. D) Enroulement de la fibre ribosomale compactée autour de la fibre AT riche de 250 nm de diamètre. E) et F) Les deux NORs forment une structure caractéristique en forme de croissant lors du rapprochement des deux chromatides sœurs au cours de la métaphase. Une vue frontale de cette structure (en E) et une vue du profil arrière (en F) sont montrées. G) et FI) Structure tridimensionnelle du chromosome 13 humain et identification de la constriction primaire (centromère) et secondaire (NORs) et des bras court et long du chromosome. G) Vue frontale et H) vue arrière. La chromatine ribosomale est en jaune et le chromosome porteur en rouge. Les deux bandes référence représentent 250nm. Modifié de Fleliot et al., 1997.

Longtemps considéré comme étant simplement le lieu de la transcription ribosomale

et de l'assemblage des sous-unités ribosomales, le nucléole est maintenant connu pour avoir

10

plusieurs autres fonctions. Par exemple, plusieurs courts ARN, dont PUôsnRNA, transitent

par le nucléole pour subir des transformations (Gerbi et al., 2003). Chez S. cerevisiae, il a

été montré que les différentes régions chromosomales codant pour les ARN de transfert

sont localisées dans le nucléole (Thompson et al., 2003). Les auteurs ont montré que cette

organisation est dépendante de la transcription ribosomale, de la transcription des tRNAs et

d'un nucléole intact (Thompson et al., 2003). Cette organisation tridimensionnelle permet

peut-être la formation d'un sous-domaine nucléolaire dédié à la transcription par l'ARN

polymérase III. La localisation, chez la levure, de la séquence codant pour l'ARN 5S au

sein du locus ribosomal suggère que sa transcription soit aussi nucléolaire et pourrait avoir

lieu dans ce sous-domaine. Si cette hypothèse était vérifiée, cela suggérerait que l'intégrité

du nucléole est aussi importante pour l'activité de l'ARN polymérase III. Cependant, il

n'existe aucune preuve de l'existence de ce domaine.

Une autre fonction du nucléole est son activité régulatrice sur un des gènes

suppresseur de tumeur le plus étudié soit la protéine p53 (Prives et Hall, 1999). Rubbi et

Milner ont montré que tous les stress cellulaires qui induisent une stabilisation de p53

induisent aussi le démantèlement du nucléole (Rubbi et Milner, 2003). Il est connu que

MDM2 médie la dégradation de p53 via le sentier de l'ubiquitination et que MDM2 est à

son tour régulé par la protéine nucléolaire pl9ARF qui séquestre MDM2 au nucléole (Weber

et al., 1999). Cette séquestration entraîne la stabilisation et l'accumulation nucléaire de p53

en inhibant l'interaction de cette dernière avec MDM2. De plus, Rubbi et Milner ont

montré qu'en absence du démantèlement du nucléole, des bris dans l'ADN génomique

n'induisent pas la stabilisation de p53 (Rubbi et Milner, 2003). À l'opposé, un

démantèlement du nucléole, provoqué par la microinjection d'anticorps dirigé contre UBF,

mais sans bris de l'ADN induit une stabilisation de p53. Les auteurs suggèrent que le

nucléole est un senseur de stress qui, lorsque ces activités sont altérées, médie la

stabilisation de p53. Dans ce contexte, il est intéressant de noter que la transcription

ribosomale est diminuée par p53 (section 1.2.1.2). En conclusion, les nouvelles fonctions

nucléolaires, dont quelques unes sont décrites dans cette section, renforcent le rôle central

du nucléole et, par conséquent de la transcription ribosomale, dans le bon fonctionnement

cellulaire.

11

1.2. La biogenèse des ribosomes La capacité d'une cellule à produire des protéines limite sa croissance. La

production des ribosomes, qui permettent la traduction des ARN messagers en protéines,

est quant à elle limitée par la transcription des ARN ribosomaux qui sont au centre de la

structure et de la fonction de ceux-ci (Moss et al., 2006b; Moss et Stefanovsky, 2002).

Ainsi, la transcription ribosomale est limitante pour la croissance cellulaire puisqu'elle

limite la production des protéines.

La biogenèse des ribosomes est un processus qui utilise une importante partie des

ressources cellulaires (Moss, 2004; Moss et Stefanovsky, 2002; Warner, 1999). En effet,

celle-ci requiert la participation des trois ARN polymérases et nécessite la coordination de

centaines de protéines pour l'assemblage des deux sous-unités ribosomales (Moss et

Stefanovsky, 2002). L'ARN polymérase I transcrit les ARN ribosomaux (ARNr) qui sont

les ARNs cellulaires les plus abondants, et l'ARN polymérase III (PolIII) transcrit l'ARN

5S (Moss et al., 2006a). L'ARN polymérase II (PolII) quant à elle, transcrit les ARN

messagers (ARNm) codant pour les protéines ribosomales (intégrées dans les sous-unités

ribosomales) (Hernandez-Verdun et al., 2002; Raska et al., 2004), les snoRNA et les

ARNm codant pour les protéines non ribosomales nécessaires aux transformations co et

post-transcriptionnelles des ARN ribosomaux et à l'assemblage des sous-unités

ribosomales. La transcription par l'ARN polymérase I représente entre 35% et 60% de la

transcription d'une cellule eucaryote en croissance (Moss et Stefanovsky, 2002; Warner,

1999) et les ARN ribosomaux représentent environ 80% des ARN cellulaires totaux (Moss

et al., 2006a, Warner, 1999). La transcription de l'ARN 5 S et des snoRNA représente

quant à elle de 10 à 20% de la transcription cellulaire (Moss et Stefanovsky, 2002). Ainsi,

le pourcentage de la transcription cellulaire dédiée à la biogenèse des ribosomes représente

près de 80% de l'activité transcriptionnelle chez la levure et près de 50% chez une cellule

mammifère (Moss et Stefanovsky, 2002). De plus, la formation même du nucléole est

dépendante de la transcription des ARNr (section 1.1) (Dimario, 2004; Moss et

Stefanovsky, 2002). Le rôle essentiel de la transcription ribosomale dans la biogenèse des

ribosomes et dans la fonction du nucléole suggère que la régulation de cette transcription

est cruciale pour le fonctionnement et le devenir cellulaire.

12

1.2. 1 Les gènes ribosomaux et leur transcription Les gènes ribosomaux, ADNr, sont présents en plusieurs centaines de copies dans le

génome haploïde de la majorité des organismes eucharyotes étudiés (Moss et Stefanovsky,

2002). Il y en aurait entre 200 et 250 copies chez l'humain (Dimario, 2004), environ 150

copies chez Saccharomyces cerevisiae (Warner, 1999) et entre 150 et 200 copies chez

Drosophila melanogaster (Dimario, 2004). Ces gènes sont généralement répétés en tandem

selon une organisation tête à queue (figure 1.4A) et sont répartis sur un ou plusieurs

chromosomes selon les espèces. Ainsi, chaque chromosome porteur contient un nombre

élevé de gènes ribosomaux. La présence de ces gènes répétés définie une région

chromosomique nommée région organisatrice nucléolaire (section 1.1) (Moss et al., 2006a;

Raska et al., 2004).

A ) Tandem rRNA n . n i •*C=J*M=> I c3«c=3 OczwKrrj I

gène repeats • J" T T ^ •

Transcribed segment 13.7 kb Intergenic spacer 29.3 kb

B) rRNA gène -4kzZMH l [k § 18S 5.8S 28S

C) rRNA gène r transcript

5' 18S 5.8S 28S 3' B Promotor D External transcribed spacer □ rRNA-coding séquence and

RNA séquence ■ Internai transcribed spacer ■ External norvtranscribed spacer

Figure 1.4 Les gènes ribosomaux.

Représentation du locus ribosomal en A, d'un gène ribosomal en B et de l'ARN ribosomal précurseur en C. La région promotrice est en mauve, les ETS sont en jaune, les régions codant pour les ARN ribosomaux sont en rose, les ITS sont en rouge et l'IGS ou « external non-transcribed spacer » est en bleu. Tiré de Raska, 2003.

Chaque unité ribosomale est constituée d'une région transcrite et d'un espace

intergénique ou IGS pour « Intergenic spacer » (figure 1.4B) (Raska et al., 2004). La région

transcrite comprend : une région externe en 5', 5'ETS pour 5' « External Transcribed

Spacer», la séquence codant pour l'ARN 18S, une première région interne ou ITS 1

13

pour « Internai Transcribed Spacer », la séquence codant pour l'ARN 5,8S, une deuxième

région interne, 1TS 2, la séquence codant pour l'ARN 28S et finalement une région externe

en 3', 3'ETS (figure 1.4C). La région promotrice, d'environ 150pb, contient deux

éléments : l'UCE pour « Upstream Control Elément » et le « core » (figure 1.6) (Moss et

Stefanovsky, 2002). L'ARN polymérase I transcrit un seul ARN ribosomal, l'ARNr

précurseur, qui contient tous les éléments de la région transcrite (figure 1.4C). L'ARN

ribosomal précurseur est nommé 47S chez l'humain, 45S chez la souris et 35S chez la

levure S. cerevisiae. Cet ARN est clivé pour former les trois ARN ribosomaux matures

nommés 18S, 5,8S et 28S (25S chez S. cerevisiae) (Moss et al., 2006a; Moss et

Stefanovsky, 2002). L'assemblage des deux sous-unités ribosomales se fait de façon co-

transcriptionnelle et requiert l'association de l'ARN 5S et des protéines ribosomales avec

les ARN ribosomaux (figure 1.5) (Fatica et Tollervey, 2002; Fromont-Racine et al., 2003;

Moss et al., 2006a; Tschochner et Hurt, 2003). La petite sous-unité, nommée 40S, contient

l'ARNr 18S et environ 32 protéines ribosomales. La grosse sous-unité, nommée 60S,

contient les ARNr 5,8S et 28S, l'ARN 5S et environ 46 protéines ribosomales (figures 1.5

et 1.8) (Johnson et al., 2002; Moss et al., 2006a). Puisque l'assemblage des sous-unités

ribosomales se fait de façon co-transcriptionnelle et non de façon séquentielle, la situation

est plus complexe que celle présentée à la figure 1.5et elle sera discutée à la section 1.3.

La transcription ribosomale commence par la formation du complexe de pré­

initiation, PIC, au promoteur. La figure 1.6 présente la formation du PIC chez les

mammifères. Le complexe de pré-initiation contient la protéine UBF, pour « Upstream

Binding Factor », le complexe SL1, « Selectivity Factor 1 » ou TIF-IB chez la souris, la

protéine Rrn3, ou TIF-IA chez la souris, et l'ARN polymérase I (Moss, 2004; Moss et

Stefanovsky, 2002). Ces protéines seront décrites individuellement un peu plus loin

(sections 1.2.1.1 à 1.2.1.4). L'événement initiateur de la formation du PIC est le

recrutement du facteur architectural UBF au promoteur ribosomal (figure 1.6A et B) (Moss

et al., 2006a; Moss et Stefanovsky, 2002). UBF, qui contient six boîtes HMG1, possède la

capacité de s'attacher à l'ADN et d'en modifier la conformation Bazett-Jones et al., 1994)

Stefanovsky et al., 1996; Stefanovsky et al., 2001a). Il est suggéré que deux dimères d'UBF

s'attachent au promoteur ribosomal de façon à placer côte à côte les deux éléments du

14

promoteur soit l'UCE et le « core » (figure 1.7, section 1.2.1.4) (Stefanovsky et al., 1996;

Stefanovsky et al., 2001a). Par la suite, UBF recrute le complexe SLl (Hempel et al., 1996;

Kihm et al., 1998; Tuan et al., 1999) qui contient la protéine TBP, pour « TATA-Binding

Protein », et trois facteurs associés ou TAFis, pour « TBP Associated Factors I» (Cornai et

al., 1992; Eberhard et al., 1993). Il n'est pas encore connu si un ou deux complexes SLl

est/sont recruté(s) au promoteur (figure 1.6C) (Moss et Stefanovsky, 2002; Stefanovsky et

al., 1996). UBF et SLl sont tous deux en contact avec les séquences de l'UCE et du «core »

(Tuan et al., 1999). La dernière étape de la formation du PIC est le recrutement de l'ARN

polymérase I en association avec la protéine Rrn3 phosphorylée (figure 1.6D). Ce

recrutement se fait via les interactions d'UBF avec PAF53, pour « Polymérase associated

Factor 53 » (Hanada et al., 1996; Seither et al., 1997) et de SLl avec Rrn3 (Miller et al.,

2001; Yuan et al, 2002).

5 E T S CH3 CH3 CH3 CH3 CH3 CH3 3 F T S

B)

C)

às£5vx 41!',

Figure 1.5 Principales étapes de la production du ribosome. A) Représentation de l'ARN ribosomal précurseur mammifère aussi nommé 47S. La localisation des séquences codant pour les trois ARNr matures, des ETS, des ITS, de certaines méthylations (CH3) et de quelques pseudouridinations (trident) est indiquée. Les séquences 5'ETS, ITS 1, ITS 2 et 3'ETS sont aussi indiquées. B) Représentations de certains intermédiaires de clivages chez les mammifères. C) Composition des deux sous-

15

unités ribosomales à gauche et représentation d'un ribosome à droite. S pour « Small subunit », le 40S, et L pour «Large subunit », le 60S. Tiré de Ruggero et Pandolfi, 2003.

—es*— ^ ^ U B F d l m e r

Figure 1.6 La formation du complexe de pré-initiation chez les mammifères. A et B) Recrutement d'un ou de deux dimères d'UBF aux éléments UCE et « core » du promoteur ribosomal. C) Recrutement d'un ou de deux complexes SLl via une interaction de ce complexe avec UBF et l'ADN ribosomal. D) Recrutement de l'ARN polymérase I associée à la protéine Rrn3 via des interaction avec SLl et UBF. La sous-unité RPA43 de Poli est aussi indiquée puisqu'elle est essentielle pour l'interaction de la polymérase avec Rrn3. Le facteur Paf53 est responsable de l'interaction entre UBF et Poli. Tiré de Moss, 2004; Moss et al., 2006a; Moss et Stefanovsky, 2002.

1.2.1.1. L'ARN polymérase I L'ARN polymérase I transcrit exclusivement les ARN ribosomaux (Moss, 2004;

Moss et al., 2006a; Moss et Stefanovsky, 2002). Il s'agit d'un large complexe qui contient

11 sous-unités chez les mammifères et 14 chez la levure S. cerevisiae (Moss et al., 2006a).

Deux formes de Poli, Pollct et PolIp\ ont été isolées chez la levure et les cellules

mammifères (Milkereit et Tschochner, 1998; Miller et al., 2001). Ces deux complexes sont

16

capables de transcription sur des substrats non spécifiques, mais seul Poli[3, qui représente

environ 2% des complexes, est compétente pour initier la transcription in vitro (Milkereit et

Tschochner, 1998; Miller et al., 2001). La diminution de Polip a été associée, chez la

levure, à la phase stationnaire de croissance et, chez les mammifères, à une longue période

de famine ou à un traitement à la cycloheximide, un inhibiteur de la traduction (Yuan et al.,

2002). Ainsi, une diminution de la transcription ribosomale est associée à une réduction du

nombre de polymérases compétentes à initier la transcription. La protéine responsable de

cette compétence est Rrn3 (section 1.2.1.3) (Miller et al., 2001; Yuan et al., 2002), une

protéine conservée chez les cellules eucaryotes (Bodem et al., 2000; Moorefield et al.,

2000). L'interaction de Rrn3 avec RPA43, une sous-unité de Poli, est aussi conservée dans

les cellules eucaryotes. Chez les mammifères, la protéine Rrn3 interagit aussi avec PAF67

qui est une protéine associée à Poli (Yuan et al., 2002). La présence de Rrn3 est essentielle

pour la formation du premier lien phosphodiester (Schnapp et Grummt, 1991; Schnapp et

al., 1993) et sa phosphorylation est requise pour la compétence de la polymérase chez les

mammifères (Yuan et al., 2002).

L'ARN polymérase I a été purifiée chez le brocoli (Brassica oleracea), la souris, la

grenouille Xenopus et le rat sous la forme d'une holoenzyme contenant la majorité des

facteurs nécessaires à l'initiation de la transcription in vitro en plus de contenir différentes

enzymes (Albert et al., 1999; Hannan et al., 1999; Saez-Vasquez et Pikaard, 1997; Seither

et al., 1998). Ce modèle d'holoenzyme est remis en cause par des résultats qui suggèrent

plutôt que les facteurs impliqués dans la transcription Poli, ainsi que les sous-unités de

Poli, entrent dans le nucléole individuellement et non en complexes pré-asssemblés (Dundr

et al., 2002). Les auteurs de cette étude suggèrent que la stabilité des complexes de pré­

initiation est augmentée suite à l'association des différentes protéines ce qui explique la

purification de ces complexes en tant qu'holoenzymes (Dundr et al., 2002). Ces mêmes

expériences ont aussi suggéré que Poli se désassemble à la fin d'un cycle de transcription

pour se réassembler avant la prochaine initiation (Dundr et al., 2002). Des études ont aussi

montré que Poli et Rrn3 sont recyclées alors que SL1 et UBF restent associés au promoteur

(Panovetal., 2001).

17

1.2.1.2. SLl Le complexe SLl contient la protéine TBP et trois TAFis (Cornai et al., 1992;

Eberhard et al., 1993; Heix et al., 1997). Il s'agit de TAF,48, de TAF^S/63 et de

TAF|95/110, souris/humain, nommés ainsi selon leur poids moléculaire apparent. Les

protéines TAF|68/63 et TAFi48 présentent une certaine homologie avec respectivement les

protéines Rrn7p et Rrnllp de la levure Schizosaccharomyces pombe (Boukhgalter et al.,

2002). D'autres TAFj existent probablement car l'utilisation d'un complexe SLl

recombinant contenant TBP et ses trois TAF| n'est pas compétent pour initier la

transcription (Moss et Stefanovsky, 2002). Le complexe SLl a la particularité d'être

espèce-spécifique. Il n'est pas possible de transcrire un gène ribosomal humain dans des

cellules somatiques hybrides souris/humain ou dans des extraits nucléaires de cellules de

souris si le complexe SLl humain n'est pas présent et vice-versa (Heix et Grummt, 1995).

La présence de TBP dans les trois complexes d'initiation, SLl, TFIID et TFIIIB, suggère

que cette protéine puisse être utilisée pour permettre la corégulation des trois systèmes de

transcription. Toutefois, avant 2004, peu de modifications avaient été identifiées comme

visant uniquement TBP.

Les modifications connues des protéines du complexe SLl ont toutes la

caractéristique d'influencer la formation du complexe de pré-initiation. TBP et TAFi95/l 10

sont phosphorylés par la kinase cdkl/cycline B au début de la mitose (Heix et al., 1998).

Ces phosphorylations inhibent l'interaction de SLl avec UBF et sont importantes pour

l'arrêt de la transcription (Heix et al., 1998). L'acétylation de TAF[68/63 par PCAF

augmente la transcription ribosomale en favorisant l'interaction de cette protéine avec le

promoteur ribosomal (Muth et al., 2001). À l'inverse, la déacétylation de TAFi68/63 par

Sir2, une histone déacétylase, inhibe la transcription ribosomale (Muth et al., 2001). Les

protéines interagissant avec SLl, tout comme les modifications de ce complexe, ont aussi la

caractéristique d'influencer la formation du PIC. La protéine p53 diminue la transcription

ribosomale en s'associant avec SLl et en inhibant son interaction avec UBF (Zhai et

Cornai, 2000). Finalement, la protéine Rrn3 interagit avec le complexe SLl, via TAFi68/63

et TAFi95/110, lors du recrutement de l'ARN polymérase au promoteur (Miller et al.,

2001; Yuan et al., 2002). Les résultats disponibles suggèrent que la fonction du complexe

18

SL1, dans la transcription ribosomale, est limitée à l'étape de la formation du complexe de

pré-initiation.

1.2.1.3 Rrn3 Rrn3, aussi appelée TIF-IA chez la souris, est une protéine conservée de la levure à

l'humain (Bodem et al., 2000; Moorefield et al., 2000). Comme il a été mentionné dans la

section sur TARN polymérase I (1.2.1.1), l'interaction de Poli avec la protéine Rrn3

phosphorylée détermine la compétence de la polymérase à initier la transcription

ribosomale chez les mammifères (Miller et al., 2001; Yuan et al., 2002). L'influence de la

phosphorylation de Rrn3 sur l'activité de Poli chez la levure n'a pas été démontrée. Rrn3

interagit avec une sous-unité de la polymérase, RPA43 (Fath et al., 2001; Yuan et al.,

2002), et, chez les mammifères, avec un facteur associé à la forme compétente de la

polymérase PAF67 (Seither et al., 2001). Rrn3 est relargué de la polymérase et inactivé

lorsque débute l'étape de l'élongation des transcrits chez les mammifères (Hirschler-

Laszkiewicz et al., 2003; Milkereit et Tschochner, 1998).

Comme il a été mentionné précédemment, la forme PolI(3 varie en fonction de la

croissance cellulaire. Ces résultats suggèrent que Rrn3 est la cible de cette régulation

croissance-spécifique puisque cette protéine est responsable de la compétence de la

polymérase (Grummt, 2003). Le fait que Rrn3 contienne plusieurs sites de phosphorylation

supporte ce rôle de régulateur potentiel de l'initiation de la transcription. Il a été montré que

des cellules mammifères traitées avec des inhibiteurs de la synthèse protéique, affamées ou

encore dont la croissance est inhibée par des contacts cellule-cellule contiennent la protéine

Rrn3 sous sa forme hypophosphorylée (Yuan et al., 2002). Ces résultats suggèrent que

Rrn3 est phosphorylée de façon croissance-dépendante. Cette hypothèse est aussi supportée

par le fait que la voie des MAP kinases, une voie de signalisation importante dans la

régulation de la croissance, phosphoryle Rrn3 (Zhao et al., 2003). Il a été montré que Rrn3

est phosphorylée sur les serines 633 et 649 de façon ERK dépendante (Extracellular signal-

Regulated Kinase) (Zhao et al., 2003). Des expériences de mutation de la serine 649

suggèrent que cette serine est essentielle pour l'activation ERK-dépendante de la

transcription ribosomale (Zhao et al., 2003). Par contre, trois groupes différents ne

19

parviennent pas à reproduire ces résultats que ce soit chez le rat ou chez la drosophile

(communications personnelles de L.I Rothblum, R.Hannan et de B. Edgard) ce qui

relativise le rôle essentiel de la phosphorylation de Rrn3 dans la régulation de la

transcription ribosomale en réponse à la croissance.

1.2.1.4 UBF La protéine UBF, pour « Upstream Binding Factor », est une protéine architecturale

qui comprend un domaine amino-terminal de dimérisation, une queue carboxy-terminale

acide et une région centrale constituée, chez les mammifères, de six boîtes HMG1

caractéristiques de la famille des « High Mobility Group proteins 1» (figure 1.7A) (Moss et

al., 2006a). Il existe deux variants d'épissage, nommés UBF1 et UBF2 (Hisatake et al.,

1991; O'Mahony et al., 1992) qui sont exprimés dans des proportions équivalentes dans les

cellules en croissance (Hisatake et al., 1991). UBF2 contient un exon de moins qu'UBFl ce

qui a pour effet de tronquer sa deuxième boîte HMG (Hisatake et al., 1991; Kuhn et al.,

1994). Cette troncation a comme effet de diminuer la capacité d'UBF2 à activer la

transcription puisque cette protéine ne compense pas pour la perte d'UBFl dans des essais

de transcription in vitro (Klein et Grummt, 1999). Comme il a été mentionné

précédemment, l'attachement d'UBF au promoteur chez les mammifères est considéré

comme étant nécessaire in vivo au recrutement du complexe SL1 et de l'ARN polymérase I,

via son interaction avec PAF53, et donc à la formation du PIC (Grummt, 2003; Moss,

2004; Moss et al., 2006a; Moss et Stefanovsky, 2002).

La région contenant le domaine de dimérisation ainsi que les trois premières boîtes

HMG1 est la région minimale, nommée « Core UBF », nécessaire à l'activation de la

transcription ribosomale in vitro (figure 1.7A) (Jantzen et al., 1992; Stefanovsky et al.,

1996; Sullivan et McStay, 1998). Le laboratoire du Dr Moss a montré que cette région

minimale induit la formation d'une structure semblable au nucléosome qu'ils ont nommé

« enhancesome » (figure 1.7B à gauche). Cette structure contient environ 140 paires de

bases qui sont repliées sur un seul tour de 360° (Bazett-Jones et al., 1994; Stefanovsky et

al., 1996; Stefanovsky et al., 2001a). La formation de deux de ces structures pourrait

permettre le rapprochement des deux éléments du promoteur, l'UCE et le « core », de façon

20

à présenter une surface d'interaction optimale à l'ARN polymérase I (figure 1.7B) (Bazett-

Jones et al., 1994; Stefanovsky et al., 1996; Stefanovsky et al., 2001a). Le contenu en ADN

et en protéines de « l'enhancesome » est semblable à celui des nucléosomes, mais sa

structure tridimensionnelle est différente car, contrairement aux nucléosomes, le cœur de

«l'enhancesome» est vide (figure 1.7B) (Bazett-Jones et al., 1994; Stefanovsky et al.,

1996; Stefanovsky et al., 2001a). Il s'agit de la première démonstration de la capacité

d'UBF à modifier la structure de l'ADN.

A)

mUBF

B)

HMQI-bomt HOOC * h

Ai *A

y Cor* UBF

Enhancaaoma

M^L

Superhallcal modal for RPI promolar

Figure 1.7 UBF et l'« enhancesome ». A) Structure schématique de la protéine UBF mammifère. Le domaine amino terminal de dimérisation, les six boîtes IIMG1, la région minimale pour l'activation de la transcription, « core UBF », et la queue carboxy terminale acide et les nombreuses serines qu'elle contient sont représentés. B) Représentation de « l'enhancesome » formé par un dimère d'UBF à gauche et de la structure du promoteur ribosomal formé par deux dimères de « cores » UBF à droite. Les trois premières boîtes HMG1 d'UBF sont numérotées et son domaine de dimérisation est aussi illustré. UBF est en bleu, l'ADN en rouge, le « core » en jaune et l'UCE est représenté par un damier jaune et noir. Les étoiles représentent les sites phosphorylés par ERK. Provient de Moss et al., 2006a.

La formation de « l'enhancesome » par UBF et l'implication de cette protéine dans

la formation du PIC suggère qu'une structure particulière de l'ADN au promoteur soit

21

importante pour l'initiation de la transcription. Cette hypothèse est supportée par le fait que

la machinerie associée à la transcription ribosomale est relativement bien conservée au

cours de l'évolution alors que la séquence du promoteur ne l'est pas (Grummt, 2003; Moss

et al., 2006a). Cependant, malgré le fait que la séquence du promoteur ribosomal ait divergé

au cours de l'évolution, les éléments UCE et « core » qui sont importants pour la

transcription sont conservés (Grummt, 2003; Moss et al., 2006a). Il est suggéré que deux

dimères d'UBF s'attachent au promoteur ribosomal de façon à rapprocher ces deux

éléments et à présenter une surface d'interaction optimale pour l'ARN polymérase I (figure

1.7B) (Bazett-Jones et al., 1994; Stefanovsky et al., 1996; Stefanovsky et al., 2001a). De

plus, des études ont montré que la structure tridimensionnelle du promoteur est

relativement bien conservée et est importante pour l'interaction de SL1 avec le promoteur

(Marilley et Pasero, 1996; Marilley et al., 2002). Ces résultats suggèrent que la structure du

promoteur est importante pour la formation du PIC. Par contre, l'activité de SL1 n'est

conservée qu'entre les espèces ayant une séquence promotrice semblable ce qui suggère

que la séquence de la région promotrice est aussi importante pour la formation du complexe

d'initiation (Heix et Grummt, 1995).

Des études d'immunoprécipitation de la chromatine ont montré qu'UBF n'est pas

seulement localisé au promoteur ribosomal, mais qu'il est aussi présent sur toute la

séquence ribosomale (O'Sullivan et al., 2002). Puisqu'UBF a la capacité de modifier la

structure de l'ADN (Bazett-Jones et al., 1994; Stefanovsky et al., 1996; Stefanovsky et al.,

2001a; Stefanovsky et al., 2006b;), il est suggéré qu'UBF définit une structure

chromatinienne ribosomale différente de la structure nucléosomale classique (Bazett-Jones

et al., 1994; Stefanovsky et al., 1996; Stefanovsky et al., 2001a). De plus, UBF

compétitionne avec des protéines qui s'attachent à l'ADN de façon non spécifique comme

l'histone Hl (Kuhn et Grummt, 1992; Kuhn et al , 1993). Cette observation suggère aussi

qu'UBF est un régulateur de la chromatine ribosomale. La présence d'UBF sur toute la

région transcrite et sa capacité à modifier la structure de la chromatine suggère qu'il puisse

non seulement réguler la formation du PIC, mais aussi les étapes subséquentes de la

transcription comme la libération du promoteur et l'élongation de la polymérase. À ce sujet,

une étude faite par Panov et collègues suggère que la libération du promoteur est limitante

22

pour la transcription et que cette étape dépend d'UBF (Panov et al., 2001). Cet article

suggère qu'il existe possiblement d'autres niveaux de régulation que celui de la formation

du complexe de pré-initation. Cependant, ces données sont contestées et le mécanisme

moléculaire impliqué dans cette régulation n'est pas défini. Les chapitres 2 et 3 de ma thèse

explorent ces niveaux de régulation et sont spécifiquement consacrés à la fonction d'UBF

dans la régulation de la transcription ribosomale suite à une stimulation par un facteur de

croissance.

La fonction architecturale d'UBF pourrait être aussi impliqué dans le maintien de

l'intégrité du nucléole. Une étude faite sur le nucléole et la stabilisation de p53 (section 1.1)

a montré qu'une microinjection d'anticorps dirigés contre UBF induisait le démantèlement

du nucléole (Rubbi et Milner, 2003). Cependant, ces résultats ne permettent pas de conclure

que le bris du nucléole est dû uniquement à la perte de la fonction architecturale d'UBF

puisque le démantèlement pourrait aussi être causé par l'arrêt de la transcription ribosomale

comme c'est le cas lors de la mitose. Il est toutefois intéressant de noter qu'UEÏF reste

associé aux gènes ribosomaux lors de la mitose (Roussel et Hernandez-Verdun, 1994) et il

est suggéré que sa présence soit importante lors de la réactivation de la transcription

ribosomale en Gl (Klein et Grummt, 1999; Voit et al., 1999). Ainsi, la fonction

architecturale d'UBF pourrait être essentielle pour le rétablissement d'une chromatine

favorable à la transcription ribosomale.

UBF est aussi un facteur important pour la régulation de la transcription ribosomale.

Certaines études montrent qu'UBF est phosphorylé différentiellement selon les phases du

cycle cellulaire ce qui aurait comme effet de réguler la transcription ribosomale au cours du

cycle cellulaire (section 1.1) (Klein et Grummt, 1999; Voit et al., 1999). Le complexe SL1

(Heix et al., 1998) et UBF sont inactivés par une kinase en mitose (Klein et Grummt,

1999). L'activité d'UBF est progressivement rétablie en phase Gl par l'action combinée,

ou séquentielle, d'une phosphatase et des kinases cdk4/cycline D et cdk2/cycline E qui

phosphorylent la serine 484 (Klein et Grummt, 1999; Voit et al., 1999). L'identité de la

kinase mito tique, la caractérisation de son effet, l'identité de la phosphatase et la

caractérisation de l'effet activateur de la phosphorylation de la serine 484 ne sont pas

23

caractérisées. La phosphorylation de la queue C-terminale d'UBF est nécessaire pour le

recrutement du complexe SL1 au promoteur (Kihm et al., 1998; Tuan et al., 1999) et il est

suggéré que la Caséine Kinase II est responsable de cette phosphorylation (Voit et al.,

1992). La phosphorylation d'UBF par LRK et l'activation transcriptionnelle qui en résulte

fut la première démonstration in vivo de la régulation de la transcription ribosomale en

fonction de la croissance (chapitres 2 et 3) (Stefanovsky et al., 2006a; Stefanovsky et al.,

2001b). UBF est aussi phosphorylé suite à l'activation de la voie TOR (Target Of

Rapamycin) et cette phosphorylation est importante pour l'augmentation de la transcription

(Hannan et al., 2003). L'acétylation d'UBF par CBP a aussi été montrée, par le laboratoire

du Dr Moss, comme augmentant la transcription ribosomale in vitro (Pelletier et al., 2000).

Les protéines CBP et p300 augmentent aussi la transcription in vivo probablement par

l'acétylation combinée d'UBF et des histones (Hirschler-Laszkiewicz et al., 2001). La

protéine pRb, et peut-être aussi pi 30, réprime la transcription ribosomale en inhibant

l'interaction UBF-SL1 (Hannan et al., 2000). Le laboratoire du Dr Moss a aussi montré que

pRb peut catalyser la déacétylation d'UBF (Pelletier et al., 2000). De plus, les auteurs ont

montré que pRb compétitionne avec CBP pour un site qu'ils ont en commun sur UBF et

suggèrent un modèle flip-flop entre l'acétylation et la déacétylation d'UBF (Pelletier et al.,

2000). La protéine nucléolaire p204, induite par l'interféron, interagit aussi avec UBF et

diminue la transcription ribosomale par un mécanisme inconnu (Liu et al., 1999). La

diversité des interactions d'UBF et de ses modifications post-traductionnelles en fait

clairement une cible importante dans la régulation de la transcription ribosomale.

L'ensemble des résultats présentés suggère qu'UBF est une protéine essentielle pour

la transcription ribosomale in vivo. Bien que certains systèmes de transcription in vitro,

chez la souris et l'humain, soient activés par l'ajout d'UBF ces systèmes ne sont activés

que dans certaines conditions expérimentales mal définies (Moss et al., 2006a). Il est donc

possible qu'UBF ne soit pas essentiel in vivo. À ce jour, aucun homologue d'UBF n'a été

identifié chez la levure. Cependant, il est suggéré que la protéine Hmolp est un homologue

fonctionnel d'UBF puisque cette protéine contient une boîte HMGl et est impliquée dans la

transcription ribosomale (Gadal et al., 2002). La perte de cette protéine (lignée hmol-A)

entraîne un grave défaut de croissance, mais n'est pas létale (Gadal et al., 2002). Par contre,

24

le croisement de cette souche (hmol-A) avec une souche qui ne contient pas RPA49 (une

sous-unité non essentielle de la polymérase) est létal. Cette sous-unité est l'homologue de

PAF53 qui est la sous-unité de la polymérase avec laquelle interagit UBF chez les

mammifères. Le croisement de la lignée hmol-À avec deux autres lignées portant des

mutations dans des sous-unités non essentielles de la polymérase est aussi létal (Gadal et

al., 2002). Ainsi, la présence d'une protéine contenant une boîte HMG1 n'est pas

essentielle chez la levure, mais elle le devient en absence de sous-unités non essentielles de

la polymérase. Chez les mammifères, la situation est probablement différente puisque la

structure d'UBF est plus complexe que celle d'Hmolp (Gadal et al., 2002). En conclusion,

la fonction essentielle in vivo d'UBF n'est pas encore établie, mais sa capacité à modifier la

structure de F ADN ainsi que sa présence sur toute la séquence ribosomale en font une

protéine à haut potentiel régulateur.

1.2.2. De TARN ribosomal précurseur aux ribosomes matures Une multitude de transformations co et post-transcriptionnelles sont impliquées

dans la transformation de l'ARN ribosomal précurseur en sous-unités ribosomales matures

(figures 1.8 et 1.10) (Dimario, 2004; Fromont-Racine et al., 2003; Tschochner et Hurt,

2003). Ces transformations et cet assemblage commencent de façon co-transcriptionnelle

dans le nucléole et se terminent dans le cytoplasme (figure 1.8). Cette section a pour but de

tracer les grandes lignes de cette situation complexe et non de décrire en détail toutes les

protéines impliquées. Contrairement au reste de cette thèse, où l'emphase est mise sur le

système mammifère, les résultats présentés ici proviennent d'études faites chez S.

cerevisiae car c'est le système dans lequel ces étapes ont été le plus étudiées.

L'assemblage co-transcriptionnel des protéines ribosomales avec l'ARN ribosomal

précurseur mène à la formation du premier complexe pré-ribosomal nommé 90S (figure

1.8). Ce complexe ribonucléoprotéique comprend l'ARNr précurseur en élongation et le

complexe SSU (pour « Small Subunit processome ») (Dimario, 2004; Fromont-Racine et

al., 2003; Grandi et al., 2002; Tschochner et Hurt, 2003). Chez la levure, ce complexe de

plus de 2MDa est constitué par FU3 snoRNA et environ 40 protéines et est impliqué dans

la formation de la petite sous-unité ribosomale nommée 40S (figure 1.8 et 1.9) (Dragon et

25

al., 2002; Grandi et al., 2002). Ce complexe, aussi présent chez les mammifères, est visible

par microscopie électronique suite à des expériences de « Miller spread ». Ces expériences

produisent l'image classique de la transcription ribosomale communément apellée « arbre

de Noël » ou « Christmas tree » dans laquelle le tronc de l'arbre représente l'ADN

ribosomal et les branches les transcrits en élongation (figure 1.9). Ces expériences

permettent de visualiser les polymérases (voir agrandissement de la figure 1.9) et les

transcrits en élongation qui se terminent en 5' par une région plus ronde qui est le complexe

SSU « processome ». Des études faites sur l'effet de mutations dans l'U3 snoRNA

(Mougey et al., 1993), dans un système murin, ou dans des constituants essentiels du SSU

(Dragon et al., 2002), chez la levure, montrent que ces mutations entraînent la perte de cette

structure. Ces deux études suggèrent que les constituants du SSU « processome » qui ont

été testés sont nécessaires pour l'intégrité de ce complexe.

Le complexe 90S contient des protéines non ribosomales importantes pour le pré-

40S, mais presqu'aucune protéine du pré-60S (Grandi et al., 2002; Tschochner et Hurt,

2003). Ceci suggère que les premiers événements de transformation de l'ARNr précurseur

sont liés à la formation du pré-40S et que les protéines nécessaires au pré-60S viennent se

joindre de façon plus tardive au complexe (Grandi et al., 2002). C'est effectivement ce qui

est observé puisque les trois premiers événements de clivage se font dans le 5'ETS ou dans

1TTS1 et sont importants pour la formation du pré-40S. Le premier événement de clivage

nommé Ao (figure 1.10) est un événement précoce qui survient dans le 5'ETS. Le

deuxième, Ai, coupe à la frontière entre le 5'ETS et le 18S et le troisième événement, A2,

mène à la formation des pré-40S et pré-60S (figure 1.10) (Kressler et al., 1999). Le pré-40S

est exporté rapidement dans le cytoplasme contrairement au pré-60S qui est exporté plus

tardivement (Fromont-Racine et al., 2003; Tschochner et Hurt, 2003). Les étapes de

maturation de la petite sous-unité nécessitent la participation d'une dizaine de protéines

non-ribosomales et d'un clivage en 3' de l'ARN 18S (clivage D sur la figure 1.10)

(Fromont-Racine et al., 2003). Des mutations inhibant les clivages Ao à A2 n'affecte que la

synthèse du 18S (40S) car le 5.8S et le 25S (60S) peuvent toujours être produits en utilisant

le site de clivage A3 (figure 1.10) (Kressler et al., 1999).

26

La maturation du pré-60S en 60S mature est complexe et requiert la participation

d'environ 70 protéines non ribosomales ainsi qu'un court séjour dans le noyau avant d'être

exporté dans le cytoplasme (figure 1.8) (Fromont-Racine et al., 2003). L'export des deux

sous-unités ribosomales se fait donc selon des cinétiques différentes. En plus de devoir

modifier ses deux ARN ribosomaux, 5.8S et 25S, le pré-60S doit incorporer l'ARN 5S qui

est transcrit par l'ARN polymérase III. Chez la levure, contrairement à la situation chez les

mammifères, la séquence codant pour l'ARN 5S est située au même locus que celle codant

pour les ARNr (figure 1.10) et il est donc probablement transcrit dans le nucléole. Chez les

mammifères, contrairement à la situation de la levure présentée dans la figure 1.8, l'ARN

5S doit être exporté hors du noyau pour s'associer au pré-60S dans le cytoplasme

(Dechampesme et al., 1999). Un autre niveau de complexité vient du fait que le pré-60S,

contrairement au pré-40, est associé à des GTPases et à des ATPases de type AAA pour

« ATPases associated with various cellular activities ». Leur présence suggère que cet

intermédiaire doit subir des réarragements structuraux qui requièrent une consommation

d'énergie (Brown, 2001; Tschochner et Hurt, 2003).

Finalement, il existe deux types de 60S matures qui contiennent des ARNr 5.8S de

longueurs différentes (figure 1.10). Le plus long des ARN 5.8S, nommé 5.8SL, est produit à

partir de l'intermédiaire 27SA2 suite à un clivage en 5' de la séquence codant pour l'ARN

5.8S (figure 1.10). Le deuxième ARN 5.8S, nommé 5.8Ss, est produit à partir de

l'intermédiaire 27SA3 par la digestion de sa région 5' par exonucléation (figure 1.10)

(Kressler et al., 1999). La différence entre les intermédiaires 27SA2 et 27SA3 est la présence

dans 27SA2 d'une séquence en 3' de celle codant pour le 25S. La différence entre les ARN

5.8SI, et 5.8Ss est la présence, chez l'ARN 5.8SL, d'une région en 5' de la séquence codant

pour l'ARN 5.8S qui est absente chez l'ARN 5.8Ss (figure 1.10). Aucune différence

fonctionnelle entre ces deux types de sous-unités n'a été identifiée et ce, malgré le fait que

la sous-unité 60S contenant un court 5.8S représente 85% des sous-unités (Fromont-Racine

et al., 2003).

27

A) N 11 ( L É ()

E

1$)

"

N U ( L É <) L E

N () Y A U

CYTOPLASME

>"N*

o (4

A , A ,

*

■r

> 5 l f * » nascem rflNA

t ^ U 3 ©

jf-* P

•ua yCteavagesAQ,Ai,A2 x

pre-40S (20S)

pro-60S (25S\ 6S, 5S)

A pre-60S (258, 6.8S, 5S)

0S(2OSJ^^

Export compétence

Late steps

■:

! !

pre-40S ,_. A O «»A

pre^OS (25S. S.8S. SS) £

4 * 408 MBS) 608 (25S. 5.8S SS) Figure 1.8 Schématisation des étapes de transformation de l'ARN ribosomal précurseur et

des sous-unités ribosomales chez la levure S. cerevisiae.

A) Assemblage co-transcriptionnel des sous-unités ribosomales sur l'ARN ribosomal naissant. B) Assemblage et maturation des sous-unités ribosomales. Cette figure est la suite de celle présentée en A. Les protéines de la petite sous-unité, S, sont en vert et celles de la grosse, L, en bleu. Les facteurs de transformation de la petites sous-unité sont en jaune et ceux de la grosse en mauve et leur nom sont indiqués dans les bandes grises; à gauche pour la sous-unité 40S et à droite pour la sous-unité 60S. Les facteurs de contrôle sont surmontés d'un R dans un cercle. Le SSU « processome » contenant le snoRNA U3 est schématisé. Le

28

compartiment cellulaire où ont lieu les transformations est indiqué complètement à gauche de la figure. Modifié de Fromont-Racine et al., 2003.

Figure 1.9 Image de microscopie électronique de la transcription ribosomale d'un gène mammifère. Expérience de « Miller Spread » faite sur une cellule de souris Ltk" en culture. L'agrandissement permet de voir les polymérases au centre (sphères noires), les transcrits en élongation et le SSU « processome » en 5' des transcrits. Provient de U. Scheer.

En plus d'être exportées hors du noyau avec des cinétiques différentes, les deux

sous-unités pré-40S et pré-60S utilisent des mécanismes distincts pour traverser les pores

nucléaires. La protéine Nmd3, qui contient un signal d'export nucléaire, serait une des

protéines adaptatrices entre le pré-60S et Crml, le récepteur pour l'export de la famille des

importine « 6-likc » (Gadal et al., 2001; Ho et al., 2000). Cependant, Nmd3 n'interagit pas

directement avec Crml ce qui suggère que d'autres protéines sont impliquées dans l'export

de la sous-unité 60S (Johnson et al., 2002). Nmd3 interagirait avec la protéine RpllO du

pré-60S. La faible association entre RpllO et le pré-60S (Dick et al., 1997) suggère que

cette protéine est sensible à la conformation du pré-60S. L'export du pré-60S serait donc

dépendant de sa bonne configuration et fournirait un mécanisme de contrôle de la qualité.

La protéine Crml serait aussi impliquée dans le transport du pré-40S, mais pour l'instant,

29

aucune protéine adaptatrice n'a été identifiée. Des mutants entraînant une accumulation de

pré-60S dans le noyau n'entraînent pas d'accumulation de pré-40S (Moy et Silver, 1999;

Stage-Zimmermann et al., 2000). Ces résultats suggèrent que le transport des deux sous-

unités utilise des mécanismes d'export qui sont indépendants l'un de l'autre. 11 est connu

que le transport des deux sous-unités nécessite une nucléoporine, une karyophérine

(protéine adaptatrice) et les facteurs impliqués dans le cycle Ran GTP-GDP ( l'schochner et

Hurt, 2003). Une fois dans le cytoplasme, l'association des sous-unités individuelles 40S et

60S, d'un ARN de transfert initiateur MET et d'un ARNm entraîne la formation d'un

ribosome 80S compétent pour la traduction (Preiss et Hentze, 2003).

1.3. L'établissement et le maintien des gènes ribosomaux silencieux.

Une des caractéristiques des gènes ribosomaux est qu'ils sont présents en plusieurs

copies dans le génome. Cependant, seule une portion de ces gènes est transcrite tandis que

l'autre est maintenue silencieuse. Dans les cellules eucaryotes, ces deux états d'activité sont

associés à deux types de chromatine ribosomale présentant des accessibilités différentielles

pour un agent intercalant nommé le psoralen (Conconi et al., 1989). Le premier type

contient des nucléosomes et représente les gènes silencieux. Le deuxième est dépourvu de

nucléosomes et représente les gènes transcrits (Conconi et al., 1989). Ces deux types de

chromatine sont maintenus de façon stable au cours du cycle cellulaire. Il a aussi été montré

que ces deux états de chromatine sont maintenus indépendamment de l'activité de la

transcription ribosomale (Conconi et al., 1989). Ainsi, une plus forte transcription

n'augmente pas le nombre de gènes transcrits. Dans les lignées cellulaires utilisées dans

notre laboratoire, c'est environ 50% (2 lignées murines) et 80% (2 lignées humaines) des

gènes ribosomaux qui sont silencieux (chapitre 4 et Stefanovsky et Moss, 2006). Cette

section expliquera brièvement ce qui est connu des mécanismes responsables de l'inhibition

de la transcription chez la levure Saccharomyces cerevisiae et chez les plantes. Par la

suite, le système mammifère sera abordé plus en détail.

30

B

h - -pre-5S rRNA

rDNA repeat (9.1 kb) -

368 pro-rRNA

- H pre-5S rRNA

■*■ * " *

NT81 HTS2AQ A^ ^

\ BlL Bis A 2 AgS/ E C2 q |

B2 NTS1 NTS2 B lL O I S

rrsi ITS2

Prlmary RNA pol I transcript

Prlmary RNA pol III transcript

PseudourkJylatlon (Y)

2'-O-rlb0M methylatlon

Basa methylatlon?

H/ACA-box anoRNPs

C/D-box anoRNPs Rntlp RHAB2? RNA82?

3 5 S

Cleavage

3 3 S

Cleavage Ai

3 2 S

Cleavage Ag

^ J Rntlp? U3 anoRNP

U3 U14 tnRIO anR30 tnoRNPa Dbp4p Dbptp Fahp Roklp Rrp3p

,, Dlm1pRrp5p

U3 U14 «nRIO tnRSO anoRNPs Dbp4p DbpBp Fatlp Roklp Rrp3p PltnlpRrpSp

27SA2

Dlmlp

Dbpfip DbpSp Nop4p NopSp

Dbp7p MakSp

DralpT

Base methylatlon?

Cleavage A3 RNaae MRP Dbp3p

I RrpSp

Processing B2

I RNAB2?

27SA3 < Exonucleatie

. R a t l p X m l p A 3 - * B 1 S | 27SBS ■

Processing B2

RNA82?

Processing Bj |_

20S 7SS

CblSpNop2pSpb1p DbplOp Spb4p Processing Nlp7p C | + C j

2 7 S B L I

y » ■aa > I

5 . 8 S 5

Exoaome Doblp

7SL Exonuclease

E « - C 2

25S S.8SL

CbfbpNop2pSpb1p DbplOp Spb4p Nlp7p

Exosome Doblp

25S 5 S

Figure 1.10 Production des ARN ribosomaux matures. A) Organisation du locus ribosomal chez la levure et identification de certains sites de clivage. B) Présentation des différents intermédiaires des ARN ribosomaux et identification des transformations qu'ils subissent et des protéines ou des snoRNA impliqués. Provient de Kressler et al., 1999.

31

1.3.1. Mécanisme d'inactivation des gènes ribosomaux chez la levure S. cerevisiae

La levure S. cerevisiae est un bon modèle expérimental car les manipulations

génétiques sont plus faciles à réaliser dans ce système que chez les organismes eucaryotes

plus évolués. Cependant, il s'est avéré que les mécanismes impliqués dans l'inhibition de la

transcription chez S. cerevisiae sont différents de ceux utilisés chez les plantes ou chez les

mammifères. Une différence importante est l'absence de méthylation de l'ADN chez S.

cerevisiae (Li, 2002). Cette modification est parmi les marques épigénétiques utilisées par

les plantes et les mammifères pour réguler leur expression génique y compris celle des

gènes ribosomaux (Santoro, 2005). De plus, l'organisation des gènes ribosomaux est

différente entre S. cerevisiae et les plantes ou les mammifères. En effet, S. cerevisiae ne

contient qu'un locus ribosomal donc un NOR ce qui suggère que les gènes transcrits et les

gènes silencieux sont présents sur le même NOR. Chez les plantes et les mammifères,

l'existence de NORs actifs et inactifs suggère une séparation physique entre les gènes

ribosomaux transcrits et silencieux. Cependant, il n'existe aucune preuve expérimentale de

cette séparation. Il est donc possible qu'un NOR contienne les deux types de gènes comme

c'est le cas chez la levure.

Finalement, chez S. cerevisiae, l'expression « rDNA silencing » ne concerne pas

l'inhibition de la transcription des gènes ribosomaux. Cette expression englobe plutôt

l'étude des mécanismes menant à l'inhibition de la transcription de gènes rapporteurs

transcrits par l'ARN polymérase II et insérés dans ou près du locus ribosomal (Moss,

2004). Il s'agit donc de l'étude de l'inhibition de la transcription PolII dans le locus

ribosomal, d'où le terme « rDNA silencing » et non de l'inhibition de la transcription des

gènes ribosomaux par l'ARN polymérase I. Les résultats des études qui sont présentés dans

cette introduction concernent ce type d'inhibition de la transcription par l'ARN polymérase

II dans le locus ribosomal. Il est essentiel de présenter ces résultats car ils permettent de

comprendre l'importance des résultats et la logique de l'argumentation présentés au

chapitre 4 de cette thèse.

Chez S. cerevisiae, c'est le complexe RENT, pour « REgulator of Nucleolar

silencing and Telophase exit », qui est responsable de l'inhibition de la transcription par

32

TARN polymérase II au locus ribosomal (Shou et al., 1999; Straight et al., 1999). Ce

complexe est formé de trois protéines. La première, Sir2p, pour « Silent Information

Regulator 2 », est une histone déacétylase dépendante du NAD+ qui est aussi essentielle

pour l'inhibition de la transcription aux télomères et au locus du « mating type ». Alors que

Sir2p forme un complexe avec Sir3p et Sir4p à ces deux régions, Sir2p est associée à Netlp

et à la phosphatase Cdcl4p au locus ribosomal (Ghidelli et al., 2001; Guarente, 2000;

Straight et al., 1999). La protéine Netlp est directement impliquée dans la transcription

ribosomale et est nécessaire pour la localisation nucléolaire de Sir2p et de Cdcl4p (Shou et

al., 2001). La phosphatase Cdcl4 n'a pas de rôle connu dans la transcription ribosomale,

mais son relarguage du complexe RENT à la fin de la mitose permet la sortie de la

télophase par la déphosphorylation de cibles spécifiques (Shou et al., 2001; Shou et al.,

1999; Straight et al., 1999; Visintin et al., 1998; Zachariae et Nasmyth, 1999). Cette

protéine joue aussi un rôle dans la cytokinèse lors de la mitose (Kaiser et al., 2002; Mailand

et al., 2002). Une autre activité reliée au complexe RENT est l'inhibition de la

recombinaison des gènes ribosomaux. La protéine Foblp est responsable de la

recombinaison au locus ribosomal et de la formation de cercles d'ADNr épisomaux dont

l'accumulation entraîne le vieillissement des cellules de levure (Guarente, 2000; Guarente

et Kenyon, 2000; Sinclair et Guarente, 1997). L'activité de Sir2p antagonise l'activité de

Foblp, diminue la recombinaison au locus ribosomal et augmente la durée de vie (Benguria

et al., 2003; Sinclair et Guarente, 1997). Les activités diversifiées des protéines du

complexe RENT sont d'autres exemples éloquents de l'importance de la transcription

ribosomale dans le fonctionnement cellulaire.

L'inhibition de la transcription par TARN polymérase II au locus ribosomal peut

paraître surprenante car, contrairement à la situation aux télomères et au locus du « mating

type », elle a lieu dans un locus à haute activité transcriptionnelle. Trois modèles ont été

proposés pour expliquer cet état de fait (Cioci et al., 2003). Le premier modèle suggère que

la chromatine ribosomale est permissive à la machinerie Poli, mais restrictive à la

machinerie PolII et à celle impliquée dans la recombinaison. Ce modèle, baptisé

« inactivation indépendante », prédit que l'inhibition de la transcription PolII est

indépendante de l'activité Poli. Le deuxième modèle propose que, puisqu'il y a des gènes

33

ribosomaux transcrits et d'autres silencieux, l'inhibition des gènes rapporteurs PolII est

dépendante de leur insertion dans ou à proximité de ces gènes silencieux. Ce modèle prédit

que la perte de gènes ribosomaux silencieux entraînerait la transcription des gènes

rapporteurs (Cioci et al., 2003). Il s'est avéré que ces deux premiers modèles sont faux. En

effet, il a été montré qu'une des caractéristiques de l'inhibition de la transcription par

RENT est que non seulement cette inhibition est dépendante de Sir2p, mais qu'elle est aussi

dépendante de la transcription par Poli. Une étude surprenante a montré l'importance du

complexe UAF dans l'inhibition de la transcription par PolII. Le complexe UAF est, avec

le complexe CF, impliqué dans la formation du complexe de pré-initiation chez la levure.

Malgré le fait que ce complexe soit nécessaire pour obtenir une forte activité

transcriptionnelle, il n'est pas essentiel pour l'activité transcriptionnelle de base et donc

pour la survie des délétants. Cependant, suite à la délétion de différentes protéines de ce

complexe, les clones qui au départ poussaient difficilement se mettent à pousser plus

rapidement. De façon étonnante, les clones utilisent maintenant la machinerie PolII et non

plus la machinerie Poli (Oakes et al., 1999; Vu et al., 1999). Ces études montrent que la

machinerie PolII peut transcrire les gènes ribosomaux, mais que la cellule a développé des

mécanismes pour la maintenir hors du locus ribosomal.

Des études plus récentes ont confirmé que l'inhibition de la transcription PolII est

dépendante d'une machinerie Poli fonctionnelle (Buck et al., 2002) et d'un promoteur

ribosomal intact (Cioci et al., 2003). De plus, la propagation de l'inhibition est dépendante

de l'orientation de la transcription Poli (Buck et al., 2002). Finalement, Cioci et collègues

ont vérifié si la perte des gènes ribosomaux silencieux levait l'inhibition de la transcription

par PolII (Cioci et al., 2003). Pour ce faire, ils ont produit des lignées contenant moins de

gènes ribosomaux qui sont alors tous transcrits. À titre comparatif, la moitié des gènes

d'une cellule de levure sauvage sont dans un état silencieux. Dans ces nouvelles lignées,

l'inhibition de la transcription par la PolII est augmentée et non diminuée comme le

prédisait le deuxième modèle. À la même époque, il a été montré qu'une diminution du

nombre de gènes ribosomaux entraînait une augmentation de la densité des complexes Poli

engagés (French et al., 2003). Ces résultats suggèrent que non seulement la transcription

ribosomale inhibe la transcription PolII, mais qu'en plus cette inhibition est dépendante de

34

la densité des complexes Poli engagés. L'ensemble de ces résultats a mené à l'élaboration

du modèle de « réciprocité » qui stipule que l'activité de PollI au locus ribosomal est

inversement proportionnelle à celle de Poli et vice-versa. En conclusion, ces études ont

montré que la cellule de levure utilise deux mécanismes, l'activité du complexe RENT et la

transcription ribosomale, pour s'assurer de l'exclusion de la PollI hors du locus ribosomal.

Jusqu'à ce jour, aucun complexe pouvant s'apparenter au complexe RENE n'a été identifié

chez les mammifères.

1.3.2. Mécanisme d'inactivation des gènes ribosomaux : la dominance nucléolaire chez les plantes.

Les études réalisées sur les gènes ribosomaux silencieux chez les plantes ont porté

sur le phénomène de la dominance nucléolaire. La dominance nucléolaire est observée dans

des hybrides interspécifiques où le(s) NOR(s) provenant d'une de deux espèces est (sont)

dominant(s) sur le(s) NOR(s) provenant de l'autre espèce (figure 1.11). Ce phénomène

n'est pas exclusif aux plantes (Brassica et Arabidopsis) car il a aussi été observé chez des

hybrides de Drosophila megalonaster et de Xenopus (McStay, 2006). Une certaine forme

de dominance nucléolaire est aussi observée dans des cellules somatiques hybrides

souris/humaines. Comme il a été discuté à la section 1.2.1.2, cette dominance est due à la

spécificité du complexe SL1.

La dominance nucléolaire chez les plantes est dépendante de la méthylation de

l'ADN et des modifications de la queue des histones. L'utilisation de 5-aza-2'-

dexocytidine, un inhibiteur des ADN méthyltransférases, de sodium butyrate ou de

trichostatine A, un inhibiteur des HDACs, induit la transcription des gènes ribosomaux

provenant de l'espèce dominée (Chen et Pikaard, 1997). De plus, la transcription de mini

gènes ribosomaux provenant de l'espèce dominée dans des cellules hybrides où l'ADNr

chromosomique est non transcrit suggère que des régions flanquantes sont impliquées dans

l'inactivation des gènes ribosomaux (Frieman et al., 1999). Ainsi, la cellule hybride a la

capacité de transcrire les gènes ribosomaux de l'espèce dominée, mais le contexte

chromosomique de cette espèce, dans la cellule hybride, a un effet négatif sur sa

transcription, probablement par l'établissement d'une structure hétérochromatinienne.

35

Donc, selon l'ensemble de ces résultats, l'inhibition de la transcription des gènes

ribosomaux de l'espèce dominée est dépendante de l'établissement d'une structure

chromatinienne réfractaire à la transcription. Les mécanismes utilisés dans le choix du

génome dominé sont encore inconnus, mais ils sont sélectifs puisque ce sont toujours les

gènes de la même espèce qui sont dominés dans les cellules hybrides étudiées. Il est

suggéré que l'espèce qui s'approprie la première ou de façon plus efficace la machinerie

transcriptionnellc sera l'espèce dominante. Des résultats montrant qu'il est possible de

renverser la dominance en changeant le ratio NOR dominant/NOR dominé (Chen et al.,

1998) renforce cette hypothèse. Ainsi, la présence d'un plus grand nombre de NORs

conférerait maintenant un avantage à l'espèce classiquement dominée puisqu'elle pourrait

recruter de façon plus efficace la machinerie transcriptionnelle. Cependant, il est possible

que d'autres mécanismes existent chez d'autres espèces.

Les études faites chez les hybrides de plantes ont permis de montrer l'importance de

la méthylation de l'ADN et des modifications des histones dans l'établissement et le

maintien de gènes ribosomaux silencieux (Pikaard, 2000). Il s'est avéré que les mécanismes

utilisés dans ces systèmes sont semblables à ceux utilisés chez les mammifères (Grummt et

Pikaard, 2003) (voir section suivante). De plus, ce sont des NORs entiers qui sont

silencieux chez les plantes tout comme chez les mammifères. Il est toutefois vraisemblable

que les NORs actifs contiennent aussi des gènes ribosomaux silencieux en plus de ceux

transcrits. Finalement, l'inactivation systématique de gènes ribosomaux d'une espèce dans

les cellules hybrides souligne de façon dramatique l'importance de ne pas transcrire tous les

gènes ribosomaux. Il semble donc, dans ce système, que le maintien de gènes ribosomaux

silencieux soit essentiel au bon fonctionnement de la cellule hybride.

36

'RNA gBfies active — — ^ ^ - ~ — fRNA gènes active

rRNA gènes reprnb&ed

Inrubilors of DNA methylution or hislane ctedcetyluhun

Hybnd (A-.B)

Figure 1.11 La dominance nucléolaire.

Chez l'hybride A+B, seuls les gènes ribosomaux provenant de l'espèce A sont transcrits. Ceux de l'espèce B sont maintenus silencieux. L'inhibition de l'activité des histones déacétylases et de la méthylation de l'ADN dans la cellules hybride restaure la transcription des gènes silencieux. Tiré de Grummt et Pikaard, 2003.

1.3.3. Mécanisme d'inactivation des gènes ribosomaux chez les mammifères.

La méthylation de l'ADN est aussi suggérée comme jouant un rôle central dans le

maintien des gènes ribosomaux dans un état silencieux chez les mammifères. Des études

descriptives chez l'humain ont montré que, dans les unités ribosomales, les régions non

transcrites sont méthylées alors que les régions transcrites sont majoritairement non

méthylées (Brock et Bird, 1997). Brock et Bird suggèrent aussi que les régions

normalement transcrites qui sont méthylées représentent les gènes silencieux (Brock et

Bird, 1997). Ces résultats sont en accord avec l'affinité différentielle des ces deux régions

pour le psoralen (Conconi et al., 1989) et suggèrent une corrélation entre la présence de

nucléosomes et la méthylation de l'ADN. Une autre étude a montré que la région

promotrice des gènes silencieux est méthylée sur un dinucléotide CpG chez le rat

(Stancheva et al., 1997). L'ensemble de ces résultats suggère que la méthylation de l'ADN

joue un rôle dans l'inactivation et le maintien des gènes ribosomaux dans un état silencieux.

Malgré ces observations, aucune étude mécanistique reliant la méthylation de l'ADN à

l'inhibition de la transcription des gènes ribosomaux n'avait été faite dans des cellules

37

humaines au début de mon doctorat. C'est donc le modèle développé dans des cellules de

souris qui est présenté ici. Cependant, il est possible que l'effet de la méthylation de l'ADN

diffère entre les systèmes humain et souris puisque la séquence du promoteur ribosomal

humain contient un îlot CpG alors que la séquence du promoteur de souris ne contient que

3 dinucléotides CpG. Les résultats plus récents obtenus dans le système humain seront

discutés dans la conclusion de cette thèse. Les enzymes responsables de la méthylation de

l'ADN et la formation de l'hétérochromatine qui lui est associée sont décrites dans les

sections 1.3.3.2 et 1.3.3.3 respectivement.

1.3.3.1. Mécanisme d'inactivation des gènes ribosomaux chez la souris C'est paradoxalement en cherchant à identifier un complexe de remodelage de la

chromatine nécessaire à l'activation de la transcription que le laboratoire de I. Grummt a

identifié le complexe NoRC, pour « Nucleolar Remodeling Complexe ». Ce laboratoire

avait préalablement montré que l'activation de la transcription ribosomale suite à

l'attachement de TTF1, pour « Termination Transcription Factor I », à son site en aval du

promoteur ribosomal, To, nécessitait un remodelage ATP dépendant des nucléosomes au

niveau du promoteur (Langst et al., 1998; Langst et al., 1997). Par la technique du double

hybride, ils ont identifié Tip5, pour « TTFI Interacting Protein 5 », comme étant un

partenaire de TTFI (Strohner et al., 2001). L'analyse de la séquence de Tip5 a permis de

constater qu'il possédait plusieurs domaines ainsi qu'un arrangement de ces domaines

similaires à celui des protéines de la famille de complexes de remodelage de la chromatine

ISWI. De plus, les auteurs ont montré que Tip5 est associé avec SNF2 (la sous-unité

catalytique de cette famille de complexes) et que l'activité de NoRC, tout comme celle des

autres membres de la famille ISWI, est dépendante de l'ATP et de la queue N-terminale de

l'histone H4 (Strohner et al., 2001).

Des expériences de surexpression de Tip5 ont montré l'activité répressive de NoRC

sur la transcription de gènes rapporteurs contenant un promoteur ribosomal (Zhou et al.,

2002). L'inhibition de cette transcription est aussi associée à une disparition d'un marqueur

de Peuchromatine, l'acétylation de l'histone H4, et l'apparition de marqueurs de

l'hétérochromatine, soit la protéine HPl, « Heterochromatin Protein 1 », et la méthylation

38

de la lysine 9 de l'histone H3 (Santoro et al., 2002). Ce groupe a aussi montré qu'un

inhibiteur des histones déacétylases, TSA, bloque l'activité répressive d'une surexpression

de Tip5 suggérant que des histones déacétylases sont essentielles à l'action de NoRC (Zhou

et al., 2002). Toutefois, la quasi totalité des expériences ont été faites sur des gènes

rapporteurs et l'activité répressive de NoRC sur les gènes endogènes n'a pas été clairement

démontrée.

Ce même laboratoire avait publié l'année précédente un article sur l'importance de

la méthylation de l'ADN dans l'inactivation des gènes ribosomaux (Santoro et Grummt,

2001). Ils ont montré qu'un traitement avec un inhibiteur des ADN méthyltransférases (5-

aza-2'dexoxycytidine) augmentait la transcription in vivo. Des essais de reconstitution de

chromatine ont montré que la méthylation du promoteur inhibait la transcription ribosomale

dans un contexte chromatinien, mais non sur de l'ADN nu. Finalement, ils ont déterminé,

par mutagenèse dirigée, que la méthylation du dinucléotide CpG localisé à la position -

133pb, par rapport au site d'initiation, est essentiel à l'inhibition de la transcription et que

cette méthylation bloque l'attachement d'UBF à l'ADN plasmidique (Santoro et Grummt,

2001). L'effet de la méthylation de ce site sur les gènes endogènes n'a pas été étudié. Tip5

a aussi la capacité d'interagir avec les ADN méthyltransférases 1 et 3b, DNMT1 et 3b, et

avec le complexe co-répresseur Sin3A qui contient, entre autres, les histones déacétylases 1

et 2, HDAC 1 et 2 (Santoro et al., 2002; Zhou et al., 2002).

L'ensemble de leurs résultats les ont menés à l'élaboration du modèle présenté à la

figure 1.12. La séquence d'événement menant à l'inactivation des gènes ribosomaux chez

la souris commence par l'attachement de TTFl à son site To en aval du promoteur

ribosomal. Par la suite, TTFl recrute le complexe de remodellage NoRC via son interaction

avec Tip5. Cependant, il n'est pas encore connu si l'attachement de TTFl est essentiel à

l'action répressive de NoRC. NoRC remodèle alors les nucléosomes au niveau du

promoteur, cela n'a pas été démontré, et recrute les HDAC 1 et 2 ainsi que les DNMT1 et

3b de façon à établir une structure hétérochromatinienne. La méthylation du dinucléotide

CpG en position -133pb empêche alors l'attachement d'UBF au promoteur et par

conséquent bloque la formation du complexe de pré-initiation. Selon ce modèle, la

39

méthylation de la position -133pb est essentielle puisque c'est elle qui bloque la formation

du PIC. Cette marque est donc importante pour le maintien des gènes ribosomaux dans un

état silencieux (Santoro, 2005). Toutefois, comme mentionné, la majorité des expériences

qui ont menées à l'élaboration de ce modèle ont été faites sur des gènes rapporteurs.

Transcription

Historié deacetylatlon

Nucleosoma 7 remodelllng —

* Transcription

DNA méthylation

X ► No transcription

Figure 1.12 Modèle d'inactivation des gènes ribosomaux chez la souris.

Le recrutement de NoRC au promoteur ribosomal se fait via son interaction avec TTFI. NoRC recrute ensuite HDACl, qui déacétyle la queue N-terminale des histones, et les ADN Méthyltransférases 1 et 3b. La méthylation de l'ADN inhibe la transcription en empêchant le recrutement d'UBF au promoteur et donc la formation du complexe de pré­initiation. Tiré de Grummt et Pikaard, 2003.

1.3.3.2. Les ADN méthyltransférases mammifères La méthylation de l'ADN fut observée pour la première fois en 1948 par Hotchkiss

dans de l'ADN de thymus de veau. Chez les mammifères, c'est le carbone en position 5 de

la cytosine qui est méthylé lorsque celle-ci est suivit d'une guanine, 5'-CG-3'. Chez

d'autres organismes, comme chez les plantes et les eubactéries, l'azote en position 6 de

40

l'adénine et l'azote en position 4 de la cytosine peuvent aussi être méthylées (Hermann et

al., 2004; Klose et Bird, 2006). Dans tous les cas, le groupement méthyl se retrouve dans le

sillon majeur de l'ADN où il n'interfère pas avec l'appariement des bases (figure 1.13)

(Hermann et al., 2004).

La méthylation de l'ADN est la seule modification covalente de l'ADN connue chez

les métazoaires. La 5-methylcytosine, m5C, est souvent considérée comme étant la

cinquième base de l'ADN puisqu'entre 3 et 8% des cytosines sont méthylées dans l'ADN

génomique des mammifères (Turek-Plewa et Jagodzinski, 2005). De plus, si l'on ne

considère que les cytosines localisées dans les îlots CpG c'est entre 60 et 90% d'entres-

elles qui sont méthylées. Cependant, l'existence même de la 5-methylcytosine est

intriguante puisque cette base est sujette à une déamination spontanée qui résulte en une

transition d'une m5cytosine vers une thymine et en une progressive perte de m5C au cours

de l'évolution (Jones et Baylin, 2002). Les cytosines non méthylées sont aussi sujettes à

une déamination, mais l'uracil qui en résulte est reconnu comme étant étranger dans l'ADN

et est remplacé par une cytosine via le sentier de l'uracil-deglycosylase (Hermann et al.,

2004). Cette situation fait en sorte que les îlots m CpG sont des sites sensibles aux

mutations et que les transitions m5C vers thymine représente près de 30% des mutations

ponctuelles retrouvées dans les cellules germinales et dans les cellules somatiques (Jones et

Baylin, 2002). Cependant, certaines régions riches en îlots CpG, fréquemment localisées en

5' des régions codantes ou dans des régions promotrices, sont maintenues au cours de

l'évolution parce qu'elles sont généralement déméthylées.

Figure 1.13 Structure de l'ADN-B méthylé. L'ADN est coloré selon le type atome et les groupements méthyl sont en vert. Cette figure provient de Hermann et al., 2004.

41

Le patron de méthylation de l'ADN forme une partie du bagage épigénétique, ou de

l'épigénome, d'une cellule qui complète le contenu génétique. Il est maintenant connu que

la méthylation de l'ADN joue un rôle déterminant dans le fonctionnement cellulaire. Par

exemple, la méthylation de l'ADN est impliquée dans la régulation de l'expression génique

(Bird, 2002 et références incluses), dans le développement et la différenciation cellulaire

(Li, 2002 et références incluses), dans l'empreinte génétique parentale (Murphy et Jirtle,

2003 et références incluses), dans l'inactivation liée au chromosome X (Chow et Brown,

2003 et références incluses) et dans l'inhibition de la transcription de rétrotransposons

(Yoder et al., 1997 et références incluses). Il est suggéré que la méthylation de l'ADN

inhibe la transcription de façon directe en bloquant le recrutement de facteurs de

transcription ou de façon indirecte en recrutant des protéines impliquées dans la formation

de l'hétérochromatine (Hermann et al., 2004; Klose et Bird, 2006). La réalisation de souris

dans lesquelles les gènes codant pour Dnmtl, 3A et 3B ont été inactivés génétiquement,

descriptions plus loin, ainsi que l'existence de patrons de méthylation anormaux dans

certaines maladies humaines et dans une large variété de cancers (Esteller, 2002; Esteller et

Herman, 2002; Feinberg et Tycko, 2004; Jones et Baylin, 2002) renforcent l'importance de

la régulation épigénétique, via la méthylation de l'ADN, sur l'expression génique dans les

cellules de mammifères. Enfin, puisque l'épigénome d'une cellule doit être conservé pour

assurer le fonctionnement normal de la cellule, celui-ci doit être rétabli lors de la réplication

de l'ADN.

Les enzymes responsables de la méthylation de l'ADN sont nommées ADN

méthyltransférases ou DNMT pour « DNA Methyltransferase ». Il y en a cinq chez les

mammifères : Dnmtl, 2, 3a, 3b et 31 (figure 1.14). Cependant, seules Dnmt 1, 3a et 3b ont

à ce jour été montrées comme étant catalytiquement actives (Hermann et al., 2004).

Classiquement, Dnmt 1 était considérée comme l'ADN methyltransferase nécessaire au

maintien de la méthylation de l'ADN tandis que Dnmt 3a et 3b étaient considérées comme

nécessaires pour la méthylation de novo. Cependant, plusieurs observations suggèrent que

la séparation entre ces deux fonctions n'est pas si stricte. Toutes les DNMTs de

mammifères, exception faite de Dnmt 31, se caractérisent par un domaine catalytique qui

contient dix motifs d'acides aminés très semblables à ceux présents chez les enzymes

42

procaryotiques (figure 1.14) (Hermann et al., 2004). Le domaine catalytique, constitué de 6

feuillets p flanqués d'hélices a, peut être divisé en deux sous-domaines. Le premier forme

une pochette dans laquelle se logera la base à modifier. Le deuxième forme une pochette

avec laquelle le S-adenosyl-L-Methionine, le donneur du groupement méthyl, interagira. La

structure de ces deux sous-domaines est similaire et résulte probablement d'une duplication

de gène (Hermann et al., 2004). Dnmt 1, 3a et 3b contiennent aussi un domaine de

régulation situé en N terminal (figure 1.14) qui contient, entre-autre, des domaines

d'interactions protéine-protéine. Chacune des ADN méthyltransférases sera décrite

individuellement dans les sections suivantes.

Kcgulatonr Domain

DNMT1

Keplicarion Cyn-rich Fuel Targtttag

Catalyric Domain

< OIIM-rtril n i r l l l l l l i Kii>li I..M motif*

SAM .

DNMT3A

DNMT3B

DNMT2

Kvigulalory Domain

I IV VI IX X

Calalylk Domain

Cy*-ffcn PHD Domain

Homology to ATRX

DNMT3L

I IV V! IX X

I IV VI

Figure 1. 14 Les ADN méthyltransférases mammifères. Les cinq DNMTs de mammifères sont présentées. Les régions régulatrices ainsi que les domaines connus d'interactions qu'elles contiennent sont indiqués. Le domaine catalytique avec ces différents motifs conservés, indiqués en chiffres romains, est aussi représenté. La région d'attachement au S-adenosyl-L-Methionine, SAM, est délimitée. Seule DNMT31 est considérée comme étant théoriquement inactive puisque son domaine catalytique est tronqué. Tiré de Robertson, 2001.

La totalité des ADN méthyltranférases utilisent le S-adenosyl-L-Methionine, SAM

ou AdoMet, comme source de groupement méthyl (Goll et Bestor, 2005; Hermann et al.,

43

2004). La réaction enzymatique de transfert entre l'AdoMet et la cytosine est présentée à la

figure 1.15A. Le groupement méthyl de l'AdoMet est lié à un atome sulphonium ce qui

déstabilise thermodynamiquement la molécule et la rend très réactive face à une attaque

nucléophilique par un atome d'azote, d'oxygène, de souffre ou encore par un carbone

activé. La réaction de transfert débute par l'attaque du carbone 6 de la cytosine par le

groupement thiol de la cystéine du motif conservé PCQ, motif IV, de l'enzyme. La

formation du lien covalent qui résulte de cette attaque active le carbone 5 et mène au

transfert du groupement méthyl de l'AdoMet à ce carbone. Par la suite, il y a élimination du

proton en position 5 et résolution du lien covalent intermédiaire entre l'enzyme et la

cytosine en position 6. L'acide glutamique du motif ENV, motif VI, de l'enzyme est

nécessaire à la stabilisation du complexe ADN protéine (Hermann et al., 2004).

Il a aussi été montré que ces enzymes tirent leur base cible hors de la double hélice

d'ADN avant la réaction de méthylation (figure 1.15B). Cette particularité mécanistique

des ADN méfhyltransférases, appelée « base flipping » en anglais, ne leur est pas unique

puisque certaines enzymes interagissant avec l'ADN, par exemple lors de la réparation de

l'ADN, en sont aussi capables (Goll et Bestor, 2005). Suite à cette extraction, la base n'est

plus cachée dans l'ADN, mais est plutôt projetée dans le domaine catalytique de l'enzyme.

Cette étroite proximité est nécessaire pour la reconnaissance de la base et pour la réaction

chimique de transfert du groupement méthyl. Au cours de cette étape, les liens hydogènes

nécessaires à l'appariement entre bases sont brisés et les interactions d'empilement, ou

« stacking », sont aussi perdus. Suite à la modification, la base reprendra sa place dans la

double hélice (Goll et Bestor, 2005).

1.3.3.2. J Dnmtl Dnmtl fut la première ADN méthyltransférase identifiée chez les mammifères et

c'est aussi la mieux caractérisée. C'est le groupe du Dr Razin qui identifia le premier son

activité en 1982 (Gruenbaum et al., 1982). Son homologue murin fut aussi le premier à être

clone (Bestor, 1988) et à être exprimé de façon recombinante. Dnmtl possède le plus long

domaine amino-terminal des DNMTs de mammifères (figure 1.14). Ce domaine est

important pour la régulation de l'activité catalytique de l'enzyme et pour sa localisation car

44

il contient un signal de localisation nucléaire ou NLS (figure 1.14) (Goll et Bestor, 2005).

De plus, ce domaine contient un site d'attachement aux ions de Zinc qui sont importants

pour l'interaction de cette enzyme avec l'ADN (figure 1.14) (Fatemi et al., 2001). Cet aussi

ce domaine N-terminal qui médie la majorité des interactions protéine-protéine (Goll et

Bestor, 2005; Hermann et al., 2004).

A ) «»ouf vi

Figure 1.15 Transfert du groupement méthyl. A) Réaction de transfert du groupement méthyl du S-adenosyl-L-Methionine à la cytosine. La réaction est décrite dans le texte. Tiré de Hermann et al., 2004. B) Représentation d'une cytosine tirée hors de la double hélice d'ADN. Tiré Goll et Bestor, 2005.

Dnmtl a été montré comme ayant de 10 à 40 fois plus d'affinité pour l'ADN double

brin hémiméthylé versus l'ADN double brin non méthylé ce qui lui a valu le qualificatif

d'ADN méthyltranférase de maintien (Pradhan et al., 1999; Pradhan et al., 1997). Cette

première qualification basée sur l'activité de l'enzyme fut renforcée par des analyses de

localisation nucléaire au cours du cycle cellulaire. Alors que Dnmtl a une localisation

nucléoplasmique diffuse lors des phases Gl ou G2, elle s'associe aux foyers de réplication

de l'ADN lors de la phase S (Leonhardt et al., 1992; Liu et al., 1998). Ces observations ont

mené à l'élaboration d'une hypothèse qui stipule que Dnmtl, en tant qu'enzyme de

maintien, est nécessaire pour la conservation de l'information épigénétique lors de la

division cellulaire. Ainsi, Dnmtl serait requise pour copier le patron de méthylation des

45

chromosomes mères aux chromosomes filles (Robertson, 2002). Si cette hypothèse est

vraie, l'exclusion de Dnmtl des foyers de réplication résulterait en une perte passive de la

méthylation de l'ADN. C'est effectivement ce qui est observé lors de la fertilisation. À

cette étape du développement, seule Dnmtlo, isoforme spécifique au sexe féminin voir plus

loin, est exprimée dans les embryons de stades pré-implantatoires. Ce variant est

activement exclu du noyau (Carlson et al., 1992) ce qui entraîne une perte progressive de la

méthylation du génome féminin. Cette perte résulte du découplage entre la réplication et la

méthylation de l'ADN pendant les trois premiers cycles de réplication dans les œufs

fertilisés soit jusqu'au stade de 8 cellules.

Malgré le fait que Dnmtl soit clairement importante pour le maintien de la

méthylation de l'ADN, plusieurs résultats expérimentaux suggèrent qu'elle possède

d'autres rôles dans la cellule. Si Dnmtl avait comme unique rôle de maintenir l'information

épigénétique lors de la division cellulaire, alors son expression devrait être forte dans les

cellules prolifératives et faible ou nulle dans les cellules terminalement différenciées.

Toutefois, le niveau de transcription du gène codant pour Dnmtl est élevé dans les cellules

de poumons ou de placenta, mais aussi dans les cellules cardiaques ou nerveuses

(Robertson et al., 1999). Ces résultats suggèrent que Dnmtl possède un autre rôle que celui

du maintien de la méthylation de l'ADN aux foyers de réplication. De plus, lorsque Dnmtl

est sous sa forme activée, c'est-à-dire qu'elle a reconnu son substrat et qu'elle est en

réaction de transfert, sa préférence pour l'ADN hémiméthylé diminue en faveur de l'ADN

non méthylé (Bacolla et al., 1999; Fatemi et al., 2001; Pradhan et Esteve, 2003). Ceci

suggère que Dnmtl puisse aussi être capable de de novo méthylation.

Dnmtl possède trois variants (figure 1.16) qui sont mal caractérisés et qui

pourraient avoir des fonctions cellulaires importantes. Dnmtlb contient 16 acides aminés

de plus que Dnmtl qui sont insérés en 5' du domaine amino-terminal (figure 1.14). Cette

isoforme, initialement identifiée chez l'humain, mais aussi présente chez la souris, est

exprimée dans tous les tissus étudiés à ce jour et son activité in vitro est comparable à celle

de Dnmtl (Bonfils et al., 2000). Sa fonction spécifique n'est toujours pas connue. Les deux

autres isoformes de Dnmtl utilisent un promoteur et un exon 1 qui sont spécifiques aux

46

lignées germinales et aux embryons de stades pré-implantatoires (figure 1.16). Dnmtlo est

spécifique aux oovocytes ainsi qu'aux embryons de stades pré-implantatoires, tandis que

Dnmtlp est spécifique aux spermatozoïdes pachytènes. Dnmtlo a été montrée comme étant

transportée du cytosol au noyau de façon dépendante du stade de développemental

(Mertineit et al., 1998). Le phénotype résultant de l'inactivation génique de Dnmtlo montre

que cette enzyme n'est pas requise pour l'établissement de l'empreinte génomique

maternelle, mais suggère qu'elle puisse être essentielle pour le maintien de la méthylation

de ces régions chez les individus adultes (Howell et al., 2001). L'ARNm mâle, codant pour

Dnmt lp, est plus long que l'ARNm femelle et sa fonction n'est pas connue.

■I IHNA I1««rin

OMAPI. PC'NA-hinJiii" huidmj

»K site

E*<m l« 2 i U N M T I n u i T: =>, r 3

IIWfflflUI'WIH YHH

NUS

III 1 ^i ~<=ï E«m I I 2 A Alternative jpllccd 5

e*(K)4ii

D N M T I H + U u u )

gggKSgrë l I L a M m

**»lp | » ^£j Dnmtlo (IIKiu»

in Oi^yic.kjvcif»; ««on to 2 )

Figure 1.16 Variants d'épissage de DNMT1. Les ARN messager sont à gauche et les protéines produites sont à droite. Tiré de Robertson, 2002.

L'inactivation génique de Dnmtl a été faite chez la souris et les individus

homozygotes meurent in utero au stade de mi-gestation, jours embryonnaires 9.5-10.5, et

présentent un retard de croissance comparativement aux embryons sauvages de la même

portée (Li et al., 1992). L'étude des embryons mutants montre une réduction de l'empreinte

génomique parentale (Li et al., 1992), de l'inactivation du chromosome X (Beard et al.,

1995) et une réactivation de la transcription de rétrovirus endogènes (Walsh et al., 1998).

L'inactivation de cette protéine dans les cellules embryonnaires n'est pas létale et ne nuit

47

pas à leur prolifération puisque les ES Dnmtl-/- peuvent être maintenues en culture (Li et

al., 1992).

L'inactivation génique de Dnmtl a aussi été faite dans une lignée de cellules

humaines provenant d'un carcinome colorectal (HCT 116) qui peuvent être maintenues en

culture (Rhee et al., 2000). Des études faites sur d'autres lignées tumorales à l'aide d'ARN

antisens, qui diminuent fortement l'expression de Dnmtl, permettent aussi la survie

cellulaire (Robert et al., 2003). Ces différentes études dans les lignées tumorales ont été

faites pour analyser l'importance de cette Dnmt dans l'hyperméthylation du promoteur de

certains gènes suppresseurs de tumeurs (Rhee et al., 2002; Rhee et al., 2000). L'inactivation

de gènes suppresseurs de tumeur peut être causée par une mutation génétique sur chacun

des allèles, par une mutation génétique d'un allèle combinée à l'hyperméthylation du

promoteur du second allèle, ce qui inhibe sa transcription, ou par l'hyperméthylation du

promoteur des deux allèles (Esteller, 2002; Feinberg et Tycko, 2004; Jones et Baylin,

2002). L'inactivation génique de Dnmtl dans les cellules HCT 116 entraîne une réduction

de 96% de l'activité méthyltransférase cellulaire. Par contre, cette drastique réduction

entraîne une diminution de seulement 20% dans le contenu de cytosines méthylées (Rhee et

al., 2000). De plus, cette déméthylation de l'ADN est localisée dans les régions satellites

juxtacentromériques (Rhee et al., 2000). Il ne s'agit donc pas d'une réduction globale de la

méthylation de l'ADN. L'inactivation de Dnmtl dans ces cellules n'entraîne pas la

réactivation des gènes suppresseurs de tumeurs car leurs promoteurs sont toujours

hyperméthylés (Rhee et al., 2000). Ces résultats suggèrent que Dnmtl n'est pas la seule

Dnmt à être capable de maintenir les patrons de méthylation.

Un nombre croissant de protéines sont maintenant connues comme interagissant

avec Dnmtl. Toutes ces interactions ne seront pas décrites en détail car ce n'est pas le

propos de cette introduction. Cependant, il est intéressant de noter que Dnmtl interagit avec

un ensemble de protéines connues pour leur régulation de l'expression génique comme par

exemple les histones déacétylases 1 et 2 (HDAC1 et 2), MBD3 (« Methyl-CpG Binding

Domain Protein 3 »), DMAP1 (« DNA Méthyltransférase 1-Associated Protein ») et pRb

(protéine du rétinoblastome) (Robertson, 2001). L'interaction de Dnmtl avec ces

48

régulateurs connus renforce l'hypothèse suggérant que Dnmtl est importante pour la

régulation de la transcription. De plus, les interactions avec différents partenaires sont

probablement importantes pour conférer une spécificité de substrat aux DNMTs

(Robertson, 2002). En effet, malgré le fait que Dnmtl, 3a ou 3b n'aient pas d'affinité

particulière pour des séquences d'ADN spécifiques in vitro, leur inactivation individuelle in

vivo résulte en des variations de patron de méthylation qui diffèrent selon l'enzyme

inactivée. Ces résultats suggèrent que les Dnmts sont des enzymes spécifiques en raison de

l'utilisation de partenaires différents (Robertson, 2002).

1.3.3.2.2. Dnmt2 Dnmt2 est l'ADN méthyltransférase la plus conservée et la plus largement

distribuée dans le règne animal et ce même chez des espèces où l'on ne détecte peu,

Drosophila megalonaster, ou pas, Schizosaccharomyces pombe, de méthylation de l'ADN

(Goll et Bestor, 2005). Dnmt2 fut identifiée en 1998, mais aucune activité ADN

méthyltransférase n'a été détectée à l'époque. Plus récemment, une faible activité de

l'enzyme recombinante a été démontrée in vitro (Hermann et al., 2003 ; Kunert et al., 2003;

Tang et al., 2003) et in vivo (Liu et al., 2003). L'absence d'activité enzymatique est

surprenante puisque cette enzyme est constituée presqu'exclusivement du domaine

catalytique propre aux ADN méthyltransférases (figure 1.14). Cependant, il est possible

que cette enzyme possède une activité méthyltransférase, mais que les systèmes

expérimentaux utilisés ne permettent pas sa détection. Par exemple, l'activité de cette

enzyme nécessite peut-être un partenaire non identifié ou encore un substrat qui ne soit pas

l'ADN comme par exemple TARN.

1.3.3.2.3. Les membres de la famille Dnmt3 La famille des DNMT3 comprend trois membres : Dnmt3a, 3b et 31 (figure 1.14).

Comme il a été mentionné précédemment, Dnmt31 est considérée comme étant

catalytiquement inactive et Dnmt3a et 3b, sont considérées comme étant des ADN

méthyltransférases capables de de novo méthylation puisque ces enzymes ne présentent pas

de préférence pour l'ADN hémiméthylé ou non méthylé (Goll et Bestor, 2005; Hermann et

al., 2004). Ces deux enzymes sont essentielles puisque leur inactivation génique

49

individuelle et combinée chez les souris est létale et entraînent des retards de croissance

(Okano et al., 1999). Les souris DNMT3B-/- meurent in utero au jour embryonnaire 14.5-

18.5 alors que les souris DNMT3A-/- meurent environ quatre semaines après leur naissance

(Okano et al., 1999). La double inactivation DNMT3A-/-3B-/- entraîne la mort des

embryons au jour embryonnaire 8,5 (Okano et al., 1999). Malgré leur grande similitude

structurale (figure 1.14) ces enzymes ont des activités qui leur sont propres. C'est

particulièrement évident chez les individus atteints du syndrome ICF, pour

« Immunodeficiency, Centromeric instability and Facial abnormalities », qui est une

maladie récessive très rare. Les patients atteints sont immunodéficients, présentent des

anomalies faciales, un mauvais fonctionnement de l'intestin, des problèmes neurologiques

et un développement retardé (Franceschini et al., 1995). Ce syndrome est causé par une

mutation dans le gène codant pour Dnmt3b et une des particularités de cette maladie

génétique est que les individus atteints ne sont porteurs que d'un allèle muté (Robertson et

Wolffe, 2000). Ceci suggère que, comme chez la souris, la perte des deux allèles codant

pour Dnmt3b est incompatible avec la vie. Toutes les mutations identifiées à ce jour, sauf

une, se retrouvent dans le domaine catalytique de l'enzyme (Robertson et Wolffe, 2000).

L'existence de cette maladie démontre que Dnmt3a ne peut pas compenser complètement

pour la perte de Dnmt3b.

Un autre exemple de la spécificité des enzymes Dnmt3a et b est le fait l'inactivation

génique combinée de Dnmtl et 3b, dans la lignée de cellules humaines HCT 116, entraîne

la perte quasi totale de l'activité méthyltransférase cellulaire en plus de diminuer de 95% le

contenu en cytosines méthylées (Rhee et al., 2002). Contrairement à l'inactivation unique

de Dnmtl, cette double inactivation entraîne la réactivation de gènes suppresseurs de

tumeur maintenus silencieux par une hyperméthylation anormale de leur promoteur dans

les cellules parentales non manipulées génétiquement (Rhee et al., 2002). Ces deux

exemples, les patients ICF et les cellules DNMT1-/-DNMT3B-/-, montrent que Dnmt3a ne

peut pas compenser pour la perte de l'activité de Dnmt3b.

Des analyses de l'activité catalytique montrent que ces deux enzymes diffèrent

mécanistiquement. Dnmt3a possède une activité plus restreinte que Dnmt3b qui, elle, a une

50

activité processive (Gowher et Jeltsch, 2002). Ainsi, Dnmt3b est efficace pour méthyler des

régions riches en dinucléotides CpG, comme la région péricentromérique, alors que

Dnmt3a est plus efficace pour la méthylation ciblée, par exemple sur des gènes uniques

(Hâta et al., 2002). Le fait que Dnmt3a doive être dirigée à sa cible permet probablement

une régulation plus fine de son activité comparativement à celle de Dnmt3b. Ces deux

protéines, contrairement à Dnmtl, possède un domaine d'interaction avec l'ADN PWWP et

un domaine PHD dans leur région amino terminale (figure 1.14). Le domaine PHD est

souvent retrouvé dans les protéines impliquées dans la régulation de la transcription chez

les eucaryotes. Le domaine PHD de Dnmt3a a été montré comme pouvant inhiber la

transcription et ce indépendamment de son activité méthyltransférase (Bachman et al.,

2001; Fuks et al., 2001). Plusieurs protéines ont été montrées comme interagissant avec

Dnmt3a et/ou 3b. Par exemple ces deux enzymes interagissent avec Dnmtl (Kim et al.,

2002) et avec des HDAC (Fuks et al., 2001). Dnmt3a a aussi été montrée comme

interagissant avec Suv39 qui est une histone méthyltransférase spécifique à la lysine 9 de

l'histone H3 (Fuks et al., 2003). L'ensemble de ces interactions soutient le rôle important

de ces deux enzymes dans la régulation de l'expression génique vraisemblablement par

l'établissement d'une structure hétérochromatinienne.

Dnmt3a possède deux isoformes (figure 1.17). L'isoforme Dnmt3a2 possède 223

acides aminés de moins que Dnmt3a en N-terminal (Chen et al., 2002). Cet isoforme est

produit par l'utilisation d'un site de transcription interne sous le contrôle d'un promoteur

intronique (Chen et al., 2002). Cette région N-terminale tronquée est probablement

importante pour la localisation de l'enzyme. En effet, Dnmt3a est localisée dans

l'hétérochromatine et elle est la forme majeure dans les tissus adultes alors que Dnmt3a2

est localisée dans l'euchromatine et est la forme prédominante lors de l'embryogenèse

(Chen et al., 2002; Hermann et al., 2004). Finalement, il existe aussi une utilisation

différentielle du premier exon lors de la production de Dnmt3a (Weisenberger et al., 2002).

L'impact de l'utilisation différentielle de l'exon 1 sur l'activité de Dnmt3a n'est pas

connue, mais il a été montré que l'exon lli est utilisé dans les cellules embryonnaires

souches alors que l'exon l a est utilisé dans les cellules somatiques (Weisenberger et al.,

2002).

51

I 1 1 1 1 1 1 1 1 1 0 MO 600 nwA^o»— l '► « 0

I)nmt3a ■ ' M ■ " H I 1 ■■ MB

I)nmt3a2 i u » ¥"HI ■ Il ■■

Figure 1.17 Les deux isoformes de Dnmt3a.

Les isoformes Dnmt 3a et Dnmt 3a2 sont représentées avec leurs principaux domaines d'interactions ainsi qu'avec certains motifs conservés du domaine catalytiques. L'exon 1 a et B, utilisé différentiellement lors de la transcription de Dnmt3a, sont aussi indiqués. Tiré de Hermann et al., 2004.

L'enzyme Dnmt3b possède aussi plusieurs isoformes (figure 1.18) qui sont

exprimés à différents niveaux au cours du développement. Ce sont des variants d'épissage

utilisant différentiellement les exons 10, 21 et/ou 22. Les variants Dnmt3bl et 3b2

possèdent une activité catalytique (Aoki et al., 2001) alors que les variants Dnmt3b3, 3b4

et3b5, qui possèdent des domaines catalytiques tronqués, sont probablement inactifs. Ces

derniers pourraient avoir une activité de régulateurs positifs ou négatifs sur les formes

actives ou alors posséder une activité qui soit indépendante de l'activité méthytransférase

(Hermann et al., 2004).

On ml Mil r

OnmLM>2 rj

DnmObJ £

DiimKI.I

DnmUbS

RL'"i(in-1

PWWP ATRX I I I 1 1

.WMMin Ihmmt T////////A

Reeion-2

IV VI IX X

■ ■ m P A I M A H Ml

:, — ■ . m

T T

■ ■ mini

figure 1.18 : Les cinq isoformes de Dnmt3b.

Les cinq isoformes de Dnmt3b et les domaines importants pour la fonction de l'enzyme sont indiqués. Tiré de Robertson, 2002.

52

Le dernier membre de la famille Dnmt3, Dnmt31, est uniquement exprimé dans les

cellules germinales (Aapola et al., 2000). Aucune activité méthyltransférase n'a été

identifiée chez cette protéine et elle aurait plutôt un rôle de protéine régulatrice puisqu'elle

peut interagir avec Dnmt3a (Chedin et al., 2002) et Dnmt3b (Ilata et al., 2002).

L'inactivation de cette protéine dans les souris permet la naissance de souris viables, mais

infertiles (Bourc'his et al., 2001).

1.3.3.3. La méthylation de l'ADN et la formation de l'hctcrochromatine. Si la méthylation de l'ADN est une marque caractéristique de l'inhibition de la

transcription c'est qu'elle est étroitement associée à l'hétérochromatine. De multiples

évidences soutiennent que les modifications de la queue des histones et la méthylation de

l'ADN sont mécanistiquement étroitement liés (Bird, 2002; Li, 2002 et références

incluses). Trois modèles sont suggérés pour expliquer l'interdépendance de la méthylation

de l'ADN et des modifications des histones (figure 1.19). Le premier suggère que la

méthylation de l'ADN entraîne la méthylation des histones (figure 1.19A). Dans ce modèle,

la méthylation de l'ADN serait établi par les de novo méthyltransférases de la famille

Dnmt3, Dnmt 3a et b, et maintenu par Dnmtl. La méthylation des cytosines serait reconnue

par des protéines de la famille MBD pour « Methyl Binding Domain Protein ». Des

histones déacétylases seraient ensuite recrutées par leurs interactions avec les Dnmts et/ou

les MBD et déacétyleraient les histones. Des histones méthyltransférase, HMT,

stabiliseraient ensuite l'hétérochromatine en méthylant la lysine 9 de l'histone H3, H3K9.

En support à ce modèle, des études ont montré que la méthylation de l'ADN était liée à la

déacétylation des histones via MeCP2 une protéine de la famille des MBD (Jones et al.,

1998; Nan et al., 1998). Les interactions des Dnmts avec des protéines modifiant les

histones comme, par exemple, les HDACs ou Suv 39, renforce aussi ce modèle (Robertson,

2001).

Le deuxième modèle (figure 1.19B) suggère que les modifications des histones

induisent la méthylation de l'ADN. La présence de la méthylation H3K9 permettrait le

recrutement de HPl qui à son tour recruterait de façon directe ou indirecte, par un facteur X

sur la figure 1.19B, les Dnmts qui stabiliseraient l'hétérochromatine en méthylant l'ADN.

53

Ce modèle fut développé à partir de résultats expérimentaux qui montrent que la

méthylation H3K9 est nécessaire pour la méthylation de l'ADN (Jackson et al., 2002;

Tamaru et Selker, 2001). Ce modèle peut expliquer la reformation de l'hétérochromatine

suite à la réplication de l'ADN.

a Modal al DNA méthylation direcUng Mttone méthylation

Aî:tNe chromatin rnethytaton ineltlylatiDn inw:tPve chromalm

Figure 1.19 Trois modèles expliquant la formation de l'hétérochromatine. A) La méthylation de l'ADN induit la méthylation des histones. Le recrutement de DNMTs et la méthylation de l'ADN serait la première étape de la formation de l'hétérochromatine. Par la suite, des HDAC et des MBD sont recrutées et recruteraient à leur tour HP1 et des HMT qui méthyleraient entre-autre la lysine 9 de l'histone 3. B) La méthylation des histones induit la méthylation de l'ADN. Suite à la réplication de l'ADN, la présence de la marque hétérochromatinicnne H3K.9 entraînerait le recrutement de HP1 et de Dnmts sur l'ADN hémiméthylé. C) Le remodelage ATP-dépendant de la chromatine induirait la méthylation de l'ADN. Dans ce modèle, le remodelage de la chromatine faciliterait l'accès des Dnmts à l'ADN. La méthylation de l'ADN permettrait ensuite ou simultanément la méthylation de H3K9. Tiré de Li, 2002.

Le troisième modèle (figure 1.19C) suggère que le remodelage de la chromatine

induit la méthylation de l'ADN. Les complexes de remodelage de l'ADN ATP-dépendant

pourraient faciliter l'accès des Dnmts et des modificateurs de la queue des histones,

HDACs et HMTs, à l'ADN et aux histones ce qui stabiliseraient alors l'hétérochromatine.

Ce modèle est supporté par des expériences qui montrent qu'une mutation dans les gènes

54

codant pour ATRX ou Lsh, deux membres de la famille Swi/Snf2, diminue le niveau de

méthylation de l'ADN (Dennis et al., 2001; Gibbons et al., 2000). De façon intéressante, la

mutation dans ATRX mène à une déméthylation des gènes ribosomaux (Gibbons et al.,

2000). Le modèle suggéré pour l'inactivation des gènes ribosomaux silencieux (figure 1.12)

pourrait probablement être classé dans cette catégorie car le remodelage de la chromatine

par NoRC est suggéré comme étant la première étape de l'inactivation, même si cela n'a

pas été démontrée expérimentalement (Santoro, 2005). Il est raisonnable de penser que le

mécanisme de formation endogène de l'hétérochromatine n'est pas uniforme et qu'il s'agit

plutôt d'un mélange de tous ces modèles que la cellule utilise selon l'exigence de la

situation. Par exemple, le mécanisme utilisé varie peut-être en fonction du type de séquence

à inactiver, gène unique ou ADN répétitif, ou de l'état original de la chromatine, active ou

inactive. Ainsi, la méthylation de l'ADN joue peut-être un rôle de stabilisateur et/ou

d'initiateur dans la formation de la l'hétérochromatine et il apparaît évident que sa présence

est essentielle au développement (Li, 2002) et au fonctionnement normal de la cellule

(Robertson, 2001; Robertson et Wolffe, 2000).

En conclusion, Dnmtl est importante pour le maintien du bagage épigénétique lors

de la réplication de l'ADN puisque des analyses de localisation ont montré que Dnmtl est

associée à la fourche de réplication tout au long de cette phase du cycle cellulaire

(Leonhardt et al., 1992; Liu et al., 1998). Des études ont aussi montré que Dnmtl peut

maintenir à lui seul l'information épigénétique des régions euchromatiniennes, répliquées

au début de la phase S, alors qu'il ne peut pas le faire pour les régions

hétérochromatiniennes, répliquées tardivement en phase S (Liang et al., 2002). Ceci

pourrait peut-être s'expliquer par le fait que l'information épigénétique de l'euehromatine

correspond à un patron précis de méthylation contrairement à l'hétérochromatine où les

dinucléotides CpG sont méthylés par défaut. Puisque Dnmt3a et 3b sont insensibles à l'état

de méthylation de l'ADN (hémiméthylé ou non), il est peu probable que ces enzymes sont

assez précises pour permettre la conservation exacte du patron de méthylation de

l'euehromatine (Hermann et al., 2004) ce qui explique que Dnmtl puisse, ou doive, le faire

seul. Cependant, l'activité de Dnmtl n'est pas suffisante pour copier le grand nombre de

méthylation présentes dans l'hétérochromatine et cette étape de réplication nécessiterait

55

aussi la participation de Dnmt3a ou 3b (Hermann et al., 2004). De plus, la méthylation en

cis de dinucléotides CpG a été montrée comme activant Dnmtl en diminuant sa préférence

pour l'ADN hémiméthylé en faveur de l'ADN non méthylé (Pradhan et Esteve, 2003). La

coopération des trois enzymes est probablement essentielle pour copier la méthylation de

l'hétérochromatine sur le brin nouvellement répliqué (Hermann et al., 2004). Le maintien

du patron de méthylation des gènes ribosomaux requiert probablement la participation de

Dnmtl et 3a et/ou 3b puisqu'une partie des gènes est transcrite et donc répliquée au début

de la phase S et l'autre partie est silencieuse et donc répliquée de façon plus tardive.

Cependant, cette hypothèse n'a pas encore été vérifiée et la participation essentielle des

différentes Dnmts dans l'établissement et le maintien des gènes ribosomaux n'est pas

encore connue.

1.4. Contenu de la thèse Au début de mon doctorat, il existait plusieurs questions non résolues concernant les

mécanismes de régulation de la transcription ribosomale. Mon doctorat a tenté de répondre

à trois grandes questions contenant chacune un ensemble de sous-questions. La première de

ces questions consistait à comprendre comment la transcription ribosomale est régulée en

fonction de la croissance. Plus précisément, j 'ai voulu savoir quelle est la voie de

signalisation impliquée dans cette régulation? Qu'elle est la cible de cette voie de

signalisation? Comment cette cible est modifiée et quel est l'effet de cette modification?

Les chapitre 2 et 3 répondent à ces premières questions. Le chapitre 2 montre que la

protéine UBF est phosphorylée par ERK1/2 suite à une stimulation par le facteur de

croissance EGF. Cette étude montre que l'intégrité de la voie ERK et la phosphorylation

d'UBF par ERK sont nécessaires à l'augmentation de la transcription ribosomale suite à

une stimulation par un facteur de croissance. Ces résultats sont les premiers à montrer que

la transcription ribosomale est directement et immédiatement régulée en fonction de la

croissance. L'ensemble de ces résultats et la démonstration de la présence d'UBF sur toute

la longueur des gènes ribosomaux suggèrent qu'UBF puisse avoir un rôle de régulateur de

la transcription en modifiant la structure de la chromatine ribosomale ce que confirme les

résultats présentés au chapitre 3. En effet, cette étude montre qu'UBF non phosphorylé

56

bloque l'ARN polymérase I et qu'UBF phosphorylé permet l'élongation de la polymérase.

Il s'agit de la première démonstration de l'existence d'une régulation au niveau de

l'élongation des transcrits. Ces deux études mettent en évidence le rôle central d'UBF et de

la voie ERK dans la régulation de la transcription ribosomale en réponse à la croissance.

Le deuxième objectif de mon doctorat était de déterminer si UBF est essentiel à la

transcription ribosomale et à la survie cellulaire. En effet, au début de mon doctorat, il

existait une différence entre les résultats obtenus in vitro versus in vivo qu'en a la

nécessaire présence d'UBF. Or, cette protéine est potentiellement importante dans la

régulation de la formation du complexe de pré-initiation puisqu'elle est essentielle pour la

formation de « l'enhancesome » au promoteur. Mon projet de doctorat consistait donc à

faire l'inactivation géniquc d'UBF dans la souris pour déterminer la réelle nécessité de

cette protéine dans la transcription ribosomale. Pour pouvoir comparer le phénotype

résultant de la perte d'UBF avec un phénotype Poli-spécifique, j 'ai aussi produit les outils

nécessaires pour l'inactivation génique de TAF168 et de TAFi95. Ces résultats sont

présentés au chapitre 5.

La troisième partie de mon doctorat consistait à étudier un autre niveau de

régulation possible de la transcription ribosomale soit une variation du nombre de gènes

ribosomaux transcrits. Il était déjà connu qu'une large partie des gènes ribosomaux sont

silencieux en tout temps, mais plusieurs questions étaient toujours sans réponse concernant

cette catégorie de gènes chez les mammifères. Par exemple, comment les gènes sont

maintenus silencieux et pourquoi? Malgré le rôle proposé de la méthylation de F ADN dans

l'établissement et le maintien de gènes ribosomaux silencieux chez les mammifères, son

importance dans les cellules humaines n'avait pas encore été déterminée. L'étude présentée

au chapitre 4 a permis d'obtenir des résultats surprenants et inattendus. De plus, l'ensemble

de ces résultats permet d'émettre une hypothèse concernant la nécessaire conservation, au

cours de l'évolution, d'une grande partie des gènes ribosomaux dans un état silencieux.

Finalement, les résultats présentés dans cette thèse permettent de suggérer que les différents

mécanismes de régulation de la transcription ribosomale entraînent une conséquence

commune indispensable à une biogenèse efficace des ribosomes.

Chapitre 2 : Une stimulation par le facteur de croissance EGF entraîne, chez les mammifères, une réponse immédiate de la transcription ribosomale via la phosphorylation d'UBF par ERK;

« An Immédiate Response of Ribosomal Transcription to Growth Factor Stimulation in Mammals is Mediated by ERK Phosphorylation of UBF »

* * + *

Victor Y. Stefanovsky , Guillaume Pelletier , Ross Hannan , Thérèse Gagnon-Kugler ,

Lawrence I. Rothblum , and Tom Moss . *

Cancer Research Centre and Department of Médical Biology, Laval University, Hôtel-

Dieu de Québec, 11 côte du Palais, G1R 2J6 Québec, Canada. ^ Baker Médical Research

Institute, St.Kilda Road Central, Melbourne, Victoria 8008, Australia. Henry Hood Research

Program, Sigfried and Janet Weis Center for Research, Geisinger Clinic, 100 N. Academy

Ave., Danville, PA 17822-2618. ^LBME - CNRS UMR 9006, 118, Route de Narbonne,

31062 - Toulouse Cedex, France.

2.1. Avant-propos Cet article a été publié dans la revue « Molecular Cell » en novembre 2001 (vol

8 :1063-1073) et est reproduit avec la permission de l'éditeur. Victor Stefanovsky a fait la

majorité des expériences. Guillaume Pelletier a participé à la production des mutants

d'UBF et à l'expression des boîtes HMG individuelles. Ross Hannan a fait les essais

préliminaires d'activation du plasmide rapporteur avec les mutants d'UBF. Lawrence I

Rothblum était le directeur de recherche de Ross Hannan. Les expériences que j 'ai réalisées

ont été faites lors de la révision de l'article. J'ai fait une série d'essais de Run On nucléaires

préliminaires pour mesurer le nombre de polymérases engagées. Malgré le fait que ces

données ne sont que mentionnées dans l'article, elles ont été essentielles pour convaincre le

comité de pairs de la qualité et de l'importance des résultats présentés dans cet article. Par

58

la suite, les résultats de ces expériences ont formé la base des travaux qui sont présentés au

chapitre 3.

J'ai jugé nécessaire d'insérer cet article comme chapitre et non en annexe car les

résultats publiés ont permis de changer complètement la façon de concevoir la transcription

ribosomale. Cette étude avait pour but de vérifier si la transcription ribosomale était régulée

en fonction de la croissance. Nos résultats montrent, pour la première fois, que la

transcription ribosomale augmente, et ce en 10 minutes, suite à un traitement avec le facteur

de croissance épidermal EGF ou « Epidermal Growth Factor ». Ainsi, la transcription

ribosomale répond à une signalisation de croissance provenant de l'extérieur de la cellule.

De plus, il est nécessaire de reclassifier les gènes ribosomaux dans la catégorie des gènes de

réponse immédiate. Nous avons aussi montré que l'activation de la transcription ribosomale

est dépendante de la phosphorylation du facteur architectural UBF par ERKl/2. Fait

intéressant, cette phosphorylation diminue l'interaction d'UBF avec l'ADN et un cycle de

phosphorylation/déphosphorylation est nécessaire pour l'augmentation de la transcription.

Cet article a montré que la transcription ribosomale ainsi que la capacité d'UBF à modifier

la structure de la chromatine sont régulées suite à une stimulation par un facteur de

croissance. Ces résultats ont mené aux études présentées aux chapitre 3 et 5 de cette thèse.

2.2. Résumé La transcription ribosomale est régulée, chez les mammifères, en réponse à la

croissance, à la différenciation, aux maladies et au vieillissement. Cependant, les

mécanismes impliqués dans cette régulation sont toujours inconnus. Nos résultats montrent

que l'EGF induit une activation immédiate de la transcription ribosomale endogène qui est

dépendante de ERKl/2. De la même façon, l'inactivation de ERKl/2 cause une réduction

immédiate de la transcription à un niveau basai. Nous avons trouvé que la cible de ERKl/2

est le facteur architectural UBF. ERK 1/2 phophoryle les acides aminés 117 et 201 d'UBF

qui sont localisés respectivement dans les boîtes HMG1 et 2 et prévient ainsi leur

interaction avec l'ADN. La mutation de ces sites inhibe l'activation de la transcription et la

réponse transcriptionnelle à ERKl/2. Par conséquent, la régulation de la transcription

ribosomale par les facteurs de croissance doit probablement agir de façon cyclique sur la

59

structure de l'ADN. Nos résultats suggèrent que la biogenèse des ribosomes joue un rôle

central dans la régulation de la croissance.

2.3. Summary Ribosomal transcription in mammals is regulated in rcsponse to growth,

differcntiation, disease and ageing but the mechanisms of this régulation hâve remained

unresolved. We show that epidermal growth factor induces immédiate, ERK1/2 dépendent,

activation of endogenous ribosomal transcription, while inactivation of ERK1/2 causes an

equally immédiate reversion to the basai transcription level. ERK1/2 was found to

phosphorylate the architectural transcription factor UBF at a.a. 117 and 201 within HMG

boxes 1 and 2 preventing their interaction with DNA. Mutation of thèse sites inhibited

transcription activation and abrogated the transcriptional response to ERK1/2. Thus, growth

factor régulation of ribosomal transcription likely acts by a cyclic modulation of DNA

architecture. The data suggest a central rôle for ribosome biogenesis in growth régulation.

2.4. Introduction The primordial rôle of ribosomal RNA transcription as regulator of ribosome

production makes it a key target for growth régulation (Warner, 1999). It has been shown

that changes in growth and prolifération occurring during processes such as organ

régénération, (Castle et al., 1978; Friedman et al., 1984; Loeb and Yeung, 1975; Nikolov et

al., 1991), differentiation (Cavanaugh et al., 1995; Stoykova et al., 1983; Zahradka et al.,

1991), hypertrophy (Hannan et al., 1996; Luyken et al., 1996), ageing (Castle et al., 1978)

and starvation (Grummt et al., 1976) are ail correlated with changes in the rate of ribosomal

transcription and nucleolar size and number has for many years been used as a measure of

tumour quality, e.g. (Derenzini et al., 2000). Ribosomal transcription rates hâve been found

to vary by up to a factor of four. Though such changes are small when compared to the

régulation of many mRNA gènes, they are very significant in the context of cell growth and

prolifération rates. Given our présent knowledge, it would seem reasonable to suppose that

stimulators of cell growth and/or prolifération directly regulate transcription of the

ribosomal gènes.

60

In addition to the dedicated RNA polymerase I (Poli), ribosomal transcription

requires the so-called Upstream Binding Factor UBF, a multiple HMG box architectural

factor, and the "selectivity" factor SLl which contains the TATA-box binding protein

(TBP), see (Paule, 1998). UBF catalyses the recruitment and binding of SLl, or of the SLl

containing holo-polymerase (Hannan et al., 1999a; Saez-Vasquez and Pikaard, 1997;

Seither et al., 1998) to the Poli promoter, making it a potential regulator of pre-initiation

complex assembly and promotion. The minimal or core région of UBF necessary for

transcription in vitro spans its N-terminal dimerization domain and the three HMG boxes

constituting its DNA binding région (Jantzen et al., 1992; McStay et al., 1991). A dimer of

this "core" UBF forms the ribosomal enhancesome, a structure in which about 140 bp of

ribosomal DNA is looped into a single 360 deg. turn, (Bazett-Jones et al., 1994;

Stefanovsky et al., 1996), suggesting that DNA architecture plays a key rôle in the

promotion of transcription by Poli. Previous studies hâve suggested rôles for both casein

kinase II (CKII) (Kihm et al., 1998; O'Mahony et al., 1992a; O'Mahony et al., 1992b; Tuan

et al., 1999; Voit et al., 1992) and cyclin dépendent kinase 2/4 (CDK2/4) (Voit et al., 1999)

in regulating UBF activity. Differentiation, sérum deprivation and glucocorticoids hâve

been found to induce a régulation of Poli activity probably via the polymerase associated

RRN3 or TIF-1A factor, (Bodem et al., 2000; Miller et al., 2001; Moorefield et al., 2000).

Thèse régulations occur over extended time periods, reinforcing the established view that

ribosomal transcription responds indirectly to changes in cellular metabolism. In contrast,

we show that activation of the classic MAP-kinase (ERK) pathway induces an immédiate

up-regulation of ribosomal transcription via UBF modification. As such, our data add yet

another ERK regulated transcription factor to a growing list (Sharrocks et al., 2000), but

more importantly demonstrate the existence of a direct link between growth factor

signalling and ribosome biogenesis in mammalian cells.

2.5. Results To better understand the relationship between growth stimulation and ribosomal

gène transcription we asked whether the treatment of mammalian cells with a typical

growth factor, in this case human epidermal growth factor (EGF), activated ribosomal

61

transcription, and if so was this activation a slow or a rapid cvent. Treatment of the human

neuroepithelioma cell line SKF-5 (van Weering et al., 1995) with EGF resulted in an

increase in endogenous 45S pre-ribosomal RNA (45S pre-rRNA) transcription by 2.5

times, Figure 2.1 A. This degree of activation is typical of the ribosomal gènes, see

Introduction. Already 10 min. after EGF addition, 45S pre-rRNA transcription had risen to

1.4 times the basai level and by 30 min had increased 2.5 times. Pre-rRNA transcription

remained about 2 fold the basai level for 360 min., but had nearly returned to the

unstimulated level after 24 h. Neither the changes in 45S half-life nor the uptake of 32P-

orthophosphate could explain thèse data. The 45S half-life was found to decrease by around

20%, while cellular uptake of 32P-orthophosphate was reduced by less than 20% after EGF

stimulation, see Expérimental Procédures. Since both effects tended to mask an increase in

transcription rate, our measurements somewhat underestimated the true degree of

stimulation of 45S transcription by EGF. The extrême rapidity of the response to EGF was

completely unexpected. It compared closely with the earliest nuclear responses to growth

factors, e.g. the activation of the c-fos, c-jun and actin gènes, which occur within 10 to 30

min. of growth factor addition, e.g. see (Quantin and Breathnach, 1988) and références

therein. On this basis, the ribosomal gènes should, in fact, be re-classified as immédiate

response gènes.

No significant changes in cell cycle distribution of the SKF-5 cells were observed

during the EGF treatment. The percentage of Gl cells remained unchanged, while a modest

8% increase in G2 cells and an équivalent decrease in S phase cells was noted after 6h of

EGF treatment, see Expérimental Procédures for détails. Thèse changes were far too small

to account for the two fold increase in transcription either by an increase gène copy number

or by mitotic régulation of the Poli machinery, see Expérimental Procédures. Further,

measurements of CDK2 and CDK4 activities showed that the latter did not change while

the former activity decreased by 20%. Since CDK2 and/or CDK4 activity is required for

ribosomal gène transcription (Voit et al., 1999), thèse changes again tended to mask its

EGF activation.

62

4 * fc«Jh D)

■ « 10 30 «0 M MO Uh Titt*» o( CGf trwrtnwM. min

MBP

boic2

0 M M M M MO" M» IGr InMlmwtt, min

fô&OSO O 0 1 *

0 10 30 40 60 360 4-»iT traaliMwtt, mm

UBP

v jb Atf i Addlbon o< POM04Q

P U Stop

.<KI_ BOml»

0 M U> «T Tlm« 01 POMOS» «ddilloo. min.

Figure 2.1 ERK activation leads to an immédiate increase in ribosomal RNA transcription. 32, A) Time course of transription activation by EGF in SKF-5 cells. Inlayed panel: top; P

pulse-labelled 45 S precursor rRNA before and lh after stimulation, bottom; EtBr-stained 28S and 18S rRNA. "-" no EGF. B) Corresponding time course of ERK2 activity in the SKF-5 cells measured on MBP and UBF HMG box 2 (box2) substrates. Inlayed panel: the

63

autoradiographs of typical assays. C) SKF-5 cells were mock treated or treated for 1 hr with EGF or a combination of EGF and 5 (+) or 50 (++) \M PD 98059 or 1 uM U0126 and then 32P puise labelled as in A). D) 32P pulse-labelling of the 45S-rRNA precursor and E) ERK2 activity in mouse NIH3T3ARaf:ER cells before and after stimulation with 4-hydroxytamoxifen (4-HT). F) Puise labelling of endogenous 45S precursor rRNA in the untreated and 4-HT or 4-1 IT plus 50uM PD98059 or 4-HT plus luM U0126 treated NIH3T3ARaf:ER cells. Puise labelling was started 1 hr after addition of 4-HT or 4-HT plus inhibitor. G) ARafER cells were stimulated with 4-HT and one hour later were puise labelled to reveal the 45S precursor rRNA transcription rate as in F). PD98059 was added at the indicated times before puise labelling to inhibit ERK activation, see inset. In A), C), D), F) and G) labelling was for 20 min and 32P incorporation into 45S RNA was normalised tobulk 18SRNA.

2.5.1. ERK activation is necessary and sufficient to activate ribosomal transcription in mammalian cells

The rapid response to EGF suggested a direct signalling mechanism via the MAP-

kinase (ERK1/2) cascade. In the SKF-5 cells, ERK1/2 activity was rapidly elevated 10 fold

by EGF treatment and remained more than 6 fold above the basai level for 6 h, dropping to

about 5 fold by 24 h, Figure 2.1B. Thus, the endogenous ribosomal transcription rate

qualitatively followed ERK1/2 activity. When ERK1/2 activation was specifically blocked

by the MEK inhibitors PD98059 or U0126 (Dudley et al., 1995; Favata et al., 1998), EGF

activation of 45S transcription was already significantly inhibited at 5uM PD98059 and

completely inhibited by 50uM PD98059 or at 1 piM U0126, Figure 2.1C. Both PD98059

and U0126 even suppressed basai ribosomal transcription, suggesting that a level of ERK

activity was necessary for its maintenance. To facilitate comparison with subséquent data,

the transcriptional response of the ribosomal gènes to ERK 1/2 activation, or "Poli

Activation Ratio" (PAR), was defined as the ratio between transcription at maximal ERK

activation, (i.e. hère with EGF) and at minimal ERK activation (i.e. with 50uM PD98059).

1 h after EGF addition, the "PAR" for endogenous transcription in the SKF-5 cells was 3.3,

Figure 2.1C. Cellular uptake of 32P-orthophosphate was unaffected by PD98059. 50uM

PD98059 treatment somewhat inhibited endogenous CDK2/4 kinases, but to a very similar

level as did EGF treatment, i.e. less than 30% for PD98059 as compared with 20% for

EGF. Thus changes in CDK2/4 activities did not affect the PAR and could not hâve

explained the inhibitory effects of PD98059. It was concluded that, in the EGF stimulated

human SKF-5 cells, the ribosomal transcription rate not only qualitatively followed, but

64

was also dépendent on ERKl/2 activation. Nuclear Run-On experiments in SKF-5 cells

treated as in Figure 2.1C repeatedly gave a PAR of 1.6, see Expérimental Procédures. Thus,

EGF stimulated 45S transcription in part by increasing Pol 1 loading onto the ribosomal

gènes, but also by another mechanism, presumably leading to an increase in elongation and

termination rates.

ERKl/2 activation was clearly necessary for endogenous ribosomal transcription

activation, but was it also sufficient? Treatment of NIFI3T3ARaf:ER cells with 4-

hydroxytamoxifen (4-HT) directly induces Raf activation, resulting in the activation of

ERKl/2 (Pritchard et al., 1995). Already 10 min. after 4-HT addition, 45S transcription had

increased by about 2 times and remained at approximately this level for the next 6 hr,

Figure 2.1D, ciosely following the activated levels of ERKl/2, Figure 2.1E. Inhibition of

MEK with 50uM PD98059 or 1 \xM U0126 not only prevented this activation but, as in the

SKF-5 cells, very significantly repressed the basai transcription rate (40%), Figure 2.1F,

giving an overall PAR of 2.3. As seen below, dominant négative ERK2 also eliminaled the

response of Poil transcription to 4-HT, Figure 2.5B. Pulse-chase experiments showed that

the rate of 45S pre-rRNA processing remained unchanged (± 3%) during 4-HT or PD98059

treatments, while cellular uptake of 32P-orthophosphate remained contant (data not shown).

The ARaf:ER cells did not detectably change their cell cycle distribution during the 1 h

incubation used in Figure 2.1F. CDK2 activity remained constant during 4-HT treatment,

while a 40% réduction in CDK4 activity was observed and treatment with 50uM PD98059

did not inhibit CDK2 and inhibited CDK4 by less than 25%. Thus the stimulatory effects of

ERK activation were, if anything, underestimated in Figure 2.1F. It was concluded that

ERK activation was both necessary and sufficient for the activation of ribosomal

transcription in mouse NIH3T3 cells.

2.5.2. Sustained ERK activity is required to maintain the activated level of ribosomal transcription

ERKl/2 activity could regulate ribosomal transcription at gène activation or could

be required to maintain stimulated transcription. In the former case, stimulation should be

maintained for some time after ERK inactivation, while in the latter case ERK inactivation

65

should lead to a rapid decrease in transcription. Cells were treated as in Figure 2.1 F, but the

MEK inhibitor PD98059 was added at various times after 4-HT addition and before a 20

min. puise labelling was started, Figure 2.1 G. Addition of the inhibitor as little as 3 minutes

before puise labelling reduced transcription to basai level. Idcntical results were obtained in

EGF stimulated human SKF-5 cells (data not shown). Thus, in both mouse and human

cells, sustained ERK1/2 activity was required to maintain the activated state of the gènes.

2.5.3. UBF is a target of ERK phosphorylation both in vitro and in vivo UBF is limiting for transcription in vivo (Hannan et ai., 1999b; Hannan et al., 1996)

and thus represents a potential target for régulation by ERK 1/2. Inspection of the UBF

séquence revealed only three possible sites for phosphorylation by ERK1/2 (PXS/TP)

(Aitken, 1999). One of thèse sites, (a.a. 482 to 487) is recognised by CDK2/4 (Voit et al.,

1999). The other two novel sites, PLTP (a.a. 115 to 118) and PKTP (a.a. 199 to 202), are

conserved in UBFs from Xenopus to man (Moss et al., 1998), Figure 2.2A. ERKl

efficiently phosphorylated HMG boxes 1 and 2, Figure 2.2B. The in vitro [32P]

phosphorylated HMG box 1 was gas phase sequenced and the major peak of [ P] released

with threonine 117, the predicted ERK site, Figure 2.2C. Both ERKl and ERK2 were

found to robustly phosphorylate HMG boxes 1 and 2, while mutation of the putative ERK

sites from threonine to glutamic acid (Tl 17E and T201E) eliminated this phosphorylation,

Figure 2.2D. Neither Stress Activated Protein Kinase 1/Jun N-terminal Kinase,

(SAPK1/JNK), SAPK2/p38/HogSAPKl nor activated CDK2/cyclinE detectably

phosphorylated the HMG boxes of UBF, Figure 2.2D & E.

The MAP-kinases interact with their substrates via docking sites, D-domains and F-

X-F-P motifs (Sharrocks et al., 2000). A search of UBF revealed six potential D-domain

homologies including K-H-P-E-L-N-I, in the C-terminal of HMG box 2. Three imperfect

matches to the F-X-F-P motif were found, F-R-F and P-D-F-P both flanking the EiRK site

of HMG box 1. Pull-down assays showed that both recombinant HMG box 1 and box 2

were selectively retained on immobilised recombinant ERKl, Figure 2.2F. This interaction

was not inhibited by prebinding the HMG boxes with an excess of a natural UBF DNA

binding séquence, (data not shown).

66

A) NHI

HMG-boxas

4 4 Potentlal MAP- b o x l : NH-PKKPLIPJFRFF kinase Sites :- b o x 2 : N H - P E K P K I R Q Q L N Y

C ) i . 5 l

irs—i—ÏÏ-JTS HMG box substrate, «g

D) L ^

117

<P V * ' SAPK1/JNK

Activ. ■»

T201E 80X2

E) Substrat*: H1 BOXl B0X2

H 1 ^

Box»» ::'■:<[

cdk2 cycE

+ + + + + + + - + - +

F) y<*r^ WSIlia \ Matrix K

Boxl». B0X2». fe *" Figure 2.2 The MAP kinase ERK phosphorylates HMG boxes 1 and 2 of UBF. A) Schematic présentation of UBF1 and the positions of the putative MAP kinase sites in HMG boxes 1 and 2. B) Box 1 and box 2 were phosphorylated in vitro with 32P-ATP and recombinant GST-ERKl. C) Box 1 was ERK phosphorylated and sequenced, Expérimental Procédures. The determined amino acid séquence is shown below a histogram of the corresponding released counts. D) ERK1, ERK2, SAPKlp7JNK2 and SAPK2/p38, expressed in NIH3T3 cells and activation by EGF (ERKs) or UV (SAPKs), were used to in

67

vitro phosphorylate wt and Tl 17E or T201E mutants of HMG boxes 1 and 2. The upper panel shows the control kinase assays and lower panels the phosphorylation of the HMG box substrates. E) CDK2 and CDK2/cyclin E (cyc E) phosphorylation of 111 and HMG box substrates. F) Pull-down assays of HMG boxes 1 and 2 using ERK1 and control (GST) matrices, "load" indicates the applied HMG box peptide.

We next asked if UBF was also phosphorylated by ERK1/2 in vivo. SKF-5 cells,

transfected with FLAG tagged wild type or mutant (T117/201E) UBFT, were stimulated

with EGF, labelled with |32P] ortho-phosphate and FLAG-UBF isolated. Tryptic peptides of

the gel purified, endogenously labelled wt FLAG-UBF were fractionated by HPLC on C-ix

and peaks containing 32P analysed by MALDI-ToF (Castro et al., 1992). The HMG-box

ERK site tryptic peptides KPLTPYFR (a.a. 114-121) and TPQQLWYTHEK (a.a. 201-

211), see Figure 2.2A, were both identified. To follow ERK phosphorylation of UBF after

various cell treatments, homogeneous, endogenously 32P labelled wt or T117E mutant

FLAG-UBFs were subjected to partial trypsin digestion to release a.a. 2 to 176 ± 6 of wt

UBF while reducing the rest of UBF to peptides of 10 kD and smaller, Figure 2.3A and B.

This peptide, containing a unique ERK site, was separated from the other digestion

products on SDS-PAGE and subjected to Western blotting and phosphoimaging. EGF

stimulated phosphorylation of this peptide was found to be reduced some 4 times by the

MEK inhibitor PD98059 and by greater than 2 fold when the threonine of the ERK site was

mutated to glutamic acid (T117E), Figure 2.3C. Since the exogenous T117E mutant UBF

hetero-dimerised with endogenous wt UBF (Hannan et al., 1999b), recovery of the N-

terminal polypeptide from the endogenous UBF could not be avoided, explaining the

residual phosphorylated peptide with Tl 17E-FLAG-UBF, Figure 2.3C. Interestingiy, the

ratio of UBF phosphorylation with and without MEK inhibitor, Figure 2.3C, equalled the

PAR for thèse cells, see Figure 2.1C. That is, the level of ribosomal transcription was

proportional to the level of ERK phosphorylated UBF.

68

A A A A Partial Trypsin Cleavaga

<lOkUpeplldfls

B) Antl-FLAG

-«FLAG-UBF1

-» FLAG+2-176

0 1 4 10 Trypsin digestion, min.

C) Ant i -FLAG

32p

% B

1

_L IL1 'm. m -*20kD

-*»20kD

UBF: 117E 117E WT WT

EGF * ■*" + ♦ PD9H05» " *

Figure 2.3 In vivo phosphorylation of UBF by ERK. A) Schematic of experiment. B) Analytical time course of trypsin digestion of FLAG-UBF1. The N-terminal partial trypsin product of 22 kD (± 0.8 kD) corresponded to a.a. 2 to 176 ± 6 of UBFl. C) Phosphorylation of peptide 2 to 176 from wt UBFl or UBF1-T117E after treatment for 2h with EGF or EGF plus PD98059. Upper panel shows the FLAG-peptides and the lower panel and histogram, the 32P labelling observed by phosphoimaging.

69

2.5.4. Mutation of the ERK phosphorylation sites on UBF inhibits transcription activation

Poli transcription frorn the pSMECAT reporter, as measured by chloramphenicol

acetyl transferase (CAT) levcls, faithfully reilects régulation of the endogenous ribosomal

gènes (Hannan et al., 1996). The ERK sites of UBF were mutated to glutamic (T117E and

T201E) or aspartic acid (T117D and T201D), simulating phosphorylation, or to alanine

(T117A and T201A), preventing phosphorylation. UBF1 was found to activate Poli

transcription of pSMECAT some 5 fold relative to ÀN-xUBF, a control for possible

squelching (Pelletier et al., 2000), while T to A mutation of HMG boxes 1 & 2

significantly reduced or abrogated this activation (to 0.9 - 2.4 times), Figure 2.4B.

Surprisingly, the T to E and T to D mutations also severely reduce activation (to 1.4 - 2.5

times).

Overexpression of wt UBF gave only weak activation of the human Poli reporter in

SKF-5 cells as estimated by primer extension, Figure 2.4C. However, when both ERK

phosphorylation sites of UBF were mutated (Tl 17/201 E), activation was not only abolished

but transcription suppressed well below the basai level. Thus, mutation of the ERK

phosphorylation sites of UBF1 abrogated or significantly reduced activation of Poli

transcription both in mouse and in human cells.

2.5.5. UBF phosphorylation site mutants act as dominant négatives, preventing the activation of Poli transcription by the ERK signalling cascade

Given that Poli transcription is exclusively nucleolar, it was possible that ERK

phosphorylation was necessary for UBF targeting. However, the mutant UBFs co-localised

with the nucleolar marker ANA (Sigma), see examples in Figure 2.5A. If then UBF was a

direct major target for ERK 1/2 dépendent régulation of the Poli transcription machinery, its

ERK site mutants should act as "dominant négatives", rendering Poli transcription non-

responsive to ERK1/2. Mouse NIH3T3ARaf:ER cells were co-transfected with the wt or

TU7/201E and TU7/201A UBFs and Poli reporter gène (pSMECAT) transcription

monitored, see Figure 2.5B. In the absence of transfected UBF, reporter gène transcription

was activated -1.4 times by the addition of 4-HT and was inhibited to below basai level

70

(-0.7 times) by PD98059, giving a PAR of 2, similar to endogenous ribosomal transcription

in thèse cells (2.3), Figure 2.1F. Co-transfection of an expression vector for dominant

négative ERK2 (ERK2-K52R, (Frost et al., 1994)), Figure 2.5B, and dominant négative

MEKl (MEK1-S218/222V, data not shown) abrogated the transcriptional rcsponse of

pSMECAT to 4-HT as effectively as did the MEK inhibitor PD98059.

Co-transfection of wt UBF activated Poli transcription of pSMECAT in the

ÀRaf:ER cells by approximately 1.4 times, Figure 2.5B. This was a weaker response than

for the parent NIH3T3 line, see Figure 2.4B, possibly due to the elevated basai level of

ERK activity, e.g. see Figure 2.1E, or an elevated endogenous UBF level (Glibetic et al.,

1995). However, addition of 4-HT caused a further activation of transcription of 1.6 times

and addition of PD98059 reduced this to below the unstimulated level. The PAR after co-

transfection of wt UBF was then essentially the same as that in unco-transfected cells (1.8

compared with 2 times) and again very similar to that for the endogenous gènes. When the

double mutants UBF-Tll7/201E or UBF-Tll7/201A were co-transfccted, transcription of

the reporter gène was found to be below that observed in the unco-transfected cells. In fact,

the transcription level more closely resembled transcription in the unco-transfected cells

treated with PD98059. Most importantly, transcription no longer responded to 4-HT nor

was it repressed by PD98059. This was especially clear when the PAR values were

compared. After co-transfection of UBF-Tll7/201E or UBF-Tll7/201 A, the PAR was

<1.1. Again hère, treatment with 4-HT or PD98059 did not induce significant changes in

cell cycle distribution nor CDK2/4 levels, see Expérimental Procédures. Thus, the

activation of Poli transcription via the MAP-kinase cascade required the direct

phosphorylation of HMG boxes 1 and 2 of UBF.

71

A ' Pol I Promolor pSMECAT

UBFl: k à

117 201 point mutations T to A/E/D

B)

* 3

22

C)

^ . * * ■ * * ? ■

I

122t>p » . ., «**"■*— § 1.S

C 1 £ i o.s

««porter: ♦ ♦ ♦ UBF: . . ♦ . T117ffl01E:- - - ♦

Figure 2.4 Transactivation of the mouse and human Poli promoter by UBFl and ERK site mutants.

A) Schematic of the pSMECAT reporter gène construct and of the UBFl mutants. B) Activation of the mouse Poli promoter in NIH3T3 cells. Cells were transfected with 0.5 jxg of pSMECAT and 0.5, 1 and 2 \xg of appropriate UBF vector made up to a total of 3jxg DNA with the empty pCDNA3 plasmid. The data shown is for 2u.g of UBF vector only, this being représentative. Poli transcription from the pSMECAT reporter was quantified by CAT assays as previously described (Hannan et al., 1996), see Expérimental Procédures. C) Activation of the human Poli promoter in SKF-5 cells. Cells were transfected with 4(xg of the human Poli promoter construct (pHV3CAT) and 6 (xg of either wt UBFl or Tl 17/201E -UBFl expression vector. The upper panel shows the results of primer extension

72

generating the expected 122 bp fragment and the histogram, the signais quantified by phosphoimaging.

4-HT: - ♦ ♦ ♦ ♦ • ♦ ♦ - ♦ ♦ - ♦ ♦ PD90OS9: - - ♦ . . - • ♦ - - ♦ - • ♦ ERK2 K62R: . . . ♦ 4+

Figure 2.5 ERK-site mutant UBFs localise correctly to the nucleolus but block MAP-kinase activation of Poli.

A) Nucleoli were revealed using the AN A antibody (Sigma) while the wt and mutant UBFs were revealed using anti-FLAG antibody (a-FLAG). B) ARaLER cells were transfected with pSMECAT and with the empty pCDNA3 vector (No UBF), dominant négative ERK2 ("+" 1 and "++" 1.5[xg), (No UBF+ERK2K52R), wt UBF1 (WT UBF) or the UBF1-Tl 17/201E mutant. Each transfection was either mock treated, treated with 4-HT or treated

73

with both 4-HT and PD98059. The histogram présents the data from four independent experiments, each in duplicate. The expression levels of the wt and mutant UBFs were shown to be équivalent by Western Blot.

2.5.6. ERK phosphorylation of the HMG boxes of UBF diminishes their interaction with DNA

The ERK phosphorylation sites of UBF lie within the N-terminal peptides of HMG

boxes 1 and 2. By analogy with other HMG boxes, thèse peptides are probably implicated

in DNA binding and lie in the minor DNA groove (Read et al., 1995), e.g. see Figure 2.6A.

Since the sites in the HMG boxes of UBF were accessible to ERK, their phosphorylation

might be expected to regulate DNA binding. To test whether this was the case, recombinant

HMG box 1 of UBF was partially phosphorylated (1 to 2%) with ERK1 and allowed to

bind the immobilised human UCE duplex. The unphosphorylated box bound the UCE and

was efficiently eluted only in 0.6M NaCl, see lower panel in Figure 2.6B, while the

phosphorylated box did not significantly bind to the UCE, >90% being found in the flow

through, see upper panel in Figure 2.6B. A minor phosphorylated fraction (<10%),

consistently retained by the UCE, may hâve represented rétention via HMG box homo-

dimerization (Bianchi et al., 1992) or non-specific interaction with the column matrix. In

vitro, phosphorylation of the HMG box was inhibited by prebinding to the UCE duplex but

not to polydA-T, to which it bound poorly, Figure 2.6C. Thus, phosphorylation of the ERK

site of HMG box 1 abrogated interaction with DNA, while prior binding of the box to DNA

inhibited ERK phosphorylation. HMG box 1 bound cruciform DNA with a KD of 1 u,M and

as expected the T to A mutant had the same affinity, Figure 2.6D. However, the T to E

mutation reduced the affinity for cruciform DNA by 2 to 4 times, suggesting that it partially

simulated the phosphorylation event.

2.6. Discussion Our data demonstrate that ribosomal transcription responds immediately to EGF

signalling via ERK activation. The rapidity of this response is surprising and suggests a

direct parallel with the stringent response in microorganisms (Nomura, 1999; Warner,

1999). ERK phosphorylation of UBF at two novel sites was found to be a necessary step in

the stimulation of ribosomal transcription via the ERK pathway. Interestingly, both ERK

74

and UBF activation of transcription was inhibited whether the ERK sites of UBF were

mutated to alanine, to prevent their phosphorylation, or to glutamic or aspartic acid, to

simulate their phosphorylation. Further, ERK phosphorylation of UBF prevented the

interaction of its two essential N-terminal HMG boxes with DNA. Thus, stimulation of

ribosomal transcription by ERK would seem to require a cyclic phosphorylation event.

Intriguingly, ERK phosphorylates the same two I IMG boxes of UBF that recruit CBP

(Pelletier et al., 2000), while the activity of CBP is enhanced by its ERK phosphorylation

(Ait-Si-Ali et al., 1999; Espinos et al., 1999; Liu et al., 1999), suggesting a coordinate

action ofERK and CBP.

Increased ERK activity might lead to an increase in the number of active ribosomal

gènes. However, preliminary experiments using differential psoralen accessibility (Conconi

et al., 1989) tend to exclude this as a possibility (T. Moss, M. Muller and J. M. Sogo,

unpublished data). A more plausible hypothesis is that ERK phosphorylation of UBF plays

a rôle in the initiation of Poli transcription. Two adjacent dimers of UBF are believed to

bind at the Poli promoter, suggesting that they juxtapose the UCE and Core éléments on the

surface of two consécutive enhancesomes (Bazett-Jones et al., 1994; Moss et al., 1998;

Stefanovsky et al., 1996). While this model accounts for what is known of pre-initiation

complex formation, it appears incompatible with promoter clearance by the polymerase.

For example, it is difficult to see how the polymerase could transcribe the first 60 or so

bases downstream of the initiation site that are bound by UBF (Leblanc and Moss, 1993).

ERK phosphorylation of UBF might, however, transiently release DNA from the grip of

individual UBF HMG boxes, thereby enhancing promoter clearance. This model predicts

that UBF phosphorylation would be dynamic and less than stochiometric. Indeed, we found

that on average UBF was phosphorylated at less than one of its two ERK sites, see

Expérimental Procédures for détails. Further, it was recently shown in vitro that Poli

transcription is limited at the level of promoter clearance (Panov et al., 2001). Clearly,

elucidation of the précise mechanism by which the ERK phosphorylation of UBF leads to

increased ribosomal gène transcription will require significant further work. However, the

présent study places ribosomal transcription under the direct control of extracellular growth

signalling.

75

A)

DNA Minor groove

B) Wash Elute

O t 1 2 3 4 5 6

32p

Box

M

Competitor: S S jtf c < s.

32p

Box

D)

Stltftfr Crfm*

boxl tJOXl T117A tH>XlT117Ë

• • • • • • • • * • * • •

protàîn: 0 2 5 » 12 2 5 8 12 2 5 8 12

Figure 2.6 ERK phosphorylation of HMG box 1 diminishes its interaction with DNA.

A) The binary structure of HMG-D (Murphy et al., 1999) indicating that ERK phosphorylation of UBF (PO4) would lie in the minor DNA groove. The peptide chain is indicated in light gray while the DNA backbone is shown in dark grey. B) Phosphorylation of HMG box 1 by ERK1 prevents its binding to the human UCE promoter duplex. "Cntrl"; total protein load, "FT/4"; one quarter of the flow-through, tracks 1 to 6 wash and elution.

76

C) HMG box 1 was phosphorylated with ERK1 in the absence of DNA, (none), the présence of 2 u.g of hUCE duplex or of 2 u.g of poly dA-T. In both b) and C), autoradiogram "32P" upper panel, Western blot "Box" lower panel. D) DNA mobility shifts of wt and mutant (T117E/A) HMG box 1 on 100 fmoles of 3iP-labelled cruciform (Crfm).

2.7. Acknowledgements We thank D.J. McKay of the Protein Sequencing Facility, University of Calgary for

séquence analyses, S. Pelech, J. Grose, J. Landry, N. Colburn, M. Young, J. Charron, S.

Meloche, N. Rivard and L. Poitras for supplying kinase reagents and protocols, D. H. van

Weering and M. McMahon for cell Unes, L. Cornai for the human Poli reporter, N. Roberge

for flow cytometry and L. Heliot for immunocytochemistry. We also thank B. Leblanc for

preliminary studies and J. Coté for critical reading of the manuscript. This work was

supported in Canada by an operating grant from the Médical Research Council of Canada

(MRC-C, now the CIHR), a FCAR-FRSQ Santé (Québec) scholarship to G.P. and a MRC-

Canada Scientist award to T.M. and in France by the Centre National de la Recherche

Scientifique (CNRS) with grants from the Fondation pour la recherche médicale (FRM)

and the Association pour la Recherche sur le Cancer (ARC). L.I.R. acknowledges support

from the NIH, grant no. GM46991. The Centre de Recherche en Cancérologie de

l'Université Laval is supported by the Fonds de Recherche sur la Santé de Québec (FRSQ).

2.8. Expérimental Procédures

2.8.1. Plasmids and constructs Rat UBF1 and ANxUBFl mammalian expression vectors and the pSMECAT

reporter construct hâve been described elsewhere (Hannan et al., 1999b; Hannan et al.,

1996). Point mutations were introduced by PCR using mutated primers and were each

sequenced. UBF HMG box 1 (amino-acids 110-184 or 110-189) and box 2 (amino-acids

194-266) as well as the box 1-T117E and Tl 17A and the box 2-T201E and T201A mutants

were expressed in E. coli after sub-cloning from Xenopus UBF2 into the BamHI /EcoRI

site of pGEX2T (Amersham Pharmacia Biotech). Mammalian expression vectors for the

MAP kinases pEXV-ERK2tag, pCDNA-HA-p38 and pMT2-HA-SAPKlp were a gift from

J. Landry, CDK2 (E. Harlow) and cyclin E (R. Weinberg) were gifts from S. Meloche and

77

pERK2-K52R (Frost et al., 1994) from N. Colburn and M. Young, pMEKIVV (R.

Erikson), pGEX-ERKl from S. Pelech. The pHV3CAT was a gift from L. Cornai.

2.8.2. Tissue culture and transfection NIH 3T3 cells (ATCC) and neuroepithelioma SKF-5 cells were grown in

Dulbecco's Modified Essential Médium (DMEM, BRL) containing 10% fetal bovine sérum

(Wisent). NIH-3T3-ARaf:ER cells were grown in DMEM without Phénol Red. NIH3T3

cells (1 x 106) were transfected with a constant amount of 3p.g total plasmid DNA by

Lipofectamine (BRL) as described elsewhere (Hannan et al., 1999b; Pelletier et al., 2000).

SKF-5 cells (1 x 106) were transfected by the standard CaPCVChloroquine method with a

constant amount of 10 ug total plasmid DNA. Where indicated, SKF cells were treated with

50 ng/ml human EGF (BRL), and NIH-3T3-ARaf:ER cells with 1 jxM 4-hydroxytamoxifen

(4-HT, Sigma). The concentrations of the MEK inhibitors were 5 or 50uM (PD 98059) and

luM (U0126, Calbiochem). The chloramphenicol acetyl-tranferase (CAT) assays of

pSMECAT reporter plasmid expression were performed as described (Ausubel et al., 1991;

Gorman et al., 1982). Primer extensions assays of pHV3CAT reporter transcription in SKF-

5 cells assays were performed as described elsewhere (Kingston, 1991; Zhai and Cornai,

2000)

2.8.3. In vivo puise labelling, RNA extraction and analysis SKF-5 cells (lxl06 cells per pétri) were grown for 24 hours, starved for 18hr in

serum-free médium, then washed twice with phosphate-free médium and treated for

différent periods of time with human EGF. NIH3T3ARaf:ER cells were treated as the SKF

cells but 4-HT was added in place of EGF. Where indicated, MEK inhibitor PD98059 or

U0126 was added. Labelling was performed for 20 min. immediately preceding harvesting

by the addition of 0.5 or 1 mCi 32P-orthophosphate (NEN). The uptake of 32P-

orthophosphate was measured as the total 32P per cell, (Granick, 1975). The RNA was

extracted by the acidic phénol method (Chomczynski and Sacchi, 1987). 10|ag of total RNA

was resolved on a 1% agarose gel, EtBr stained and digitally recorded, then dried and

subjected to phosphorimaging using a STORM 860 (Molecular Dynamics). The 28S and

78

18S RNAs served as an internai control for the total amount of RNA loaded. 45S half-lifes

were estimated from the ratios of 45S to 32S and to 18S after a 1, 2 and 3 hr treatment of

SKF-5 or ÀRaf:ER cells respectively with EGF or 4-HT. Cells were puise labelled as above

for 20 min. then chased by replacing the médium with DMEM containing unlabelled

phosphate and EGF or 4-11T.

2.8.4. Nuclear Run-On The relative loading of Poli onto the ribosomal gènes of SKF-5 cells was

determined one hour after their stimulation with EGF in the présence or absence of the

inhibitor PD98059 following the procédure of Affolter et al. (Affolter and Ruiz-Carrillo,

1986). The RNA recovered from the Run-On reaction was then hybridised in duplicate to

the immobilised human 45S coding séquence DNA fragment spanning nucleotides +2086

to +4244 at 42°C in 5xSSC, 50% formamide, 0.1%SDS and washed to a final stringency of

0.3xSSCor0.1xSSC, 65°C.

2.8.5. Flow Cytometry Cells were stained with propidium iodide and the DNA content estimated in an

EPICS Elite ESP (Beckman Coulter) using standard techniques (Dressler, 1990). The data

were analyzed with the EXP02 Cytometer Software (Applied Cytometry Systems)

imposing no constraints on any of the three cell cycle phases. The proportion of tetraploid

SKF cells in G2 increased by only 8% during EGF treatment while the proportion of S-

phase cells decreased in proportion, typical G1:S:G2 ratios before, 2h after and 6h after

addition of EGF being respectively 54:45:1, 60:34:6 and 54:37:9. 18 h treatment of the

NIH3T3ARaf:ER cells with 4-HT only led to a 10% increase of cells in Gl at the expense

of those in G2, while treatment with 4-HT in combination with PD98059 led to a 10%

increase of cells in S phase at the expense of those Gl. Typical G1:S:G2 ratios for

untreated cells and cells treated for 18h with 4-HT or 4-HT plus PD98059 were

respectively 34:27:39, 46:25:29 and 24:39:37% after 18h of treatment. Thèse variations and

the concomitant changes in gène copy number would lead to an underestimate rather than

an overestimate of the effects of ERK.

79

2.8.6. Endogenous kinase assays and in vitro UBF phosphorylation reactions

Endogenous ERK2 activities were determined as follows: cells were extracted in

100 ul MGE (20mM MOPS, pH 7.0, 10% glycerol, ImM Na3V04 lmM EDTA, lOmM

EGTA, 5mM sodium pyrophosphate, 50mM NaF, 80mM beta-glycerophosphate, 1%

Triton X-100, ImM benzamidine, ImM DTT, ImM PMSF) and centrifuged at 12000 g for

10 min. Immunoprecipitations were performed on 15 p.1 of cell extract to which was added

45ul MIKI (20mM Tris-HCl, pH 7.5, 150mM NaCl, O.lmM EDTA, ImM EGTA, ImM

MgCl2 ,lmM Na3V04, 1% Triton X-100, ImM PMSF) and excess anti-ERK2 antibody (J.

Grose). After 1 h shaking at 4°C, 20 (il of 50% Protein A-Sepharose in MIKI was added

and incubated for 30 min at 4°C. The Sepharose beads were washed three times with 0.5

ml MIKI and then used for kinase assays. The 20 ul final reaction volume contained 5 pg

MBP (BRL) or 2 u.g homogeneous recombinant UBF HMG box2, 150 uM y[32P] ATP (3

u€i), 0.5 x MPM30 (30mM MOPS, pH 7.0, 10% glycerol, 120mM p-nitrophenyl

phosphate, 30mM MgCl2, 2mM DTT, 0.2mM PMSF). Kinase reactions were incubated for

30 min at 30°C, SDS gel-fractionated and subjected to phosphoimaging

Assays of endogenous CDK2/4 were performed as follows. The kinases were

immunoprecipitated from cell extracts as described above using anti-CDK2 (M2) and anti-

CDK4 (C22) (Santa Cruz Biotech.). The immunoprecipitated kinase was incubated for 30

min. at 30 °C in a 40 ul final volume with 5 pg histone Hl (Calbiochem) for CDK2 or 2 u.g

pRb(769) (Santa Cruz Biotech.) for CDK4 in 20 mM HEPES pH 7.4, 10 mM MgCl2,

0.5mM DTT, 5 mM para-nitro-phenyl phosophate (ICN), 1 mM y[32P] ATP (2 uCi). The

reactions were gel fractionated and subjected to phosphoimaging.

A typical in vitro UBF phosphorylation reaction contained 2 ug of HMG box 1 or

box 2 in 50ul of MAP-kinase buffer (12mM MOPS, pH 7.2, lOmM MgCl2, ImM DTT), 10

pCi y-[32P]ATP, 50 uM ATP and either 10 ul GST-ERK1 agarose suspension (Kinetek

Biotechnology Corporation) or 20 pi of the appropriate immunoprecipitated kinase as a

50% Protein A-Sepharose (Pharmacia) suspension in MIKI, see below. The reaction was

incubated at 30°C for 30 min before gel fractionation and subséquent quantification by

phosphoimaging.

80

In order to obtain the various kinases in activated form, NIH 3T3 cells were

transfected with the MAP-kinase or with CDK2 and cyclin E or control expression vectors.

After 18 hours the kinases were activated in cell by either i) overnight sérum deprivation

followed by sérum addition for 10 min. for ERK1 and ERK2, ii) UV irradiation (UVP

UVGL-25, 100 J/m2) for SAPK2/p38 and SAPK1/JNK or iii) no further treatment for

CDK2/cyclinE. The cells were then extracted and immunoprecipitated as for ERK2 above

except for SAPK1 where Sepharose immobilised GST-cJun was added instead of Protein

A-Sepharose . The Sepharose beads were washed three times with 0.5 ml MIKI and then

used for kinase assays with spécifie substrates. For SAPK2: 20 ul final reaction volume

containing 1 ug GST-ATF-2 , 75 uM Y[3 2P| ATP (3 u€i), 0.5 x MPM45 (as MPM30 but

45mM MgCl2); For SAPK1: 20 ul final reaction volume containing 150 uM y[32P] ATP

(3u Ci), 0.5 x MPM30; For ERK-1 and ERK-2: 20 ul final reaction volume containing 5 iig

MBP (BRL), 150 uM y[32P] ATP (3 uCi), 0.5 x MPM30; For CDK2/cyclinE: 40 ul final

volume containing 5 u.g histone Hl (Calbiochem) or 2u.g UBF HMG box 1 or box 2 and 20

mM HEPES pH 7.4, 10 mM MgCl2, 0.5mM DTT, 5 mM para-nitro-phenyl phosophate

(ICN), 1 mM y[32P] ATP (2 uCi). AU kinase reactions were incubated for 30 min at 30°C,

gel fractionated, transferred to nitrocellulose and subjected to phosphoimaging.

2.8.7. Phospho-polypeptide sequencing The phosphorylated recombinant HMG box was covalently attached to Sequelon-

AA (Millipore) by carbodiimide coupling. The sequencing was performed on an Applied

Biosystems 470A protein sequencer using 90% aqueous acetonitrile as the

amlinothiazolinone extraction solvent (Shively et al., 1987; Xu and Shively, 1988). Each

total cycle product was used for Cerenkov cpm measurement and then PTH-amino-acid

analysis.

2.8.8. Pull-down assays Recombinant GST-ERK1 was expressed in E. coli and bound to G-Sepharose.

Recombinant UBF HMG box 1 or box 2 was applied to 5u,g of immobilised GST-ERK-1

or GST in HBS (50mM HEPES, pH 7,9, 150mMNaCl) and incubated for lhr at 4°C. After

81

three washes with HBS, the matrix bound proteins were resolved on 18% Tris-Tricine-SDS

PAGE, the gel transfered to nitrocellulose and stained with SYPRO Ruby protein blot stain

(Molecular Probes) according to the manufacurer's protocol. The membrane was then

analysed by blue fluorescent imaging (STORM 860, Molecular Dynamics).

2.8.9. In vivo phosphorylat ion of UBF1 IJBF1 and UBF1-T117/201E were expressed in SKF-5 cells, see above. The cells

were labelled with 3 mCi 32P-orthophosphate (NEN) in phosphate-free médium for three

hours. Crude nuclear extracts were prepared from cell pellets by homogenization in 10mM

Tris-HCl, pH 7.9, 0.3M sucrose, 3mM CaCl2, 2mM Mg- acétate, 0.1% NP-40, 0.1%

TritonX-100, ImM DTT, O.lmM EDTA and the inhibitors lOOuM Na3V04, 50mM NaF

and 1 ug/ml each Leupeptin and Pepstatin (Roche). After a brief centrifugation, the nuclear

pellets were resuspended in 0.42M NaCl, 50mM Tris-HCl, pH 7.9, 12.5mM MgCl2, ImM

EDTA, 20% glycerol plus inhibitors. After the addition of 1/8 volume 8M urea, the samples

were briefly sonicated and centrifuged in a SW 50.1 rotor (Beckman) for 30 min at 30000

rpm. The supernatants were incubated with 50 ul of M2 anti-FLAG affinity matrix (Sigma)

and the FLAG-UBF1 protein was recovered according to the manufacturera protocol. The

labelled UBF1 was then analysed by one of two approaches. 1) After standard SDS PAGE

fractionation it was subjected in gel to total tryptic digestion. The peptides were then

fractionated by reverse phase HPLC on a Ci s microbore column and radioactive fractions

analysed by MALDI-ToF mass spectrometry (Eastern Québec Proteomic Service). 2) A

limited digest of the labelled UBF protein with 10 \xg of trypsin (Sigma) was performed on

ice for 3-5 min. The products were then separated on a 5-15% Tris-Tricine SDS PAGE,

transferred on a nitrocellulose filter, subjected to phosphoimaging and probed with anti-

FLAG antibody.

It was estimated by phosphoimaging that no more than 2% of the total P32 counts

incorporated in vivo into FLAG-UBF were recovered with the N-terminal peptide. UBF

has 21 potential CKII sites in the C-terminal acidic domain and several other phospho-

peptides hâve been detected (O'Mahony et al., 1992a; Voit et al., 1999; Voit et al., 1992). A

gross upper limit for the number of UBF phosphorylation sites, including the ERK sites,

82

thereforc lies at around 24. Assuming ail sites were equally modified, each site should hâve

accounted for 4% of UBF counts, while if fewer than 24 sites were actually modified, or

some sites only partially modified, the percentage of counts per stochiometrically modified

site should increase in inverse proportion. On this basis, our measurements showed that the

ERK site of HMG box 1, accounting as it did for only 2% of the total counts, was at most

only 50% phosphorylated evcn when ERK was maximally activated.

2.8.10. Immunocytochemistry NIH 3T3 cells were transfected with UBF1, UBF1-T117E, UBFl-T201E,or UBF1-

Tl 17/201E in 4 mm2 chamber slide wells and immunostained as described elsewhcre

(Hannan et al., 1999b). Spécimens were viewed on a BioRad MRC-1024 confocal imaging

System and treated with CytoFish and ED1T3D software developed at RFMQ, Grenoble,

e.g. see (Panse et al., 1999).

2.8.11. DNA Binding experiments The human UCE was affinity labelled with Klenow polymerase and biotin-7 dATP

(BRL) and bound to Streptavidin-Agarose (BRL). Typically, 0.5ug of in vitro labelled

protein was then incubated with 50 ul of this Streptavidin-Agarose-hUCE affinity matrix in

100 JLXI binding buffer (80mM NaCl, 50mM Tris, pH 7.9) for 2 hours at 4°C. After three

washes in binding buffer, the protein was recovered with 0.6M NaCl and gel fractionated,

transferred to a nitrocellulose membrane, autoradiographed and probed with anti-HMG box

1 antibody. Cruciform DNA mobility shift assays were performed as previously described

(Pôhler et al., 1998).

2.9. Références Affolter, M., and Ruiz-Carrillo, A. (1986). Transcription Unit of the Chicken Histone H5

Gène and Mapping of H5 Pre-mRNA Séquences. J. Biol. Chem. 261, 11496-11502.

Ait-Si-Ali, S., Carlisi, D., Ramirez, S., Upegui-Gonzalez, L. C., Duquet, A., Robin, P.,

Rudkin, B., Harel-Bellan, A., and Trouche, D. (1999). Phosphorylation by p44 MAP

83

Kinase/ERKl stimulâtes CBP histone acetyl transferase activity in vitro. Biochem.

Biophys. Res. Commun. 262, 157-62.

Aitken, A. (1999). Protein consensus séquence motifs. Mol. Biotechnol. 12, 241-53.

Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A., and

Struhl, K. (1991). Current Protocols in Molecular Biology, 1 Edition (New York: Greene

Publishing Associates & Wiley-Interscience).

Bazett-Jones, D. P., Leblanc, B., Herfort, M., and Moss, T. (1994). Short-range DNA

looping by the Xenopus HMG-box transcription factor, xUBF. Science 264, 1134-1137.

Bianchi, M. E., Falciola, L., Ferrari, S., and Lilley, D. M. J. (1992). The DNA binding site

of HMG1 protein is composed of two similar segments (HMG boxes), both of which hâve

counterparts in other eukaryotic regulatory proteins. EMBO J. 77, 1055-1063.

Bodem, J., Dobreva, G., Hoffmann-Rohrer, U., Iben, S., Zentgraf, H., Delius, H., Vingron,

M., and Grummt, I. (2000). TIF-IA, the factor mediating growth-dependent control of

ribosomal RNA synthesis, is the mammalian homolog of yeast Rrn3p. EMBO Rep. 7, 171-

5.

Castle, T., Katz, A., and Richardson, A. (1978). Comparison of RNA synthesis by liver

nuclei from rats of various âges. Mech. Ageing Dev. 8, 383-95.

Castro, J. A., Koster, C, and Wilkins, C. (1992). Matrix-assisted laser desorption/ionization

of high-mass molécules by Fourier-transform mass spectrometry. Rapid Commun. Mass

Spectrom. 6,239-41.

Cavanaugh, A. H., Hempel, W. M., Taylor, L. J., Rogalsky, V., Todorov, G., and

Rothblum, L. I. (1995). Activity of RNA polymerase I transcription factor UBF blocked by

Rb gène product. Nature 374, 177-180.

Chomczynski, P., and Sacchi, N. (1987). Single-step method of RNA isolation by acid

guanidium thiocyanate-phenol-chloroform extraction. Anal. Bioch. 162, 156-159.

84

Conconi, A., Widmer, R. M., Koller, T., and Sogo, J. M. (1989). Two différent chromatin

structures coexist in ribosomal RNA gènes throughout the cell cycle. Cell 57, 753-761.

Derenzini, M., Trerè, D., Pession, A., Govoni, M., Sirri, V., and Chieco, P. (2000).

Nucleolar size indicates the rapidity of cell prolifération in cancer tissues. J. Pathol. 191,

181-186.

Dressler, L. G. (1990). Controls, standards, and histogram interprétation in DNA flow

cytometry. Methods Cell. Biol. 33, 157-71.

Dudley, D. T., Pang, L., Decker, S. J., Bridges, A. J., and Saltiel, A. R. (1995). A synthetic

inhibitor of the mitogen-activated protein kinase cascade. Proc. Natl. Acad. Sci. U S A 92,

7686-9.

Espinos, E., Le van Thaï, A., Pomiès, C, and Weber, M. J. (1999). Coopération between

phosphorylation and acetylation processes in transcriptional control. Mol. Cell. Biol. 19, 3474-3484.

Favata, M. F., Horiuchi, K. Y., Manos, E. J., Daulerio, A. J., Stradley, D. A., Feeser, W. S.,

Van Dyk, D. E., Pitts, W. J., Earl, R. A., Hobbs, F., Copeland, R. A., Magolda, R. L.,

Scherle, P. A., and Trzaskos, J. M. (1998). Identification of a novel inhibitor of mitogen-

activated protein kinase kinase. J. Biol. Chem. 273, 18623-32.

Friedman, J. M., Chung, E. Y., and Darnell, J. E. (1984). Gène expression during liver

régénération. J. Mol. Biol. 179, 37-53.

Frost, J. A., Geppert, T. D., Cobb, M. H., and Feramisco, J. R. (1994). A requirement for

extracellular signal-regulated kinase (ERK) fonction in the activation of AP-1 by Ha-Ras,

phorbol 12-myristate 13-acetate, and sérum. Proc. Natl. Acad. Sci. USA 91, 3844-3848.

Glibetic, M., Taylor, L., Larson, D., Hannan, R., Sells, B., and Rothblum, L. (1995). The

RNA polymerase I transcription factor UBF is the product of a primary response gène. J.

Biol. Chem. 270, 4209-4212.

Gorman, C. M., Moffat, L. F., and Howard, B. H. (1982). Recombinant génomes which

express chloramphenicol acctyltransferase in mammalian cells. Mol. Ccll. Biol. 2, 1044-51.

Granick, D. (1975). Nucleolar necklaces in chick embryo fïbroblast cells. I. Formation of

necklaces by dichlororibobenzimidazolc and other adenosine analogues that decrease RNA

synthesis and dégrade preribosomes. J. Cell Biol. 65, 398-417.

Grummt, L, Smith, V. A., and Grummt, F. (1976). Amino acid starvation affects the

initiation frequency of nucleolar RNA polymerase. Cell 7, 439-45.

Hannan, R. D., Cavanaugh, A., Hcmpel, W. M., Moss, T., and Rothblum, L. (1999a).

Identification of a mammalian RNA polymerase I holoenzyme containing components of

the DNA repair/replication System. Nucl. Acids Res. 27, 3720-3727.

Hannan, R., Stefanovsky, V., Arino, T., Rothblum, L., and Moss, T. (1999b). Cellular

régulation of ribosomal DNA transcription: both rat and Xenopus UBF1 stimulate rDNA

transcription in 3T3 fibroblasts. Nucl. Acids Res. 27, 1205-1213.

Hannan, R. D., Stefanovsky, V., Taylor, L., Moss, T., and Rothblum, L. I. (1996).

Overexpression of the transcription factor UBF1 is sufficient to increase ribosomal DNA

transcription in neonatal cardiomyocytes: Implications for cardiac hypertrophy. Proc. Natl.

Acad. Sci. USA 93, 8750-8755.

Jantzen, H. M., Chow, A. M., King, D. S., and Tjian, R. (1992). Multiple domains of the

RNA polymerase I activator hUBF interact with the TATA-binding protein complex hSLl

to médiate transcription. Gènes Dev. 6, 1950-1963.

Kihm, A. J., Hershey, J. C, Haystead, T. A. J., Madsen, C. S., and Owens, G. K. (1998).

Phosphorylation of the rRNA transcription factor upstream binding factor promûtes its

association with TATA binding protein. Proc. Natl. Acad. Sci. USA 95, 14816-14820.

Kingston, R. E. (1991). Préparation and Analysis of RNA. In Current Protocols in

Molecular Biology, F. M. Ausubel, R. Brent, R. E. Kingston, D. D. Moore, J. G. Seidman,

86

J. A. Smith and K. Struhl, eds. (New York: Greene Publishing Associates & Wiley-

Interscience), pp. 4.0.1-4.10.11.

Leblanc, B., Read, C., and Moss, T. (1993). Récognition of the Xenopus ribosomal core

promoter by the transcription factor xUBF involves multiple HMG box domains and leads

to an xUBF interdomain interaction. EMBO J. 12, 513-525.

Liu, Y. Z., Thomas, N. S. B., and Latchman, D. S. (1999). CBP associâtes with the p42/p44

MAPK enzymes and is phosphorylated following NGF treatment. Neuroreport. 10, 1239-

1243.

Loeb, J. N., and Yeung, L. L. (1975). Synthesis and dégradation of ribosomal RNA in

regenerating liver. J. Exp. Med. 142, 575-87.

Luyken, J., Hannan, R. D., Cheung, J. Y., and Rothblum, L. I. (1996). Régulation of rDNA

transcription during endothelin-1-induced hypertrophy of neonatal cardiomyocytes -

Hyperphosphorylation of upstream binding factor, an rDNA transcription factor. Cire. Res.

78, 354-361.

McStay, B., Frazier, M. W., and Reeder, R. H. (1991). xUBF contains a novel dimerization

domain essential for RNA polymerase I transcription. Gènes Dev. 5, 1957-1968.

Miller, G., Panov, K. L, Friedrich, J. K., Trinkle-Mulcahy, L., Lamond, A. L, and

Zomerdijk, J. C. (2001). hRRN3 is essential in the SLl-mediated recruitment of RNA

Polymerase I to rRNA gène promoters. EMBO J. 20, 1373-82.

Moorefield, B., Greene, E. A., and Reeder, R. H. (2000). RNA polymerase I transcription

factor Rrn3 is functionally conserved between yeast and human. Proc. Natl. Acad. Sci.

USA 97, 4724-4729.

Moss, T., Stefanovsky, V. Y., and Pelletier, G. (1998). The structural and architectural rôle

of Upstream Binding Factor, UBF. In Transcription of ribosomal gènes by eukaryotic RNA

polymerase I, M. R. Paule, éd. (Austin, TX USA: Landes Bioscience), pp. 75-94.

87

Murphy, F. V., Sweet, R. M., and Churchill, M. E. (1999). The structure of a chromosomal

high mobility group protein-DNA complex reveals sequence-neutral mechanisms important

for non-scqucnce-specific DNA récognition. EMBO J. 18, 6610-8.

Nikolov, E. II., Nankova, B. B., and Dabeva, M. D. (1991). Activatcd ribosomal RNA

synthesis in regenerated rat liver upon inhibition of protcin synthesis. Mol. Biol. Rep. 15,

45-52.

Nomura, M. (1999). Régulation of ribosome biosynthesis in Escherichia coli and

Saccharomyces cerevisiae: diversity and common principles. J. Bacteriol. 181, 6857-64.

O'Mahony, D. J., Smith, S. D., Xie, W., and Rothblum, L. I. (1992a). Analysis of the

phosphorylation, DNA-binding and dimerization properties of the RNA polymerase I

transcription factors UBF1 and UBF2. Nucl. Acids Res. 20, 1301-1308.

O'Mahony, D. J., Xie, W., Smith, S. D., Singer, H. A., and Rothblum, L. 1. (1992b).

Differential phosphorylation and localisation of the transcription factor UBF in vivo in

response to sérum deprivation. J. Biol. Chem. 267, 35-38.

Panov, K. L, Friedrich, J. K., and Zomerdijk, J. C. (2001). A step subséquent to

preinitiation complex assembly at the ribosomal RNA gène promoter is rate limiting for

human RNA polymerase I-dependent transcription. Mol. Cell. Biol. 21, 2641-9.

Panse, S. L., Masson, C, Heliot, L., Chassery, J. M., Junera, H. R., and Hernandez-Verdun,

D. (1999). 3-D organization of ribosomal transcription units after DRB inhibition of RNA

polymerase II transcription. J. Cell Sci. 112, 2145-54.

Paule, M. R. (1998). Transcription of ribosomal gènes by eukaryotic RNA polymerase I, 1

Edition (Austin, TX USA: Landes Bioscience).

Pelletier, G., Stefanovsky, V. Y., Faubladier, M., Hirschler-Laszkiewicz, L, Savard, J.,

Rothblum, L. L, Côté, J., and Moss, T. (2000). Compétitive recruitment of CBP and Rb-

HDAC régulâtes UBF acetylation and ribosomal transcription. Mol. Cell 6, 1059-1066.

XX

Pôhler, J. R. G., Norman, D. G., Bramham, J., Bianchi, M. E., and Lilley, D. M. J. (1998).

HMG box proteins bind to four-way DNA junctions in their open conformation. EMBO J.

77,817-826.

Pritchard, C. A., Samuels, M. L., Bosch, E., and McMahon, M. (1995). Conditionally

oncogenic forms of the A-Raf and B-Raf protein kinases display différent biological and

biochemical properties in NIH 3T3 cells. Mol.Cell. Biol. 15, 6430-6442.

Quantin, B., and Breathnach, R. (1988). Epidermal growth factor stimulâtes transcription of

the c-jun proto-oncogene in rat fibroblasts. Nature 334, 538-9.

Read, C. M., Cary, P. D., Crane-Robinson, C, Driscoll, P. C., Carrillo, M. O. M., and

Norman, D. G. (1995). The structure of the HMG box and its interaction with DNA. In

Nucleic Acids and Molecular Biology, vol 9, F. Eckstein and D. M. J. Lilley, eds. (Berlin

Heidelberg: Springer-Verlag), pp. 222-250.

Saez-Vasquez, J., and Pikaard, C. S. (1997). Extensive purification of a putative RNA

polymerase I holoenzyme from plants that accurately initiâtes rRNA gène transcription in

vitro. Proc. Natl. Acad. Sci. USA 94, 11869-11874.

Seither, P., Iben, S., and Grummt, I. (1998). Mammalian RNA polymerase I exists as a

holoenzyme with associated basai transcription factors. J. Mol. Biol. 275, 43-53.

Sharrocks, A. D., Yang, S. H., and Galanis, A. (2000). Docking domains and substrate-

specificity détermination for MAP kinases. Trends Biochem. Sci. 25, 448-453.

Shively, J. E., Miller, P., and Ronk, M. (1987). Microsequence analysis of peptides and

proteins. VI. A continuous flow reactor for sample concentration and séquence analysis.

Anal. Biochem. 163, 517-29.

Stefanovsky, V. Y., Bazett-Jones, D. P., Pelletier, G., and Moss, T. (1996). The DNA

supercoiling architecture induced by the transcription factor xUBF requires three of its five

HMG-boxes. Nucl. Acids Res. 24, 3208-3215.

89

Stoykova, A. S., Dudov, K. P., Dabeva, M. D., and Hadjiolov, A. A. (1983). Différent rates

of synthesis and turnover of ribosomal RNA in rat brain and liver. J. Neurochem. 41, 942-

9.

Tuan, J. C, Zhai, W. G., and Cornai, L. (1999). Recruitment of TATA-binding protein-

TAF, complex SL1 to the human ribosomal DNA promoter is mediated by the carboxy-

terminal activation domain of upstream binding factor (UBF) and is regulated by UBF

phosphorylation. Mol. Cell. Biol. 19, 2872-2879.

van Weering, D. H., Medema, J. P., van Puijenbroek, A., Burgering, B. M., Baas, P. D., and

Bos, J. L. (1995). Ret receptor tyrosine kinase activâtes extracellular signal-regulated

kinase 2 in SK-N-MC cells. Oncogene 11, 2207-14.

Voit, R., Hoffmann, M., and Grummt, I. (1999). Phosphorylation by Gl-spécifie cdk-cyclin

complexes activâtes the nucleolar transcription factor UBF. EMBO J. 18, 1891-9.

Voit, R., Schnapp, A., Kuhn, A., Rosenbauer, H., Hirschmann, P., Stunnenberg, H. G., and

Grummt, I. (1992). The nucleolar transcription factor mUBF is phosphorylated by casein

kinase II in the C-terminal hyperacidic tail which is essential for transactivation. EMBO J.

77,2211-2218.

Warner, J. R. (1999). The économies of ribosome biosynthesis in yeast. Trends Biochem.

Sci. 24, 437-40.

Xu, Q. Y., and Shively, J. E. (1988). Microsequence analysis of peptides and proteins. VIII.

Improved electroblotting of proteins onto membranes and derivatized glass-fiber sheets.

Anal. Biochem. 770, 19-30.

Zahradka, P., Larson, D. E., and Sells, B. H. (1991). Régulation of ribosome biogenesis in

differentiated rat myotubes. Molecular and Cellular Biochemistry 104, 189-194.

Zhai, W., and Cornai, L. (2000). Repression of RNA polymerase I transcription by the

tumor suppressor p53. Mol. Cell. Biol. 20, 5930-8.

90

Chapitre 3 : Un facteur de croissance régule la vitesse d'élongation de la transcription par TARN Polymérase I, chez les mammifères, via la phosphorylation d'UBF et le remodelage de la chromatine ribosomale;

« Growth Factor Signaling Régulâtes Elongation of RNA Polymérase I Transcription in Mammals via UBF Phosphorylation and r-Chromatin Remodeling »

Victor Stefanovksy1, Frédéric Langlois1, Thérèse Gagnon-Kugler1, Larry I. Rothblum2 and

TomMoss1.

1 Cancer Research Center Center and Département of Mediacl Biology, Laval University,

Hôtel-Dieu de Québec, 9 rue MacMahon, G1R 2J6, Québec, Canada. 2 Henry Hood Research Program, Sigfried and Janet Weis Center for Research, Geisinger

Clinic, 100 North Academy Avenue, Danville, Pennsylvania , USA.

3.1. Avant-propos Cet article a été publié dans la revue « Molecular Cell » le 3 mars 2006 (vol 21, pp.

629-639) et est reproduit avec la permission de l'éditeur. J'ai réalisé, en collaboration avec

Dr Moss, les expériences de Run-On nucléaires. J'ai déterminé les conditions optimales et

mis au point l'expérience lors de travaux précédents dont les résultats ont été publiés dans

Molecular Cell par Stefanovsky et al en 2001, chapitre 2. De plus, j 'ai préparé le matériel

que nous avons utilisé lors de ces expériences : culture cellulaire, clonage, amplification et

linéarisation des plasmides utilisés pour l'hybridation et transcription de l'ARN trié utilisé

comme contrôle lors de l'extraction des ARN et de l'hybridation. Les expériences de ChIP

ont été faites par Frédéric Langlois et Larry I Rothblum a fourni une grande quantité

d'anticorps utilisés pour faire ces expériences. Les autres expériences ont été faites par Dr

Victor Y. Stefanovsky.

()1

Les résultats présentés dans cet article sont très importants dans le domaine de la

transcription par l'ARN polymérase I car ils modifient de façon dramatique la

compréhension de la régulation de la transcription. En effet, c'est la première fois qu'une

régulation au niveau de l'élongation des transcrits est démontrée. De plus, les expériences

de Run-On nucléaires que j'avais réalisées pour Stefanovsky et al (2001) montraient qu'il

n'y avait pas de recrutement massif de Pol I aux gènes ribosomaux suite à une stimulation

par l'EGF. Ces observations étaient en contradiction avec une régulation au niveau de

l'initiation exclusivement et ont encouragé la poursuite des travaux pour déterminer cet

autre mécanisme de régulation. De plus, les Run-On nucléaires que j'ai fait avec le Dr

Moss pour Stefanovsky et al (2006) ont été un élément clé de la preuve qu'il n'y avait pas

d'augmentation de recrutement de Pol I sur les gènes ribosomaux. Puisque l'absence de

recrutement significatif de Pol I est à la base de l'argumentation d'une régulation au niveau

de l'élongation, l'existence de résultats indépendants provenant de deux expériences

distinctes, Run-On nucléaires et ChIP, montrant la même chose a été importante dans la

démonstration de ce nouveau niveau de régulation de l'ARN polymérase I.

3.2. Résumé La synthèse de l'ARN ribosomal 45S par l'ARN polymérase I limite la croissance

cellulaire. L'identification du mécanisme responsable de sa régulation est conséquemment

un élément clé dans la compréhension du contrôle de la croissance. La régulation de la

transcription ribosomale est considérée comme intervenant seulement aux niveaux de

l'initiation de la transcription et du relarguage du promoteur. Cependant, nous avons trouvé

que la synthèse endogène du 45 S suite à une stimulation par le facteur de croissance

épidermal (EGF) ou par l'ajout de sérum n'entraînait pas une augmentation du nombre

d'ARN polymérases I engagées sur les gènes ribosomaux et ce même si la synthèse du 45S

augmentait de façon significative. De plus, la vitesse d'élongation des polymérases en

transcription fut mesurée et montrée comme étant directement proportionnelle à la synthèse

du 45S. Conséquemment, l'élongation est une étape limitante dans la synthèse des ARN

ribosomaux in vivo. La phosphorylation des boîtes HMG l et 2 d'UBF, facteur nécessaire

pour l'augmentation de la transcription ribosomale, par ERK a été montrée comme régulant

92

directement la vitesse d'élongation de TARN polymérase I en induisant un remodelage de

la chromatine ribosomale. Ces données suggèrent l'existence d'un mécanisme de

coordination dans l'assemblage co-transcriptionel des particules pré-ribosomales.

3.3. Summary Synthesis of the 45S rRNA by RNA polymérase I limits cell growth. Knowledge of

the mechanism of its régulation is thereibre key to understanding growth control. rRNA

transcription is believed to be regulated solely at initiation/promoter release. However, we

found that stimulation of endogenous 45S rRNA synthesis by EGF and sérum failed to

induce an increase in RNA polymérase I engagement on the rRNA gènes, despite robust

enhancement of 45S rRNA synthesis. Further, endogenous transcription elongation rates

were measured and found to be directly proportional to 45S rRNA synthesis. Thus,

elongation is a rate-limiting step for rRNA synthesis in vivo. ERK phosphorylation of the

HMG boxes of UBF, an RNA polymérase I factor essential for transcription enhancement,

was shown to directly regulate elongation by inducing the remodeling of ribosomal gène

chromatin. The data suggest a mechanism for coordinating the co-transcriptional assembly

of pre-ribosomal particles.

3.4. Introduction Hyperproliferative and hypertrophie diseases such as cancer and cardiac

hypertrophy are due to dysfunctions in the control of cell growth. However, we still

understand very little of how cell growth rate and cell size are determined. It has become

clear over the last few years that the capacity to produce ribosomes détermines the cell's

ability to grow and proliferate. Régulation of the ribosomal protein (r-protein) and

ribosomal RNA (rRNA) gènes has been shown to control cell size and cell cycle

progression (Jorgensen et al., 2004; Rudra et al., 2005). But given the multiple signaling

pathways shown to impinge upon the expression of thèse gènes, the mechanisms of in vivo

régulation are still far from clear. The rRNA gènes are transcribed by RNA polymérase I

(RPI) and a dedicated set of factors (Moss and Stefanovsky, 2002), and hâve been shown to

be growth-regulated via the MAPK-kinase (ERK) (Stefanovsky et al., 2001b) and the

93

insulin-mTOR pathways (Claypool et al., 2004; Hannan et al., 2003; James and Zomerdijk,

2004; Mayer et al., 2004). The ERK, mTOR and JNK pathways regulate RPI transcription

via the activities of UBF, SL1 and Rrn3(TIF-IA) (Hannan et al., 2003; Mayer et al., 2005;

Mayer et al., 2004; Stefanovsky et al., 2001b; Zhang et al., 2005; Zhao et al., 2003). In

vitro studies hâve implicated Rrn3 and UBF in initiation complex assembly and in

promoter release and it is now generally accepted that régulation of rRNA gène

transcription occurs exclusively during the initiation process, see (Moss, 2004; Russell and

Zomerdijk, 2005). Flowever, this has never been demonstrated in vivo.

UBF is a multiple HMG box architectural protein that is able to induce a near 360

deg. looping in about 140 bp of DNA via multiple in-phase bends (Bazett-Jones et al.,

1994; Stefanovsky et al., 1996; Stefanovsky et al., 2001a). This nucleoprotein structure,

called the rRNA gène enhancesome, resembles the nucleosome in both overall mass and

DNA-protein composition, but contains only one turn of DNA. UBF binds throughout the

rRNA gène repeats, including the transcribed régions (O'Sullivan et al., 2002), strongly

suggesting that it defines a specialized rRNA gène chromatin (Moss, 2004). Réversible

ERK phosphorylation of the two N-terminal HMG boxes of UBF is essential for growth

factor stimulation of rRNA gène transcription (Stefanovsky et al., 2001b). Recently, we

showed that this phosphorylation modulâtes DNA looping within the enhancesome

(Stefanovsky et al., 2006). Thus, growth factor activation of ERK most probably leads to

the remodeling of the rRNA gène chromatin (r-chromatin).

Hère, we hâve reinvestigated the mechanisms by which the endogenous ribosomal

gènes respond to growth factor stimuli and the rôle UBF plays. The results are both

unexpected and provocative. We reveal that a major regulatory control occurs at the level

of transcription elongation and that unmodified UBF unexpectedly présents a formidable

barrier to transcription elongation by RPI, a barrier that is overcome by its ERK

phosphorylation. The data may further provide insight into how the hundreds of

components required for ribosome biogenesis are efficiently assembled.

94

3.5. Resuifs

3.5.1. Growth Factors Enhance Transcription without Increasing RNAPI Loading on the Mammalian rRNA Gènes

If ribosomal gène transcription is regulated at initiation, the number of RPI

molécules engaged in transcribing the rRNA gènes (RPI loading) should be directly

proportional to 45S rRNA production. We compared 45S transcription induced by EGF and

by sérum with MEK1/2 inhibition by U0126 or PD98059. Both sérum and EGF enhanced

the 45S rRNA production rate by 5 fold or more as compared to the MEK inhibitors

(Figure 3.1 A and B). As we round previously (Pelletier et al., 2000; Stefanovsky et al.,

2001b), primary 45S rRNA processing intermediates were not apparent during puise

labeling and processing rates were not affected by cell treatment (Figure 3.1D). Parallel

cultures were analyzed for RPI loading levels by Nuclear Run-On (Figure 3.1C) see

Expérimental procédures. Quite unexpectedly, we found that neither EGF/serum

stimulation nor MEK/ERK inhibition caused significant changes in RPI loading (±0.2).

Thus, the number of RPI molécules engaged in transcribing the rRNA gènes remains

constant despite 5 fold changes in 45S rRNA synthesis rates. Similar results were obtained

in an unrelated human cell Une, see Supplementary Data (Figure 3.8).

3.5.2. ChIP Measurements Confirm that RNAPI Loading Remains Constant over a Broad Range of 45S rRNA Synthesis Rates

Given that Nuclear Run-On suggested a fondamental rethinking of rRNA gène

régulation, we first substantiated our data by an independent technique. NIH3T3 cells were

again stimulated with EGF or sérum or inhibited with PD98059 and either puise labeled or

RPI loading analyzed by ChIP and quantitative PCR (Q-PCR) (O'Sullivan et al., 2002)

using an antibody against the RPI-(3' (194kD) subunit (Hannan et al., 1998), (Figure 3.1A

and Expérimental Procédures). Despite the 3 to 5 fold enhancement in 45S synthesis, no

significant increase in RPI loading (± 0.2) was discerned (Figure 3.2A and B). Thus, the

RPI loading data obtained by ChIP was fully consistent with the Nuclear Run-On data

(Figures 3.1 and 3.8).

95

A) EGF, Sérum or MEKInhlbltor

Nuclear Run-On/ChIP

15

[3H] -urldlne

B)

45S M m

«î, 5

Treatment :- 0 ^

3Qfr time In . minutes:

[3H] -RNA analysls or

Urldlne Chase

+3101 b.p.

Ctl

CI)

EGF Probe 3 Probe 2 Probe 1 Vector

/ Mi mu

7

6

- 5

4

3

2

I

B => g <" S G ^ g <» S 3 => g <" o Probe 1 Probe 2 Probe 1+2

D)

4SS

PD Sérum

fiWn EGF

111 I 1 il!'™™"

32S «■ifièi •■ « M M H '' H « . M * 288 ; J ig n ,,, N j |

30 45 60 75 90 30 45 60 75 90 30 45 60 75 90 Time from start of labellng, min.

o 4

tt 3

§2

O PD D S O EGF

O PD D S O EGF

:^+4 ~ik 40 50 60 70 80 90 1 0°

Time from start of labellng, min.

Figure 3.1 Growth factor enhancement of rRNA gène expression does not increase in RPI loading as determined by Nuclear Run-On.

A) Expérimental plan. At zéro time, mouse NIH3T3 cells were stimulated by the addition of EGF or sérum (S), the ERK pathway blocked with the MEK1/2 inhibitors PD98059 (PD) and U0126 (U) or cells untreated (Ctl). The 45 S rRNA was determined by puise labeling and RPI loading was determined at mid-pulse either by Nuclear Run-On or chromatin immunoprecipitation (ChIP). Maximal/minimal transcription rates occur at 10 to 30 minutes post-stimulation/-repression (Stefanovsky et al., 2001b). B) 45S rRNA transcription rates are indicated relative to the level in PD98059 treated cells. Panels above show incorporation of [3H]-uridine into the 45S rRNA precursor (fluorogram) and 18S levels (EtBr staining) in one analysis. C) Upper panels; nuclear Run-On hybridizations to 5'-ETS probes +1 to +293bp (Probe 2) and +1 to +3101bp (Probe 1) compared with -167 to - lbp (Probe 3) and vector. Histogram shows the [32P] nuclear Run-On signais relative to

96

those for the PD98059 for Probes 1 and 2 and the mean of thèse two probes. D) NIH3T3 cells were treated as in A) but then chased and labeled RNA analyzed, Expérimental Procédures. The right panel shows the 45S to 32S rRNA ratio. B) and C) represent the average of 7 independent experiments each performed in triplicate, error bars represent ±one standard error.

Actinomycin D (ActD), a DNA interealating agent, inhibits transcription elongation

and destabilizes RPI transcription complexes, (Wilmanska et al., 2001). 45S synthesis was

rapidly repressed (90% at 30min) by ActD in NIH3T3 cells proliferating maximally in full

médium (Figure 3.2C). A very similar réduction was seen for RPI loading levels

immediately preceding the 45S rRNA termination site; 85% after 30min. Hencc, should

RPI loading hâve been modulated in responsc to growth factor stimulation in Figure 3.2B,

this change would certainly hâve been detected. Consistent with a stochastic ActD block to

elongation, the réduction in RPI loading was more pronounced at the 3' than at the 5' of the

transcribed région.

Together the data of Figures 3.1 and 3.2 provide an unequivocal démonstration that

the increased rRNA gène transcription induced by growth factors was not the resuit of an

increase in RPI transcription complexes. This is clearly inconsistent with régulation of

rRNA synthesis solely at the level of initiation.

3.5.3. Direct Measurements of RNAPI Elongation Rates In Vivo Reveals Growth Factor Régulation

During the initial phase of puise labeling, the accumulation of 45S rRNA will be

proportional to e.t2, (where e is the elongation rate and t the time from the start of labeling

(to)), since most transcripts hâve already initiated (Figure 3.3A). This holds within the time

range t = to to t = lg/e, the time to transcribe a full length 45 S rRNA, where lg is the gène

length. Subsequently, since ail transcripts are initiated after the start of labeling, each will

be uniformly labeled and accumulation of label into 45 S rRNA will be linear, varying as

e.t. Estimâtes of RNA polymerase elongation rates hâve ranged from 30 to 100 nucleotides

per second (Cavanaugh and Thompson, 1985; Dundr et al., 2002; French et al., 2003).

Slowing of elongation should lead to an observable lag before the linear phase of label

incorporation is reached (Figure 3.3B, upper panel). On the other hand, if the elongation

97

rate is constant, no lag should be apparent, and the normalized incorporation curves for

stimulated and unstimulated cells should superimpose (Figure 3.3B, lower panel).

3078 -►♦3221

45S-5' ETS ITS-2 PCR probes

13154-»- -«-13295 I 15401 -*-■•-15567

45S-3' IGS1

M

PCR probe:-47S-5' ETS ITS-2 47S-3' IGS1 47S-5'

Treatment:- S " H £ <° S £ m « S " S £ <" S °> <" <" LU " ■ LU

^Ao II ' : RPI|V if 0*r «N <?

c)

g> 1.6-]

rJ

L!

,L,

Minutes 0 15 30 0 15 30 0 15 30 0 15 30 0 15 30 after ActD:-

PCR probe:- 47S-51 ETS ITS-2 47S-3'

Figure 3.2 ChIP assays of RPI loading also show that growth factor enhancement of ribosomal RNA gène expression in NIH3T3 cells does not resuit from an increase in RPI loading.

A) Position of the five mouse ribosomal gène segments, (PCR probes), 45S-5', ETS, ITS-2, 45S-3' and IGSl, used to analyse yield of immunoprecipitated DNA by quantitative PCR

98

(Q-PCR). B) Comparison of 45S transcription rates determined by puise labeling, as in Figure 3.1, with RPI loading determined as the yield of various gène segments after immunoprecipitation with an antibody raised against the |3' (194kD) subunit of RPI. The yield of 45S-5' gène segment in control immunoprecipitations with a pre-immune (Pre-Immun) and a sérum against the phosphorylated CTD of the elongating form of RNA polymerase II (RPII) (Vincent et al., 1996) are also shown. Data has been normalized to the results for PD98059 treated cells. Cell treatments, labeling and harvesting were as indicated in Figure 3.1 A. C) 45S production rates and RPI loading were determined in parallel respectively by puise labeling and ChIP in rapidly proliferating, un-starved NIH3T3 cells before (0 min.) or after inhibition for 15 or 30 min. with 50 ng.ml"1 actinomycin D (ActD). Data has been normalized to the results for 0 min. In B) and C) the insets show examples of 45S labelings and the 18S levels on corresponding electrophoretic analyses. Both the data in Figure 3.2B and in 3.2C are the mean of three independent experiments, each performed in triplicate, error bars represent ±one standard error.

When 45S rRNA labeling was followed with time in NIH3T3 cells, a significant lag

was observed in the MEKl/2 inhibited cells as compared to the EGF stimulated cells. As

expected, incorporation into 45 S rRNA increased to a steady state level that was some five

times lower in PD than in EGF treated cells (Figure 3.3C). The initial phase of 45S labeling

in the PD treated cells closely followed the predicted dependence on t2, (corrélation

coefficient of 0.99) (Figure 3.3D) (the steady state 45S rRNA levels for PD and EGF

treatments hâve both been normalized to 1.0). It also lagged significantly on that in the

EGF stimulated cells, which was already linear by 5 min. This lag was not due to a delay

in [H3]-uridine uptake, which was unaffected by cell treatment (Figure 3.3C, right panel).

Extrapolation of the maximum incorporation rates onto the time axis gave a direct measure

of the transit time for RPI to complète a 45S transcript (Figure 3.3D, dashed lines). This

gave ~2.2 min for EGF treated cells, corresponding to an elongation rate of 100 nucl.s"1, the

same as that measured in CMT3 cells (Dundr et al., 2002), and ~12 min for PD treated

cells, équivalent to 19 nucl.s"1. (Correction of the curves in Figure 3.3D for [3H]-uridine

uptake gave the same elongation rate for PD treatment, while the apparent elongation rate

for EGF was somewhat increased.) Thus, RPI elongation was ~5 fold faster in growth-

stimulated NIH3T3 cells as compared to growth-inhibited cells. This explains our inability

to detect a change in RPI loading by Run-On or ChIP and agrées with the elongation rate

régulation of 5 to 6 fold inferred from thèse measurements, i.e. (45S rRNA synthesis

rate)/(RPI loading).

99

A) : ■ | : |

fWfWt i fWfWWWffW

è i l l l l A a l l HMMMtfbMMHM I '

•ii UUi Timo of puise labellinQ

i .1 . ii. slimuliilutl Growth Inhibiled

f % È8F 0.6 <£

0 20 40 60 80 100 0 20 40 60 Time of [H ] labeling, min. Time of ( HJ labeling. min.

I» 5 7.5 10 12.5 15 17.5 20 25 30 min.

0 A 5 10 A 15 20 25 30 35

Time of (3H] labeling, min.

Figure 3.3 Direct détermination of elongation rates in growth-stimulated and inhibited NIH3T3 cells.

A) Diagram describing the expected time course of incorporation of radioactive label into 45 S rRNA during and just folio wing the time taken to complète the synthesis of a full

100

transcript. RPI molécules are indicated as solid ellipses transcribing along the rRNA gène. Spécifie RPI molécules are shown by différent open symbols to indicate the progression of transcription. When labeling is started at time to (top panel) ail initiated transcripts are as yet unlabeled (vertical grey Unes). As the early phase of labeling proceeds (centre panels) nascent transcripts initiated before labeling began will become partially labeled (vertical black Unes) dépendent on their extent of synthesis at the start of labeling, to. During this early labeling phase the rate of incorporation into full-length transcripts (indicated as vertical lines after the + sign) will be proportional to et2 (labeled (shaded) area of full-length transcripts in centre panels), where E is the RPI elongation rate. The RPI molécule indicated by an open cross will complète a full-length transcript when labeling has been continued for a time t, where t = lg.e, lg being the gène length and e the RPI elongation rate. Subséquent to this, ail transcripts will be labeled from end to end and incorporation of label into 45S rRNA will increase linearly with time (bottom panel). B) Indicates the expected curves for 45S rRNA labeling under i) variable elongation rates, ii) constant elongation rate. Normalizing the steady state 45 rRNA levels permits visual comparison of the incorporation curves, dashed lines. C) Expérimental détermination of the time course of f3H]-uridine incorporation into 45S rRNA, left panel, and of [3H]-uridine uptake, right panel, in N1H3T3 cells treated with EGF or PD98059 (PD). D) The 45S data in C) were normalized to the steady state levels of labeling and the early time points analyzed by curve fit. Upper panels show an example electrophoretic analysis of 45S and 32S rRNA labeling in EGF treated cells. 45S labeling and [3H|-uridine uptake data are the mean respectively of five and three independent experiments in duplicate, error bars represent ±one standard error.

3.5.4. High DNA template Occupancy by UBF Represses RNAPI Transcription In Vitro

UBF has been shown to rcgulate rRNA gène transcription dépendent on its steady

state level (Brandenburger et al., 2003; Hannan et al., 1999) and its modification state

(Luyken et al., 1996; Stefanovsky et al., 2001b; Voit et al., 1995). It is now generally

agreed that UBF is found bound throughout the rRNA gènes and especially throughout the

45S coding région (Grandori et al., 2005; O'Sullivan et al., 2002; Poortinga et al., 2004)

(Figure 3.7A). Thus, UBF defines a specialized r-chromatin through which RPI

transcription complexes must pass.

In vitro, UBF has been found variously to activate RPI transcription at pre-initiation

(Jantzen et al., 1992) or promoter release (Russell and Zomerdijk, 2005), to relieve Hl

repression (Kuhn and Grummt, 1992) or to be completely unnecessary (Smith et al., 1993).

However, at the concentrations generally used in vitro, UBF bound no more than a few

percent of the DNA template. Given thèse uncertainties, it was necessary to establish the

101

effect of near stochiometric levels of DNA template occupancy by UBF, as occur in vivo

(Leblanc et al., 1993). Though lOOng of wt UBF1 gave a slight increase in RPI

transcription, 300ng had no positive effect and 600ng, corresponding to an UBF-DNA

Template Occupancy (UTO) of -90%, very significantly inhibited transcription (Figure

3.4A and B). Since UBF interacts with both RPI and SLl (mTIF-IB) (Moss and

Stefanovsky, 2002), this inhibition could hâve been caused by squelching. Core UBF

(NBoxl23) is responsible for enhancesome formation (Stefanovsky et al., 1996) but lacks

the RPI and SLl interaction domains, (op. cit.). Nboxl23 (44.5kD) inhibited transcription

to the same degree as UBF1 (89.5kD), (compare 300ng Nboxl23 with 600ng UBF1, -90%

UTO) (Figure 3.4C and D). Thus, repression of RPI transcription was not due to squelching

and required only the architectural core of UBF responsible for DNA binding and looping.

3.5.5. ERK Phosphorylation of HMG Boxes 1 and 2 of UBF Derepresses Transcription

ERK phosphorylation of amino acids 117 and 201 within ITMG-boxes 1 and 2 of

UBF prevents enhancesome formation (Stefanovsky et al., 2006). Mutational simulation of

this phosphorylation (Nbox 123-Tl 17/201E) also abrogated repression of RPI transcription

(Figure 3.4C). Similarly, ERK2 phosphorylation of Nbox 123 abrogated repression in a

quantitative manner (Figure 3.4E and F). Thus, transcription repression is mediated by the

two DNA binding domains of UBF essential for enhancesome formation and is relieved by

ERK phosphorylation, shown to unfold the enhancesome (Stefanovsky et al., 1996;

Stefanovsky et al., 2006).

3.5.6. Unmodified, but Not ERK-Phosphorylated UBF Impedes Transcription Downstream of+34

Re-initiation during the standard run-off transcription reaction (-15 rounds in Figure

3.4) made it impossible to décide whether UBF inhibited promoter release or elongation.

Promoter release can be a rate-limiting step in transcription, but after transcribing 14 or so

nucleotides the polymerase enters a stable elongation phase (Dvir, 2002).

102

To résolve this problem, the first 34bp of the 45S transcribed région were replaced

by a G-less cassette, see Expérimental Procédures. Incubation of this template with a

DEAE280 fraction (Schnapp and Grummt, 1996) in the présence of ATP, [32P]-UTP and

CTP, but no GTP allowed transcription to initiate and proceed to the end of the G-less

séquence, where it paused (Figure 3.5A) and left-hand tracks in Figure 3.5B and C).

Addition of GTP (and exccss UTP to suppress radiolabel incorporation) then "chased-out"

the paused complexes to yield full length transcripts of 320bp. Addition of increasing

quantities of full-length UBF1 to the paused transcription complexes before the chase

inhibited subséquent elongation in a dose dépendent manner, such that by 1200ng UBF1

the yield of full length transcripts had dropped by 40%, (Figure 3.5B). Addition of 600ng

of core UBF (Nboxl23) inhibited elongation to a similar degree, (Figure 3.5C) and this

inhibition was reversed by heparin release of Nboxl23 (data not shown, but see Figure

3.6B). ERK2 phosphorylation of Nboxl23 relieved inhibition in a stepwise manner, while

mock-phosphorylation had no effect. Thus, core UBF reversibly impeded continued

elongation of late initiation/early elongation complexes dépendent on its state of ERK

phosphorylation.

3.5.7. UBF Reversibly Impedes Passage of Elongating RNAPI Core UBF (Nboxl23) also inhibited continued elongation of RPI complexes paused

at +64bp (Figure 3.6A and B). As on the 34bp G-less template, this inhibition was not due

to the disruption of transcription complexes, since addition of heparin, which removes UBF

(data not shown), restored the yield of full-length transcripts. Both full length UBF1 and

core UBF (Nboxl23) inhibited elongation from +64 bp in a dose dépendent manner, while

the corresponding Tl 17/201 double mutants had only minor effects (Figure 3.6C and D).

Increasing ERK phosphorylation of Nboxl23 relieved this inhibition in a stepwise manner,

while mock-phosphorylation had no effect (Figure 3.6E). Maximally, half of the ERK sites

on Nboxl23 were phosphorylatcd in thèse experiments. Thus, the level of UBF

phosphorylation corresponded closely with the degree of de-repression of RPI elongation.

103

A) UBF1 H)

297b I

template + + + + UBF - 100 300 600 ng

C) 297b ►

template

Nbox123 Nbox123 T117/201E Predlcted template. occupancy

D)

Nbox123 Moasured template occupancy

+ + + + 50 100 300

60 75 90

+ + - ng

50 100 300 ng 60 75 90 %

50 100 300 600 ng „„ 80±10 (100) % 50±5 85±10

E) 297b I

template Nbox123

-T117/201E Erk2, min

F)

+ +

+ +

10 30 +

10 i

30

Erk2, min. 0 10 20 30

Nbox123 w «■»■•■•

-P04

0 10 ERK Incubation, min.

Figure 3.4 High template occupancy by UBF inhibits RPI transcription in vitro.

A) Comassie stained SDS-PAGE analysis of baculovirus-expressed mammalian UBFl. B) In vitro run-off transcription from the mouse RPI promoter with addition of increasing amounts of UBFl. The DNA template was transcribed in a reaction containing nuclear extract supplemented with UBFl, Expérimental Procédures. C) A transcription reaction as in B was also performed but UBFl was replaced by Nboxl23 or Nboxl23-T117/201E. Template occupancy was estimated using a binding constant (Kd) of 2 x 10"8M (Leblanc et al., 1993; Stefanovsky et al., 2006). D) Expérimental détermination of template occupancy

104

by Nboxl23. Characteristic changes in the UBF DNasel cleavage pattern around the initiation site were used to estimate template occupancy under the conditions of in vitro transcription, (Leblanc and Moss, 2001; Leblanc et al., 1993), left panel shows représentative DNasel cleavage patterns and right panel the corresponding cleavage profiles from phosphoimager analyses at increasing Nboxl23 amounts, see Expérimental Procédures. E) In vitro transcription as in B and D but in the présence of 300 ng Nboxl23 or Nboxl23-T 117/203 phosphorylated with ERK2 for increasing times. F) Left: upper panel, Coomassie-stained SDS-PAGE gel of a time course of ERK2 phosphorylation of Nboxl23, lower panel, Phospholmager analysis of the same gel. Right: Phospholmager quantitation of transcription levels in E plotted against time of Nboxl23 reaction with ERK2.

A) 34b G-less template, DEAE fraction +

ATP,[32P]UTP,CTP

i UBF

addition UTP + GTP | f Elongation

t2ûmin t35m|n »- 20 min.

B) G-less Temp: + GTP Chase: -FL-UBF1 ng:- - S

Full length. 320b

C) G-less Temp: - + + + + + + + + + + GTP Chase: + - + + + + + + + + + Nbox1Z3-ERK: - - 0102030- - - min. NbOXl23-Mock: 010 2030min.

0 300 600 900 1200 UBF1,ng

0 10 20 30

Nbox123, min. ERK Incubation

Figure 3.5 UBF inhibits RPI transcription downstream of+34 bp dépendent on its phosphorylation state.

105

A) Order of addition of reaction components on the 34 bp G-less cassette template. B) In vitro transcription reactions using the 34 bp G-less cassette linked to the mouse RPI promoter. Full length UBFl (FL-UBFl), see (Figure 3.4A), was added in increasing amounts to reactions in which the RPI transcription complexes had previously been paused at +34 bp by withdrawal of GTP. C) Nboxl23 was ERK2- or mock-phosphorylated for increasing times, e.g. see Figure 3.4F, and 600ng added to reactions in which the RPI transcription complexes had been paused at +34 bp. In both B and C transcription was continued by the addition of GTP and 32P incorporation suppressed by the addition of excess UTP (GTP chase). "5S" refers to end labeling of endogenous 5S RNA during the transcription reaction. In B and C the gel analyses are shown above the corresponding graphie of the phosphoimager quantification, error bars give estimated error of ± 0.075. In C, phosphorylation levels of Nboxl23, determined by y[ P]-ATP tracing, exceeded 1 mole per mole Nboxl23 in a 30 minutes reaction. Thus, greater than 50% of the ERK sites of Nboxl23 were phosphorylated.

3.5.8. In Vivo Activation of ERK Does Not Displace UBF from the Mouse rRNA gènes

In vitro data has suggested that ERK phosphorylation does not displace UBF from

rDNA, (Stefanovsky et al., 2006). However, it was possible that such a displacement could

occur in vivo as a necessary part of enhancing transcription elongation. Therefore, we

performed ChIP analyses to détermine the UBF levels bound on the rRNA gènes after

growth factor stimulation or ERK repression (Figure 3.7A). Consistent with our in vitro

observations, growth factor stimulation caused no significant réduction in the levels of UBF

bound throughout the transcribed gène région.

3.6. Discussion The ability of a cell to assemble new ribosomes is both a direct measure of

its capacity to proliferate and the déterminant of its prolifération rate. Indeed, it is

becoming clear that a cell décides to commit to cell division or to withdraw from the

proliferative cell cycle based on its ability to assemble ribosomes. In simple organisms,

ribosome biogenesis is regulated by the availability of nutrients, while in more complex

organisms, and in particular in mammals, it is also regulated by endocrine and paracrine

regulatory signaling. Such régulation must coordinate the synthesis of the more than two

hundred ribosomal and accessory proteins required for ribosome biogenesis with the

production of the 5.8S, 18S and 28S rRNAs. While protein synthesis can be regulated at

106

both the transcriptional and translational levels, this is not the case for the 45 S precursor

rRNA, whose synthesis can only be regulated at the level of RPI transcription. To date, it

has been generally accepted that régulation of RPI transcription occurs solely at the level of

initiation via pre-initiation complex formation and the availability of initiation compétent

RPI, see (Moss, 2004; Moss and Stefanovsky, 2002) for reviews. In contrast, we show hère

that RPI elongation rates are growth regulated by r-chromatin remodeling.

A) B) Nbox123: - - - + ♦

Heparin: • ■ + - + GTP chase: • + -f + +

64b G-less template, DEAE traction + ATP,[32P]UTP,CTP

,'o

UBF addition

T ~ t20mln -

UTP + GTP (Comp. template)

{Heparin) | Elongation

t3Sm(n »- 20 min

C) G-less Temp: +

Somp.Temp: -T P C h a s f

UBF1: UBF1-T117/201E:

V" flr T!Î

0 400 800 1200 UBF1 (-T117/S01E), ng

E)

t iË° X !S??P K , ; - - - 0 10 20 30 - - - -Nbox123Mock:

(6S).

64b - « ~

t 'Ul. -»4t> -

( 5 S ) - -_

6 4 b -

D)

8-less Temp: -omp.Temp: +

GTP Chase: . Nbox123: * * Nbox123-T117/20tE

Nbox123(-T117/201E), rg

0 10 20 ERK Incubation, min.

Figure 3.6 UBF reversibly inhibits the full elongation phase of RPI transcription.

107

A) Order of addition of reaction components on the 64bp G-less cassette template. B) Elongation of preformed transcription complexes paused at + 64 was inhibited by 600ng of Nboxl23. Heparin, 80 ug.ml"1, released Nboxl23 and relieved this inhibition. C) Increasing amounts of full-length UBF1 or double mutant -T117/201E were added to preformed transcription complexes paused at +64 bp before continuing elongation by GTP addition. D) Alternatively, increasing amounts of Nboxl23 or the -T 117/201 mutant were added to complexes paused at +64 bp before continuing elongation. In the experiment shown in D, a +155bp template was added inlO fold excess (Comp. Template) along with the chase. E) ERK2- or mock-phosphorylated NBoxl23 was added to complexes paused at +64 bp before continuing elongation. Phosphorylation levels of Nboxl23 exceeded 1 mole per mole Nboxl23 in a 30 minutes reaction. The graphie plots in D and E display the mean of three independent experiments, error bars represent ±one standard error.

I s-

^ ■ ^ ^ aafl ^^____»^HIghERKactlvatlon; I ^Ê H * Elongation facllltatsd

core UBF dimer

Figure 3.7 Engagement of UBF throughout the 45S rRNA coding région is not significantly affected by growth factor stimulation.

108

A) UBF ChIP and Q-PCR assays, data is the average of 5 independent experiments. Error bars represent ±one standard error. B) Model for elongation régulation by UBF phosphorylation. In the upper panel multiple enhancesomes are shown impeding the passage of the polymerase, while in the lower panel local unfolding of the enhancesomes by ERK phosphorylation of UBF (*) permits elongation. The DNA template is shown in light grey, dimers of core UBF in darker grey shading and the transcript as a black wavy Une.

Using both Run-On and ChIP we determined that RPI loading on the rRNA gènes

remains constant regardless of 45S rRNA synthesis rates, in contradiction with presently

held views on the régulation of thèse gènes. Using a kinetic analysis to measure elongation

rates in vivo, we determined that between stimulated and inhibited growth conditions RPI

elongation rates varied from -100 nucl.sec"1 to -20 nucl.sec"1. This degree of régulation

corresponded precisely with the 5 fold change in 45S rRNA synthesis rate observed. Thus,

not only are RPI elongation rates regulated in response to growth factor stimulation, they

also quantitatively account for rRNA gène régulation. In fact, thèse data are consistent with

carly biochemical studies suggesting régulation at the level of transcription elongation

(Cavanaugh and Thompson, 1985; Dauphinais, 1981; Dembinski and Bell, 1984; Fuhrman

and Gill, 1975). They are also consistent with the many examples of EM spreads from

vertebrates showing a near constant, high density packing of RPI molécules on the rRNA

gènes, see (Moss et al., 1985) for discussion.

Studies of the transcriptional rôle of UBF hâve also revealed a completely

unrecognized function for this architectural protein, that of an ERK responsive transcription

elongation regulator. UBF binds throughout the ribosomal RNA gènes of mammals, where

it defines a spécifie r-chromatin (O'Sullivan et al., 2002). Its phosphorylation by ERK is

essential for growth factor enhancement of 45 S rRNA synthesis (Stefanovsky et al.,

2001b). Stochiometric levels of UBF binding to ribosomal rDNA, as occur in vivo (Leblanc

et al., 1993), inhibit RPI transcription elongation, while ERK-phosphorylated UBF is

permissive to elongation. Transcription inhibition requires only the N-terminal "core" of

UBF, which encompasses its DNA binding HMG boxes and is responsible for bending

~135 bp of DNA into a near 360 deg. loop to produce a nucleosome-like structure called

the enhancesome (Stefanovsky et al., 1996; Stefanovsky et al., 2001a). In a parallel study

we hâve shown that ERK phosphorylation of UBF unfolds the enhancesome (Stefanovsky

109

et al., 2006). Thus, passage of elongating RPI complexes appears to be blocked by the

enhancesome structure and remodeling of this structure by HMG-box phosphorylation

allows continued elongation (Figure 3.7B). This is the first évidence showing that UBF

controls the accessibility of r-chromatin and that it does so in a growth-regulated manner.

The chromatin nucleosome and the core histones also rcgulate r-chromatin, but their rolc in

regulating RPI transcription is still unclear, see (Moss, 2004) for discussion.

Does UBF also play a rôle in enhancing or repressing initiation and/or promoter

release? Though, our experiments do not address this question directly, it is possible to

deduce from the data that if this does occur the effects must be quite small. On short

templates, able to establish at most two enhancesomes within the transcribed région, we

found that UBF repressed multi-round transcription (15±2 rounds) by ~10 times and

arrested -50% of elongation complexes in single-round transcription, (Figures 3.4, 3.5 and

3.6). Consistent with its low binding constant, UBF exchange is rapid compared with the

time of the in vitro experiment, (Chen and Huang, 2001; Dundr et al., 2002; Leblanc et al.,

1993). Thus, each template that missed UBF arrest in the first round of transcription due to

lack of effective UBF binding would be subject to -50% chance of arrest at each re-

initiated round of transcription. The final yield of transcripts should then be -15 times less

than in the absence of UBF (21/2n, n=l to 15). We observed -10 times inhibition in a

multi-round reaction, perhaps due to slightly less than a 50% chance of elongation arrest

per round or to a small number of initially arrested complexes restarting elongation. A

kinetic analysis of multi-round in vitro transcription further supported this conclusion, the

ratio of transcript yield in the absence and présence core-UBF increasing over 10 fold

during the reaction, (data not shown). This said, the in vitro System is an imperfect

représentation of the in vivo reality, leaving open the possibility that UBF is required for

multiple steps in rRNA gène transcription.

The surprising ability of the rRNA gènes to maintain a constant level of RPI loading

despite large transcription rate changes, clearly infers that a mechanism must exist to

exactly correlate initiation and elongation rates. How might this be achieved? Our data

show that relative changes in 45S rRNA synthesis and elongation rates are quantitatively

110

indistinguishable. Hence, given excess active RPI, elongation régulation alone can explain

the full dynamic range rRNA gène transcription we observe. The simplest mechanism to

co-ordinate initiation with elongation would then be to allow elongation détermine the

initiation rate, lower elongation rates backing up polymerases and slowing initiation. RPI

initiation must of course be sufficient to match the enhanced elongation rates we observe.

However, we were unable to detect significant changes in the capacity of RPI to

specifically initiale 47S rRNA transcription in vitro in nuclear extracts made from NIH3T3

cells treated with EGF or PD exactly as for Figures 3.1 and 3.2 (Langlois and Stefanovsky,

unpublished observation). Consistent with this, we found that inactivation of the ERK

pathway almost always led to a slight increase in RPI loading, compare PD and sérum

treatments (Figure 3.2B) and U0126 treatment with Control (Figures 3.8). Such an increase

could only hâve occurred if the elongation rate were limiting.

A limiting rôle for elongation provides an explanation for the élégant orchestration

of the co-transcriptional phase of ribosome biogenesis. Assembly of ribosomal and

accessory proteins onto the 45S rRNA begins soon after initiation and continues throughout

the production of the full-length precursor (Tschochner and Ilurt, 2003). Control of 45S

rRNA synthesis at the level of RPI initiation does not provide an obvious means by which

this assembly process could be coordinated. On the other hand, coupling RPI elongation

rates to the assembly of the nascent ribosomes would provide a direct feedback mechanism

that is continuously able to adapt 45S rRNA synthesis to the availability of r-proteins.

Reduced production of thèse proteins would automatically lead to a proportional slowing in

45S elongation, while any potential increase in 45S rRNA synthesis would always be

maintained in step with available protein. This would not only directly couple rRNA

synthesis rates to ribosome assembly, but also might act as a "proof reading" mechanism by

minimizing the assembly of non-functional ribosomes. Consistent with thèse possibilities,

inhibition of processome assembly was shown to reduce rRNA synthesis in yeast

(Gallagher et al., 2004).

Clearly, future work is required to détermine how ERK phosphorylation of UBF fits

into the complex network regulating ribosome biogenesis and whether RPI elongation and

111

pre-ribosome assembly are in fact coupled. However, the présent study describes a

completely unforeseen mechanism by which the mammalian rRNA gènes are regulated. As

such, it should provide the basis for a signifïcantly better understanding of the mechanisms

that coordinate ribosome biogenesis with growth.

3.7. Supplementary Data A) EGF/HT/U

/ - 0 — -15-

[3H] -uridlne

Nuclear Run-On/ChIP

* 30 -45 ► time in

minutes:

RNA analysis

B)

5? 3.5

~ 2.5

!

S 1.5

% 0.5

oc 0

47S

18S

C)

— 3.5 p.

t. o 1 2.5 8 8 * I 1 - 5 ë 1 0)

1 °-5

I DE r,

)7 +1 1 /

+316 I

+2087 I

+4245 b.p

■ k / S p r e - i i i i i

- 2 Probes - 2 Probes

Ctl «HP — • u

EGF ^ | ^ V W^^V

Probe 2 Probe 1 Vector

I III C3 g LU

CD S LU w

a m

Cell Treatment Probe 1 Probe 2 Probe 1 + 2

Figure 3.8 EGF stimulation of 45S rRNA transcription in human cells does not lead to an increase in the number of RPI transcription complexes on the ribosomal gènes. Human SKF-5 neuroepithelioma cells were subjected to short-term (16h) sérum starvation and then ERK activity either stimulated with human EGF or suppressed with 10 ^M of the MEK inhibitor U0126 (Stefanovsky, Pelletier et al. 2001). A) The 45S production rates were then determined by puise labeling with [3H]-uridine for 30min, the puise beginning 15min after hEGF addition and RPI loadings determined by nuclear Run-On. B) EGF induced a three-fold increase in 45 S rRNA synthesis as compared to the fully repressed ERK state (U0126, U). C) Parallel cultures were treated identically, but analysed for RPI loading levels by nuclear Run-On, see Expérimental Procédures. Neither EGF stimulation nor MEK inhibition caused a significant change in the number of RPI molécules engaged in transcribing the ribosomal gènes. The data in B and C was derived from 4 independent experiments, each performed in triplicate. Inset in C indicates the human rRNA gène

112

probes used in Run-On experiments and shows examples of slot-blot hybridization. Probe 1 + 2 refers to the mean of the normalized data for the separate probes.

3.8. Acknowledgments The authors wish to thank Dr M. Vincent for an anti-RPII-CTD antibody, Dr R.

Hannan for the pFastBaclrUBFl clone, Drs M. Cobb and C. Chrestensen for the ERK-

MEK vector and Dr J.-Y. Masson for critical reading. This work was supported by an

operating grant from the Canadian Institutes of Health Research (CIHR) and a CIHR

scholarship to T. G.-K. The Research Centre of the CHUQ, in which the Cancer Research

Centre is housed, is supported by a grant from the FRSQ (Québec).

3.9. Expérimental procédures

3.9.1. Plasmid constructs and protein expression Point mutations T117E and T201E were introduced into core UBF (Nboxl23)

(Leblanc et al., 1993) and FLAGrUBFl. The différent forms of Nboxl23 were expressed as

GST-fusion proteins in E. coli, purified from the soluble protein fraction on G-Sepharose

(Amersham Biosciences) and recovered by thrombin cleavage (Smith and Corcoran, 1991).

Full-length UBF1 and UBF1-T117201E were expressed in Sf9 cells (ATCC) using the

Bac-to-Bac Baculovirus Expression System (Invitrogen) following the manufacturer's

instructions and subsequently purified on Anti-Flag M2 Affinity Resin (Sigma).

Templates used for in vitro transcription contained the mouse rRNA gène séquences

from -168 to + 297 subcloned into Sali site of pUC9 and were either linearized at +155

with Smal or +297 with PvuII. The G-less template GL34 was created by replacing the

nucleotides between the Rsal site ( +1) and the Bsal site +41 by the synthetic séquence 5'-

ACTCACACTCTCTCCTTTCCCAAATTTCACTAAAGGACACT-3'. The GL64

séquence was created in two steps, first the Rsal to Bsal séquence was replaced by the

synthetic séquence 5'-ACTCACACTCTCTCCTTTCCCAAATTTCACTAAACCACACT-

3', resulting in a G-less cassette to + 46. The Bsal to Mwol was then replaced by the

synthetic séquence 5'-ATAAACACACCCTTTCCATTTAA-3' resulting in a G-less

113

cassette to +64. The G-less cassette templates were linearized at +320 within the cloning

vector by HindiII.

3.9.2. In vivo puise labeling, RNA extraction and analysis NIH 3T3 cells (ATCC) were grown in 6 cm pétri dishes (seeded at 0.8x106 cclls per

pétri) for 24 hours in DMEM-high glucose (Invitrogen), 10% fetal bovine sérum (Wisent)

and then washed with DMEM, high glucose, no sérum and maintained for 18 hr in serum-

free médium. At time zéro (to) the médium was replaced with 1 ml of fresh pre-warmed

médium containing either 10% sérum, 100 ng/ml recombinant EGF (Invitrogen), 50u.M

PD98059 (Calbiochem), 10u.M U0126 (Calbiochem) or no addition and the cells incubated

for 15 min. before the addition of lOOul DMEM, no sérum, containing 2.5 uCi [3Hj-uridine

(Perkin Elmer) (or 10u,Ci in the time course experiments) and incubation for a further 30

min. For pulse-chase experiments, the médium was replaced with unlabeled médium

containing 0.5 mM uridine. RNA was extracted in 1ml Trizol (Invitrogen) according to the

manufacturera protocol. 2-4 |ig of RNA was loaded on 1% formaldehyde gel and

electrophoresed at 2.5V/cm in MOPS running buffer overnight. The EtBr stained gels were

subsequently photographed, UV-irradiated in a UV-crosslinker (Hoefer) for 5 min at

maximum energy and transferred to a Biodyne B membrane (Pall) by a Vacu-Blot transfer

(Biometra) at 200 mbar for 90 min in 2 x SSC. The membrane was UV-crosslinked at 70

J.cm" , washed in water, dried and sprayed with ENHance (Perkin-Elmer) and exposed to

Kodak-XAR-5 film at -80 °C for 24 to 72 hr. The [3H]-labeled bands were eut from the

membrane and counted in ScintiVerse (Fisher) and external standard corrected counts

normalized to 18S rRNA using Image J (NIH). Actinomycin D (Calbiochem) treatments

were performed on un-starved NIH3T3 cells proliferating at maximum rate in full sérum

médium. [3H]-uridine uptake was determined as (Granick, 1975). Labeled cells were

rapidly washed with unlabeled médium containing EGF or PD98059, solubilized with

Scintigest (Fisher) in pétri and counted in Scinti Verse (Fisher). Total RNA labeling was

<5% of [3H] total uptake during the experiments.

114

3.9.3. Nuclear Run-On Nuclear Run-On assays were performed as (Cavanaugh and Thompson, 1986;

Hannan et al., 1995). Triplicate cultures of cells were treated in parallel with puise

labclings, but then rapidly scrape-harvested into ice-cold PBS, pelleted and stored at -80°C.

Frozen cells were lysed in lOmM Tris-HCl pH7.4, lOmM NaCl, lOmM MgCl2, 0.1%

IGEPAL (Sigma), nuclei recovered, 4°C, 1300 rpm, 5 min, and resuspended in 50mM Tris-

HCl pH8.3, 5mM MgCl2, 0.1 mM EDTA, 40% Glycerol and DNA concentrations

determined (Labarca and Paigen, 1980). Equal aliquots of nuclei were subjected to Nuclear

Run-On for 30min at 37°C in 28mM Tris-HCl pH8.3, 147mM KC1, 5mM MgCl2, 0.45mM

EDTA, 18% Glycerol, 80fag/ml Heparin, 0.49mM ATP, UTP and GTP, 1.2uM CTP (100

uGi), 2.4mM DTT, 26U RNAguard (Amersham Biotech) and 20 to 50u.g genomic DNA

(nuclei). Run-On reactions were terminated by addition of 2.5M Guanidinium

isothiocyanate, 0.1 M Na sarkosyl, 0.1 M |3-mercapto-EtOH, 20mM Na citrate pH 7, 10

Hg.ml"1 tRNA (Sigma), 60.103 cpm of [3H] labeled RNA transcribed from pT7T319U.

RNA recoveries were estimated by [3H] counting in ScintiVerse (Fisher). RNA

hybridization was performed on duplicate membranes (Biodyne B, Pall, 5\xg per slot of

probe in pT7T319U) in 50% formamide, 5 x SSC, 0.1% SDS at 55 °C, 4 h. After washing,

filters were analyzed for [32P] on a STORM 860 (Molecular Dynamics). Each membrane

slot was counted in ScintiVerse (Fisher) and [3H] RNA hybridization used to normalize the

Phospholmager data.

3.9.4. ChIP and Q-PCR assays 15 x 150mm pétri dishes of NIH 3T3 cells (5.8xl06 cells each) were treated in

parallel with puise labelings. Cells were rapidly (<2min.) washed with serum-free médium

and fixed in 0.25% formaldehyde for 10 min at room température. Nucleolar chromatin and

nucleolar ChIP assays were as (O'Sullivan et al., 2002).

PCR reactions (20u.l) were performed in triplicate with 2jxl of sample DNA, 20pmol

of each primer, and lOul Quantitect™ SYBR Green PCR Master Mix (QIAGEN). 45

reaction cycles of 15s at 94°C, 20s at 58°C (51°C for IGS1 primers), and 10s at 72°C were

carried on a Lightcycler (Roche) and quantified as (Frank et al., 2001). The primer

115

coordinates (relative to 45S rRNA initiation) were: 45S-5', 159-178 and 301-320; ETS,

3078-3095 and 3204-3221; ITS-2, 7482-7499 and 7624-7641; 45S-3', 13154-13171 and

13278-13295; IGS1, 15401-15417 and 15551-15567.

3.9.5. In vitro Phosphorylation of core UBF Active ERK.2 was prepared as (Wilsbacher and Cobb, 2001). 2-5 ug Nboxl23 was

phosphorylated at 37°C in 20ul of kinase buffer (20mM HEPES, pH 7.3, lOmM MgCl2,

ImM benzamidine, ImM DTT, 0.5mM ATP) with 4ul activated ERK2. Mock reactions

used heat-inactivated ERK2, (5 min at 95 °C).

3.9.6. In vitro transcription Reactions (25ul) contained 30ng of linearized template, 7ul S100 extract (Miller

and Sollner-Webb, 1981), lui RNAguard (Amersham Biotech), 0.5mM ATP, UTP and

GTP and 0.05mM CTP (Amersham Biotech) and lOuCi a-[32P]-CTP (Amersham Biotech),

in 4mM HEPES-KOH, pH 7.9, 8mM Glycerol, 7mM DTT, 0.05mM EDTA, 4mM MgCl2,

98mM KC1, 250ug.ml"' a-amanitin (Sigma). 1 to 3ul of UBF protein or buffer was added

in TBS or in ERK2 reaction buffer. Reactions were stopped with 180ul TE, pH 8.3, 0.1%

SDS, O.lmg/ml Proteinase K (Sigma), RNA resolved on 5% TBE-urea sequencing gels and

analyzed on a STORM 860 (Molecular Dynamics).

For G-less cassettes, 12 ul of DEAE 280 fraction (Schnapp and Grummt, 1996) was

used in place of the S100 extract. Initial reaction was for 20 min at 30°C with 0.5mM ATP

and CTP and 0.05mM a-[32P]-UTP (Perkin Elmer). After addition of UBF protein or

buffer, incubation was continued for another 15 min at 30°C. Subsequently, 0.5mM of GTP

and UTP were added and where indicated excess competitor template or 80 ug.ml"1 heparin

(Sigma) and incubation continued at 30°C for 20 min. The samples were resolved on a

15% TBE-urea sequencing gel and analyzed as above. 5S RNA contained in the DEAE 280

fraction was end-labeled by endogenous enzymes during the transcription reaction.

116

3.9.7. Dnasel footprinting Reactions (25u.l) contained 30ng of linearized template, 7ul S100 extract (Miller

and Sollner-Webb, 1981), luJ RNAguard (Amersham Biotech), 0.5mM ATP, UTP and

GTP and 0.05mM CTP (Amersham Biotech) and lOuCi a-f32P]-CTP (Amersham Biotech),

in 4mM HEPES-KOH, pH 7.9, 8mM Glycerol, 7mM DTT, 0.05mM EDTA, 4mM MgCl2,

98mM KO, 250u,g.mr' a-amanitin (Sigma). 1 to 3ul of UBF protein or buffer was added

in TBS or in ERK2 reaction buffer. Reactions were stopped with 180uJ TE, pH 8.3, 0.1%

SDS, 0. lmg/ml Proteinase K (Sigma), RNA resolved on 5% TBE-urea sequcncing gels and

analyzed on a STORM 860 (Molecular Dynamics).

For G-less cassettes, 12 ul of DEAE 280 fraction (Schnapp and Grummt, 1996) was used in

place of the S100 extract. Initial reaction was for 20 min at 30°C with 0.5mM ATP and

CTP and 0.05mM a-[32P]-UTP (Perkin Elmer). After addition of UBF protein or buffer,

incubation was continued for another 15 min at 30°C. Subsequently, 0.5mM of GTP and

UTP were added and where indicated excess competitor template or 80 jig.ml"1 heparin

(Sigma) and incubation continued at 30°C for 20 min. The samples were resolved on a

15% TBE-urea sequencing gel and analyzed as above. 5S RNA contained in the DEAE 280

fraction was end-labeled by endogenous enzymes during the transcription reaction.

3.10. Références Bazett-Jones, D. P., Leblanc, B., Herfort, M., and Moss, T. (1994). Short-range DNA

looping by the Xenopus HMG-box transcription factor, xUBF. Science 264, 1134-1137.

Brandenburger, Y., Arthur, J. F., Woodcock, E. A., Du, X. J., Gao, X. M., Autelitano, D. J.,

Rothblum, L. I., and Hannan, R. D. (2003). Cardiac hypertrophy in vivo is associated with

increased expression of the ribosomal gène transcription factor UBF. FEBS Lett 548, 79-

84.

Cavanaugh, A. H., and Thompson, E. A., Jr. (1985). Hormonal régulation of transcription

of rDNA: glucocorticoid effects upon initiation and elongation in vitro. Nucl. Acids Res.

13, 3357-3369.

117

Cavanaugh, A. H., and Thompson, E. A., Jr. (1986). Hormonal régulation of transcription

of rDNA. Initiation of transcription by RNA polymerase I in vitro. J. Biol. Chem. 261,

12738-12744.

Chen, D. Y., and Huang, S. (2001). Nucleolar components involved in ribosome biogenesis

cycle between the nucleolus and nucleoplasm in interphase cells. J. Cell Biol. 153, 169-

176.

Claypool, J. A., French, S. L., Johzuka, K., Eliason, K., Vu, L., Dodd, J. A., Beyer, A. L.,

and Nomura, M. (2004). Tor pathway régulâtes Rrn3p-dependent recruitment of yeast RNA

polymerase I to the promoter but does not participate in altération of the number of active

gènes. Mol Biol Cell 15, 946-956.

Dauphinais, C. (1981). The control of ribosomal RNA transcription in lymphocytes.

Evidence that the rate of chain elongation is the limiting factor. Eur. J. Biochem. 114, 487-

492.

Dembinski, T. C, and Bell, P. A. (1984). Glucocorticoids modify the rate of ribosomal

RNA synthesis in rat thymus cells by regulating the polymerase elongation rate. J. Steroid

Biochem. 21, 497-504.

Dundr, M., Hoffmann-Rohrer, U., Hu, Q., Grummt, L, Rothblum, L. L, Phair, R. D., and

Misteli, T. (2002). A kinetic framework for a mammalian RNA polymerase in vivo.

Science 298, 1623-1626.

Dvir, A. (2002). Promoter escape by RNA polymerase II. Biochim Biophys Acta 1577,

208-223.

Frank, S. R., Schroeder, M., Fernandez, P., Taubert, S., and Amati, B. (2001). Binding of c-

Myc to chromatin médiates mitogen-induced acetylation of histone H4 and gène activation.

Gènes Dev. 15, 2069-2082.

French, S. L., Osheim, Y. N., Cioci, F., Nomura, M., and Beyer, A. L. (2003). In

exponentially growing Saccharomyces cerevisiae cells, rRNA synthesis is determined by

118

the summed RNA polymerase I loading rate rather than by the number of active gènes.

Mol. Cell Biol. 23, 1558-1568.

Fuhrman, S. A., and Gill, G. N. (1975). Adrenocorticotropic hormone stimulation of

adrenal RNA polymerase I and III activities. Nucleotide incorporation into internai

positions and 3' chain termini. Biochemistry 14, 2925-2933.

Gallagher, J. E. G., Dunbar, D. A., Granneman, S., Mitchell, B. M., Osheim, Y., Bcyer, A.

L., and Baserga, S. J. (2004). RNA polymerase I transcription and pre-rRNA processing are

linked by spécifie SSU processome components. Gènes Dev. 18, 2506-2517.

Grandori, C., Gomez-Roman, N., Felton-Edkins, Z. A., Ngouenet, C., Galloway, D. A.,

Eisenman, R. N., and White, R. J. (2005). c-Myc binds to human ribosomal DNA and

stimulâtes transcription of rRNA gènes by RNA polymerase I. Nat. Cell Biol. 7, 311-318.

Granick, D. (1975). Nucleolar necklaces in chick embryo fibroblast cells. I. Formation of

necklaces by dichlororibobenzimidazole and other adenosine analogues that decrease RNA

synthesis and dégrade preribosomes. J. Cell Biol. 65, 398-417.

Hannan, K. M., Brandenburger, Y., Jenkins, A., Sharkey, K., Cavanaugh, A., Rothblum, L.,

Moss, T., Poortinga, G., McArthur, G. A., Pearson, R. B., and Hannan, R. D. (2003).

mTOR-dependent régulation of ribosomal gène transcription requires S6K1 and is

mediated by phosphorylation of the carboxy-terminal activation domain of the nucleolar

transcription factor UBF. Mol. Cell Biol. 23, 8862-8877.

Hannan, R., Stefanovsky, V., Arino, T., Rothblum, L., and Moss, T. (1999). Cellular

régulation of ribosomal DNA transcription: both rat and Xenopus UBF1 stimulate rDNA

transcription in 3T3 fibroblasts. Nucl. Acids Res. 27, 1205-1213.

Hannan, R. D., Hempel, W. M., Cavanaugh, A., Arino, T., Dimitrov, S. I., Moss, T., and

Rothblum, L. (1998). Affinity purification of mammalian RNA polymerase I -

Identification of an associated kinase. J. Biol. Chem. 273, 1257-1267.

119

Hannan, R. D., Luyken, J., and Rothblum, L. I. (1995). Régulation of rDNA transcription

factors during cardiomyocyte hypertrophy induced by adrenergic agents. J. Biol. Chem.

270, 8290-8297.

James, M. J., and Zomcrdijk, J. C. (2004). Phosphatidylinositol 3-kinase and mTOR

signaling pathways regulate RNA polymerase I transcription in rcsponse to IGF-1 and

nutrients. J. Biol. Chem. 279, 8911-8918.

Jantzen, H. M., Chow, A. M., King, D. S., and Tjian, R. (1992). Multiple domains of the

RNA polymerase I activator hUBF interact with the TATA-binding protcin complex hSLl

to médiate transcription. Gènes Dev. 6, 1950-1963.

Jorgensen, P., Rupes, L, Sharom, J. R., Schneper, L., Broach, J. R., and Tyers, M. (2004).

A dynamic transcriptional network communicates growth potential to ribosome synthesis

and critical cell size. Gènes Dev. 18, 2491-2505.

Kuhn, A., and Grummt, I. (1992). Dual rôle of the nucleolar transcription factor UBF:

trans-activator and antirepressor. Proc. Natl Acad. Sci. (USA) 89, 7340-7344.

Labarca, C , and Paigen, K. (1980). A simple, rapid, and sensitive DNA assay procédure.

Anal Biochem 102, 344-352.

Leblanc, B., and Moss, T. (2001). DNase I footprinting. In DNA-Protein Interactions:

Principles and Protocols, T. Moss, éd. (Totowa, N.J., Humana), pp. 31-38.

Leblanc, B., Read, C , and Moss, T. (1993). Récognition of the Xenopus ribosomal core

promoter by the transcription factor xUBF involves multiple HMG box domains and leads

to an xUBF interdomain interaction. EMBO J. 12, 513-525.

Luyken, J., Hannan, R. D., Cheung, J. Y., and Rothblum, L. I. (1996). Régulation of rDNA

transcription during endothelin-1-induced hypertrophy of neonatal cardiomyocytes -

Hyperphosphorylation of upstream binding factor, an rDNA transcription factor. Circ.Res.

78, 354-361.

120

Mayer, C, Bierhoff, H., and Grummt, 1. (2005). The nucleolus as a stress sensor: JNK2

inactivates the transcription factor TIF-IA and down-regulates rRNA synthesis. Gènes Dev.

79,933-941.

Mayer, C, Zhao, J., Yuan, X., and Grummt, I. (2004). mTOR-dependent activation of the

transcription factor TIF-IA links rRNA synthesis to nutrient availability. Gènes Dev. 18,

423-434.

Miller, K. G., and Sollner-Webb, B. (1981). Transcription of mouse ribosomal RNA gènes

by RNA polymerase I in vitro and in vivo initiation and processing sites. Cell 27, 165-174.

Moss, T. (2004). At the crossroads of growth control; making ribosomal RNA. Curr. Opin.

Gen. Dev. 74,210-217.

Moss, T., Mitchelson, K., and De Winter, R. F. J. (1985). The promotion of ribosomal

transcription in eukaryotes. Oxford Surveys on Eukaryotic Gènes 2, 207-250.

Moss, T., and Stefanovsky, V. Y. (2002). At the center of eukaryotic life. Cell 109, 545-

548.

O'Sullivan, A. C, Sullivan, G. J., and McStay, B. (2002). UBF binding in vivo is not

restricted to regulatory séquences within the vertebrate ribosomal DNA repeat. Mol. Cell

Biol. 22, 657-668.

Pelletier, G., Stefanovsky, V. Y., Faubladier, M., Hirschler-Laszkiewicz, I., Savard, J.,

Rothblum, L. I., Côté, J., and Moss, T. (2000). Compétitive recruitment of CBP and Rb-

HDAC régulâtes UBF acetylation and ribosomal transcription. Mol. Cell 6, 1059-1066.

Poortinga, G., Hannan, K. M., Snelling, H., Walkley, C. R, Jenkins, A., Sharkey, K., Wall,

M., Brandenburger, Y., Palatsides, M., Pearson, R. B., et al. (2004). MADl and c-MYC

regulate UBF and rDNA transcription during granulocyte differentiation. EMBO J. 23,

3325-3335.

121

Rudra, D., Zhao, Y., and Warner, J. R. (2005). Central rôle of Ifhlp-Fhllp interaction in the

synthesis of yeast ribosomal proteins. EMBO J. 24, 533-542.

Russell, J., and Zomerdijk, J. C. (2005). RNA-polymerase-1-directed rDNA transcription,

life and works. T rends Biochem. Sci. 30, 87-96.

Schnapp, A., and Grummt, 1. (1996). Purification, assay, and propertics of RNA

polymerase I and class 1-spccific transcription factors in mouse. Meth. Enzym. 273, 233-

248.

Smith, D. B., and Corcoran, L. M. (1991). Protein expression. In Current Protocols in

Molecular Biology, F. M. Ausubel, R. Brent, R. E. Kingston, D. D. Moore, J. G. Seidman,

J. A. Smith, and K. Struhl, eds. (New York, Wiley-Interscience).

Smith, S. D., O'Mahony, D. J., Kinsella, B. J., and Rothblum, E. I. (1993). Transcription

from the rat 45S ribosomal DNA promoter does not require the factor UBF. Gène Exp. 3,

229-236.

Stefanovsky, V. Y., Bazett-Jones, D. P., Pelletier, G., and Moss, T. (1996). The DNA

supercoiling architecture induced by the transcription factor xUBF requires three of its five

HMG-boxes. Nucl. Acids Res. 24, 3208-3215.

Stefanovsky, V. Y., Pelletier, G., Bazett-Jones, D. P., Crane-Robinson, C., and Moss, T.

(2001a). DNA looping in the RNA polymerase I enhancesome is the resuit of non-

cooperative in-phase bending by two UBF molécules. Nucl. Acids Res. 29, 3241-3247.

Stefanovsky, V. Y., Pelletier, G., Bazett-Jones, D. P., and Moss, T. (2006). ERK modulâtes

DNA bending and Enhancesome Structure by phosphorylating HMGl-boxes 1 and 2 of the

RNA polymerase I transcription factor UBF. Biochemistry in revision.

Stefanovsky, V. Y., Pelletier, G., Hannan, R., Gagnon-Kugler, T., Rothblum, L. E, and

Moss, T. (2001b). An immédiate response of ribosomal transcription to growth factor

stimulation in mammals is mediated by ERK phosphorylation of UBF. Mol. Cell 8, 1063-

1073.

122

Tschochner, H., and Hurt, E. (2003). Pre-ribosomes on the road from the nucleolus to the

cytoplasm. Trends Cell Biol. 13, 255-263.

Vincent, M., Lauriault, P., Dubois, M. F., Lavoie, S., Bensaude, O., and Chabot, B. (1996).

The nuclear matrix protein p255 is a highly phosphorylated form of RNA polymerase II

largest subunit which associâtes with spliceosomes. Nucl. Acids Res. 24, 4649-4652.

Voit, R., Kuhn, A., Sander, E. E., and Grummt, I. (1995). Activation of mammalian

ribosomal gène transcription requires phosphorylation of the nucleolar transcription factor

UBF. Nucl. Acids Res. 23, 2593-2599.

Wilmanska, D., Czyz, M., Studzian, K., Piestrzeniewicz, M. K., and Gniazdowski, M.

(2001). Effects of anticancer drugs on transcription in vitro. Z. Naturforsch. [C] 56, 886-

891.

Wilsbacher, J. L., and Cobb, M. H. (2001). Bacterial expression of activated mitogen-

activated protein kinases. Meth. Enzym. 332, 387-400.

Zhang, C , Cornai, L., and Johnson, D. L. (2005). PTEN represses RNA Polymerase I

transcription by disrupting the SL1 complex. Mol. Cell Biol. 25, 6899-6911.

Zhao, J., Yuan, X., Frodin, M., and Grummt, I. (2003). ERK-Dependent Phosphorylation of

the Transcription Initiation Factor TIF-IA Is Required for RNA Polymerase I Transcription

and Cell Growth. Mol. Cell 11, 405-413.

123

Chapitre 4 : La perte de la méthylation de l'ADN dans une lignée de cellules de carcinome colorectal entraîne des défauts dans la transformation de l'ARNr et une transcription par l'ARN polymérase II des gènes ribosomaux;

« Loss of CpG méthylation in human colorectal carcinoma cells results in rRNA processing defects and RNA polymérase II transcription of the rDNA»

Thérèse Gagnon-Kugler, Frédéric Langlois, Victor Stefanovky and Tom Moss

Cancer Research Centre and Department of Médical Biology, Laval University, CHUQ Research Centre, Pavillon St Patrick, 9 rue McMahon, Québec, G1R 3S3, QC, Canada.

4.1. Avant-propos Dans cet article, les expériences de ChIP et de Run-On nucléaires ont été réalisées

par Frédéric Langlois avec ma participation. Les expériences de Run-On/ChIP ont été

mises au point par F. Langlois et moi-même. La figure 4.10B présente les résultats

provenant d'expériences faites par nous deux et 4.1 OC par F. Langlois. Les expériences de

marquages du 47S endogène ont été réalisées majoritairement par V. Stefanovsky et en

partie par moi-même. J'ai réalisé les autres expériences, y compris celles présentées dans la

section des données supplémentaires. J'ai aussi fait la quasi totalité de la culture cellulaire.

La transcription ribosomale peut être régulé à différents niveaux. Il a été montré que

l'initiation et l'élongation, chapitre 3 de cette thèse, peuvent être régulées. De plus,

puisqu'il existe un grand nombre de gènes ribosomaux silencieux, il est théoriquement

possible que le nombre de gènes transcrits puisse aussi être régulé. Cependant, des études

réalisées dans notre laboratoire en utilisant des cellules de souris et des cellules humaines

montre qu'une augmentation de la transcription n'entraîne pas de changement dans le

nombre de gènes transcrits (Stefanovsky and Moss, 2006). Ainsi, ce niveau de régulation

124

n'existe pas chez les mammifères, à tout le moins pas en réponse à des facteurs de

croissance. Cet état de fait soulève deux questions. Si les gènes ribosomaux silencieux ne

sont pas transcrits lors d'une augmentation de la transcription, alors pourquoi sont-ils

conservés au cours de l'évolution et comment sont-ils maintenus silencieux? Ce chapitre

est le premier article qui s'intéresse aux mécanismes endogènes qui régulent l'état de

silence des gènes ribosomaux chez les humains. Plusieurs mécanismes ont jusqu'à

maintenant été proposés comme étant impliqués dans l'inactivation des gènes ribosomaux

et dans leur maintien dans un état silencieux, décrits dans l'introduction. Cependant, aucun

de ces résultats ne permet d'expliquer pourquoi une grande partie des gènes sont maintenus

silencieux et pourquoi cette caractéristique ribosomale est conservée au cours de

l'évolution. Nos résultats proposent, pour la première fois, une réponse à cette question

fondamentale dans le domaine de la transcription ribosomale.

4.2. Résumé Le génome haploïde humain contient environ 200 gènes ribosomaux répétés en

tandem et localisés sur cinq chromosomes. Ces gènes sont transcrits par l'ARN polymérase

I (RPI) et la majorité d'entrés eux sont maintenus silencieux par la méthylation des

dinucléotides CpG. Toutefois, la nécessité mécanistique de cette inactivation était

inconnue. Nous montrons dans cet article que la perte de la méthylation de l'ADN, suite à

l'inactivation génique de DNMT1 et 3b dans une lignée de cellules de carcinome colorectal

humain (HCT116), entraîne une réactivation partielle des gènes ribosomaux silencieux. De

plus, cette perte de méthylation permet la transcription des gènes ribosomaux par TARN

polymérase II (RPII) et entraîne un défaut dans les étapes précoces de transformation de

l'ARN ribosomal précurseur. Ces résultats suggèrent qu'une des fonctions principales de la

méthylation des dinucléotides CpG est d'augmenter la densité des complexes RPI sur les

gènes ribosomaux de façon à exclure RPII du locus. Cette étude suggère que les effets

cytotoxiques causés par la perte de la méthylation de l'ADN soient majoritairement dus à

une biogenèse anormale des ribosomes. Ces résultats ont une implication importante quant

à l'utilisation d'inhibiteurs des DNMTs dans le traitement contre le cancer.

125

4.3. Summary The human génome contains around 200 ribosomal RNA gènes arranged as tandem

repeats at 5 chromosomal sites. Thèse gènes are transcribed by a dedicated polymerase,

RNA polymerase I. A majority of the ribosomal gènes are subject to silencing by CpG

methylation. But why this should be so is unknown. We now show that loss of CpG

methylation in a colorectal carcinoma cell Une, by the inactivation of DNA

methyltransferases 1 and 3b, causes a partial loss of ribosomal gène silencing, active

transcription of the gènes by RNA polymerase II and a disruption of early ribosome

biogenesis. The data suggest that a major function of CpG methylation is to prevent RNA

polymerase II ingression by enhancing the density of RNA polymerase I complexes on the

rRNA gènes. They suggest that the cytotoxic effects of loss of CpG methylation are due in

large part to disruption of normal ribosome biogenesis. Thèse data hâve important

implications for the use of DNA methyltransferase inhibitors in the treatment of cancer.

4.4. Introduction CpG methylation of DNA is an essential function in animalia used to ensure

epigenetic silencing of telomeres, X chromosome inactivation and genetic imprinting (Li,

2002). Both hypo- and hyper-CpG methylation hâve been implicated in human disease

(Robertson, 2005; Ting et al., 2006). Three active DNA methyltransferases, DNMT1, 3a

and 3 b, hâve been identified in mammals, of which DNMT1 is predominantly responsible

for maintaining existing CpG methylation patterns, and DNMT 3a and 3b are responsible

for de novo methylation (Hermann et al., 2004). Combined inactivation of the DNMT1 and

DNMT3b gènes demonstrated that they cooperate in epigenetic silencing at several loci and

together are responsible for greater than 95% of DNA methylation in thèse cells (Rhee et

al., 2002; Rhee et al., 2000). Methylation imprinting of both growth and cell cycle

regulatory gènes, including IGF2, P16lNK4A/CDKN2A, TIMP-3, GATA4 and -5 and SFRP1,

hâve been shown to occur in the colorectal çarcinomas (Akiyama et al., 2003; Cui et al.,

1998; Myohanen et al., 1998; Suzuki et al., 2002). In the colorectal çarcinomas cell Une

HCT116, imprinting of the single Wùd-type pl6/NK4A/CDKN2A allele is believed to permit

an enhanced prolifération rate (Rhee et al., 2002). Thèse and other studies hâve lead to the

126

suggestion that DNMT inhibitors may represent an important source of new anti-cancer

drugs.

A less well-studied function of DNA methylation in this context is the silencing of

the ribosomal RNA gènes. The haploid human génome contains about 200 ribosomal RNA

(rRNA) gène units arranged in tandemly repeated arrays at the secondary constrictions of

acrocentric chromosomes 13, 14, 15, 21 and 22 (Henderson et al., 1972; Schmickel, 1973).

Thèse gènes are responsible for the production of 80% of cellular RNA, are essential for

ribosome biogenesis and are regulated in précise coordination with growth (Stefanovsky et

al., 2006; Stefanovsky et al., 2001), see (Grummt, 2003; Moss, 2004; Moss et al., 2006;

Moss and Stefanovsky, 2002) for reviews. A large proportion, if not the majority of the

human rRNA gènes has been found to be silent (Haaf et al., 1991; Stefanovsky and Moss,

2006), and some of thèse gènes are found to be heavily methylated (Brock and Bird, 1997;

Ghoshal et al., 2004; Majumder et al., 2006). In other animal and in plant Systems rRNA

gène silencing has been shown to dépend at least in part on DNA methylation status

(Grummt and Pikaard, 2003). In particular, the induced loss of DNA methylation has been

shown to suppress nucleolar dominance (Lawrence et al., 2004), and in vitro methylation of

the mouse rRNA gène promoter has been shown to silence a linked transgene (Santoro and

Grummt, 2001). However, the biological significance of rRNA gène silencing is uncertain.

Since drug thérapies employing DNMT inhibitors will inevitably target rRNA gène

methylation and hence affect silencing, it is essential that we understand the biological

conséquences. Hère we hâve used HCT116 derivative cell Unes in which the DNMT1 and

DNMT3b gènes hâve been inactivated to détermine the biological conséquences of loss of

CpG methylation on the expression of the rRNA gènes and ribosome biogenesis.

4.5. Results Previous data hâve shown that CpG methylation of a single site (-133) within the

Upstream Promoter Elément (UPE, also called the UCE) of the mouse RNA polymerase I

(RPI) promoter is suffïcient to détermine the silencing of reporter constructs (Santoro and

Grummt, 2001). While only two CpGs occur within the mouse promoter, the human RPI

promoter (-155 to +20) contains some 22 CpG sites. Thus, we used bisulphite mapping

127

(Clark et al., 1994) in order to détermine the methylation status of the RPI promoter in the

DNMT1T and 3b7" double knock-out (DKO) and parent strain (PS) HCT116 cells (Rhee et

al., 2002). The methylation status at 58 CpG sites within a 410bp région encompassing the

promoter was determined (Figure 4.1). On this basis, the rRNA gènes in the PS cells could

be classified into one of two groups, either near fully methylated or fully unmethylated.

50% (18 of 35) of the gènes showed a very high level of methylation at ail 58 CpGs (94%

average, with no site less than 60%) while the other 50% (17 of 35) of gènes showed

essentially no methylation (0.2% average, only two gènes showing single site

methylations). Thus, CpG methylation provided an unequivocal epigenetic marker of a

distinct subset of rRNA gènes in thèse cells. Inactivation of both DNMT1 and 3 b in the

DKO cells essentially eliminated ail CpG methylation, (0.5% average, 7 of 24 gènes

analyzed showed single site methylation, but only two of thèse sites lay within the RPI

promoter and neither was in the UPE).

A)

PS

DKO

B ) n-18 100

8 0

PS-M

6 0

40

20

n=17 «H

E&Un o-|

0100% § 85-95% 0 50-84% ()1-10% methylation 306bp -199bp -84bp

' HIIHIIHHH m I1WIIII Jil M II +1 +29bp

■n=17

306bp 199bp

UPE

UPE

i—t -n=24

4bp

Core

Core

f29bp

DKO n=24

a CpG

Figure 4.1 : CpG methylation status across the human RPI promoter. A) The distribution of CpG motifs between -406 to +31bp. Data for the hyper- and hypo-methylated gènes analyzed from the PS and ail analyzed gènes from the DKO cell Une are

128

shown independently, and each site is shown shaded to indicate its statistical degree of methylation. B) The data in A) is shown in the form of a histogram indicating the exact percentage of methylation observed at each site for the three gène catergories. In both A) and B), M refers to the highly methylated subset of gènes and Un to the unmethylated subset of the PS and n indicates the number of gènes analyzed. Methylation data is derived from early passage DKO cells. UPE, Upstream Promoter Elément (also known as the UCE); Core, the initiation site proximal promoter élément.

A differential susceptibility to psoralen crosslinking distinguishes actively

transcribed rRNA gènes from inactive ones, respectively "a" and "i" in Figure 4.2B

(Conconi et al., 1989; Lucchini and Sogo, 1998). In the PS cells, around 20% of gènes were

found to be hyper-accessible to psoralen and hence to be actively transcribed (Figure 4.2C).

Though this fraction is somewhat lower than we and others hâve estimated in rodent cells

(Conconi et al., 1989; Stefanovsky and Moss, 2006), it is consistent with psoralen-based

estimâtes of active gène number in the SKF human neuroepithelioma cell Une and with

earlier immunofluorescence-based estimâtes in human peripheral blood lymphocytes (Haaf

et al., 1991; Stefanovsky and Moss, 2006). As expected for a strict corrélation with active

transcription, psoralen hyper-accessibility was limited to the transcribed, 47S rRNA coding

région, the Intergenic Spacer (IGS) being uniformly hypo-accessible and showing no sign

of double banding (Figure 4.2A and B). Inactivation of DNMT1 or 3b alone had little effect

on the active gène number. However, combined inactivation of DNMT1 and 3b in the DKO

cells increased the number of actively transcribed gènes by over two fold to 50%. This

effect was not due to a change in rRNA gène copy number, which did not vary between the

various cell lines (±5%). Thus, DNMT1 and 3b activity was required to maintain the

silencing of around 30% of the rRNA gènes. However, it was striking that around 50% of

the gènes remained transcriptionally inactive even in the DKO cells. Even if, as in mouse, a

single methylation within the human promoter were to be sufficient to détermine silencing,

residual CpG methylation could not explain why 50% of gènes remain silent in the DKO

cells, since only 8% of promoters (2 of 24) contained single site methylations.

Consistent with the increase in active gène number in the DKO cells, the rRNA

gène chromatin displayed an increase in the level of histone H4 tetra-acetylation (H4ac), a

marker of active chromatin, and réduction in the level of methylation of lysine 9 on histone

H3 (H3-K9me), a marker of heterochromatin, (Jenuwein and Allis, 2001; Stancheva, 2005),

129

(Figure 4.3B and C). Interestingly, a significant increase in the recruitment of the rRNA

gène chromatin factor UBF was also observed in the DKO cells (Figure 4.3D), suggesting

that this factor may be spécifie for the active gène fraction. Thus, complète loss of CpG

methylation lead to a 250% increase in the number of active rRNA gènes and of active

rRNA gène chromatin, but was clearly not required for the silencing of the other 50% of

rRNA gènes.

A)

Probes &~

IGS i l

ri Fragments: i 2 5 i 4.9 i A2,2 i ^ 5.1 i

-8047(A) | -513(A) I |+5296(E) j

»)

-5498(A) +2087(B) +4245(B) +9347(B)

+ Psoralen

PS PS 1-/- 3b-/- DKO-E DKO-L

47S

^ ^ ^ _ ^ ^ g*m* ^ u

*mm <^^m ^^m ' ^ ^ ^ ^^r ^ ^ *

région 2.2 kb ^ ^ ^ ^ i àg^ i m | ^ ^ ^

W ^ ^ * ^^m ^ ^ ^ ^ P V^w

IGS

4.9 kb ij&|iÉiifôi' ■rfiàlià.';,: :ij*(ilfi M ,

jHH| ÊÊÊÊL ÉM É M a ^g||b m p ^ ^ » w*^^ ^ ^ F ^^^^^ ^ ^ p

région 2-5Kb ^ JJ^J. JH IL jÉ|Hb J | Ù ^ ^ ^ ^ ^^u

«i n H '-: PI C)

Wl 1-/- 3b-/- DKO-E DKO-L

Figure 4.2 : Active gène count in the PS and DNMT inactivated cell Unes. A) Map of the human rRNA gènes indicating the analysed fragments and the probes used. The numbers indicate the restriction sites for Apol (A), BamHI (B) and EcoRI (E) relative to the 47S rRNA initiation site. In most cases the probes and detected genomic fragments coincided, but for the 5.1 BamHI-BamHI fragment the probe was shorter (+4245 to +5296bp) and is indicated in grey. B) Psoralen crosslinking analysis of the genomic

130

fragments indicated in A). The left hand PS (parent strain) track shows the genomic DNA fragments before crosslinking and the other five tracks after crosslinking. "a"; active and "i" inactive gènes. C) Quantitative analyses of the cumulative psoralen crosslinking data from 2 analyses for each probe, that is 4 independent data sets. In both B) and C) 17" and 3bT refer to the corresponding DNMT inactivated cell Unes and DKO to the DNMT17" and 3b7" inactivated Une. DKO-E indicates early passage and DKO-L, late passage cells, see Materials and Methods.

5.8S

A)

-47»|U-32Prom. 952-JL-1030

47S-5' 3990» ♦ 4 0 9 2 8204 ■» U-8300

18S 28S-S' Amplicons

12855~| |*12970 27366 » | |*27477 IGS

28S-3'

47S-5' 18S 28S-5' 28S-3' Amplicon

47S-5' 18S 28S-5' 28S-3' Amplicon

47S-5' 18S 28S-5' 28S-3' Amplicon

Figure 4.3 : rRNA gène chromatin composition in the présence and absence of CpG methylation. A) Map of amplicons. B), C) and D) show average quantitative ChIP data respectively for tetra-acetyl H4, H3-K9 tri-methyl and UBF derived from 3 to 5 independent experiments each performed in triplicate. The data is shown relative to H3 levels and normalized to the recovery for the 28S-5' amplicon.

Despite the significant increase in the number of rRNA gènes, the prolifération rate

of the DKO cell Une was very significantly reduced as compared to the parent strain

(average doubling time of 38 h for DKO versus 27 h for PS), in agreement with previous

data (Rhee et al., 2002), (Figure 4.4A). However, the DKO cells showed no sign of cell

cycle arrest, the cell cycle distribution differing only slightly from that of the PS, (<10%

increase in Gl, Figure 4.4B) and no réduction in cell volume was detected, (data not

shown). Despite this, the total RNA, and hence the ribosome, content of the DKO cells was

only 60±3% of that in the parent strain (Figure 4.4C). This suggested that an unexpected

resuit of the loss of DNA methylation was to limit ribosome biogenesis.

131

The rate of synthesis of 47S rRNA in the DNMT1 or DNMT3b deleted strains was

found to be about 60% of that in the PS, while in the DKO cells synthesis was reduced to

around 35%, roughly proportionate with the réduction in the prolifération rate of thèse cells

(Figure 4.5A). However, ChIP analyses showed that the number of RPI molécules engaged

in transcribing the rRNA gènes of the DKO cells was only slightly reduced (-80% of the

PS, also confirmed by nuclear Run-On, data not shown) and did not explain the much

larger réduction in 47S synthesis (Figure 4.5B). We hâve recently shown that 47S synthesis

is regulated at the level of the RPI elongation rate (Stefanovsky et al., 2006). Consistent

with this, the elongation rate in the DKO cells was found to be around 35% of that in the

parent strain, (2.5 min per 47S transcript in the PS versus 7 min in the DKO cells),

explaining the reduced rRNA synthesis in thèse cells (Figure 4.5C).

A) H)

to 4 < >

* 3

ii) O

-e -PS - B - DNMT1 - e - DNMT3b - * - DKO-E -&- DKO-L

v/ A

5 2 4 Time,

è i days

!

C)

G1 26.7

PS

S 57.2%

LM G2 6.1%

28.3 DNMT 3b-/-

58.2 8 ^ 1 3 . 5

K.n-vMffllM. I

33.6 DNMT

1-/-

51.7

^ 1 4 . 6 .*«.—IBM.

39.6 DKO-E

50.8

9.6

DNA content

PS DKO-E

Figure 4.4 : DKO cells display a reduced prolifération rate and an RNA déficit but a normal cell cycle distribution. A) Prolifération rate of the PS and derivative strains. B) Cell cycle distribution détermine by FACS analysis. C) RNA content of the PS and DKO-E cell Unes on a per cell basis.

The DKO cells also displayed a drastically différent nucleolar morphology. While

the PS cells displayed tightly compacted UBF, RPI positive Fibrillar Centres and adjacent

fïbrillarin positive Dense Fibrillar Centres of active nucleoli (Figure 4.6A) (Moss et al.,

132

2006), in the DKO cells thèse were dispersed and thread-like and traversed large régions of

the nucleus. However, the total nuclear area occupied by the nucleoli of the DKO cells was

significantly smaller than that of the more rapidly growing parent strain, as would be

expected from their slow growth and lower synthesis rates (Figure 4.6B).

Figure 4.5 : rRNA gène transcription rates and RPI loadings in the PS and cell Unes and its derivatives. A) The rates of 47S rRNA synthesis determined by metabolic ([3H]-uridine labeling). Upper panel show an example of 47S labeling and the lower panel the corresponding 18S rRNA loadings. Below the mean data from three independent experiments each performed in duplicate is shown normalized to the PS cells. B) Quantitative ChlP analysis of RPI loadings on the rRNA gènes of the PS and DKO-E cells. The data is derived from 7 independent experiments and is shown normalized to the chromatin recovery in the PS cells

133

at the 28S-5' amplicon, see Figure 4.3A. C) Elongation rate measurements. The PS and DK.O-E cells were puise labeled for increasing times and 47S synthesis determined. The upper panel shows the data from 3 independent experiments in duplicate, each normalized to steady state 47S incorporation in the PS cells. In the middle and lower panels the data sets for the PS and DKO cells are shown each normalized to their corresponding 47S steady state levels. The différence in elongation times in the two cell lines can be determined from the horizontal shift between the uptake curves. The lower panel simply shows an enlargement of the middle panel allowing an estimate of the absolute elongation rates from the intercepts on the time axis, see (Stefanovsky et al., 2006) for a détail explanation.

«J RPI Fib DAPI Merge

PS B DKO ■ ■

PS DKO DKO

Figure 4.6 : The DKO cells display a distended nucleolar structure. A) Typical immunofluorescence staining of endogenous RPI, UBF, and fibrillarin (Fib) and of DNA (DAPI) in the PS and DKO-E cell lines. In each case the right-hand images (Merge) show superimpositions of the three corresponding left-hand panels. B) Total nucleolar area per cell estimated from endogenous UBF immunofluorescent images. The boxed area encloses the upper and lower quartiles relative to the médian value and the bars extend to the upper and lower limits defined by 1.5 times the interquartile distance from the médian or to the outer data points.

We detected no depletion of RPI transcription factors in the DKO cells, levels of

TAF,48 and TBP, (subunits of the RPI-specific TBP-complex SL1), UBF and the RPI

subunits RPA194, RPA135, PAF53 and Rrn3 ail being closely similar to those in the PS

134

(Figure 4.7). Further, the ribosome assembly protein nucleolin and the two ribosome

proteins, RPL28 and RPL39L, ail showed no détectable réduction in the DKO cells.

Microarray analysis of mRNA levels also revealed that only a small number of gènes were

affected when comparing the DKO and PS cell Unes, and none of thèse hâve been

implicated in ribosome biogenesis. Of 122 gènes down-regulated in the DKO cells, the

largest group (25 gènes) was implicated in cell cycle régulation and cell death and included

bcl2ll and box (both 0.3 of PS), while, of 61 up-regulated gènes, 22 were implicated in

growth and prolifération and included the oncogene myc (4 times PS). Levels of p53

protein, which is wild type in HCT116, were also significantly reduced in the DKO as

compared with the PS cells (Figure 4.7). Thus, key protein levels and the genome-wide

expression profile in the DKO cells were consistent with the active growth and cell cycle

distribution.

DKO PS DKO

«MM» Rpl28 ,

« • * Rpl39L « M » W

■'m p53 tfJMfc '*"""

——~ Yx-tubulin mm» 4nHP

Figure 4.7 : RPI transcription factor and nucleolar protein concentrations in PS and DKO cells. Cell extracts were probed for a range of proteins relevant to ribosome biogenesis.

When the steady state 47S rRNA levels were détermine by Northern, we were

surprised to find that thèse were identical (±10%) in ail the cell Unes, regardless of

synthesis rates (Figure 4.8B). Even the DKO cells, where transcription was some 65 %

lower than the PS, showed little or no réduction in the 47S rRNA pool. Quantitative Sl-

mapping also showed little différence in the levels of bulk 47S rRNA between the PS and

DKO cell Unes (Figure 4.8C). Thèse data suggested that 47S processing was significantly

slowed in the DKO as compared to the PS cells. This was confïrmed when we observed

that the labeled 47S rRNA pools in the DKO and PS cells approached unity as the puise

time was increased (Figure 4.8D), Subséquent pulse-chase experiments, however, revealed

PS DKO PS

Nucleolin^i«ir««ÉW» Rrn3

UBF « t t * « M * RPA194

TBP «MJ» «EH»' RPA135

TAF48i|iP*«' ■immv PAF53

135

that rather than processing being uniformly slowed in the DKO cells, thèse cells contained

two distinct 47S fractions. The major DKO fraction was processed at a very similar rate to

47S in the PS cells (ti/2 = 157±17 min versus 148±23 min for the PS), but a minor fraction

(15±3%) was processed (or perhaps degraded) much more slowly (estimated tin >20h)

(Figure 4.9A to C). Due to its long half-life, this fraction accumulated over time in the

DKO cells, and hence made up the major component (-65%) of the steady state 47S rRNA

pool. We also found that the DKO cells displayed a 20 to 25% molar déficit of newly

synthesized 28S rRNA, suggesting that rRNA processing problems were not restricted to

the primary precursor level (Figure 4.9D). An early step in 47S rRNA processing is its

methylation, which occurs soon after transcription (Maden, 1988). We found that uniike the

newly synthesized 47S rRNA, the methylated 47S fraction was rapidly processed with

similar kinetics in PS and DKO cells (Figure 4.9E). Further, the ratio of methylated 18S

and 28S rRNA was closely similar in both cell types (data not shown). This strongly

suggested that the slowly processed 47S component in the DKO cells was predominantly

unmethylated. Hence, the rRNA processing defects in thèse cells very probably derived

from errors preceding methylation and could hâve been transcription coupled.

A) +1 B)

h47S (-70.+230)

•//• PS 1-/- 3b-/- S S C)

Beb(-504,+1647)

Bbh (+2085.+2923)

D) 1.2- 1 W /

1.0- ^1-®' ^*<^ £0.8- yf^ co rS^

S M" y :* r/ ?» 04 -t c~ J ■ *

0.2- -

0 - > r-//—A 0 5 10 Steady Pulse-Labeling time, h State

h47S

GADPH

Bbh

GADPH

Bob

GADPH

DKO/PS:-

«mip ^ ^ w ^^^w~ ^ P ^ ^ ^^w

■^^^^ ^^^* imiP T^iff ^^p^

^ ^ ^ _ ^ ^ ^ » ^ ^ ^ M ^ ^ ^ ^

W^P ' P?7 w™^ 1PHF ;^H>

JÊ^ ■ÉJÊk m | mm. ^ ^ F S P P ^ H F ^IK T(P)r

1.0 1.0 1.0 1.0 ±0.1 ±0.1 ±0.1 ±0.2

* #° 9 2 8 b * <É» *

6 2 2 b *

5 2 7 b *

47S + 1 * *» m» 404b- *

3 0 9 b *

DKO/PS:- 1 0.93 ±0.02

Figure 4.8 : 47S rRNA turn-over is affected by loss of DNA methylation. A) and B) Northern blot analyses of 47S rRNA levels in the PS and derivative cell Unes. The probes used are shown in A), while the blots for each of the probes and for the

136

GADPH mRNA control are shown in B). The estimâtes of relative 47S levels are from four independent experiments, three in duplicate. C) S1 mapping in the région surrounding the 5' of the 47S rRNA. The upper band represents undigested probe DNA. D) Metabolic labeling of 47S rRNA for increasing puise times from 30 min to 8h. Data is shown as a ratio between incorporation in 47S in the DKO-E and PS cell Unes. The steady state 47S levels were determined from Northern blot as in B). An exponential curve fit to the data is shown.

A) - * PS ► ~+ DKO-0 0.5 2 4 6 8 0 0.5 2 4 6 8h

47S

32S m 28s m

18S

;•$#•'••us

1 " 1 1 T^ 0 2 4 6 8 10

Chase time, h

i i i 0 2 4 6 8 10

Chase time, h

T " 2 4 6 8

Chase time, h

i i

1 h pHl-methionine puise - © - P S - S - D K O

, ^ » P ■ 8 ■ $ 1 2 3 4 5 6

Chase time, h

Figure 4.9 : Processing of the 47S rRNA is disrupted in the DKO cells. A) Typical pulse-chase experiment, a 3h puise labeling is shown. B) and C) The raw 47S rRNA incorporation curves for the 1 and 3h [3H]-uridine pulse-chase experiments. Each data set is the mean of 2 labelings performed in duplicate. Data was fitted to the exponential decay function cpm = a + b*exp(-c*t) were "a" would be the fraction of 47S rRNA subject to slow processing/degradation. D) The 28S to 18S ratio for the

137

accumulation of thèse rRNAs during the chase-time after a 3h [ H]-uridine puise labeling. The 18S rRNA is processed and exported more rapidly that the 28S, hence the lower ratios at early time points. The curve fit was made to an exponential rise function. The expected steady state ratio (Expected, 1.88) was calculated as the ratio of uridines in 28S rRNA (750) to that in 18S rRNA (400). E) 47S rRNA was labeled with [3H-methyl] methionine for lh and then chased for 6h. The [3H] incorporation inlo 47S rRNA for both the PS and DKO cells was normalized to the zéro time point and the curve fit was as in B) and C).

We had observed a 250% increase in the number of active rRNA gènes in the DKO

cells, but little or no réduction in the total RP1 loading. Thus, the DKO cells had less than

half the number of RPI transcription complexes per gène as compared to the PS cells. In

yeast, active RPI transcription of the rRNA gènes is in large part responsible for the

exclusion of RNA polymerase II (RPII) from this locus, a phenomenon referred to as rDNA

silencing (Cioci et al., 2003). It was, therefore, possible that the decrease in density of RPI

transcription complexes permitted ingression of RPII. ChIP assays for the elongating form

of RPII revealed a significant signal throughout the rRNA gènes of the DKO cells, while

the PS cells displayed only a signal indistinguishable from background (Figure 4.10A). To

détermine if the RPII signal truly represented active transcription of the rRNA gènes, we

performed an RPII spécifie ChIP after a nuclear Run-On reaction (Run-On/ChIP, see

Materials and Methods) and then quantitated the immunoprecipitated rRNA gène

transcripts by slot-blot hybridization (Figure 4.10B). While the signal from the PS cells was

close to background, that from the DKO cells was some 6 to 8 fold higher. The inclusion of

a-amanitin, a spécifie inhibitor of RPII, completely eliminated this signal, confirming that

RPII actively transcribed the rRNA gènes in the DKO cells (Figure 4.10C). This to our

knowledge is the first observation of loss of ribosomal DNA silencing in any organism

other than yeast.

4.6. Discussion Silencing of a large fraction of the rRNA gènes is a well documented phenomenon.

However, why a cell should need to silence thèse essential gènes has for many years been

unknown. Hère we show that rRNA gène silencing is essential to prevent RPII from

transcribing thèse gènes. Loss of CpG methylation leads to a large increase in the number

of active gènes in human colorectal carcinoma cells and the active transcription of thèse

138

gènes by RPII. Our data suggest that loss of methyl-CpG silencing opens up a larger

number of gènes to a limited RPI transcriptional machinery, causing a dispersion of this

machinery over a larger gène pool and hence a réduction in the density of RPI transcription

complexes on thèse gènes. The less dense packing of RPI complexes then allows the

ingression of RPII and the aberrant transcription of the rRNA gènes by this polymerase.

Prom. 47S-5' 18S 28S-5' 28S-3' IGS Amplicon

I!)

PS rDNA

DKO #«» — *

PS DKO

C) E rDNA *i. .

PS + PS -DKO + DKO-

6

,y^* ' , ;

"& ri

3-

a-Am. + PS DKO

Figure 4.10 : RPII silencing is lost in the DKO cells. A) RPII loadings on the rRNA gènes of the PS and DKO cells determined by quantitative ChIP assays using a phospho-CTD spécifie antibody. The respective amplicons, 47S-5', etc, are as indicated in Figure 4.3A. The data are mean of 7 experiments and are shown normalized to the 28S-5' signal for the PS. B) Run-On/ChIP assays. After standard nuclear Run-On, phospho-RPII associated chromatin was immunoprecipitated, the nascent transcripts isolated and then hybridized to the 2.2kbp BamHI-BamHI rRNA gène fragment

139

shown in Figure 4.2A, see Materials and Methods. "Vector" refers to the négative control plasmid DNA. Upper panel shows a typical phosphoimage of a slot-blot hybridization and lower panel the quantitation from two experiments in duplicate normalized to the signal for the PS cells. C) Run-On/ChIP was performed as in B), but the RPII inhibitor a-amanitin was included in duplicate Run-Ons. Data show the mean of two experiments in duplicate normalized to the data for the PS cells in the présence of a-amanitin.

Ribosome biogenesis starts with the transcription coupled assembly of a large

number of proteins onto the nascent 47S rRNA, leading to the formation of a 90S ribosome

precursor. We were surprised to find that an important resuit of loss of CpG methylation is

the disruption of an early step in this process, leading to the aberrant accumulation of

unprocessed and unmethylated precursor rRNA. This disruption occurs during or soon after

47S rRNA synthesis, suggesting that it may affect transcription-coupled processes. Since

loss of CpG methylation also leads to RPII transcription of the rRNA gènes, it is possible

that the aberrantly processed rRNAs are actually RPII transcription products. In support of

this possibility, it has been shown in yeast that RPII is able to produce functional precursor

rRNAs from a cryptic IGS promoter (Siddiqi et al., 2001). However, SI analyses of total

DKO cell RNA failed to detect any aberrant initiation within a région of one kilobase

surrounding the normal 47S initiation site. An alternative scénario is that heterogeneous

transcription of the rRNA gènes by RPII, interfères with assembly of proteins onto the

nascent 47S rRNA either by locally competing for thèse proteins or by affecting the folding

of adjacent nascent RPI transcripts.

4.7. Acknowledgements We would like to thank B. Vogelstein for making the HCT116 derivative cell lines

available to us, K. Schuebel for advice and DNMT constructs, A. R. Macleod for AS-

RNAs, L. I. Rothblum and A. Cavanaugh for RPA194, RPA135, UBF, PAF53 and Rrn3

antibodies, J. Coté and A. Nourani for advice on appropriate antibodies for ChIP, and the

Clinical Interface and DNA Microarray Units of the Cancer Centre for FACS and DNA

microarray analyses.

140

4.8 Material and methods

4.8.1 Cell Culture. The HCT116 colorectal carcinoma cell Une (parental strain, PS) and the DNMT1T,

DNMT3bT and the double knockout DNMT1T, DNMT3b7" cell Une (clone DKOl) were

obtained from B. Vogelstein (Rhee et al., 2002; Rhee et al., 2000) and were grown in

McCoy's 5A médium (Invitrogen) complemented with 10% FBS sérum (Wisent) and

Antibiotic/Antimicotic (Wisent). In the case of the DKOl clone, experiments were

performed with passage numbers from 15 to 25, refered to in the text as DKO-E, and

passages 30 to 43 refered to as DKO-L.

4.8.2 Psoralen crosslinking and Southern blot. Psoralen crosslinking was performed as described elsewhere (Conconi et al., 1989;

Lucchini and Sogo, 1998). Cells were grown in 60 mm pétri dishes (0.8 x 106 cells/petri)

and washed once with serum-free médium before adding 1.5ml ice-cold serum-free

médium immediately before crosslinking. 1/20 volume (75ul) of 200iig/ml Trioxsalen (4,

5',8- trimethylpsoralen, Sigma) in methanol was added and after 5 min. incubation, the

cells irradiated for 5 min. with a 366nm UV lamp (BlackRay model B-100A, 440W) placed

at a distance of 6-7cm. Crosslinking was repeated four more times, each time adding fresh

Trioxsalen. The crosslinking procédure was performed in a dark room and the cells were

maintained throughout on ice. Subsequently, the cells were washed with PBS and lysed in-

petri with 0.5 ml of lysis buffer (lOmM Tris, pH 7.5, 50mM NaCl, 25mM EDTA, 2% SDS)

and transferred to an Eppendorf tube. Pétri dishes were rinsed with 0.5 ml lOmM Tris-HCl,

1 mM EDTA, pH 7.5 (TE), this combined with the cell lysate, lmg/ml Proteinase K (from

lOmg/ml stock, Sigma) was added and the combined lysate incubated overnight at 50C,

adding fresh Proteinase K after the first 2 hrs. The genomic DNA was then deproteinized

twice with phenol/chloroform, ethanol-precipitated and resuspended in 250ul TE, pH7.5,

0.1% SDS. 20ug of RNAse A (Sigma) was added and after 30 min at 37°C, 0.5mg/ml

Proteinase K was also added and incubation continued for another lh at 50 C. After twice

extracting with phenol/chloroform the genomic DNA was ethanol precipitated, resuspended

in 300 ul TE and quantified by EtBr staining on an agarose gel. About lOug of genomic

141

DNA was digested overnight cither with Apo I or BamHI, deproteinized, ethanol

precipitated and finally resolved on a 1% TEA agarose gel at 2V.cm"' for 18 hr in the

absence of EtBr. The gel was subsequently EtBr-stained, photographed, the crosslinks

reversed by UV irradiation of 4000J.cm"2 at 254nm in a UVC 500 crosslinker (Hoefer),

transferred on a Biodyne B membrane (Pall) and the membrane crosslinkcd (70J.cm"2). o

Hybridization was performed overnight at 65 C with random-primed labeled DNA

(Rediprime II DNA labelling System, GE Healthcare) in 250mM NaxHyP04 pH 7.2, 7%

SDS, 225u.g/ml salmon sperm DNA. Subsequently membranes were washed five times (30

min each time) with 20mM NaxHyP04 pH 7.2 and 5% SDS and two times with 20mM

NaxHyP04 pH 7.2 and 1% SDS. Data were revealed by phospho-imaging on a STORM 860

and analyzed with ImageQuant software (GE Healthcare).

4.8.3. Sodium Bisulfite Genomic Sequencing. Sodium bisulfite genomic sequencing was performed as described (Clark et al.,

1994). Genomic DNA from PS and DKO cells was digested over night with EcoRI,

phenol/chloroform extracted, ethanol-precipitated and resuspended in 1/10 TE. The DNA

was then denatured in freshly prepared 0,3M NaOH for 15 minutes at 37°C. Sodium

bisulfite and hydroquinone were added to final concentrations of 3.1 M and 0,5mM

respectively from stock solutions of sodium bisulfite (3,6M) and hydroquinone (lOmM)

prepared immediately beforehand. The samples were then incubated at 55°C for 16 hours.

The treated DNA was purified using the Wizard DNA clean-up system (Promega)

according to the manufacturer's instructions and eluted in H2O, denatured with fresh 0,3 M

NaOH and ethanol-precipitated. PCR amplification was achieved using the noncoding

strand primer hBFl.for; (-406pb) GATTTTTTTTGGAGAGTTTT (-387pb) and the coding

strand primer hBFl.rev: (+54pb) CATCCTAAAACCCAACCTCTCCA (+31pb). The

amplication cycle was 94C, 2min. foliowed by 5 cycles of: 94°C, 30s; 49C, 30s; 72°C,

lmin.; then 25 cycles of: 94°C, 30s; 49°C, 30s; 72°C, 30s; then 72°C, 6min. The amplified

products were then cloned in pBSKS+ using HindIII linkers and sequenced.

142

4.8.4. Prolifération rate assay and FACS analyses. For prolifération rate assays 50 x 103 cells of PS and DKO cells were seeded in

triplicate in 100 mm pétri dish. For FACS analyses 250 x 103 cells of PS and DKO cells

were seeded in 150 mm pétri dish. Cells were collected at days 2, 4, 6 and 8 and viable

cells, as assayed by trypan blue exclusion, counted in a Bright-line hemocytometer

(Reichert-Jung). For FACS analyses, cells were stained with propidium iodine and

analyzed in an Epie Elite ESP flow-cytometer (Beckman Coulter). Data analysis was

performed using the Muticycle software from Phoenix Flow Systems Inc.

4.8.5. In vivo puise labeling, RNA extraction and analysis. PS (5.5 x 105cells/petri dish) and DKO cells (7.2 x 105 cells/petri dish) were seeded

in 60mm pétri dishes 16h before labeling. Labeling was performed by adding 10 uCi [3H]-

uridine (Perkin Elmer)/petri dish or 10 |iCi [methyl- H] methionme (Amersham)/petri dish

and left at 37°C for 30 min., 1, 3, 6 or 9 hours as indicated. For pulse-chase experiments,

the médium was replaced with unlabeled médium containing 0.5 mM uridine (Sigma) or

0,5 mM methionine (Sigma) and incubation continued for the times indicated. RNA was

extracted in 1ml Trizol (Invitrogen) according to the manufacturer's protocol. 2-4 jag of

RNA was loaded onto a 1% formaldehyde gel and electrophoresed overnight at 2.5V/cm in

MOPS running buffer. The EtBr stained gels were subsequently photographed, UV-

irradiated for 5 min at maximum energy in a UV-crosslinker (Hoefer) and transferred to

Biodyne B membrane (Pall) by a Vacu-Blot transfer (Biometra) at 200 mbar for 90 min in

2 x SSC. The membrane was UV-crosslinked at 70 J.cm"2, washed in water, dried, sprayed

with ENHance (Perkin-Elmer) and exposed to Kodak-XAR-5 film at -80 °C for 24 to 72 h.

The [3H]-labeled bands were eut from the membrane and counted in ScintiVerse (Fisher)

and external standard corrected counts normalized to 18S rRNA band intensities measured

using ImageJ (NIH).

4.8.6. RNA meaurements. PS (4.1 x 105 or 5.5 x 105cells/petri dish) and DKO cells (5.4 x 105 or 7.2 x 105

cells/petri dish) were seeded in 60mm pétri dishes the day prior the experiment. Three

143

pétris were prepared for each seeding density for both cell Unes. The cells were trypsinized

and a sample was count in a Z2 Coulter Particle Count and Size Analyzer (Beckman

Coulter). The remaining cells were centrifuged and resuspended in 1ml of Trizol

(Invitrogen) for RNA extraction according to the manufacturera instructions. In vitro

transcribed 3H RNA was added with the chloroform during the phenol/chloroform

extraction to control for recovery. RNA amounts were dose in a UV absorbance

spectrometer and 3H RNA recoveries estimated by scintillation counting in ScintiVerse

(Fisher).

4.8.7. Northern blot. PS (1.7 x 106cells/petri dishe) and DKO cells (2 x 106 cells/petri dish) were seeded

in 100mm pétri dishes. 16 h later RNA was extracted in 1ml Trizol (Invitrogen) according

to the manufacturer's protocol. 2-4 |ig of RNA was loaded onto a 1% formaldehyde gel and

electrophoresed overnight at 2.5V/cm in MOPS running buffer. The EtBr stained gels were

subsequently photographed, UV-irradiated in a UV-crosslinker (Hoefer) for 5 min at

maximum energy and transfered to a Biodyne B membrane (Pall) by a Vacu-Blot transfer

(Biometra) at 200 mbar for 90 min in 2 x SSC. The membrane was UV-crosslinked at 70

J.cm"2. Hybridization was performed overnight at 65 C with random-primed labeled probe

DNA (Rediprime II DNA labelling system, GE Healthcare) in 250mM NaxHyP04 pH 7.2,

7% SDS and lOmM EDTA. Subsequently membranes were washed at 65 C four times (30

min each time) with 20mM NaxHyP04 pH 7.2, 5% SDS followed by two washes with

20mM NaxHyP04 pH 7.2, 1% SDS. Data were revealed by phospho-imaging on a STORM

860 and analyzed with ImageQuant software (GE Healthcare). Probe 1 : -70 to +230, probe

2 : -500 to +1446, probe 3 : +2086 to +4244. RNA amounts were normalized to 18S (4

experiments) or GADPH (1 experiment).

4.8.8. Western blot. Proteins levels, in total or nuclear fractions, were analyzed by standard procédures.

Protein loadings were normalized using the DC Protein Assay (Biorad) and where possible

compared to gênerai cell proteins such as a-tubulin. The antibodies used were as follows;

144

UBF, RPA194, RPA135, PAF53 and Rrn3 (kindly provided by A. Cavanaugh and L.

Rothblum), Rpl39L (clone 4A8-1B6, Abnova, cat #H00116832-M01), fibrillarin (Covance,

cat#MMS-581S), Rpl28, nucleolin, TAF(48 and TBP (Santa Cruz, cat #sc-14151, #sc-8031,

#sc-6572, #sc-273), p53 (ab-6, Oncogene) and a-tubulin (B-5-1-2, Sigma T5168).

4.8.9. Microarrays. PS (1.7 x 106 cells/pctri dish) and DKO cells (2 x 106 cells/petri dish) were seeded

in 100mm dishes. RNA was extracted 16h later using Trizol (Invitrogen) according to the

manufacturera protocol and treated with DNase I. The quality of ail RNA samples was

examined by capillary electrophoresis using the Agilent 2100 Bioanalyzcr (Agilent).

Duplicate samples of fluorcscently labeled cDNAs were generated from 20 ug of total

RNA in each reaction using the Agilent Fluorescent Direct Label Kit and 1.0 mM Cyanine

3- or 5-labeled dCTP (PerkinElmer) following the manufacturer's instructions. Cyanine-

labeled cDNA from one control sample was mixed with the same amount of reverse-color

cyanine labeled cDNA from one treated sample. Hybridization was performed in duplicate

following Oligonucleotide Microarray Hybridization User's Manual using the

Hybridization Plus kit A (Agilent). Combined cyanine 3- and cyanine 5-labeled cDNAs in

a volume of 200 ul of H2O were denatured at 98 °C for 3 min and cooled to room

température. They were mixed with 50 ul of 10 x control probe, followed by the addition

250 ul of 2 x hybridization buffer. The 500 (al reaction mix was applied to Agilent 22K

Human Oligonucleotide Microarray (containing about 44,000 gènes and ESTs), and

hybridized at 60 °C for 17 h. The slides were disassembled in 6 x SSC, 0.005% Triton X-

102, washed with 6 x SSC, 0.005% Triton X-102 for 10 min at room température, then with

0.1 x SSC, 0.005% Triton X-102 for 5 min on ice, and dried using a nitrogen-filled air gun.

The arrays were scanned using a dual-laser DNA microarray scanner (Agilent). The data

was extracted from images by the Feature Extraction software 6.1 (Agilent) and analyzed

with the GeneSpring software (Agilent). An Intensity-Dependent Normalization (known as

"Lowess Normalization") was applied to correct for artifacts caused by nonlinear rates of

dye incorporation as well as inconsistencies of the relative fluorescence intensity between

some red and green dyes. Lists of differentially expressed gènes using a 2 fold expression

145

cutoff were generated using the data from the independent experiments. The resulting "up"

and "down" regulated gène lists were compared with the Nucleolar Data Base (NOPbd)

(http://www.lamondlab.com/NoPDB) (Leung et al., 2006) and analyzed using the Ingenuity

Pathways Analysis (IPA) software (Ingenuity Systems).

4.8.10. Immunofluorescence staining. Antibodies against UBF and RPA194 were kindly provided by A. Cavanaugh and

L. Rothblum and the anti-fïbrillarin antibody was MMS-581S (Covance). PS (2 x

105cells/petri dish) and DKO cells (2.5 x 105 cells/petri dish) were seeded in 35 mm pétri

dishes containing a cover glass treated with 0.2% gelatin. The following day, cells were

washed once with PBS, 0.5 raM CaCb (pre-warmed at 37°C) and fix at room température

for 20 minutes in 1 ml of 3.7% formaldehyde in PBS, 0.5mM CaCl2 (pre-warmed at 37°C).

Cells were then washed 2 times with PBS, ImM MgCb and permeabilized at room

température for 5 minutes in 1 ml 0.2% Triton X-100 diluted in PBS, ImM MgCb. They

were then washed 2 times with PBS, ImM MgCb, 0,05% Tween 20 and blocked in 1 ml of

5% milk in PBS, ImM MgCl2 for lh at 37°C and incubated at 37°C for lh with the first

antibody. The cells were then washed 5 times with PBS, ImM MgCb, 0,05%) Tween 20

and again blocked for 15 minutes at 37°C in incubated inl ml 5% milk in PBS, ImM

MgCl2. Cells were incubated for 45 minutes with Alexa Fluor 488 anti-rabbit IgG

(Invitrogen, #cat A-11008) for UBF or RPA194 and with Alexa Fluor 568 anti-mouse IgG

(Invitrogen, #cat A-l 1004) for fibrillarin. Cells were then washed 6 times with PBS, ImM

MgCl2, 0,05%) Tween 20 before coloration with DAPI in PBS, ImM MgCl2 for 20 minutes

at 37°C and then washed 6 times with PBS, ImM MgCl2, 0,05% Tween 20, 1 time with

PBS, ImM MgCb and one time with H2O. Cover slips were mounted in 50% glycerol/50%

glycin buffer (0,2 M glycin, 0,3 M NaCl, NaOH, sodium azide) and observed under a Leica

DMI6000 B using OpenLab software (Improvision). Sériai optical sections were volume

deconvoluted using Volocity (Improvision). Nucleolar areas were measured using the

OpenLab software (Improvision).

146

4.8.11. ChIP and Q-PCR assays. PS (3 x 106cclls/petri dish) and DKO cells (4.5 x 106 cells/petri dish) werc seeded in

150mm pétri dishes 18h prior to the experiment. Cells were rapidly washed with DMEM

high glucose serum-frce médium and fixed in 0.25% formaldehyde for 10 min at room

température. Nucleolar chromatin préparation and and nucleolar ChIP were performed as

described previously (O'Sullivan et al., 2002). Q-PCR reactions (20(0.1) were performed in

tripicate on 2\x\ of DNA, with 20pmol of cach primer, and 10u.l Quantitect™ SYBR Green

PCR Master Mix (QIAGEN). 45 reaction cycles of 15s at 94°C, 20s at 58°C (51°C for

IGS1 primers), and 10s at 72°C were carried on a Lightcycler (Roche) and quantified as

(Frank et al., 2001). The primer coordinates (relative to 47S rRNA initiation) were: Prom., -

47-32; 47S-5', 952-1030; 18S, 3990-4092; 28S-5', 8204-8300; 28S-3', 12855-12970 and

IGS, 27366-27477. The antibodies used were H3 (abl791, Abcam), H3-K9-trimethyl (07-

442, Upstate) and H4-tetra-acetyl (06-946, Upstate), UBF and RPA194 (kindly provided by

A. Cavanaugh and L. Rothblum) and RPII (H-224, Santa Cruz, sc-9001 ).

4.8.12. Run-On/ChlP. PS (1.7 x 106cells/petri dish) and DKO cells (2 x 106 cells/petri dish) were seeded in

100mm pétri dishes 18h prior to the experiment. Nuclear Run-On experiments were

performed as described in (Cavanaugh and Thompson, 1986; Hannan et al., 1995;

Stefanovsky et al., 2006). Cells were rapidly scrape-harvested into ice-cold PBS, pelleted

and stored at -80C. Frozen cells were lysed briefly in lOmM Tris-HCl pH7.4, lOmM NaCl,

lOmM MgCl2, 0.1% Igepal (Sigma), nuclei recovered, 4°C, 1300 rpm, 2 min, and

resuspended in 50mM Tris-HCl pH8.3, 5mM MgCl2, O.lmM EDTA, 40% Glycerol and

DNA concentrations determined in lOmM Tris-HCl, 2.5M NaCl, ImM EDTA, pH 7.4,

using Hoechst 33258 (Biorad) in a Versafluor (Biorad). Equal aliquots of nuclei,

representing 20 to 50|ag genomic DNA, were subjected to Nuclear Run-On for 30min at

37°C in 28mM Tris-HCl pH8.3, 147mM KC1, 5mM MgCl2, 0.45mM EDTA, 18% Glycerol,

80(xg/ml Heparin (Sigma), 0.49mM ATP, UTP and GTP, 1.2uM CTP (100 uCi), 2.4mM

DTT and 26U RNAguard (Amersham Biotech). After the reaction the nuclei were washed

two times with cold PBS and crosslinked with formaldehyde and whole nuclear chromatin

147

was prepared essentially following the protocols from the Young and Farnham laboratories,

see http://www.ircm.qc.ca/microsites/francoisrobert/en/317.html. After normalization of

the chromatin levels, ChIP was performed using an anti phospho-RPII-CTD Ser2 or 5

antibody (H5 (MMS-129R) Covance) (Jones et al., 2004) following the same protocol,

except that the RNA not the DNA was recovered and in vitro transcribed [3H] RNA was

added to normalized extraction efficiencies and hybridization efficiency (Stefanovsky et al.,

2006). Duplicate membranes (Biodyne B, Pall) with 5u.g/slot of linearized probe DNA

(HhrRNA gène fragment 2087-4245bp) cloned in pBSKS+ or of linearised pBSKS+ vector

were prepared using the Bio-Dot SF slot blot apparatus (BioRad). RNA hybridization was

performed in 50% formamide, 5 x SSC, 0.1% SDS, 5 x Denhardt's, 0,1 mg/ml denatured

Torula RNA and 0,1 mg/ml denatured Salmon Sperm DNA at 45C for 18 hours. The filters

were then washed three times at 65°C in 3 x SSC, 0,1% SDS and, if necded, one time in 1 x

SSC, 0,1% SDS (20 minutes each). Data were revealed by phospho-imaging on a STORM

860 and analyzed with the ImageQuant software (GE Healthcare). Each membrane slot was

counted in ScintiVerse (Fisher) and the [3H] RNA hybridization efficiency used to

normalize the Phospholmager data.

4.8.13. SI Mapping. SI Mapping was performed as described (Berk and Sharp, 1977). RNA was

extracted with Trizol following the manufacturer's instructions and dissolved in

formamide. An HhrRNA gène fragment encompassing -550 to +424, was 5' end labeled at

+424 with y[32P]-ATP. 15-30U£ (24u.l) of total cellular RNA were mixed with 0.05pmol

(6(4,1) of labeled probe in hybridization buffer (4M NaCl, 0,4M PIPES, lOmM EDTA, pH

6,4) to give a final volume of 30ul. The mix was denatured at 100°C for 1 min. and

incubated for 18h at 53°C. The next day, 270u.l of ice cold SI buffer (30mM NaAc, 2mM

ZnSÛ4, 200mM NaCl) containing 25U of SI nuclease (Invitrogen) were added to each

reaction. The reactions were vortexed, briefly centrifuged and incubated at 45 °C for 1 h.

The hybrids were phenol/chloroform extracted, ethanol-precipitated and resuspended in 6(xl

of 6M urea loading buffer. The samples were denatured 1,5 min. and resolved on a 6%

polyacrylamide sequencing gel.

148

4.9. Références Akiyama, Y., Watkins, N., Suzuki, H., Jair, K. W., van Engeland, M., Esteller, M., Sakai,

H., Ren, C. Y., Yuasa, Y., Herman, J. G., and Baylin, S. B. (2003). GATA-4 and GATA-5

transcription factor gènes and potential downstream antitumor target gènes are

epigenetically silenced in colorectal and gastric cancer. Mol Cell Biol 23, 8429-8439.

Berk, A. J., and Sharp, P. A. (1977). Sizing and mapping of early adenovirus mRNAs by

gel electrophoresis of SI endonuclease digested hybrids. Cell 12, 721-732.

Brock, G. J. R., and Bird, A. (1997). Mosaic methylation of thc repeat unit of the human

ribosomal RNA gènes. HumMolGenet 6, 451-456.

Cavanaugh, A. H., and Thompson, E. A., Jr. (1986). Hormonal régulation of transcription

of rDNA. Initiation of transcription by RNA polymerase I in vitro. Journal of Biological

Chemistry 261, 12738-12744.

Cioci, F., Vu, L., Eliason, K., Oakes, M., Siddiqi, 1. N., and Nomura, M. (2003). Silencing

in yeast rDNA chromatin: reciprocal relationship in gène expression between RNA

polymerase I and II. Mol Cell 12, 135-145.

Clark, S. J., Harrison, J., Paul, C. L., and Frommer, M. (1994). High sensitivity mapping of

methylated cytosines. Nucleic Acids Res 22, 2990-2997.

Conconi, A., Widmer, R. M., Koller, T., and Sogo, J. M. (1989). Two différent chromatin

structures coexist in ribosomal RNA gènes throughout the cell cycle. Cell 57, 753-761.

Cui, H., Horon, I. L., Ohlsson, R., Hamilton, S. R, and Feinberg, A. P. (1998). Loss of

imprinting in normal tissue of colorectal cancer patients with microsatellite instability. Nat

Med4, 1276-1280.

149

Frank, S. R., Schroeder, M., Fernandez, P., Taubert, S., and Amati, B. (2001). Binding of c-

Myc to chromatin médiates mitogen-induced acetylation of histone H4 and gène activation.

Gènes Dev 75, 2069-2082.

Ghoshal, K., Majumder, S., Datta, J., Motiwala, T., Bai, S., Sharma, S. M., Frankel, W.,

and Jacob, S. T. (2004). Rôle of human ribosomal RNA (rRNA) promoter methylation and

of methyl-CpG-binding protein MBD2 in the suppression of rRNA gène expression. J Biol

Chem 279, 6783-6793.

Grummt, I. (2003). Life on a planet of its own: régulation of RNA polymerase I

transcription in the nucleolus. Gènes Dev 17, 1691-1702.

Grummt, I., and Pikaard, C. S. (2003). Epigenetic silencing of RNA polymerase 1

transcription. Nat Rev Mol Cell Biol 4, 641-649.

Haaf, T., Hayman, D. L., and Schmid, M. (1991). Quantitative détermination of rDNA

transcription units in vertébrale cells. Exp Cell Res 193, 78-86.

Hannan, R. D., Luyken, J., and Rothblum, L. I. (1995). Régulation of rDNA transcription

factors during cardiomyocyte hypertrophy induced by adrenergic agents. J Biol Chem 270,

8290-8297.

Henderson, A. S., Warburton, D., and Atwood, K. C. (1972). Location of ribosomal DNA

in the human chromosome complément. Proc Natl Acad Sci U S A 69, 3394-3398.

Hermann, A., Gowher, H., and Jeltsch, A. (2004). Biochemistry and biology of mammalian

DNA methyltransferases. Cell Mol Life Sci 61, 2571-2587.

Jenuwein, T., and Allis, C. D. (2001). Translating the histone code. Science 293, 1074-

1080.

150

Jones, J. C, Phatnani, H. P., Haystead, T. A., MacDonald, J. A., Alam, S. M., and

Greenleaf, A. L. (2004). C-terminal repeat domain kinase I phosphorylates Ser2 and Ser5

of RNA polymerase II C-terminal domain repeats. J Biol Chem 279, 24957-24964.

Lawrence, R. J., Earley, K., Pontes, O., Silva, M., Chen, Z. J., Neves, N., Viegas, W., and

Pikaard, C. S. (2004). A concerted DNA methylation/histone methylation switch régulâtes

rRNA gène dosage control and nucleolar dominance. Mol Cell 13, 599-609.

Leung, A. K., Trinkle-Mulcahy, L., Lam, Y. W., Andersen, J. S., Mann, M., and Lamond,

A. I. (2006). NOPdb: Nucleolar Proteome Database. Nucleic Acids Res 34, D218-220.

Li, E. (2002). Chromatin modification and epigenetic reprogramming in mammalian

development. Nat Rev Genêt 3, 662-673.

Lucchini, R., and Sogo, J. M. (1998). The dynamic structure of ribosomal RNA gène

chromatin, In Transcription of ribosomal gènes by eukaryotic RNA polymerase I, M. R.

Paule, éd. (Austin, TX USA: Landes Bioscience), pp. 255-276.

Maden, B. E. H. (1988). Locations of Methyl Groups in 28S rRNA of Xenopus laevis and

Man. Journal of Molecular Biology 201, 289-314.

Majumder, S., Ghoshal, K., Datta, J., Smith, D. S., Bai, S., and Jacob, S. T. (2006). Rôle of

DNA methyltransferases in régulation of human ribosomal RNA gène transcription. J Biol

Chem 281, 22062-22072.

Moss, T. (2004). At the crossroads of growth control; making ribosomal RNA. Current

Opinion in Genetics & Development 14, 210-217.

Moss, T., Langlois, F., Gagnon-Kugler, T., and Stefanovsky, V. (2006). A housekeeper

with power of attorney: the rRNA gènes in ribosome biogenesis. Cell Mol Life Sci.

151

Moss, T., and Stefanovsky, V. Y. (2002). At the center of eukaryotic life. Cell 109, 545-

548.

Myohanen, S. K., Baylin, S. B., and Herman, J. G. (1998). Hypermethylation can

selectively silence individual pl6ink4A alleles in neoplasia. Cancer Res 58, 591-593.

O'Sullivan, A. C, Sullivan, G. J., and McStay, B. (2002). UBF binding in vivo is not

restricted to regulatory séquences within the vertebrate ribosomal DNA repeat. Mol Cell

Biol 22, 657-668.

Rhee, L, Bachman, K. E., Park, B. H., Jair, K. W., Yen, R. W., Schuebel, K. E., Cui, H.,

Feinberg, A. P., Lengauer, C, Kinzler, K. W., et al. (2002). DNMT1 and DNMT3b

cooperate to silence gènes in human cancer cells. Nature 416, 552-556.

Rhee, L, Jair, K. W., Yen, R. W., Lengauer, C, Herman, J. G., Kinzler, K. W., Vogelstein,

B., Baylin, S. B., and Schuebel, K. E. (2000). CpG methylation is maintained in human

cancer cells lacking DNMT1. Nature 404, 1003-1007.

Robertson, K. D. (2005). DNA methylation and human disease. Nat Rev Genêt 6, 597-610.

Santoro, R., and Grummt, I. (2001). Molecular mechanisms mediating methylation-

dependent silencing of ribosomal gène transcription. Mol Cell 8, 719-725.

Schmickel, R. D. (1973). Quantitation of human ribosomal DNA: hybridization of human

DNA with ribosomal RNA for quantitation and fractionation. Pediatr Res 7, 5-12.

Siddiqi, I. N., Dodd, J. A., Vu, L., Eliason, K., Oakes, M. L., Keener, J., Moore, R., Young,

M. K., and Nomura, M. (2001). Transcription of chromosomal rRNA gènes by both RNA

polymerase I and II in yeast uaOO mutants lacking the 30 kDa subunit of transcription

factor UAF. Embo J 20, 4512-4521.

152

Stancheva, I. (2005). Caught in conspiracy: coopération between DNA methylation and

histone H3K9 methylation in the establishment and maintenance of heterochromatin.

Biochem Cell Biol 83, 385-395.

Stefanovsky, V. Y., Langlois, F., Gagnon-Kugler, T., Rothblum, L. I., and Moss, T. (2006).

Growth factor signaling régulâtes elongation of RNA polymerase I transcription in

mammals via UBF phosphorylation and rchromatin remodeling. Mol Cell 21, 629-639.

Stefanovsky, V. Y., and Moss, T. (2006). Régulation of rRNA synthesis in human and

mouse cells is not determined by changes in active gène count. Cell Cycle 5, 735-739.

Stefanovsky, V. Y., Pelletier, G., Hannan, R., Gagnon-Kugler, T., Rothblum, L. I., and

Moss, T. (2001). An immédiate response of ribosomal transcription to growth factor

stimulation in mammals is mediated by ERK phosphorylation of UBF. Mol Cell 8, 1063-

1073.

Suzuki, H., Gabrielson, E., Chen, W., Anbazhagan, R., van Engeland, M., Weijenberg, M.

P., Herman, J. G, and Baylin, S. B. (2002). A genomic screen for gènes upregulated by

demethylation and histone deacetylase inhibition in human colorectal cancer. Nat Genêt 31,

141-149.

Ting, A. H., McGarvey, K. M., and Baylin, S. B. (2006). The cancer epigenome—

components and functional correlates. Gènes Dev 20, 3215-3231.

4.10. Données supplémentaires Les analyses de Run-On/ChIP montrent que l'ARN polymerase II transcrit les gènes

ribosomaux dans les cellules DKO et les analyses de « pulse-chase » montrent qu'il existe

une population d'ARN 47S ayant une plus longue demi-vie (figures 4.10 et 4.9). Est-il

possible que PolII transcrive un ARN 47S complet qui ne soit ni reconnu ni transformé

correctement et qui s'accumule dans les cellules DKO? Comme mentionné dans

153

l'introduction de cette thèse (section 1.3.1), des études faites chez la levure ont montré que

des lignées ayant un défaut dans la machinerie transcriptionnelle Poli, délétions de sous-

unités du complexe UAF, compensent ce défaut en utilisant PolII pour transcrire les gènes

ribosomaux (Oakes et al., 1999; Vu et al., 1999). Ces études montrent que l'ARN

polymérase II a la capacité de transcrire les gènes ribosomaux. Pour vérifier si PolII

transcrit les gènes ribosomaux dans la lignée Dnmtl-/-, Dnmt3b-/-, deux types

d'expériences furent réalisées. La première expérience consistait à analyser le site

d'initiation de la transcription pour voir s'il existait une différence entre les cellules PS et

DKO (figure 4.8). Il est en effet théoriquement possible que PolII utilise un autre site

d'initiation de la transcription comme dans les lignées de levure déficientes pour la

machinerie Poli (M. Nomura, communication personnelle). Si tel était le cas, il serait

possible de détecter les transcrits PolII de cette façon (voir corps de l'article). La deuxième

expérience consistait à vérifier la présence d'ARN ribosomaux polyA+ parce que transcrit

par PolII (section 4.10.1). Dans les deux cas, un résultat négatif n'exclu pas que cette sous-

population de 47S soit transcrite par PolII car il est possible que les techniques utilisées ne

soient pas suffisamment sensibles ou spécifiques. Cependant, un résultat positif renforcirait

l'hypothèse comme quoi cette sous-population de 47S est transcrite par PolII. Même si les

résultats obtenus suite à ces expériences ne permettent pas à eux seuls de conclure hors de

tout doute dans un sens ou dans l'autre, l'intérêt de la question posée méritait que ces

expériences soient faites.

4.10.1. Détermination de la production d'ARN ribosomaux polyA+ La deuxième approche utilisée consiste à vérifier s'il y a une production d'ARN

ribosomaux polyA+ dans les cellules DKO qui n'existe pas chez les cellules parentales et

qui représenterait une population de 47S transcrits par PolII. Pour ce faire, les ARN totaux

des deux types cellulaires furent purifiés avec du Trizol (Invitrogen) et les ARN polyA+ de

ces extractions furent enrichis sur une cellulose olido-dT avec l'ensemble « GenElute

mRNA miniprep kit » de Sigma. Par la suite, ces ARN furent déposés sur gel et transférés

sur une membrane pour pouvoir faire une analyse de type Northern, protocole décrit à la

section 3.8.2, en utilisant une sonde dirigée contre le 47S, de -504pb à +1446pb par rapport

154

au site d'initiation. L'utilisation de cette technique pour la détection d'ARN ribosomaux

polyA+ comporte un désavantage majeur. En effet, les principaux contaminants lors de la

purification des ARN polyA+ sont les ARN ribosomaux puisque ce type d'ARN représente

80% des ARN cellulaires alors que les ARNm ne représente que 1 ou 2%. Pour une analyse

des ARN messagers, cette contamination n'est pas critique puisque les ARNm sont

généralement suffisamment enrichis pour permettre une bonne détection. Cependant, pour

l'analyse d'une sous-population d'ARN ribosomaux polyA+, il est nécessaire de se

débarrasser complètement de ces contaminants.

Les ARN polyA+ utilisés pour ces expériences ont été purifiés successivement par

deux passages sur cellulose oligo-dT de façon à ne plus pouvoir détecter d'ARNr 28S et

18S lorsque visualisé sur un gel d'agarose coloré au bromure d'éthidium. La quantité

d'ARN déposé sur gel a été normalisée suite à un dosage au spectromètre. De l'ARN total

provenant des deux lignées cellulaires a aussi été déposé sur gel pour pouvoir confirmer

que la bande observée est bien le 47S. Deux préparations d'ARN polyA+, purifiées deux

fois chacune, pour chaque type cellulaire ont été analysées. La figure 4.11 A présente

l'analyse Northern d'une de ces préparations. Les deux premières colonnes contiennent les

ARN totaux et les deux suivantes, les ARN polyA+. La bande supérieure est l'ARN 47S et

la bande inférieure est l'intermédiaire de transformation résultant du clivage au site A0,

nommé 45S chez les mammifères (figure 4.11B) ou 33S chez la levure (figure 1.1 OB). La

comparaison de l'intensité des bandes provenant des deux préparations d'ARN polyA+

avec le logiciel « Image Quant » montre qu'il n'existe pas de différence, entre les deux

types cellulaires, en ce qui a trait à la quantité d'ARN polyA+ purifié. Comme il a été

mentionné précédemment, le fait qu'il n'y ait pas de différence entre les deux types

cellulaires ne signifie pas que PolII ne transcrit pas un ARN 47S pleine longueur puisque

PolII ne produit pas toujours un ARN polyA+.

Une méthode alternative aurait été la purification des ARNs, provenant des deux

types cellulaire, avec un anticorps dirigé contre la structure du « guanosine cap» qu'ajoute

toujours PolII sur les ARNs qu'elle transcrit. Suite à cette purification, il aurait été possible

de faire une analyse de type Northern avec une sonde dirigée contre la séquence

155

ribosomale. Malheureusement, aucun anticorps commercial n'est disponible contre le cap et

les anticorps maison ne sont pas efficaces pour ce type de purification puisque les deux

anticorps existant ne possèdent pas une affinité suffisante pour permettre l'isolation des

ARNm. De plus, aucune protéine de reconnaissance du cap ne possède une affinité

suffisante pour permettre la purification des ARNm selon N. Sonenberg (communication

personnelle). En conclusion, le SI mapping et la détection d'ARNr polyA+ne semblent pas

assez sensibles pour permettent la détection d'une population de 47S transcrite par l'ARN

polymérase II. Il faut toutefois garder en mémoire que ces résultats ne montrent pas que

cette population est inexistante dans les cellules DKO. La mise au point d'une technique de

détection plus sensible, les Run-On/ChIP, a d'ailleurs permis de détecter des ARN

ribosomaux transcrits pas l'ARN polymérase II (figure 4.10).

B)

AO 47S I 5'ETS ITS1JTS2J 3'ETS

18S 5.8S 28S

Figure 4.11 : Analyse de type Northern faite avec une sonde dirigée contre le 47S sur une préparation d'ARN total ou d'ARN polyA+. A) Analyse de type Northern sur une préparation d'ARN total (deux premières colonnes) et d'ARN poly-A+. Le type cellulaire (PS : Parental strain, DKO : Double Knock-out), le type d'ARN ainsi que l'emplacement des ARN 47S et 45S sont indiqués. B) Représentation de l'ARN 47S et de l'intermédiaire de transformation 45S. L'ARN 45S est produit suite au clivage du 3'ETS et du site Ao. Les séquences du 5'ETS, des ITS 1 et 2, du 3'ETS, du 18S, du 5.8S et du 28S sont indiquées. Modifié de Rouquette et al., 2005.

4.10.2. Western blot avec des extraits de protéines nucléolaires Les analyses de ChIP ont montré un enrichissement de l'ARN polymérase II sur les

gènes ribosomaux dans les cellules DKO. Dans le même ordre d'idée, des analyses Western

A) total polyA+ O O

w * c/> * : O. Q CL Q

47S

m 45S

156

faites sur des extraits de protéines nucléolaires en utilisant différents anticorps dirigés

contre PolII ont été faites pour tenter de déterminer si cette protéine est enrichie dans le

nucléole des cellules DKO. Le protocole de purification nucléolaire provient du laboratoire

Lamond et il est disponible sur son site internet (www.lamondlab.com). Il consiste en une

suite de centrifugations sur coussins de sucrose qui permet d'obtenir une préparation de

nucléoles sous forme enrichie. Les cellules sont d'abord trypsinisées, lavées avec du PBS

puis resuspendues dans 10 mM Hepes, pH 7.9, lOmM KO, 1.5mM MgCh, 0.5mM DTT.

Par la suite, les cellules sont homogénéisées et centrifugées. Le culot obtenu contient les

noyaux sous forme enrichie. Ce culot est resuspendu dans 0.25 M sucrose, 10 mM MgC^.

Sous cette solution, on place le même volume de 0.35 M sucrose, 0.5 mM MgC^, pour

constituer le premier coussin de sucrose et on centrifuge. Le culot obtenu contient une

préparation de noyaux d'une plus grande pureté que la précédente. Ce culot est resuspendu

dans 0.35 M sucrose, 0.5 mM MgCb et est soniqué. Par la suite, cette solution est déposé

sur une solution de 0.88 M sucrose, 0.5 mM MgCb et centrifugée. Le culot obtenu contient

les nucléoles. Pour augmenter la pureté de cet extrait nucléolaire, le culot est de nouveau

suspendu dans 0.35 M sucrose, 0.5 mM MgCh et recentrifugé. Finalement, le culot est

resuspendu dans 0.35 M sucrose, 0.5 mM MgCb et conservé à -80°C. Le protocole publié

contient des photos et les extraits obtenus à chaque étape ont été observes au microscope et

comparés à ces photos.

Cinq anticorps différents ont été utilisés et les résultats de quatre d'entre eux sont

présentés à la figure 4.12. Les anticorps 4H8 (Active Motif) et 8WG16 (Covance)

reconnaissent le CTD indépendamment de son état de phosphorylation alors que les

anticorps H5 (Upstate), H14 (Upstate) et CC3 (M Vincent) reconnaissent spécifiquement le

CTD phosphorylé. Les phosphorylations du CTD sur les serines 2 et/ou 5 du motif répété

YSPTSPS caractérisent l'ARN polymérase II en élongation. La phosphorylation de la

serine 5 caractérise les étapes précoces de l'élongation alors que la phosphorylation de la

serine 2 caractérise les étapes tardives de l'élongation (Sims et al., 2004). Ainsi, près du

promoteur, le CTD est phosphorylé uniquement sur la serine 5 et en 3 ' du gène uniquement

sur la serine 2. Cependant, entre ces deux points, il existe un gradient de phosphorylation

serine 5/sérine 2. L'anticorps H5 est vendu comme étant spécifique pour la serine 2

157

phosphorylée, mais des études ont montré qu'il reconnaissait aussi la serine 5 phosphorylée

(Jones et al., 2004). Les anticorps 1114 et CC3 reconnaissent la serine 2 phosphorylée

(Albert et al., 2004). Les protéines nucléolaires ont été dosées avec le « DC Protein Assay »

de Biorad, essai colorimétrique basé sur l'essai de Lowry, et la majorité des gels ont aussi

été mis en présence d'antifibrillarine comme contrôle de dépôt. Trois extraits nucléolaires

indépendants par type cellulaire ont été préparés et un minimum de deux gels par anticorps

a été fait.

La figure 4.12 permet de voir que les résultats obtenus varient selon l'anticorps

utilisé. Les anticorps reconnaissant le CTD indépendamment de sa phosphorylation, 4H8 et

particulièrement 8WG16, montrent qu'il y a plus d'ARN polymérase II dans le nucléole

des cellules DKO que dans celui des cellules parentales. La détection de l'ARN polymérase

II dans les cellules parentales est probablement due au tait que la fraction nucléolaire n'est

pas pure, mais enrichie. Puisque ces anticorps reconnaissent une large variété de CTD, du

non phosphorylé au totalement phosphorylé en passant par tous les intermédiaires

possibles, leur signal est une large bande. Néanmoins, l'anticorps 4H8 semble avoir plus

d'affinité pour la forme phosphorylée que non phosphorylée car il présente un signal plus

fort dans la partie supérieure de sa bande, indiqué par une flèche sur la figure 4.12. Les

anticorps CC3 et H14 (résultats non présentés), reconnaissant le CTD phosphorylé, ne

montrent pas d'enrichissement de l'ARN polymérase II dans le nucléole des cellules DKO

versus les cellules parentales. Cependant, l'anticorps H5 présente probablement un

enrichissement lorsque normalisé avec la fibrillarine (figure 4.12). Les anticorps 4H8,

8WG16 et H5 montrent un enrichissement de PolII dans le nucléole des cellules DKO. De

plus, la répartition du signal de 4H8 et le signal de H5 suggèrent qu'il y ait plus de PolII en

élongation. Puisque les gènes ribosomaux sont majoritaires dans le nucléole, un

enrichissement de PolII en élongation, ou non, dans les cellules DKO suggère que cette

polymérase a vraisemblablement un plus grand accès aux gènes ribosomaux dans les

cellules DKO que dans les cellules parentales. Ces résultats vont dans la même direction

que les résultats des ChIP qui montraient un enrichissement de l'association de PolII avec

les gènes ribosomaux dans les cellules DKO. Cependant, contrairement aux ChIP, les

analyses Western sont une analyse des protéines dans le nucléole et non des protéines en

158

association avec les gènes ribosomaux. Le fait que cette technique ne permette pas, de

façon claire, la visualisation de l'augmentation du CTD phosphorylé ne signifie pas qu'il

n'y a pas plus d'ARN polymérase II en élongation sur les gènes ribosomaux dans les

cellules DKO. Il est en effet possible que la phosphorylation du CTD ne soit pas essentielle

à la transcription des gènes ribosomaux et/ou que la technique des Western sur des extraits

protéiques nucléolaires ne soit pas assez sensible pour permettre cette détection. C'est pour

augmenter cette sensibilité technique que les Run-On/ChIP ont été mis au point. Ces

expériences ont montré que les gènes ribosomaux sont transcrits par l'ARN polymérase II

dans les cellules DKO (figure 4.10).

4H8 8WC16 H5 CC3 PS DKO PS DKO PS DKO PS DKO

Figure 4.12: Analyse comparative de la présence d'ARN polymérase II nucléolaire dans les cellules PS et DKO.

4H8 et 8WG16 reconnaissent le CTD indépendamment de son état de phosphorylation. H5 et CC3 reconnaissent spécifiquement le CTD phosphorylé. Le nom de l'anticorps est indiqué ainsi que le type cellulaire. PS : Parental strain, DKO : Double Knock-out. Le position du CTD phosphorylé est indiqué par une flèche. Le contrôle de dépôt avec l'anti-fibrillarine est aussi présenté.

4.10.3. Inhibition de la méthylation de l'ADN avec 5-aza-2'dexoxycytidine Pour confirmer que la perte de la méthylation de l'ADN, dans les cellules DKO, est

la cause directe de l'augmentation du nombre de gènes ribosomaux actifs, les cellules

parentales furent traitées avec 5-aza-2'dexoxycytidine, un inhibiteur de la méthylation de

l'ADN. La 5-aza-2'dexoxycytidine est une base analogue à la dexoxycytidine qui est

incorporée dans l'ADN lors de la réplication. Cette base, contrairement aux cytosines, ne

peut pas être méthylée ce qui entraîne une dilution de la méthylation de l'ADN à chaque

division cellulaire. Après plusieurs divisions, la méthylation est à toute fin pratique

159

disparue. De plus, il a été montré que la cause principale de l'effet de 5-aza-

2'dexoxycytidine est son association avec Dnmtl, 3a et 3b ce qui inhibe son action

(Juttermann et al., 1994; Oka et al., 2005). L'utilisation de ce système permet de comparer

l'effet de la perte de la méthylation sur le nombre de gènes ribosomaux actifs dans les

cellules parentales et les cellules DKO.

Les cellules parentales furent traitées avec lu,m de 5-aza-2'dexoxycytidine pendant

48, 72 et 96 heures. De plus longs traitements ou de plus fortes concentrations de cette

substance sont toxiques pour les cellules (Juttermann et al., 1994; Schneider-Stock et al.,

2005). Pour vérifier que la méthylation de l'ADN est perdue suite aux traitements, l'ADN

génomique est extrait des cellules contrôles et des cellules traitées selon le protocole décrit

dans la section 5.9.2. L'ADN est ensuite digéré avec deux isoschizomères utilisés

fréquemment pour comparer les niveaux de méthylation de l'ADN. Hpall est sensible à la

méthylation de l'ADN, qui bloque son activité, tandis que Mspl est insensible à la

méthylation. Ainsi, la perte de la méthylation de l'ADN entraîne une augmentation de la

capacité de digestion d'HpalI tandis que Mspl est capable de digérer l'ADN génomique

indépendamment de la méthylation.

La figure 4.13 présente les résultats de la digestion de ces deux enzymes suite aux

différents temps de traitements. La comparaison des patrons de digestion d'HpalI des

cellules sans et avec traitement permet de constater que l'activité de l'enzyme augmente

avec la longueur des traitements, comparez les colonnes 1, 3, 5 et 7. Cette augmentation de

la digestion par Hpall est un indicateur de la perte de la méthylation. La comparaison des

patrons de digestion d'HpalI et Mspl après 96 heures de traitement, colonne 7 versus 8,

montre qu'il reste encore de l'ADN méthylé après 96 heures de traitement. Ces résultats

suggèrent que les traitements avec 5-aza-2'dexoxycytidine entraînent une perte partielle de

la méthylation de l'ADN. Cependant, comme les traitements plus longs ou à plus haute

concentration de 5-aza-2'dexoxycytidine sont toxiques pour les cellules HCT116

(Schneider-Stock et al., 2005), il semble impossible, dans ce système, d'avoir une perte

totale de la méthylation de l'ADN.

160

En parallèle de cette digestion différentielle, Hpall versus Mspl les cellules furent

pontées au psoralen et ce en duplicata (figure 4.14). Le protocole est présenté dans la

section 4.8.2. Puisque cette expérience permet de quantifier les gènes actifs et les gènes

inactifs, il est possible de vérifier si les traitements avec 5-aza-2'dexoxycytidine entraînent

une augmentation du nombre de gènes actifs. Puisque ces gènes ont une structure

chromatinienne ouverte, leur ADN sera plus accessible au psoralen, un agent intercalant,

que l'ADN des gènes inactifs. Suite au pontage, l'ADN est digéré et ce matériel est utilisé

pour faire un Southern. L'intercalation du psoralen induit des pontages, ou « crosslinks »,

interbrins qui rigidifient l'ADN et ralentissent sa migration comparativement à l'ADN peu

ou pas ponté. Ce ralentissement de migration n'est pas dû à un changement de masse

puisque l'intercalation de bromure d'éthidium, par exemple, abolit cette différence en

rendant le brin non ponté aussi rigide que le brin ponté.

xjf Oh 48h 72h 96h ,%P H M H M H M H M

Figure 4.13: Le traitement des cellules parentales avec 5-aza-2'dexoxycytidine entraîne une perte de la méthylation de l'ADN.

L'ADN génomique provenant des cellules contrôles (colonnes 1 et 2) et traitées (colonnes 3 à 8) a été digéré avec Hpall (H) ou Mspl (M). Les différents temps de traitements avec 5'aza sont indiqués.

161

Comme l'ADN des gènes actifs et inactifs est associé avec différentes quantités de

psoralen, ces deux populations vont migrer différemment sur gel. Ainsi, l'ADN des gènes

actifs va migrer plus lentement que celui des gènes inactifs car il est associé avec plus de

psoralen. La séparation de ces deux sous-population sur gel permet la quantification de

chacune d'elle. Comme ce fut le cas pour les résultats présentés à la figure 4.2, l'ADN

génomique ponté au psoralen fut digéré avec BamHI, pour visualiser la région transcrite, et

avec Apol, pour la région non transcrite (figure 4.14A et B) respectivement. Les résultats

présentés à la figure 4.14A montrent que la perte de la méthylation de l'ADN cause une

augmentation du nombre de gènes ribosomaux actifs. La première colonne contient de

l'ADN génomique non ponté et les colonnes suivantes contiennent l'ADN provenant de

chacun des duplicata. La comparaison des colonnes 2, 3, 6 et 7 avec 4, 5, 8 et 9 montre que

des traitements de 72 et 96 heures entraînent une augmentation du nombre de gènes actifs.

Le traitement de 48 heures n'a pas d'effet sur le nombre de gènes actifs, colonnes 3 et 7

versus 2 et 6. Ces résultats montrent que la perte de l'ADN, suite aux traitements avec 5-

aza-2'dexoxycytidine, augmente le nombre de gènes ribosomaux actifs. De plus, le fait que

cette augmentation ne soit pas visible avec une sonde dirigée contre la région non transcrite

(figure 4.14B) comme c'était le cas dans les cellules DKO (figure 4.2B et argument dans le

texte) permet de conclure que cette augmentation de gènes actifs représente la situation des

gènes transcrits. En effet, il ne s'agit pas de la conséquence d'un changement de la structure

générale de la chromatine puisqu'alors les régions transcrites et non transcrites d'un même

gène actif varierait de façon identique ce qui n'est pas le cas.

Cette expérience n'a été faite qu'une seule fois car il semble que la combinaison de

l'incorporation de 5-aza-2'dexoxycytidine avec le pontage au psoralen empêche une bonne

séparation des deux bandes et donc la quantification des deux sous-populations. Par

conséquent, il est impossible de comparer directement l'augmentation du nombre de gènes

actifs dans cette expérience avec la situation des cellules DKO et c'est pourquoi cette

expérience ne se retrouve pas dans l'article. Toutefois, l'ensemble des résultats obtenus

suite au traitement des cellules parentales avec la 5-aza-2'dexoxycytidine suggèrent que la

perte de la méthylation de l'ADN observée chez les cellules DKO est la cause directe de

l'augmentation du nombre de gènes transcrits dans cette lignée.

162

A)

B)

C)

+ Psoralen

PS • Oh 48h 72h 96h Oh 48h 72h 96h '

Région transcrite 2,2kb &mmMmmW '■%£*

1 2 3 4 5 6 7 8 9

Région non transcrite 2,5 kb •/M ^', ■ ■ ' . - ' , . ■ •'■, r

Sonde

10 11 12 13 14 15 16 17 18

4424MB) 437601{A) +2087{B) 1

^ 1 1 au» m •: "HT 1 *M !

Figure 4.14: Le traitement des cellules parentales avec 5-aza-2'dexoxycytidine entraîne une augmentation du nombre de gènes actifs. A et B) Analyse de type Southern fait avec l'ADN génomique des cellules traitées durant différents temps avec 5'aza et ponté au psoralen. En A) utilisation d'une sonde dirigée contre la région transcrite et en B) contre la région non transcrite. Les colonnes 1 et 10 contiennent de l'ADN non traité et non ponté. Les différents temps de traitements sont indiqués ainsi que la position des gènes actifs et inactifs. L'ADN provenant de chacun des duplicata a été déposé. C) Carte du locus ribosomal humain et identification des sites de restriction et des sondes utilisée.

4.10.4 Micropuces à ADN (« microarrays ») Puisqu'il est impossible de vérifier par Western les niveaux d'expression de toutes

les protéines impliquées dans la biogenèse des ribosomes, TARN total provenant des

cellules PS et DKO avec du Trizol (Invitrogen) fut purifié pour faire des analyses de

micropuces à ADN. Ces analyses, vont permettre de vérifier s'il y a un changement dans le

niveau d'expression d'un ou plusieurs ARN messagers qui pourrait expliquer, entre autres,

163

la diminution du niveau de synthèse du 47S, la diminution de la vitesse de croissance et les

anomalies de transformation du 47S observées dans les cellules DKO.

Les analyses de « microarrays » ont été réalisées au service de micropuces à ADN

du CRHDQ à l'aide d'un « dual-laser DNA microarrays scanner » d'Agilent. La qualité des

échantillons d'ARN a préalablement été vérifiée à l'aide du « Bioanalyser 2100 »

d'Agilent. Les micropuces utilisées contiennent 44 000 gènes et EST humains. Les

analyses ont aussi été faites en inversant les fluorochromes associés aux ARNm de façon à

s'assurer que les résultats obtenus sont indépendants de leur liaison à ces fluorochromes. Le

logiciel utilisé pour l'extraction de l'information à partir des images des micropuces est le

« GeneSpring » d'Agilent. Avec les limites utilisées, variation de plus ou moins 2 fois, 279

ARNm ont une expression réduite, dans les cellules DKO comparativement aux cellules

parentales, et 106 ARNm ont une expression augmentée. L'expression de la majorité des

ARNm varie très peu. Par exemple, seulement 12 ARNm sur 106 ont une expression

augmentée de 5 fois ou plus. Pour ce qui est des ARNm ayant une expression réduite, 70

d'entrés eux, sur 279, présentent une réduction de 5 fois ou plus. Malgré ce nombre

relativement élevé d'ARNm à expression réduite, aucun changement, augmentation ou

réduction, dans l'expression d'ARNm codant pour des composantes des ribosomes n'a été

détecté.

Une deuxième analyse a été faite sur les résultats des puces à ADN pour vérifier s'il

y avait un changement dans l'expression d'ARNm codant pour des protéines retrouvées

dans les nucléoles. Pour cette analyse, les résultats de microarrays furent comparés avec le

protéome nucléolaire publié par le laboratoire du Dr Lamond. Le protéome nucléolaire

établi par le laboratoire du Dr Lamond contient 732 protéines et est disponible sur leur site

internet (Leung et al., 2006). L'expression de seulement 7 ARNm, sur les 732, varie et seul

l'ARNm codant pour la vimentine varie fortement soit de 37 fois moins. Ces résultats sont

présentés au tableau 4.1.

164

NOM DE LA PROTEINE VARIATION DU NIVEAU D'EXPRESSION (DKO/PS)

Vimentine -37,5267

« S100 calcium binding protein A4 » -7,1942

Histone 1, Hic -3,8461

Tubuline bêta 2 -3,4246575

« S100 calcium binding protein Al6 » -2,2472

Histone 1, H2bk +3,281

Peptidylpropyle isomérase F (cyclophiline F)

+2,408

Tableau 4.1 : Variation de l'expression d'ARNm codant pour des protéines nucléolaires. Le nom de la protéine est dans la colonne de gauche et le niveau de variation de son ARNm est dans celle de droite. La variation d'expression est le rapport entre l'expression dans les cellules DKO sur celles dans les cellules parentales.

La banque de données du « Proteome Bioknowledge™ Library » a été utilisée pour

déterminer la fonction des sept protéines nucléolaires dont l'expression de l'ARNm varie

dans les cellules DKO versus les cellules PS. La vimentine est une protéine des filaments

intermédiaires du cytosquelette. Elle est impliquée dans le contrôle de la morphologie et de

la migration cellulaire, de l'organisation nucléaire et de l'apoptose. La protéine « S100

calcium binding protein A4 » est exprimée dans tous les tissus étudiés et est une protéine

qui régule la cardiomyogenèse, l'angiogenèse et la migration cellulaire. L'histone hic est le

deuxième membre de la famille Hl et il est exprimé dans tous les tissus. Cette protéine est

localisée dans le nucléosome et elle pourrait donc être impliqué dans la compaction de la

chromatine et dans la régulation de l'expression génique, mais cela n'a pas été démontré

expérimentalement. La protéine tubuline bêta 2 est impliquée dans la polymérisation de la

tubuline et dans l'interaction des microtubules avec certains drogues. La protéine « S100

calcium binding protein Al6» est exprimée de façon ubiquitaire, mais sa fonction est peu

connue. Pour l'instant, sa propriété à s'attacher au calcium est déduite par sa similarité de

séquence avec les membres de la famille « S100 binding calcium proteins ». La protéine

165

H2bk n'est pas caractérisée, mais sa similarité avec l'histone putative H2bd suggère qu'elle

est localisée dans le nucléosome. La fonction et la localisation de la cyclophiline F ne sont

pas connues. Cette deuxième analyse montre qu'il n'y a pas de changement général dans

l'expression des ARNm nucléolaires.

Comme troisième analyse, les ARNm identifiés suite aux analyses de puces à ADN

ont été regroupés par catégorie d'activité en utilisant le logiciel de « lngenuity Systems

Inc » disponible sur leur site internet. Ce logiciel classe les ARNm soumis par l'utilisateur

selon leur rôle et permet d'obtenir une distribution quantitative des ARNm dans différents

champs d'activités (figures 4.15 et 4.16). Cette analyse quantitative a été faite

indépendamment pour les ARNm dont l'expression est augmentée (figure 4.15) et pour

celle dont l'expression est diminuée (figure 4.16). Ce logiciel a permis de classer 61 des

106 ARNm dont l'expression est augmentée et 122 des 279 ARNm dont l'expression est

diminuée.

Quatre champs d'activité métaboliques contiennent 57 des 61 ARNm classés dont

l'expression est augmentée (figure 4.15). De façon intéressante, trois de ces catégories sont

reliées avec le cancer, la croissance ou la prolifération cellulaire (figure 4.15). Pour ce qui

est des 122 ARNm classés dont l'expression est réduite, 112 ont été classés dans sept

champs d'activité (figure 4.16). Certaines voies affectées sont liées à des maladies

génétiques, dermatologiques, du système sanguin ou du système reproducteur. Cependant,

comme c'était le cas pour les ARNm dont l'expression est augmentée, la majorité des

activités détectées par l'analyse comparative sont reliées avec la croissance, la prolifération

cellulaire ou le cancer (figure 4,16). Des illustrations des 11 champs d'activité

métaboliques identifiés par « lngenuity Systems Inc » sont présentées aux figures 4.17-

4.20, augmentation, et 4.21-4.27, diminution. La légende des symboles utilisés dans les

figures est présentée à la figure 4.28. Les ARNm des protéines, dont l'expression varie dans

les analyses de microarrays, sont en gris et leur localisation cellulaire est indiquée. La

figure 4.25 ne contient pas la localisation cellulaire des protéines. Il semble qu'il y ait eu un

défaut lors de l'analyse par « lngenuity Systems Inc ». Comme notre abonnement a pris fin,

il n'est pas possible de refaire cette analyse.

166

Il est intéressant de noter que ces analyses suggèrent que les cellules DKO tentent

de compenser pour leur défaut de croissance en augmentant l'expression d'ARNm codant

pour des gènes promoteur de croissance, comme par exemple myc (figure 4.17) et en

diminuant l'expression d'ARNm codant pour des régulateurs négatifs de la croissance ou

de la prolifération comme par exemple des inducteurs d'apoptose (figure 4.21). Il est aussi

important de rappeler que la protéine p53 est diminuée dans les cellules DKO (figure 4.7).

Ainsi, malgré les efforts des cellules DKO à modifier l'expression d'ARNm qui devraient

théoriquement favoriser leur croissance, celles-ci ne poussent pas plus vite. Il est

raisonnable de penser que ces cellules n'y parviennent pas parce qu'elles n'arrivent pas à

corriger ou à contourner les défauts de la biogenèse des ribosomes causés par les problèmes

de transformations de l'ARN ribosomal précurseur.

16%

20%

37%

20%

■ Cancer, croissance et prolifération cellulaire, maladie du système reproducteur

■ Transport des molécules, biochimie de petites molécules, métabolisme des glucides

□ Cancer, mort cellulaire, maladies respiratoires

l Cycle cellulaire, croissance et prolifération cellulaire, développement et fonctionnement du système sanguin

I Autres

Figure 4.15: Classement, dans différents champs d'activité métaboliques, des ARNm dont l'expression est augmentée. Le pourcentage par catégorie correspond au nombre d'ARNm pour cette catégorie sur le nombre total d'ARNm classés. 57 ARNm, sur les 61 classés, sont regroupés dans quatre champs d'activité métaboliques majeurs.

167

14%

10%

11% 12%

11%

■ Croissance cellulaire et □ Métabolisme des acides ■ Développement et fonction prolifération, mort aminés, transport du système nerveux, cellulaire, développement moléculaire, biochimie de morphologie tissulaire, et fonctionnement du petites molécules développement et fonction système sanguin B Morphologie cellulaire, du tissu conionctif

■ Maladies génétiques et du développement cellulaire, ■ Développement cellulaire, système sanguin, cancer croissance cellulaire et morphologie cellulaire - r a n c e r maladie du prolifération,

■ Lancer, maïaaie au développement et système reproducteur, fonctionnement du maladies dermatologues s y s t è m e

sanguin ■ Autres

Figure 4.16: Classement, dans différents champs d'activité métaboliques, des ARNm dont l'expression est diminuée.

Le pourcentage par catégorie correspond au nombre d'ARNm pour cette catégorie sur le nombre total d'ARNm classés. 112 ARNm, sur les 122 classés, sont regroupés dans sept champs d'activité métaboliques majeurs.

En conclusion, les différentes analyses des résultats provenant des micropuces à

ADN permettent de conclure que les défauts observés dans les cellules DKO ne sont causés

ni par une variation de l'expression d'ARNm impliqués dans la biogenèse des ribosomes,

ni par un défaut d'expression général d'ARNm de protéines nucléolaires. Ainsi, la perte de

la méthylation de l'ADN ribosomal est vraisemblablement la cause directe des problèmes

168

observés, entre-autre dans la transcription ribosomale, dans la transformation de l'ARNr

précurseur et dans la croissance cellulaire, dans les cellules déficientes pour les gènes

codant pour DNMT1 et DNMT3b.

îfxû«cëïïûi«rspM«" /■ "\ IMknown

AMOl

EC:U64)

Figure 4.17: Cancer, croissance, prolifération cellulaire et maladie du système reproducteur. Champs d'activité métabolique représentant 37% des ARNm dont l'expression est augmentée (figure 4.11). 23 ARNm sur 61 sont dans cette catégorie. MYC est indiqué par une flèche. La localisation cellulaire des différentes protéines est aussi indiquée.

169

I *n,i. l'Unijf S|).icf

Figure 4.18: Transport des molécules, biochimie de petites molécules et métabolisme des glucides. Champs d'activité métabolique représentant 20% des ARNm dont l'expression est augmentée (figure 4.11). 12 ARNm sur 61 sont dans cette catégorie. La localisation cellulaire des différentes protéines est aussi indiquée.

170

Figure 4.19: Cancer, mort cellulaire et maladies respiratoire. Champs d'activité métabolique représentant 20% des ARNm dont l'expression est augmentée (figure 4.11). 12 ARNm sur 61 sont dans cette catégorie. La localisation cellulaire des différentes protéines est aussi indiquée.

171

IF27(lnflU<l«S EGH29)

Figure 4.20: Cycle cellulaire, croissance et prolifération cellulaire, développement et fonctionnement du système sanguin. Champs d'activité métabolique représentant 16% des ARNm dont l'expression est augmentée (figure 4.11). 10 ARNm sur 61 sont dans cette catégorie. La localisation cellulaire des différentes protéines est aussi indiquée.

172

ExtracelJuïàr Space ."inkfï

^ t f * 3

CD160

"à^^N^TjS^^v MAC ( i n < l ( l î ^ * & « 9 a i _ _

Cviopl.

PFPtRlfîB

Nutlîujy

BTB12

Figure 4.21: Croissance cellulaire et prolifération, mort cellulaire, développement et fonctionnement du système sanguin.

Champs d'activité métabolique représentant 20% des ARNm dont l'expression est diminuée (figure 4.12). 25 ARNm sur 122 sont dans cette catégorie. La localisation cellulaire des différentes protéines est aussi indiquée.

173

TCOFl (lnt)urf«J EC:6949)

Ucfcd^tC3^49(

Figure 4.22: Maladies génétiques, du système sanguin et morphologie cellulaire.

Champs d'activité métabolique représentant 12% des ARNm dont l'expression est diminuée (figure 4.12). 15 ARNm sur 122 sont dans cette catégorie. La localisation cellulaire des différentes protéines est aussi indiquée.

174

AHXA.l 1

! EG:î«S)*

Figure 4.23: Métabolisme des acides aminés, transport moléculaire et biochimie des petites molécules.

Champs d'activité métabolique représentant 11% des ARNm dont l'expression est diminuée (figure 4.12). 14 ARNm sur 122 sont dans cette catégorie. La localisation cellulaire des différentes protéines est aussi indiquée.

175

r.tr.Kclki l . i i Space îlnkn CPCN

Plasma Mtmbr i [K

KÇ3M2

JOHAtA

CytoplJim

Nucl«ut

- <MA3

Figure 4.24: Morphologie cellulaire, développement cellulaire et cancer.

Champs d'activité métabolique représentant 11% des ARNm dont l'expression est diminuée (figure 4.12). 14 ARNm sur 122 sont dans cette catégorie. La localisation cellulaire des différentes protéines est aussi indiquée.

176

«•*' / \ fc

Figure 4.25: Cancer, maladie du système reproducteur et maladies dermatologiques.

Champs d'activité métabolique représentant 11% des ARNm dont l'expression est diminuée (figure 4.12). 13 ARNm sur 122 sont dans cette catégorie. La localisation cellulaire des différentes protéines n'est pas indiquée.

177

Exuaccllular Space Unkmiwn

Hïïmï MëmBVânt

Figure 4.26: Développement et fonctionnement du système nerveux, morphologie tissulaire, développement et fonctionnement du tissus conjonctif.

Champs d'activité métabolique représentant 11% des ARNm dont l'expression est diminuée (figure 4.12). 13 ARNm sur 122 sont dans cette catégorie. La localisation cellulaire des différentes protéines est aussi indiquée.

178

X:171127)

Figure 4.27: Développement cellulaire, croissance cellulaire et prolifération, développement et fonctionnement du système sanguin.

Champs d'activité métabolique représentant 10% des ARNm dont l'expression est diminuée (figure 4.12). 12 ARNm sur 122 sont dans cette catégorie. La localisation cellulaire des différentes protéines est aussi indiquée.

n Cytokines

0 Enzyme

i i * J

Facteur de croissance

Récepteur couplé aux protéines G

i i ■ :

! 1 Canal ionique

V Kinase

Récepteur nucléaire

Autre

i i

Kinase

Récepteur nucléaire

Autre o Kinase

Récepteur nucléaire

Autre

o Peptidase

/ \ Phosphatase

o> Régulateur transcriptionel

O Régulateur traductionel

0 Récepteur transmembranaire

/ ~ \ Transporteur

Figure 4. 28: Légende pour les figures 4.17 à 4.27.

180

Chapitre 5 : Inactivation génique d'UBF, de TAFj68 et de TAF]95 dans les souris

5.1. Avant-propos À l'origine, mon projet de doctorat était de poursuivre mes travaux commencés à la

maîtrise et qui consistaient en la production de souris et de lignées cellulaires dans

lesquelles les gènes codant pour UBF, TAF168 ou TAFi95 auraient été inactivés de façon

ciblée par recombinaison homologue. Ce chapitre présente la progression de ces travaux

jusqu'à ce jour ainsi que les perspectives d'avenir de ces projets qui sont poursuivis, depuis

septembre 2006, par Françoise Morin, étudiante à la maîtrise dans le laboratoire. Les

manipulations des cellules embryonnaires souches ont été faites majoritairement par Michel

Tremblay et en partie par moi-même. Michel Tremblay est aussi responsable de la gestion

des colonies de souris. Les souris chimériques ont été réalisées par Michèle Carter du

service de transgenèse du Centre de Recherche de l'Hôtel-Dieu de Québec. La totalité des

résultats présentés dans ce chapitre proviennent d'expériences que j 'ai faites et les résultats

d'une analyse faite avec la participation de Françoise Morin est mentionnée.

5.2. Introduction Les travaux de notre laboratoire dont les résultats ont été publiés dans la revue

« Molecular Cell » en 2001 et 2006 (Stefanovsky et al., 2006a; Stefanovsky et al., 2001b),

et qui sont présentés dans cette thèse aux chapitres 2 et 3, montrent l'importance de la

protéine UBF dans le contrôle de la transcription ribosomale et dans sa réponse aux

facteurs de croissance. Brièvement, ces travaux montrent que la transcription ribosomale

est rapidement augmentée suite à une stimulation par le facteur de croissance EGF ou par

l'ajout de sérum et que la phosphorylation d'UBF, par la MAP kinase ERK, est nécessaire à

cette activation (chapitre 2). Par la suite, nous avons montré que l'EGF et le sérum régulent

l'élongation de la transcription et que la phosphorylation d'UBF par ERK est nécessaire

pour permettre le passage de l'ARN polymérase I et donc la transcription (chapitre 3). Ces

résultats suggèrent qu'w vivo UBF est un régulateur de l'élongation de l'ARN polymérase I

via sa capacité à remodeler l'ADN.

181

Des études faites chez la levure suggèrent l'existence d'un homologue fonctionnel

d'UBF (Gadal et al., 2002). Cette protéine, nommée Hmol, est essentielle pour une

activation maximale de la transcription ribosomale. Son inactivation entraîne une réduction

de deux fois de la production des ARN ribosomaux ainsi qu'un défaut de croissance (Gadal

et al., 2002). La double inactivation d'Hmol et d'une sous-unité non essentielle de la

polymérase est létale. Ceci est vrai pour la double inactivation d'Hmol avec trois sous-

unités différentes soit Rpal2, Rpa43-24 et Rpa49. Rpa49 est l'homologue de PAF53 qui est

la sous-unité avec laquelle interagit UBF chez les mammifères. Cette létalité peut être

renversée si les gènes ribosomaux, portés par un vecteur, sont transcrits par l'ARN

polymérase II (Gadal et al., 2002). Ce résultat suggère que cette létalité est uniquement

attribuable à un défaut dans la transcription ribosomale. L'ensemble de ces résultats

suggèrent que la protéine Hmol n'est pas essentielle, mais qu'elle est requise pour une

transcription ribosomale maximale chez la levure S. cerevisiae. Des études plus récentes

d'analyse de localisation d'Hmol ont montré que cette protéine, comme UBF, est localisée

sur toute la région ribosomale transcrite (Hall et al., 2006). Hmol serait aussi localisée sur

les régions promotrices de la majorité des protéines ribosomales, mais elle ne serait pas

impliquée dans la régulation de la transcription de ces protéines (Hall et al., 2006). Cette

étude montre aussi que l'inactivation d'Hmol empêche la transformation de l'ARN

ribosomal précurseur en diminuant le clivage de certaines régions de près de trois fois. Les

auteurs suggèrent qu'Hmol fait partie d'un groupe de protéines important pour la

corégulation des différentes étapes de la biogenèse des ribosomes. Les résultats présentés

au chapitre 3 et 4 suggèrent aussi l'existence d'une telle corégulation.

Comme il a été mentionné dans l'introduction de cette thèse, il existe d'importantes

différences entre les résultats provenant d'expériences in vitro versus in vivo concernant le

rôle essentiel d'UBF. Trois approches sont présentées dans ce chapitre et elles permettront

vraisemblablement de déterminer la nécessité ou non de la protéine UBF dans la

transcription ribosomale et dans la vie. La première de ces approches consiste à produire

des souris dans lesquelles les deux allèles codant pour UBF seront inactivés, souris

« Knock-out » (sections 5.3 à 5.6). La deuxième approche consiste en la production de

souris « Knock-out » conditionnelle (section 5.8). C'est-à-dire qu'il est possible de choisir

182

le lieu et/ou le moment de l'inactivation génique. La troisième approche consiste à produire

une lignée de cellules embryonnaires souches UBF-/- complémentée par un transgène,

ADNc d'UBF, sous le contrôle d'un promoteur inductible (section 5.7). Comme il est

possible qu'UBF ne soit pas essentielle ou alors que cette protéine ait une fonction

différente de celle connue dans la transcription ribosomale, il est nécessaire de comparer le

phénotype obtenu, suite à son inactivation génique, avec ceux résultants de l'inactivation

génique d'autres protéines spécifiques à l'ARN polymérase 1. Les protéines choisies qui

serviront de contrôle pour la perte spécifique de la transcription ribosomale sont TAF|68,

TAFi95 et Rrn3.

Les trois TAFiS, TAFi48, TAF168 et TAFi95, en association avec TBP, forme le

complexe SL1 qui est essentiel pour l'initiation de la transcription par l'ARN polymérase I.

Malgré le fait que TBP participe aussi à la transcription par les ARN polymérases II et III,

les TAFi n'ont aucune autre activité connue ou suspectée. Conséquemment, l'inactivation

génique de TAF[68 et de TAF]95 devrait entraîner un phénotype qui soit spécifique à un

défaut dans la transcription ribosomale. De plus, l'inactivation génique de TAF168 devrait

entraîner, en plus de la perte de cette protéine, un démantèlement du complexe SL1 puisque

c'est le TAFi le plus conservé au cours de l'évolution ce qui suggère qu'il ait un rôle central

au sein de ce complexe. Ainsi, si UBF possède une activité supplémentaire à celle qu'il a

dans la transcription par l'ARN polymérase I, le phénotype lié à son inactivation devrait

être différent de celui de l'inactivation de TAF168 ou de TAF[95.

L'inactivation génique de Rrn3 a quant à elle été faite (Yuan et al., 2005). Cette

inactivation est létale et les embryons meurent in utero au jour embryonnaire 9,5.

L'analyse, par RT-PCR, de l'ARN ribosomal précurseur montre que la transcription

ribosomale n'est pas détectable dans ces embryons. Ces résultats suggèrent aussi qu'un

certain niveau de transcription ribosomale a eu lieu puisque, logiquement, la croissance

n'est pas possible sans la production de ribosomes. Les auteurs montrent que l'ARNm

maternel de Rrn3 est stable et 30 fois plus abondant dans les ovocytes que dans les cellules

somatiques. Yuan et collègues suggèrent que cet ARNm maternel est responsable de la

transcription ribosomale au début du développement et qu'une autre protéine compense

183

pour la perte de Rrn3 jusqu'au jour 9.5. Toutefois, l'absence de la protéine Rrn3 endogène

n'est pas compatible avec la vie au-delà du jour embryonnaire 9.5. De plus, les embryons

Rrn3-/- sont de 10 à 20 fois plus petits que les embryons sauvages de la même portée et

leur développement est retardé. La petite taille des individus homozygotes serait expliquée

par une augmentation de l'apoptose (Yuan et al., 2005). Finalement, l'inactivation géniquc

de Rrn3 dans des cellules MEF, pour « Mouse Embryonnic Fibroblast », entraîne un

désassemblement du nucléole, un arrêt du cycle cellulaire, une stabilisation de p53 et une

induction de l'apoptose (Yuan et al., 2005). La comparaison des phénotypes qui seront

obtenus suite à l'inactivation d'UBF, de TAF168 et de TAFi95 et de celui publié pour

l'inactivation de Rrn3 permettra de déterminer la nécessité, ou non, de ces protéines dans

l'établissement et le maintien de la vie. L'inactivation de sous-unités essentielles de l'ARN

polymérase I pourrait aussi être utilisé comme modèle auquel serait comparé les différents

phénotypes obtenus pour l'inactivation de protéines considérées comme étant Poli-

spécifiques. Pour l'instant, aucune sous-unité de l'ARN polymérase I n'a été génétiquement

inactivé chez les mammifères. La comparaison des phénotypes permettra probablement de

déterminer si UBF, Rrn3, TAF168 et TAFi95 sont impliquées dans le même processus

cellulaire, la transcription ribosomale, ou si UBF possède une fonction supplémentaire à

son rôle dans cette transcription.

5.3. L'inactivation génique chez la souris Les deux approches utilisées pour vérifier si les protéines UBF, TAF]68 et TAFi95

sont essentielles est l'inactivation génique individuelle de ces gènes dans les souris, souris

« Knock-out » ou KO, et la production de lignées cellulaires ES-/- complémentée par un

transgène. Les souris KO sont un modèle animal dans lequel il est possible d'étudier les

effets de la complète perte de fonction d'une protéine dans toutes les cellules d'un

organisme vivant. Une autre méthode fréquemment utilisée pour l'étude de perte de

fonction est l'interférence à l'ARN. Malgré l'efficacité relative de ce système, il a le

désavantage de ne cibler que l'ARNm, et non le gène lui-même, ce qui fait en sorte qu'une

disparition de 100% de la protéine est presque impossible. L'utilisation de siRNA « short

interférence RNA » permet une réduction de la protéine qui est seulement transitoire et sa

184

transfection entraîne des défauts de croissance. L'utilisation de shRNA « short hairpin

RNA» a l'avantage de produire des lignées dont l'expression de la protéine est réduite de

façon stable, mais il reste toujours à une expression résiduelle. De plus, ces systèmes,

contrairement à l'utilisation de cellules ES, ne permettent pas de produire, chez les

mammifères, un organisme dans lequel toutes les cellules sont dépourvues du gène et donc

de la protéine d'intérêt. Avec l'interférence à l'ARN, il est impossible d'étudier le rôle

d'une protéine dans le développement d'un mammifère. La production de souris KO

s'avère donc le seul modèle animal permettant de répondre de façon définitive à la question

de la nécessaire présence d'UBF, TAF168 ou TAF|95 dans les cellules mammifères.

Le deuxième système expérimental développé pour étudier les fonctions d'UBF, de

TAF168 et de TAF[95 est l'établissement de lignées cellulaires ES-/- complémentées par un

transgène inductible (section 5.7). Normalement, il est possible d'établir des lignées

cellulaires à partir de cellules provenant des souris KO. Cependant, comme il est probable

qu'UBF, TAF168 et TAFi95 soient essentielles à la survie cellulaire, il s'avérait nécessaire

de développer une lignée cellulaire qui pourrait survivre tout en permettant de faire des

études biochimiques pour étudier les effets de la perte de ces protéines. Les cellules ES+/-

qui serviront à produire les souris KO seront aussi utilisées pour produire une lignée

cellulaire ES dans laquelle les deux allèles codant pour le gène d'intérêt seront inactivés. La

protéine endogène ne sera donc plus produite. La survie de ces cellules sera assurée par la

présence d'un transgène inductible portant l'ADNc de la protéine d'intérêt. Le fait que ce

transgène soit sous le contrôle d'un promoteur inductible permettra de faire varier le niveau

d'expression de la protéine d'intérêt et d'étudier les effets de ces variations d'expression,

par exemple, sur la transcription ribosomale et la croissance cellulaire. La section 5.7 décrit

en détails l'établissement de ces lignées et les études qui seront faites à l'aide de ce

système.

La production de souris KO se fait en plusieurs étapes successives (figure 5.1). La

première étape consiste à produire un vecteur qui sera intégré de façon stable dans le

génome des cellules embryonnaires souches de souris de façon à ce que ces cellules

contiennent un allèle sauvage et un allèle muté, ou inactif, du gène d'intérêt. Les cellules

185

ES sont des cellules non différenciées et pluripotentes provenant de la masse interne d'un

blastocyte, jour embryonnaire 3,5. Ces cellules ont la capacité de participer à la formation

de tous les tissus de la souris. Suite à l'électroporation du vecteur, contenant F ADN qui

sera intégré au génome, une sélection est appliquée sur les cellules pour forcer l'intégration

de FADN exogène et la formation de colonies de cellules résistantes. Après avoir amplifié

les cellules potentiellement positives, parce que résistantes, FADN de ces cellules est

analysé pour déterminer si l'intégration du vecteur s'est faite au bon locus.

Les cellules ayant un allèle muté sont par la suite injectées dans des blastocytes

murins qui sont ensuite placés dans une femelle pseudogestante qui portera les embryons

jusqu'à leur naissance (figure 5.1). Les femelles pseudogestantes sont réceptives à

l'implantation des blastocystes parce qu'une fausse gestation a été provoquée par leur

accouplement avec des mâles vasectomisés. Les souris qui seront utilisées pour les

croisements suivants sont celles dites chimériques parce que les cellules injectées, mutées,

et les cellules du blastocystes, sauvages ont participé à leur formation (figure 5.1). Ces

souris sont normalement identifiables par leur pelage bicolore. En effet, les cellules

injectées sont d'une couleur différente des cellules du blastocytes récepteur ce qui produit

un pelage bicolore chez les individus chimériques. Puisque les cellules injectées ont

participé à la formation du pelage, elles ont théoriquement la capacité de participer à la

formation d'autres lignées cellulaires dont la lignée germinale qui est essentielle à la

production de souris KO. Ainsi, en croissant des souris chimériques entre elles ou avec des

souris sauvages, il est théoriquement possible d'obtenir des individus hétérozygotes, ou +/-

(figure 5.1). Ces individus possèdent un allèle sauvage, +, allèle sauvage provenant d'une

gamète de l'individu sauvage ou de l'individu chimérique, et un allèle muté, -, allèle muté

provenant d'une gamète de la souris chimérique. Le croisement des souris hétérozygotes

entre elles mène à la production théorique (ou mendélienne) de 25% d'individus sauvages

(+/+) de 50% d'individus hétérozygotes (+/-) et de 25% d'individus homozygotes (KO ou

-/-) (figure 5.1). Si la protéine est essentielle à la vie, il n'y aura pas de naissance

d'individus -/-. Le sacrifice des mères et des embryons viables s'avère alors nécessaire pour

analyser les individus -/- in utero et pour déterminer à partir de quel stade embryonnaire la

protéine est essentielle à la vie.

186

DNA Normal (*)gene X

1 * î

Veclor

Defeclive (-) gène X

Marker gène

ES oells with (-) gène X

Embfyo cells with (♦) gène X

Early -mouse embryo

dzO, <cO> *zO> *zO, ♦/♦ *i- ♦/- v*

V *l* *l- ♦/- -I-

Isolale gène X and insert il into a short pièce of DNA that can be easily manip­ulâtes (i e.. a veclor). Inactivate the gène by inserting a marker gène that makes the cells résistant to certain antibiotics,

Transfer vector with (-) gène X into emtxyonic stem (ES) ceHs

Grow ES celis m antibiotic-contaming médium: only cells that hâve incorpo-rated (-) gène X survive in tins médi­um.

Inject ES cells with (-) gène X into early mouse embryos

Transfer embryos into aurrogato moth-ers for embryonic development, the resulting pups (i e.. chimeras) contain (+) gène X in some oells and (-) gène X in other cells. Mate chimeras with normal mlce.

Identify pups that carry one (♦) and one (-) copy of gène X and mate those animais with each other.

Analyze mouse pups; about 25 per­cent will hâve mherited the {-) gène from bolh parents and will completely lack the (♦) gène (i e. "knockout mice')

Figure 5.1 Schématisation des étapes menant à l'inactivation génique chez la souris. Le vecteur contenant l'ADN servant à l'inactivation génique est électroporé dans les cellules embryonnaires souches de souris. Ces cellules sont ensuite sélectionnées avec un antibiotique et leur ADN est analysé pour confirmer que l'insertion s'est faite au bon locus. Par la suite les cellules provenant de clones positifs sont insérées dans des blastocystes murins et ceux-ci sont placés dans des mères porteuses. Les souris chimériques, qui contiennent des cellules sauvages et les cellules ES+/-, sont croisées entre elles ou avec des souris sauvages. Les individus hétérozygotes, +/-, qui naîtront de ces croisements seront ensuite croisés entre eux. Normalement, 25% des individus issus de ces croisements seront homozygotes, -/-, pour le gène d'intérêt. Accessible au public sur le site Internet de National Institute on Alcohol Abuse and Alcoholism (NIAAA).

187

5.4. L'inactivation génique d'UBF avec le vecteur pGKDTAmUBFKO

Le vecteur utilisé pour la construction servant à l'inactivation génique dans les

souris, pGKDTA, contient un gène de sélection positive, qui doit être inséré dans le

génome pour permettre la sélection positive de la cellule, et un gène de sélection négative,

qui est perdu par recombinaison homologue. L'utilisation d'un système avec sélection

positive et négative favorise les événements de recombinaison homologue nécessaires au

ciblage génomique de l'ADN qui doit être intégré. En ce qui concerne le vecteur utilisé, le

gène de sélection positive neo est exprimé constitutivement et il permet la résistance au

G418. Le gène de la toxine diphtérique, DTA, est aussi exprimé constitutivement, mais

comme son produit est toxique, il doit être perdu pour que les cellules survivent.

Classiquement, c'est le gène de la thymidine kinase, tk, qui était utilisé comme sélection

négative, cependant le gène DTA s'est avéré plus efficace en augmentant le pourcentage de

recombinaison homologue et conséquemment, en diminuant le nombre de faux positifs.

Le vecteur produit pour l'inactivation génique d'UBF contient des séquences

génomiques, majoritairement introniques, situées dans la partie amino terminale du gène

codant pour UBF (figure 5.2A). Le gène lacZ, qui code pour la B-galactosidase a été inséré

en phase avec les 4 premiers codons de l'exon 3 d'UBF pour produire une protéine de

fusion sous le contrôle du promoteur endogène d'UBF. Il est à noter que le premier exon

d'UBF est non codant, en vert sur la figure 5.2A, et que le deuxième, qui contient l'ATG,

est très court, 57 paires de bases, 19 acides aminés. L'insertion se situe donc à seulement 69

paires de bases de l'ATG dans la séquence de l'ADNc. À titre indicatif, l'exon huit, en bleu

dans la figure 5.2A, correspond à une partie de la deuxième boîte HMG1 de la protéine

UBF (figure 5.3). Ainsi, la protéine de fusion contient seulement l'extrémité N-terminale

d'UBF. Il est donc peu probable que la protéine UBF-B-galactosidase possède une activité

résiduelle puisqu'elle ne contient que les 23 premiers acides aminés d'UBF qui en contient

765.

L'utilisation de cette protéine de fusion permettra de déterminer, chez les souris

hétérozygotes, s'il existe une corrélation entre le niveau d'expression d'UBF et le niveau de

188

prolifération cellulaire d'un tissu donné. Les deux segments de séquences génomiques

utilisées, qui sont majoritairement introniques (figure 5.2A) sont nommées région 5', car en

amont des gènes codant pour la B-galactosidase et neo, et région 3' car en aval de ces

mêmes gènes (figure 5.2B). Le gène codant pour DTA est en aval de la région 3' (figure

5.2B). La région 5' contient la séquence génomique de -797pb à + 1382pb par rapport à

l'ATG et la région 3' comprend la séquence génomique de +1409pb à +3580pb (figure

5.2A). Comme mentionné, les 4 premiers codons de l'exon 3 sont présents dans la région

5'. La fin de l'exon 3 se retrouve hors phase au début de la séquence utilisée comme région

3' (figure 5.2B). Les 31 paires de bases situées entre les 12 utilisées dans la région 5', en

phase avec le gène lacZ, et les 130 situées dans la région 3' n'ont pas été utilisées dans la

production du vecteur pGKDTAmUBFKO.

Une fois terminé, le vecteur nommé pGKDTAmUBFKO fut linéarisé et envoyé à

l'unité transgénique de McGill pour l'électroporation dans les cellules ES, la sélection avec

le G418 et l'amplification des clones résistants. Ce vecteur fut envoyé deux fois à l'unité

transgénique de McGill. La première électroporation a produit 220 clones et la deuxième

480 clones. Suite à leur réception, les cellules furent lysées et l'ADN génomique fut extrait.

La quantité d'ADN obtenu, pour un certain nombre de clones, fut mesurée par fluorométrie

(section 5.9.2).

5.4.1. Détermination des clones positifs Normalement, l'ADN génomique des clones potentiellement positifs est analysé par

Southern. Cependant, cette expérience nécessite un minimum de 5u,g d'ADN pour avoir

une bonne résolution. Toutefois, la quantité d'ADN obtenue pour les clones contenant le

plus d'ADN n'était que de 3jxg. Ces quantités d'ADN sont insuffisantes pour faire une

analyse de type Southern. Une méthode d'analyse par réaction PCR a donc été développée.

Exon 1 : Non codant

Exon 2 : Contient l'ATG, 57 paires de bases, 9 acides aminés

Exon 3 : Insertion après les

premiers codons

189

bxon

/

Insert 4 prer

Exon 8 : Présent chez UBFl Absent chez UBF2

5'UBF -797 à+1392

3'UBF +1409 à+3580

B) 3 (les 4 premiers codons)

Figure 5.2 pGKDTAmUBFKO.

A) Schématisation du gène codant pour UBF. Les exons sont en rouge et les introns en noir. Le premier exon, non transcrit, est indiqué et est en vert. Les exons contenant l'ATG et l'insertion, exons 2 et 3, sont aussi indiqués. L'exon 8 qui est présent chez UBFl, mais non chez UBF2 est indiqué et est en bleu. Les séquences génomique utilisées pour la construction pGKmUBFKO sont indiquées par des flèches sous le graphique. B) Schématisation du vecteur pGKDTAmUBFKO. Les séquences intronique du gène codant pour UBF sont en noir et les séquences exoniques codante en rouge. La séquence codant pour le premier exon non codant est en vert. Les exons sont identifiés par des flèches et numérotés. La séquence codant pour fi-galactosidase est en jaune, celle codant pour neo en

190

magenta et celle codant pour DTA en bleue. Les séquences en cyan sont celles du vecteur. La ligne verticale représente le site de restriction utilisé pour la linéarisation du vecteur avant l'électroporation.

Cor» UBF

Figure 5.3: La protéine UBF de mammifère A) Structure de la protéine UBF mammifère. Le domaine amino terminal de dimérisation, les six boîtes HMG1, la région minimale pour l'activation de la transcription (« core UBF ») et la queue carboxy terminale acide sont indiqués. Provient de Moss et al, CMLS 2006.

5.4.1.1. Détermination des clones positifs par PCR Pour déterminer que l'insertion de l'ADN exogène s'est faite au bon locus, il faut

utiliser un oligonucléotide dirigé contre une séquence génomique d'UBF localisée à

l'extérieur de la construction et un oligonucléotide localisé dans la construction. Cet

oligonucléotide peut être dirigé contre le gène de résistance au G418 neo ou contre le gène

lacZ. L'utilisation de ces oligonucléotides permet, théoriquement, l'amplification d'un

fragment qui ne peut pas provenir d'une intégration aléatoire de l'ADN exogène puisqu'un

des deux oligonucléotides est situé hors de la construction et l'autre dans la construction.

Toutefois, il s'est avéré qu'il y avait aussi amplification de faux positifs. Cette méthode de

détection s'est avérée difficile à mettre au point vu les très faibles quantités d'ADN

utilisées, 1 jxg d'ADN, et la relative pureté de l'ADN génomique extrait. En effet, comme il

était impossible de faire une purification par phénol/chloroforme sur tous les clones, il

restait divers contaminants qui diminuaient l'efficacité de la réaction PCR. La première

étape de la mise au point de la réaction PCR a été la reproduction, dans un vecteur, du

contexte génomique nécessaire à l'amplification. Deux vecteurs furent produits. Un

reproduisant une partie de la région génomique en 5' du lieu de l'insertion de l'ADN

exogène et un reproduisant une partie de la région en 3' de l'insertion (figure 5.4).

Hincll Hlncll

EcoRI

Xmnl

Figure 5.4 Criblage par PCR des clones potentiellement positifs. A) Représentation de l'allèle muté suite à l'insertion de l'ADN exogène provenant de pGKDTAmUBFKO. B) Agrandissement de la région amplifiée. Les régions d'UBF utilisées pour la construction du vecteur sont en noires et les exons sont en rouges sauf le premier exon non codant qui est en vert et l'exon 8 qui est en bleu. L'ATG est indiqué par une tête de flèche rouge et les deux oligonucléotides utilisé pour le criblage par des têtes de flèches noires. La longueur du fragment amplifié est aussi indiquée. Le gène codant pour fi-galactosidase est en jaune et celui codant pour neo est en magenta. Les séquences provenant du vecteur sont en cyan. Les sondes et les enzymes utilisées pour l'analyse par Southern sont indiquées à titre de repère pour comparaison avec la figure 5.4.

Il est à noter que les tests ont été faits en utilisant 0,5, 1, 5 ou 10 pmol de vecteur ce

qui correspond à la quantité de fragment de 3 kb que l'on retrouve dans 500 ng, 1, 5 ou 10

\ig d'ADN génomique. La réaction a été optimisée pour l'utilisation de 1 \ig d'ADN

génomique, 1 pmol de vecteur, puisqu'il y avait peu d'ADN génomique provenant des

clones. En plus du vecteur, les réactions d'amplification contenaient aussi de l'ADN

génomique. L'utilisation du vecteur contenant une partie de la région génomique en 3' du

lieu de l'insertion, d'un oligonucléotide localisé dans le gène codant pour neo et un autre

localisé à l'extérieur de la région 3' utilisée pour produire pGKDTAmUBFKO a permis

l'amplification d'un fragment de 2,7kb (figure 5.4). Des tests ont été faits en utilisant

192

seulement de l'ADN génomique de façon à vérifier qu'il n'y avait pas d'amplification de

fragment de 2,7kb qui soit non spécifique ce qui n'était pas le cas. Malgré l'essai d'une

combinaison d'oligonucléotides, de différentes températures d'hybridation, de différentes

enzymes et de différents temps de dénaturation initiaux, la réaction d'amplification de la

région 5' d'UBF n'a pas fonctionné. Cette région est très GC riche et c'est probablement ce

qui pose problème lors de l'amplification. Les deux oligonucléotides amplifiant le fragment

de 2,7kb dans la région située en 3' de l'insertion furent donc utilisés pour le criblage des

clones. Neuf des 220 premiers clones reçus montraient une amplification du fragment de

2,7 kb. Ces clones furent réamplifiés par l'unité transgénique de McGill et l'ADN

génomique fut extrait pour être analysé par Southern (section 5.9.2). Les 480 clones de la

deuxième électroporation furent aussi analysés avec les mêmes oligonucléotides. J'ai fait

les PCR pour la première plaque et Françoise Morin a fait les PCR pour les 4 autres

plaques. 73 clones positifs furent identifiés et dix furent commandés à McGill. Françoise

Morin les analyse présentement par Southern.

5.4.1.2. Confirmation des clones positifs par analyse de type Southern Puisque la région 3 ' a été confirmée comme étant positive pour 9 des 220 clones par

PCR, le premier Southern a consisté en la vérification de la région 5' de ces clones. Le

protocole de la digestion de l'ADN génomique et l'analyse de type Southern est décrit dans

la section 5.9.4. L'ADN des 9 clones fut digéré avec l'enzyme EcoRI et une sonde dirigée

contre la région 5' fut utilisée (figure 5.4). Un seul clone, 1F2, a donné les deux bandes

attendues soit la bande provenant de la digestion de l'allèle sauvage, 7,9kb (figure 5.5A et

tableau 5.1) et la bande provenant de la digestion de l'allèle muté, 3,7kb (figure 5.5B et

tableau 5.1). Par la suite, ce clone fut digéré avec EcoRI, HincII et Xmnl en utilisant une

sonde dirigée contre la région 5' (EcoRI, Xmnl) ou 3' (EcoRI, HincII) pour confirmer que

le clone était positif (figure 5.5 et tableau 5.1). La comparaison du patron de bandes obtenu

par Southern (figure 5.6) à celui attendu (figure 5.5 et tableau 5.1) confirme que le clone

1F2 est positif. Il est à noter qu'il y a une bande non spécifique à environ 14,5 kb lors de la

digestion avec Xmnl et une à environ 4,5 kb lors de la digestion avec HincII.

A) 193

Hin?lEopRIEooRI

1,9 kb HinolU kHincll

l*rnrj<mnl Hincll ; c M Hincll Hincll >jnnnl Xffinl E c„p, E p o R |

::I,,I,I -4 7.5 kb

EcoRI ^

— ^ . Xmnl

7,9 kb EcoRI

»)

S ^ Hincll HlnclHiticll Hirjcll

12,9 kb ,<&&

1,3 kb

J

EcoRI EcoRI ^ ~

► Xmnl

Xmnkjj-

8,7 kh

iim.-ii H.I...H Mer., il j :,,,,.! : r.....i (. ..h 11. ..i i

Figure 5.5 Vérification de l'insertion de l'ADN provenant du vecteur pGKDTAmUBFKO au locus codant pour UBF.

A) Représentation du gène codant pour UBF dans son contexte génomique (allèle sauvage). Les régions en noires sont les régions d'UBF utilisées pour la construction et celles en vert sont non utilisées. Les exons sont en rouge sauf le premier exon non codant qui est en vert et l'exon 8 qui est en bleu. L'ATG est indiqué par une tête de flèche rouge. Les sondes, les enzymes utilisées ainsi que la longueur des fragments attendus sont indiqués. B) Représentation schématique de l'ADN exogène provenant de pGKDTAmUBFKO suite à son insertion dans le génome, allèle muté. Les régions en noires sont les régions d'UBF utilisées pour la construction et celles en vert sont non utilisées. Les exons sont en rouge sauf le premier exon non codant qui est en vert et l'exon 8 qui est en bleu. L'ATG est indiqué par une tête de flèche rouge. Les sondes, les enzymes utilisées ainsi que la longueur des fragments attendus sont indiqués. Le gène codant pour fi-galactosidase est en jaune et celui codant pour neo est en magenta. Les séquences provenant du vecteur sont en cyan.

5.4.2. Les souris chimériques Puisque le clone 1F2 s'est avéré positif, les cellules provenant de ce clone furent

amplifiées et injectées dans des blastocystes. Ces derniers furent ensuite placés dans des

femelles pseudogestantes. Trois séries d'injections de cellules, provenant du clone 1F2,

furent faites et des souris chimériques sont nées de chacune de ces injections (figure 5.7)

pour un total de 16 souris chimériques, 9 mâles et 7 femelles. Ces souris furent par la suite

croisées entre-elles pour tenter d'obtenir des individus hétérozygotes. Puisqu'il est possible

que la colonisation des gamètes, par les cellules 1F2, soit un événement rare, les mâles

194

chimériques n'étaient sacrifiés qu'après avoir donné naissance à 100 souriceaux.

Malheureusement, malgré la naissance de 803 souriceaux, 1 des mâles était stérile, aucune

souris hétérozygote n'est née. Cet état de fait pourrait suggérer que soit les cellules 1F2

n'arrivent pas à coloniser la lignée germinale, soit une gamète ne peut survivre et/ou être

fonctionnelle si elle ne possède pas d'UBF. Cependant cette dernière hypothèse ne peut être

vérifiée expérimentalement.

Sonde 1 (région 5') Sonde 2 (région 3')

Allèle sauvage Allèle muté Allèle sauvage Allèle muté

EcoRI 7,9 kb 3,7 kb 7,9 kb 8,7 kb

Xmnl 7,5 kb 12,9 kb Xmnl 7,5 kb 12,9 kb

llincll 1,9 kb 1,3 kb llincll 1,9 kb 1,3 kb

Tableau 5.1 Poids des fragments attendus suite à la digestion du clone 1F2. Les poids attendus de l'allèlc sauvage et de l'allèle muté sont indiqués ainsi que la sonde et les enzymes utilisées.

EcoR

I

Xm

nl

EcoR

Hin

cl

r kl> kl> 4,2 ».*« 14,5 « • 14,2 10

O

te* 12,9 7,9

«,7 7,9 •

10 8

6 7,5 6 5 4,5 — 5

%• =* 4 3,5

4 - * 3,5 3.7

5

%• =* 4 3,5

1.9

1.3

OB 3

m 2

1.5

Sonde 5' Sonc le 3'

Ligure 5.6 Southern fait avec l'ADN génomique du clone 1F2.

195

Les enzymes et les sondes utilisées sont indiquées. Le poids des différents fragments du marqueur est aussi indiqué. Les fragments de 2 kb et 1,5 kb ne sont pas visibles sur la membrane, mais sont indiqués par une ligne fine à droite de la figure. Leur emplacement est déterminé par leur distance de migration sur le gel d'agarose par rapport. Le poids des différents fragments est indiqué. Les allèles sauvages sont en vert, les allèles mutés en rouge et les fragments non spécifiques en brun.

Figure 5.7 Souris chimériques obtenues avec les cellules ES provenant du clone 1F2.

5.5. Le vecteur pGKDTAmTAF,68KO Un vecteur, nommé pGKDTAm TAF168KO, comme celui utilisé pour la production

de cellules ES UBF +/-, fut produit. Le vecteur vide utilisé, pGKDTA, et la stratégie de

clonage sont les mêmes que pour l'inactivation génique d'UBF (section 5.4). Le gène lacZ,

qui code pour la (3-galactosidase a été inséré en phase avec les quatre premiers codons de

l'cxon 2 de TAF168 pour produire une protéine de fusion sous le contrôle du promoteur

endogène de TAF|68. Comme pour la construction pGKDTAmUBFKO, l'insertion de ce

vecteur se fait à proximité de l'ATG soit à 30 paires de bases de celui-ci dans la séquence

de l'ADNc (figure 5.8). Il est donc peu probable que la protéine de fusion possède une

activité TAF168 résiduelle. La région utilisée en 5' contient la séquence de -256 à + 1817

par rapport à l'ATG. La séquence utilisée en 3' contient la séquence de +6381 à + 9805.

Cette construction n'a jamais été électroporée dans les cellules ES puisque c'est

l'inactivation génique d'UBF qui est au cœur de ce projet. Comme aucune souris

196

hétérozygote UBF +/- n'est née et que l'étape de la détermination des clones positifs

TAF168+/- demande beaucoup de temps, aucune électroporation n'a été faite.

I 2 34 A ) / / /

B) ;/ Ï C) I Région 5' Région

-256 à+1817 LacZ neo +6381 à+ 9805

f 1 77 ' 3 4

Figure 5.8: Inactivation génique de TAF168

A) Gène codant pour TAF168 dans son contexte génomique. Les exons sont en rouge, les régions utilisées pour la construction en noir et celles non utilisées en vert. La séquence en bleue sera perdue lors de la recombinaison homologue. L'ATG est représenté par une tête de flèche rouge. Les exons 1 à 4 sont indiqués par des flèches. B) ADN exogène provenant de pGKDTAm TAF168KO dans son contexte génomique (allèle muté). Le code de couleur est le même qu'en A). La bande noire sous la figure indique l'ADN exogène qui a été intégré dans le locus. La séquence du gène LacZ est en jaune, celle codant pour neo est en magenta et celles provenant du vecteur en cyan. C) Agrandissement de l'ADN exogène dans le contexte génomique. Les séquences génomiques utilisées pour la construction sont aussi indiquées.

5.6. Les clones de cellules ES+/- de l'Institut Sanger L'Institut Sanger, ou la « Wellcome Trust Sanger Institute », est connu du monde

scientifique pour sa contribution au programme de séquençage du génome de différents

organismes et pour la distribution gratuite de ces banques de données. Cet organisme

produit aussi une banque de clones de cellules ES +/- pour un ensemble de gènes, ou de

ETS, qu'ils distribuent gratuitement. Il s'agit du « Sanger Institute Gène Trap Ressource »

197

ou SIGTR. Malheureusement, ces cellules ont été disponibles alors que notre projet de

production de souris KO était très avancé. Les constructions pGKDTAmUBFKO et

pGKDTAm TAF168KO étaient faites, les cellules ES UBF+/- du clone 1F2 avait été

injectées une fois et les souris chimériques issues de cette injection étaient déjà en

croisement.

L'institut Sanger ne produit pas de cellules ES +/- de façon ciblée comme la

méthode classique, décrite précédemment, car son but n'est pas d'obtenir un clone pour un

gène en particulier, mais de constituer une banque de clones ES+/- pour le plus de gènes

possible. Pour ce faire, les cellules ES sont infectées avec un virus qui intègre, de façon

aléatoire, le vecteur qu'il transporte dans le génome, généralement dans un intron. Ce

vecteur contient un gène de fusion provenant du gène lacZ et du gène de résistance au

G418 neo, qu'ils appellent fl-geo. Le vecteur inséré contient aussi un site accepteur

d'épissage, ou « splice acceptor », ce qui produit une protéine de fusion. Ainsi, l'ARN

messager et l'ADN complémentaire produits contiendront tous les exons en 5' de l'intron

où a eu lieu l'insertion en fusion avec la protéine /3-geo. Les autres exons de la protéine,

ceux en 3' de l'insertion, ne seront pas conservés puisque la séquence codant pour fi-geo est

suivie par un codon STOP et un site de polyadénylation. Suite à un 5'RACE sur l'ADNc,

en utilisant un oligonucléotide reconnaissant le gène /3-geo, il est possible d'identifier les

exons qui précèdent fi-geo et conséquemment d'identifier le gène inactivé et le lieu

l'insertion, c'est-à-dire dans l'intron suivant le dernier exon séquence. C'est cette séquence,

nommée « TAG », qui est publiée sur leur site Internet. Cependant, il est impossible de

savoir où dans l'intron a eu lieu l'insertion. Ainsi, l'insertion du « TAG » dans un long

intron complique l'analyse par Southern car le patron de digestion est incertain du fait que

l'insertion peut avoir eu lieu n'importe où dans cet intron.

5.6.1. Le clone de cellules ES UBF +/- de l'Institut Sanger: AY0043 À cette époque, un seul clone ES UBF+/- était disponible, AY0043, chromosome

11. L'insertion du vecteur SIGTR pour le clone AY0043 a eu lieu dans le septième intron,

qui est un court intron de 343 pb. Le fait d'avoir deux clones provenant de deux lignées

cellulaires différentes devraient permettre d'augmenter les chances de réussite de

198

colonisation des cellules germinales par les cellules ES +/-. À titre d'information, la banque

contient maintenant (janvier 2007) dix-neuf autres clones d'UBF.

5.6.1.1. Confirmation du clone AY0043 comme étant UBF +/- par RT-PCR Suite à leur réception, les cellules furent amplifiées pour avoir plus de matériel avec

lequel faire les analyses. La première analyse faite a permis de vérifier la production de

l'ARN messager contenant le gène UBF en fusion avec /3-geo. Pour ce faire, l'ARN des

cellules a été extrait avec du « TRIZOL » en suivant le protocole du fabricant (Invitrogen).

Par la suite, un RT-PCR a été fait sur ce matériel pour produire de l'ADN complémentaire.

Un PCR a été fait sur cet ADN complémentaire, en utilisant les oligonucléotides identifiés

à la figure 5.9A, de manière à amplifier l'allèle sauvage ou l'allèle muté. Comme le montre

les résultats présentés à la figure 5.9B, les fragments amplifiés correspondent aux

fragments attendus. Il est aussi possible de distinguer avec cette analyse la présence des

ARNm sauvages codant pour UBF 1 et UBF 2.

A)

ARN messager sauvage ARN messager de fusion

B)

UBF UBF 979 pb

845 pb

UBF1

UBF 2

B-geo 712 pb

631 pb

A Marqueurs

B 1+3 1+4 1+2

1000 900 800

^ ^ ^ U d J | *" n mmm 1000 900 800 ■ MM

Ml

^^WWrw4H^|p

1000 900 800 mé 7SC • MM

Ml

^^WWrw4H^|p

700 fH -Ml MM Ml

^^WWrw4H^|p

600 *• mê MM

Ml

^^WWrw4H^|p

500 # • $ 0 0

MM Ml

^^WWrw4H^|p

400

# • $ 0 0

MM Ml

^^WWrw4H^|p

UBF1 UBF 2

Figure 5.9 Vérification de l'ARNm de fusion dans le clone AY0043.

A) Schématisation des ARN messagers produits à partir de l'ARN messager sauvage et de l'ARN messager SIGTR. Les oligonucléotides utilisés pour les PCR sont représentés par

199

des flèches et numérotés. La longueur des fragments est indiquée. Le poids des marqueurs A et B sont identifiés B) Fragments obtenus suite au PCR fait avec les oligonucléotides identifiés en A) sur les produits RT-PCR.

5.6.1.2. Confirmation du clone AY0043 comme étant UHF +/- par analyse de type Southern

Le clone AY0043 a aussi été confirmé comme étant UBF+/- par une analyse de type

Southern (protocole décrit dans la section 5.9.4). L'ADN génomique a été digéré avec

BamHI et EcoRI (figure 5.10 et tableau 5.2). La digestion avec BamHI donne les fragments

attendus soit un fragment provenant de l'allèle sauvage de 8,4 kb et un provenant de l'allèle

muté de 9,6 kb (figure 5.11 et tableau 5.2). La digestion avec EcoRI ne produit toutefois

pas les fragments attendus. En effet, une seule bande d'environ 9kb est présente alors que

les fragments attendus étaient de 7,9 kb, allèle sauvage, et de 4,4 kb, allèle muté (figure

5.11 et tableau 5.2). Cependant, l'ADN génomique utilisé pour faire la mise au point des

analyses de types Southern ainsi que l'ADN du clone 1F2 provenait de deux lignées

cellulaires différentes de celle analysée ici. Il est donc possible que, dans cette lignée, la

digestion EcoRI ne produise pas les fragments théoriquement prévus. Toutefois, l'analyse

des ARN messagers produits à l'aide de deux paires d'oligonucléotides différentes et les

résultats obtenus par Southern avec l'enzyme BamHI confirme que le clone AY0043 est

bien un clone de cellules ES UBF +/-.

5.6.1.3. Les souris chimériques Puisque le clone AY0043 est bien UBF+/-, il fut injecté dans des blastocystes de

souris. 6 souris chimériques sont nées, 2 mâles et 4 femelles (figure 5.12) et ont été mises

en croisement. Malgré la naissance de 221 souriceaux, aucun d'entre eux n'était

hétérozygote. Ces résultats pourraient suggérer qu'il n'est pas possible d'avoir

d'hétérozygotes UBF+/- car l'utilisation de deux lignées cellulaires différentes, provenant

de deux construction différentes, n'a pas permis la production d'individus hétérozygotes.

Cependant, la transmission à la progéniture peut être un événement très rare mais possible,

c'est pourquoi nous allons faire de nouvelles injections.

200 A) B«nH

oRI BsmHI

BarnHI

5' EooRI ^ 7,9 kb

8,4 kb EcoRI

BarnHI

B)

EcoRI SiinHI B«mH{coRlB«ffîHJ^B<mH| B«pHE<x>RI

^ 9,6 kb E°°RI ^

_boea_

. uaJ j Uli.

4,4 kb ^ EcoRI .BimHI

Figure 5.10 Vérification du clone AY0043. A) Représentation schématique du gène UBF dans son contexte génomique. Les introns sont en noir et les exons en rouge sauf le premier exon non codant qui est en vert. La région utilisée pour la production de pGKDTAmUBFKO est en bleu. L'ATG est indiqué par une tête de flèche rouge. B) Schématisation de l'insertion aléatoire, ici au milieu de l'intron, du vecteur SIGTR dans l'intron 7. Les séquences d'UBF utilisées dans pGKDTAmUBFKO sont indiquées en bleu à titre de référence. Les séquences génomiques d'UBF sont en noir et le vecteur SIGTR est en vert. Les exons sont en rouges et l'ATG est indiqué par une tête de flèche rouge. Les enzymes et la sonde utilisées pour l'analyse par Southern sont indiquées.

Sonde 1 (région 5')

Allèle sauvage Allèle muté (si l'insertion est au milieu de l'intron)

IxoRI 7,9 kb 4,4 kb

BamHI 8,4 kb 9,6kb

Tableau 5. 2 Poids des fragments attendus suite à la digestion du clone AY0043.

201

O

5 S

■•«.,1. Vjk**.

Figure 5.11 Southern fait avec l'ADN génomique du clone AY0043. Les poids des fragments des marqueurs (colonne 1) est indiqué ainsi que les enzymes utilisées. Le poids des fragments obtenus suite aux digestions sont aussi indiqués. Le fragment provenant de l'allèle sauvage est en vert, celui de l'allèle muté en rouge et le fragment obtenu suite à la digestion avec EcoRI en brun.

m . M M

. a

» i * ( ^B

Figure 5.12: Souris chimériques obtenues suite à l'injection de cellules ES UBF +/-provenant du clone AY0043.

202

5.6.2. Les clones de cellules ES TAF,68+/- et TAFi95+/- de l'Institut Sanger

Comme il a été mentionné dans la section 5.6, l'Institut Sanger produit une banque

de cellules ES +/- pour un ensemble de gènes. Trois clones de TAF|68+/-, chromosome 1,

AE0377, CC0409, XP0245, et un de TAFi95+/-, chromosome 8, AS0135, furent

commandés. Ces clones n'ont pas été confirmés comme étant TAF168 +/- ou TAF[95 +/-.

Ce sera donc la première chose à faire lorsque le projet sera relancé. Puisque l'analyse de

ces clones est plus rapide et moins coûteuse que la production de cellules ES TAF168 +/-

avec pGKDTAmTAF|68KO et puisqu'il existe trois clones Sanger ES TAF|68 +/-, il ne

sera vraisemblablement pas utile de produire des cellules ES TAF168 +/- avec le vecteur

pGKDTAmTAFi68KO. Lorsqu'il y aura des développements du côté de l'inactivation

génique d'UBF, les clones confirmés comme étant ES TAF168 +/- et TAFi95 +/- seront

injectés dans des blastocytes pour de tenter de produire des individus hétérozygotes et

homozygotes. Il sera alors possible de comparer le phénotype obtenu pour la perte d'UBF

avec ceux résultant de la perte de TAF168 et/ou de TAFi95.

5.7. Production de lignées cellulaires ES-/- complémentées par un transgène inductible

Comme il a été mentionné précédemment, il est fort probable que les individus

homozygotes UBF-/-, TAF168-/- et TAFi95-/- ne soient pas viables. Pour pouvoir tout de

même faire des analyses biochimiques en absence d'une de ces protéines, des lignées ES -/-

complémentées par un transgène sous le contrôle d'un promoteur inductible vont être

produites. Le système de promoteur inductible utilisé est le système « Tet Off »

d'Invitrogen. Pour faire ces lignées cellulaires, un clone de cellules ES +/-, le clone 1F2 ou

un des clones de l'Institut Sanger, va être cotransfecté avec un vecteur contenant l'ADNc

du gène qui sera inactivé et un vecteur exprimant de façon constitutive l'élément régulateur

Tet (figure 5.13). Dans ce système, le gène n'est pas transcrit lorsque le milieu est

complémenté avec de la tétracycline ou de la doxycycline et il est transcrit en leur absence.

La doxycycline est moins dispendieuse et c'est elle qui sera ajoutée au milieu. Une

variation de la concentration de cette substance fera donc varier le niveau de transcription

de l'ADNc sous son contrôle. Suite à cette cotransfection, les cellules seront sélectionnées

203

avec de l'hygromycine dont le gène codant pour la résistance à cette drogue est sur le

vecteur contenant l'ADNc de la protéine d'intérêt. Il est nécessaire de transfccté l'ADNc

avant la deuxième inactivation car la cellule ne devrait pas survivre sans le gène d'intérêt.

Comme Pallèle qui est inactivé est déjà porteur du gène de résistance au G418, neo,

les cellules résistantes seront exposées à de fortes concentrations de G418 pour tenter

d'obtenir un événement de réversion. C'est-à-dire que l'allèle sauvage soit transformé en

allèle muté. De cette façon, la seule protéine fonctionnelle produite proviendra de l'ADNc

sous le contrôle du promoteur inductible. Les cellules résistantes seront ensuite

ensemencées en triplicatas. Le premier ne contiendra pas de doxycycline et les deux autres

contiendront des niveaux croissant de doxycycline. Cette façon de faire favorisera

probablement l'établissement de cellules pouvant survivre à un certain niveau d'expression

de la protéine puisque le niveau de protéines permettant la survie ainsi que le niveau

toxique d'expression, s'il existe, sont inconnus.

L'utilisation de ces lignées cellulaires permettra de déterminer la nécessité de ces

protéines et le niveau minimal d'expression nécessaire pour la survie cellulaire. Dans le cas

d'UBF, ces cellules seront aussi utilisées pour vérifier l'importance de cette protéine sur

l'intégrité de la structure nucléolaire et pour comparer les activités d'UBF 1 et d'UBF2. Un

autre avantage de ces lignées est que leur différenciation et leur immortalisation pourront

être induites. Le travail a déjà commencé en ce qui concerne les lignées cellulaires UBF-/-.

Dr Stefanovsky a produit les deux ADNc codant pour UBF1 et 2. J'ai fait une partie des

sous-clonages subséquents de façon à produire des ADNc avec l'épitope FLAG. La mise

d'un épitope est importante pour s'assurer que la protéine produite provient du transgène et

non d'un allèle non inactivé. Françoise Morin continue actuellement le projet. L'utilisation

de cette gamme d'outils fournira des réponses importantes quant à la détermination in vivo

des fonctions de ces différentes protéines.

204

Cellules hétérozygotes

Cellules hétérozygotes

Cellules homozygotes

Cotransfection: 1. ADNc sous le contrôle

d'un promoteur inductlble 2. Élément régulateur Tet

Figure 5.13 Étapes menant à la production de cellules ES -/- complémentées par un transgène sous le contrôle d'un promoteur inductible.

5.8. Perspectives et Conclusion Puisqu'aucun individu hétérozygote n'est né du croisement des souris chimériques,

une approche est développée dans le laboratoire pour contourner le problème de la

transmission à la progéniture. Françoise Morin, en collaboration avec Victor Stefanovsky,

travaille actuellement à la production d'un vecteur qui permettra l'inactivation

conditionnelle d'UBF en utilisant la technique Cre-LoxP. Le vecteur produit contient,

comme l'inactivation classique, un gène de résistance devant être conservé et un gène

toxique devant être perdu pour faciliter les événements de recombinaison homologue. Ce

vecteur contient aussi des éléments Lox de part et d'autre d'un segment de la séquence

génomique d'UBF. La différence entre les deux approches c'est que l'insertion du vecteur

pour l'inactivation conditionnelle ne modifie pas la séquence génomique d'UBF

comparativement à l'insertion du vecteur pGKDTAmUBFKO qui, elle, brise le cadre de

lecture et produit une protéine de fusion UBF-/ac Z. Ainsi, la protéine UBF est produite

chez les individus hétérozygotes, +/lox, et homozygotes lox/lox, à partir des deux allèles.

Pour qu'il ait inactivation génique d'UBF, les souris homozygotes, lox/lox, doivent être

croisées avec des souris porteuses de la Cre recombinase. Les souris issues de ces

croisement vont être lox/lox et porteuses de la Cre recombinase. Cette enzyme reconnaît les

éléments Lox dans l'ADN intégré et excise le segment d'ADN placé entre ces deux courtes

205

séquences. Cette excision a pour conséquence de briser le cadre de lecture, d'empêcher la

production d'une protéine UBF fonctionnelle et donc d'inactiver le gène.

L'activité de l'enzyme Cre peut être restreinte en la plaçant sous le contrôle de

promoteurs spécifiques à un organe ou à un tissu ou encore dont l'expression est régulée au

cours du développement. Ainsi, il est possible de choisir le moment et/où le lieu de

l'inactivation. L'utilisation d'un tel système permet de produire une colonie d'individus

lox/lox sains car la protéine est produite. Par la suite, il sera possible d'inactiver de façon

conditionnelle UBF. Ce système permet aussi l'inactivation d'un seul allèle chez les

individus hétérozygotes +/lox. Ces deux systèmes permettront de répondre aux questions de

départ de ce projet. À savoir si UBF est essentiel à la vie et si les individus hétérozygotes

présentent des défauts, par exemple de croissance, causés par un effet de dosage de gène.

De plus, il sera possible de déterminer l'importance d'UBF dans des organes spécifiques et

à des moments précis du développement.

La transcription ribosomale est vraisemblablement indispensable à la vie et au

développement animal puisqu'elle produit les trois ARN ribosomaux indispensables à

l'activité des ribosomes. Comme il a été mentionné dans l'introduction de ce chapitre,

l'inactivation génique de Rrn3, un facteur de transcription de Poli, est létale au jour

embryonnaire 9,5 (Yuan et al., 2005). Cependant, pour atteindre ce stade de

développement, l'embryon s'est développé et a accru sa masse en ARN cellulaires. Comme

les ARN ribosomaux représentent 80% de la quantité totale des ARN cellulaires, la survie

et le développement de l'embryon jusqu'à ce stade suggère que la transcription ribosomale

puisse avoir lieu en l'absence de Rrn3. Toutefois, cette protéine est nécessaire pour la

survie, et vraisemblablement pour la transcription ribosomale, à partir du jour

embryonnaire 9,5. Tout comme Rrn3, TAF168 et TAF|95 participent, au sein du complexe

SL1, à la formation du complexe de pré-initiation. À ce jour, ces trois protéines ne

semblent pas impliquées dans une autre étape de la transcription ribosomale ni dans un

autre processus cellulaire. Ainsi, si le phénotype observé pour la perte de Rrn3 résulte d'un

défaut dans l'initiation de la transcription, il est logique de penser que la perte de TAF168 et

de TAFi95 entraînera un phénotype semblable. Il est cependant aussi possible que la perte

206

de TAF]68 et de TAF|95 entraîne un phénotype plus sévère que celui observé pour la perte

de Rrn3 s'il s'avère que ces protéines sont absolument essentielles pour la transcription

ribosomale.

Ce même raisonnement peut aussi s'appliquer pour l'inactivation génique d'UBF.

Malgré le fait que l'inactivation de la protéine de levure Mmolp entraîne un défaut de

croissance, celle-ci n'est pas létale. Par contre, elle le devient lorsque qu'elle est combinée

avec la perte de sous-unités non essentielles de la polymérase I (Gadal et al., 2002). Cette

protéine est considérée comme étant un homologue fonctionnel d'UBF, mais elle est

structurellement moins complexe qu'UBF. Considérant les fonctions connues d'UBF dans

la transcription ribosomale chez les mammifères, il est logique de penser, que la perte

d'UBF est incompatible avec la vie et que les embryons vont mourir in utero. De plus, une

étude montrant qu'FImolp est aussi localisée sur une majorité de promoteurs de protéines

ribosomales (Hall et al., 2006) suggère qu'UBF puisse aussi avoir un rôle différent de celui

qui lui ait connu en transcription ribosomale. La comparaison du phénotype obtenu pour la

perte d'UBF avec ceux entraînés par la perte de Rrn3, TAF168 et TAF|95 permettra de

déterminer la sévérité relative du phénotype UBF -/- ce qui fournira une indication du degré

de nécessité de cette protéine. De plus, la comparaison des phénotypes permettra

vraisemblablement de déterminer si UBF possède une fonction indépendante ou

complémentaire à celles qui lui sont connues dans la transcription ribosomale (régulation

de l'élongation, formation du complexe de pré-initiation, formation d'une chromatine

ribosomale). Un phénotype plus sévère, ou plus complexe, que ceux résultant de la perte de

Rrn3, de TAF168 ou de TAF|95 signifie qu'UBF possède une fonction supplémentaire à son

rôle dans la transcription ribosomale. L'étude des souris « Knock-out », conditionnelles ou

non, et des lignées cellulaires ES -/- complémentées par un transgène inductible permettra

de déterminer de la nécessité in vivo des protéines UBF, TAF168 et TAFi95.

207

5.9. Matériel et méthodes

5.9.1. Culture des cellules embryonnaires souches Les cellules embryonnaires souches doivent être cultivées dans un milieu

supplémenté avec un facteur d'inhibition de la différenciation, du « Leucemia inhibitory

Factor » ou LIF, pour maintenir les cellules ES dans un état non différencié. De plus, pour

favoriser leur croissance, les cellules ES poussent sur une couche de cellules fibroblastiques

embryonnaires, communément appelées cellules nourricières, qui sont résistantes au G418.

De plus, certaines de ces lignées peuvent aussi produire du LIF. Ces cellules nourricières

sont préalablement traitées à la mitomycine C ou aux irradiations UV de façon à bloquer

leur division cellulaire. Ainsi, ces cellules s'étendent pour former un tapis, mais elles ne se

divisent plus. En plus de former une matrice à laquelle s'attachent les cellules ES, ces

cellules libèrent dans le milieu des facteurs de croissance utilisés. Elles favorisent ainsi la

croissance des cellules ES. Après la sélection, les cellules ES résistantes sont amplifiées

dans des plaques de 96 puits avec des cellules nourricières. Une fois qu'une majorité de

clones sont confluents, tous les clones de cette plaque sont trypsinisés et redistribués sur

deux plaques. Les cellules de la première plaque sont utilisées pour l'extraction de l'ADN

génomique et la détermination des clones positifs. Ces cellules poussent sur de la gélatine

et non sur les cellules nourricières pour s'assurer que l'ADN servant aux analyses ne

provienne que des cellules ES. Une fois qu'il est estimé qu'il y a assez de cellules pour

permettre l'analyse de l'ADN, les cellules sont lavées deux fois avec du PBS et congelées.

Les cellules de la deuxième plaque poussent sur des cellules nourricières et sont congelées

lorsqu'une majorité de clones sont confluents. Cette plaque sert à la réamplification des

clones positifs.

5.9.2. Lyse des cellules et extraction de l'ADN génomique Les plaques arrivent de l'unité transgénique de McGill sur glace carbonique et sont

gardées à -20°C jusqu'à la lyse des cellules. Avant la lyse, les plaques sont sorties du

congélateur jusqu'à ce qu'elles soient à température ambiante. Par la suite, 50ul de tampon

de lyse (lOmM NaCl, 10 mM EDTA, 10 mM Tris, 0,5% sarcosyl, lmg/ml protéinase K)

sont déposés dans chaque puit. Les plaques sont ensuite placées à 55°C dans un

208

environnement humide pendant environ 18 heures (ou une nuit). Le lendemain, lOO il d'un

mélange d'éthanol et de NaCl (150 u.1 de 5M NaCl/10 ml d'éthanol 100%) sont ajoutés à

chaque puit. Les plaques sont laissées à température ambiante pour un minimum de 90

minutes. Par la suite les plaques sont renversées pour enlever l'éthanol et lavées trois fois

avec de l'éthanol 70%. Suite au dernier lavage, les plaques sont déposées sur un papier

absorbant et demeurent ainsi pendant une heure. Après l'étape de la précipitation et à la fin

des trois lavages, chaque puit est observé au binoculaire pour vérifier s'il y a présence

d'ADN attaché au fond du puit et/ou aux parois. Une estimation approximative et

qualificative de la quantité d'ADN présente dans chaque puit est faite visuellement et est

notée comme étant beaucoup, peu ou pas d'ADN. Cette estimation est importante car la

quantité d'ADN d'un certain nombre de clones seulement est mesurée par fluoromètre

avant la réaction de PCR ou la digestion par des enzymes de restriction. Puisque la quantité

d'ADN récupérée pour chaque clone n'est pas mesurée, l'utilisation d'un certain nombre de

clones de chacune des trois catégories, beaucoup, peu ou pas, permet d'avoir une idée plus

juste des quantités d'ADN récupérée. Après le séchage, l'ADN génomique est resuspendu

dans 30u.l de 1/10 TE (1 mM Tris pH 8,3, 0,lmM EDTA) et les plaques sont ensuite

conservées à 4°C dans un environnement humide.

5.9.3. Réaction de PCR Pour qu'il y ait une réaction d'amplification avec les oligonucléotides dirigée contre

la région 3' de l'insertion, l'ADN génomique devait être dénaturé avant la réaction PCR.

Pour ce faire, un mélange de 30 ul contenant l'ADN génomique des clones, les deux

oligonucléotides, 0,12 mM des 4 dNTPs et 3ul de 10X buffer 3+ (300 mM tricine, 20 mM

MgCL, 1% polyoxyethylene-9-lautyl-ether, 0,1% gélatine, 50mM P-mercaptoethanol,

ajuster à pH 8,4) a été préparé et dénaturé pendant 5 minutes à 95°C. Ensuite, la réaction

était mise sur glace et les 20 ul restants (0,08 mM 4dNTP, 2ul de 10X buffer 3+) étaient

ajoutés. Après 15 minutes sur glace, les réactions étaient mises dans le PCR selon le cycle

suivant: 30 fois (30 secondes à 95°C, 2,5 minutes à 61°C et 1 minute à 72°C) suivi de 10

minutes à 72°C. Un dixième de la réaction était ensuite mis sur un gel de 0,8% agarose et

les fragments amplifiés étaient visualisés par coloration au bromure d'éthidium.

209

5.9.4. Analyse de type Southern Les cellules furent lysées et l'ADN extrait comme décrit dans la section 5.9.2.

L'ADN génomique fut digéré toute la nuit avec un surplus d'enzyme (entre 5 et 10 fois) en

présence de 10 mM spermidine, 2 ug/ul de RNAaseA et de 1 ug/ul de BSA. Après avoir

vérifié que la digestion était complète, l'ADN fut migré sur un gel de 0,8% agarose à bas

voltage toute une nuit. Le lendemain, le gel fut photographié et exposé 5 minutes aux

rayons UV à l'énergie maximale dans un UVC500 UV-crosslinker (Hoefer). Par la suite, le

gel fut mis 30 minutes dans 0,2 M HC1 pour que l'ADN soit dépuriné, puis deux fois 15

minutes dans 0,5N NaOH et 1,5M NaCl pour que l'ADN soit dénaturé et finalement remis

dans la même solution pour 1 heure pour équilibration avant le transfert. Le gel fut

transféré sur une membrane Biodyne B (Pall) dans 0,5N NaOH et 1,5M NaCl à 275 mBar

pendant 2 heures en utilisant le « Vacu-Blot » de Biometra. La membrane fut ensuite traitée

aux rayons UV à 70 J.cm"2 dans UVC500 UV-crosslinker (Hoefer).

Chapitre 6 : Conclusion Les sujets des quatre chapitres de ma thèse portent tous sur les mécanismes de

régulation de la transcription ribosomale. Plus particulièrement, je me suis intéressée à

l'activation de la transcription suite à un stimulus de croissance (chapitres 2 et 3) ainsi

qu'au rôle de la méthylation de l'ADN dans l'établissement et le maintien de gènes

ribosomaux silencieux chez l'humain (chapitre 4). Finalement, j 'ai réalisé les premières

étapes menant à l'inactivation génique de la protéine UBF dans les souris de façon à

connaître son rôle in vivo dans la transcription ribosomale, dans le développement animal et

dans le maintien de la vie (chapitre 5). J'ai aussi préparé des outils pour étudier les

fonctions in vivo de TAF168 et TAFi95 pour pouvoir comparer les phénotypes obtenus avec

celui entraîné par la perte d'UBF (chapitre 5). Pour mieux comprendre l'importance des

résultats présentés dans cette thèse, il est essentiel de les placer dans le contexte actuel de

l'état des connaissances du domaine de la transcription ribosomale.

6.1. UBF est un régulateur de Pélongation de l'ARN polymérase I

6.1.1. La phosphorylation d'UBF, par ERK, modifie son attachement à l'ADN.

L'étude présentée au chapitre 2 montre, pour la première fois, que la transcription

ribosomale répond de façon immédiate à une stimulation par le facteur de croissance EGF.

Cette démonstration permet de recatégoriser les gènes ribosomaux dans la classe des gènes

de réponse immédiate. De plus, cette activation est dépendante de la kinase ERK1/2

puisque des inhibiteurs de cette voie de signalisation empêche l'augmentation de la

transcription. Deux sites de phosphorylation de ERK ont été identifiés sur UBF et ils sont

localisés dans les boîtes HMG1 1 et 2, sur les thréonines 117 et 201 respectivement. La

mutation de ces acides aminés en alanine, qui empêche la phosphorylation, ou en acide

aspartique ou glutamique, qui simule la phosphorylation, inhibe l'augmentation de la

transcription suite à l'activation de ERK. Ces résultats suggèrent que l'activation de la

transcription par UBF nécessite un cycle de phosphorylation/déphosphorylation.

Finalement, la phosphorylation d'UBF sur les boîtes HMG1 1 et 2 a été montrée comme

211

prévenant l'interaction de ces boîtes avec l'ADN (Stefanovsky et al., 2006b). Il faut aussi

noter que ce sont ces mêmes deux boîtes qui sont acétylées par CBP et que ERK est connu

comme activant CBP (Pelletier et al., 2000). Il est donc raisonnable de penser que ces deux

enzymes coopèrent dans l'activation de la transcription ribosomale.

6.1.2. UBF définit une chromatine ribosomale. La capacité d'UBF à modifier la structure de l'ADN en faisait déjà un régulateur

connu de la formation du complexe de pré-initiation. En effet, il est suggéré que la

formation de deux « enhancesomes »au promoteur permet l'alignement du core et de l'UCE

et la formation d'une surface d'interaction optimale pour l'ARN polymérase I (Bazett-

Jones et al., 1994; Moss et al., 2006a; Stefanovsky et al., 1996; Stefanovsky et al., 2001a).

La publication d'un article montrant, par ChIP, qu'UBF est associée à toute la séquence

ribosomale et pas seulement à la séquence promotrice (O'Sullivan et al., 2002) suggère que

cette protéine architecturale définit une chromatine ribosomale. Ainsi, il est possible que le

relâchement de l'ADN, suite à une phosphorylation par ERK, permette de réguler non

seulement la structure du promoteur, mais aussi l'ensemble de cette structure

chromatinienne.

6.1.3. L'élongation de la transcription par l'ARN polymérase I est régulée par UBF.

Les résultats présentés au chapitre 3, montrent qu'effectivement la transcription

ribosomale est régulée au niveau de l'élongation de l'ARN polymérase I par UBF. Comme

l'initiation de la transcription a toujours été considérée comme le seul niveau de régulation

de la transcription ribosomale, nous avons d'abord vérifié si l'augmentation de la

transcription entraîne une augmentation du nombre de polymérases engagées. En effet, si

l'initiation de la transcription est réellement le seul niveau de régulation, alors une

augmentation de la synthèse de l'ARN précurseur doit nécessairement être causée par une

augmentation proportionnelle du taux d'initiation et donc du nombre de polymérases I

engagées sur les gènes ribosomaux. Toutefois, si l'élongation de la transcription est aussi

régulée, le nombre de polymérases engagées sera relativement constant. Par contre, celles-

ci transcriront plus vite ce qui résultera en une augmentation de la synthèse de l'ARN

212

précurseur. Évidemment, si l'augmentation de la vitesse d'élongation n'est pas couplée

avec une augmentation du taux d'initiation, alors le nombre de polymérases sur les gènes

va chuter et il n'y aura pas plus d'ARN précurseur produit. L'analyse du nombre relatif

d'ARN polymérases 1 engagées sur les gènes ribosomaux, mesuré par Run-On et par ChIP,

a montré qu'il n'y avait pas plus de polymérases engagées dans les cellules traitées avec de

l'EGF, ou un ajout de sérum, comparativement aux cellules dans lesquelles la voie ERK est

inhibée et ce même si les cellules stimulées produisent 5 fois plus d'ARN précurseur.

Une mesure directe de la vitesse d'élongation a confirmé que celle-ci est augmentée

dans les cellules stimulées par l'EGF comparativement aux cellules dans lesquelles la voie

ERK est inhibée. Ces résultats sont la première démonstration de l'existence d'une

régulation au niveau de l'élongation. De plus, les cellules traitées à l'EGF ont une vitesse

d'élongation 5 fois supérieure à celle des cellules dont la voie ERK est inhibée. Cette

augmentation explique quantitativement l'augmentation de la synthèse d'ARN précurseur

et suggère que la vitesse d'élongation est limitante pour la transcription ribosomale. Des

essais de transcription in vitro ont montré qu'UBF bloque l'élongation de la transcription et

que ce blocage est levé par la phosphorylation d'UBF par ERK. Une des implications

importantes de l'existence d'un mécanisme de régulation de l'élongation qui soit limitant

pour la transcription ribosomale est que ce mécanisme pourrait permettre la co-régulation

de l'assemblage co-transcriptionnel des sous-unités ribosomales avec la transcription elle-

même.

6.1.4. L'étape limitante de l'augmentation de la transcription ribosomale en réponse à un facteur de croissance : initiation, via Rrn3, ou élongation, via UBF?

L'augmentation de la transcription ribosomale dépendante de ERK a aussi été

associée à la phosphorylation de la protéine Rrn3 (Zhao et al., 2003). Comme je l'ai

mentionné dans l'introduction, cette protéine est responsable de la compétence de l'ARN

polymérase I à initier la transcription (Miller et al., 2001; Yuan et al., 2002). La majorité

des articles associés à cette protéine considère que sa régulation est l'élément régulateur

majeur de la transcription ribosomale en réponse à des facteurs de croissance et ce pour

213

plusieurs raisons (Grummt, 2003). Premièrement, cette protéine est conservée de la levure à

l'humain (Bodem et al., 2000; Moorefield et al., 2000), contrairement à UBF. Cependant,

chez la levure, l'activité de Rrn3 ne semble pas être régulée par la phosphorylation comme

cela est suggéré chez la souris. Deuxièmement, une inhibition de la croissance entraîne une

diminution de l'association de l'ARN polymérase I avec Rrn3 et donc une réduction de la

forme compétente de la polymérase. Sachant que Rrn3 interagit avec une sous-unité de la

polymérase, avec un facteur accessoire (Fath et al., 2001; Seither et al., 2001; Yuan et al.,

2002) ainsi qu'avec deux TAFis (Miller et al., 2001; Yuan et al., 2002), il est suggéré que

cette protéine ait un rôle régulateur important dans la formation du complexe de pré­

initiation. De plus, il semble que son rôle soit limité à cette étape, puisque Rrn3 est inactivé

et relargué de la polymérase suite à l'initiation de la transcription (Hirschler-Laszkiewicz et

al., 2003; Milkereit et Tschochner, 1998). Par contre, ce relarguage n'est pas essentiel pour

l'activité de Rrn3 car une lignée de levure produisant une protéine de fusion entre Rrn3 et

une sous-unité de la polymérase I ne présente pas de défaut de croissance (Laferte et al.,

2006).

Finalement, Zhao et collègues (Zhao et al., 2003) ont montré que la présence d'UBF

n'est pas nécessaire à l'augmentation de la transcription suite à une stimulation par la voie

ERK dans leur essai in vitro (Zhao et al., 2003). Or, comme il a été discuté lors de

l'introduction de la thèse et dans l'annexe 1, ce ne sont pas tous les systèmes de

transcription in vitro qui répondent à une stimulation par UBF. Cependant, la

démonstration que nous avons faite quant à la nécessaire présence d'UBF pour l'activation

dépendante de ERK a été faite in vivo. De plus, il est connu que les systèmes de

transcription in vitro réagissent différemment à l'ajout d'UBF selon le ratio UBF/ADN.

Ainsi, les concentrations d'UBF normalement utilisées lors des essais de transcription

correspondent à une faible occupation de F ADN par UBF comparativement à une

occupation de près de 90% dans la situation endogène. L'utilisation d'une trop faible

quantité d'UBF ne bloquera pas l'élongation de la transcription (chapitre 3, figure 3.4) et

l'effet visible de l'activation de la transcription, dépendante de ERK, sera causé

uniquement par l'augmentation du taux d'initiation. De plus, les expériences de mutation

des sites de phosphorylation de Rrn3, qui inhibent l'activation de la transcription (Zhao et

214

al., 2003), ne sont pas reproductibles chez le rat ou chez la drosophile (communications

personnelles de Lawrence I Rothblum, de Ross I Iannan et de Bruce Edgard).

Personnellement, je trouve que les deux séries de résultats, concernant les rôles

d'UBF et de Rrn3, sont en accord plutôt qu'en désaccord comme le soulignent (trop)

souvent « les partisans » de Rrn3. En effet, une augmentation de la vitesse d'élongation qui

ne serait pas couplée à une augmentation du taux d'initiation ne causerait pas

d'augmentation de la synthèse de l'ARN précurseur. À mon sens, l'utilisation d'une même

voie de signalisation, la voie ERK, pour phosphoryler Rrn3 et UBF confirme la coopération

de ces deux niveaux de régulation. Cette affirmation pourrait probablement être élargie à

une majorité de signaux de croissance puisque ces deux protéines sont aussi des cibles de la

voie TOR. Toutefois, des études non publiées faites dans notre laboratoire (annexe 1 figure

A 1.8), montrent que des extraits cellulaires provenant de cellules traitées avec un inhibiteur

de la voie ERK possèdent la même capacité d'initiation de la transcription que des extraits

provenant de cellules traitées avec l'EGF. Toutefois, les cellules traitées avec l'inhibiteur

de la voie ERK produisent 5 fois moins d'ARN précurseur. Ces résultats suggèrent que la

vitesse d'élongation des transcrits, et non le potentiel d'initiation, est l'étape limitante de la

transcription ribosomale.

6.1.5. D'autres protéines pourraient aussi réguler l'élongation de l'ARN polymérase I.

La démonstration d'une régulation au niveau de l'élongation est un concept nouveau

dans le domaine de la transcription ribosomale (voir chapitre 3). Une régulation au niveau

de l'élongation de l'ARN polymérase II est connue et suggérée comme permettant le

couplage des transformations co-transcriptionnelles des ARNm avec leur transcription.

L'activité du complexe Spt4/Spt5 est importante pour cette corégulation chez la levure et

les mammifères. De façon analogue, le contrôle de la vitesse d'élongation de l'ARN

polymérase I est probablement essentiel pour le couplage de la transcription ribosomale

avec l'assemblage co-transcriptionnel des sous-unités ribosomales. Un défaut dans

l'assemblage des sous-unités, qui entraînerait nécessairement des conséquences dans la

production des ribosomes, pourrait avoir une répercution rapide sur la transcription via la

215

régulation de la vitesse d'élongation. Un article récent suggère que Spt4/5 sont aussi

impliquées dans la régulation de l'élongation de la Poli chez la levure car ces deux

protéines interagissent physiquement avec Poli et avec l'ADN ribosomal (Schneider et al.,

2006). La délétion de Spt4p, chez la levure, diminue légèrement la transcription ribosomale

en plus de causer des défauts dans la transformation de l'ARN précurseur en ARN

ribosomaux matures et dans l'assemblage des sous-unités ribosomales. Une autre étude

montre que la délétion de certaines protéines du SSU « processome », impliquées dans la

formation de la petite sous-unité, entraîne des défauts dans la transcription ribosomale et

dans la transformation de l'ARN précurseur (Gallagher et al., 2004). Ces deux études, ainsi

que nos résultats, suggèrent qu'une régulation de l'élongation existe dans la transcription

ribosomale et que cette régulation permet le couplage de la transcription avec l'assemblage

des sous-unités ribosomales.

6.1.6. Le complexe nucléoprotéique UBF-ADN. L'association de Spt4/5 avec Poli suggère qu'il est possible que certaines protéines

soient impliquées à la fois dans la régulation de l'élongation de Poli et de PolII. Il ne faut

toutefois pas oublier que ces deux systèmes de transcription sont forts différents en ce qui

concerne les protéines impliquées, le type de gènes transcrits, unique versus répété, et leur

structure chromatinienne respective. L'ARN polymérase II transcrit des gènes uniques et

doit gérer la présence des nucléosomes. L'activation génique des gènes transcrits par PolII

requiert la participation de facteurs de transcription, de complexes modificateurs

d'histones, de complexes de remodelage de la chromatine et, probablement, de la présence

de variants d'histones pour permettre les déplacements des nucléosomes du promoteur et la

formation des complexes de pré-initiation (Workman, 2006). De plus, les nucléosomes

doivent être déplacés pour permettre l'élongation de la polymérase dans la région codante

des gènes. Il est toutefois nécessaire de les replacer derrière la polymérase de façon à

bloquer la transcription résultant de l'utilisation de sites d'initiation cryptiques localisés

dans la séquence codante du gène (Workman, 2006). Des complexes modificateurs

d'histones, de remodelage de la chromatine, des variants d'histones et des protéines

chaperonnes sont impliquées dans ce mécanisme de contrôle de fidélité transcriptionnelle

216

(Workman, 2006). Il est raisonnable de penser que la présence des nucléosomes, et de leur

déplacement dynamique, est essentielle pour le contrôle d'une variété de processus qui

agissent sur le génome comme, par exemple, la transcription, la réparation et la réplication

de FADN.

L'ARN polymérase I, comparativement à PolII, transcrit des gènes répétés dont la

région transcrite est dépourvue de nucléosomes, mais pas nécessairement d'histoncs. Les

gènes ribosomaux sont toutefois associés à la protéine UBF qui forme une structure

nucléoprotéique très semblable à celle des nucléosomes. Les modifications des histoncs

sont connues comme étant impliqués dans les changements de la permissivité de la

structure chromatinienne et c'est pourquoi certaines marques sont associées à une

chromatine active, ex : hyperacétylation de l'histone H4, ou inactive, ex : triméthylation de

la lysine 9 de l'histone H3. UBF, tout comme les histones, est modifié de façon

cotranscriptionnelle. Cette modification, sa phosphorylation par ERK, ou son absence

semblent déterminer la permissivité de la chromatine ribosomale en permettant ou en

bloquant l'élongation de l'ARN polymérase I. Malgré la similitude de leur structure

chromatinienne respective, le niveau de transcription des gènes ribosomaux est

radicalement différent de celui des gènes transcrits par l'ARN polymérase IL En effet,

contrairement aux gènes PolII, les gènes ribosomaux sont intensivement et surtout

continuellement transcrits. Il est donc possible que la cellule ait développé un mécanisme

différent pour réguler l'élongation de l'ARN polymérase I.

UBF est une protéine abondante qui a un taux d'attachement/détachement de l'ADN

très rapide (Chen et Huang, 2001; Dundr et al., 2002). Nous avons montré, au chapitre 2,

que l'attachement d'UBF à l'ADN empêche sa phosphorylation par ERK. Or, cette

phosphorylation est nécessaire pour permettre le passage de la polymérase. La cinétique

d'attachement d'UBF à l'ADN pourrait permettre un remplacement rapide d'une protéine

inhibitrice, non phosphorylée, par une protéine permissive, phosphorylée, et vice-versa. De

cette façon, la vitesse d'élongation pourrait être continuellement ajustée en réponse aux

différents stimulus. De plus, il faut se rappeler qu'UBF a deux variants d'épissage qui sont

produits dans les mêmes proportions. La protéine UBF 2, dont la fonction est mal

217

caractérisée, possède une troncation dans sa deuxième boîte HMG1 ce qui pourrait avoir

comme conséquence de mimer l'effet de la phosphorylation de cette même boîte. Cette

protéine pourrait donc être plus sensible à la voie ERK car, pour devenir permissive, elle

n'aurait besoin que d'une phosphorylation. L'existence même de ce variant d'épissage est

peut-être définie par son importance dans la régulation de l'élongation. Cette hypothèse

doit toutefois être vérifiée. Un autre avantage de l'utilisation d'UBF est que cette protéine

est impliquée dans les étapes d'initiation et d'élongation des transcrits. La régulation de

l'activité de cette protéine pourrait donc permettre la corégulation de ces deux étapes.

Finalement, le déplacement d'UBF de l'ADN, contrairement aux nucléosomes, ne semble

ni nécessiter la participation de facteurs accessoires ni la consommation d'énergie. Il est

toutefois possible qu'un complexe de remodelage d'UBF existe, mais qu'il n'ait pas encore

été identifié. Sachant que la biogenèse des ribosomes requiert énormément de ressources

cellulaires, l'utilisation d'un mécanisme de régulation de l'élongation qui permette

l'économie de l'ATP, nécessaire au remodelage des nucléosomes, s'est peut-être avéré

énergétiquement indispensable.

6.1.7. Perspectives sur les fonctions d'UBF Les travaux présentés aux chapitres 2 et 3 montrent qu'UBFl est important dans la

régulation de l'élongation de la polymérase I. Plusieurs questions demeurent toutefois sans

réponse en ce qui concerne les fonctions de cette protéine. Par exemple, quelle est

l'efficacité d'UBF2 à bloquer l'élongation de la transcription. Des essais de transcription in

vitro comme ceux présentés au chapitre 3 pourraient répondre à cette question. Il serait

aussi intéressant de savoir si UBF interagit avec les protéines Spt4 et Spt5 et si cette

interaction est importante dans la régulation de l'élongation. De plus, les résultats de ChIP

contre la protéine UBF dans les cellules où DNMT1 et DNMT3b ont été inactivées

montrent une augmentation de la présence d'UBF sur les gènes ribosomaux. Ces résultats

suggèrent que l'augmentation du nombre de gènes actifs entraîne une augmentation de

l'attachement d'UBF aux gènes ribosomaux transcrits. Il serait intéressant de vérifier si

UBF est localisé spécifiquement sur ces gènes en conditions normales. Il est raisonnable de

penser qu'UBF est localisé aux gènes transcrits alors que les nucléosomes sont localisés

218

aux gènes silencieux. Ainsi, UBF pourrait réguler l'initiation et l'élongation des transcrits

alors que les nucléosomes pourraient permettre la compaction des gènes silencieux. La

caractérisation des protéines associées à chacun des deux types de gènes ribosomaux serait

très instructive.

L'étude réalisée par Hall et collègues (Hall et al., 2006) montre qu'Hmolp,

l'homologue fonctionnel d'UBF chez la levure, est associé à une majorité de promoteurs de

protéines ribosomales. Il est donc possible qu'UBF soit aussi associé aux promoteurs de

protéines ribosomales. Il serait très intéressant de vérifier, par ChIP, si UBF s'associe à ces

promoteurs chez les mammifères, mais aussi s'il est associé à la région codante de ces

gènes. Si tel était le cas, il faudrait vérifier si UBF peut aussi réguler l'élongation de l'ARN

polymérase IL La régulation par UBF de la transcription des gènes ribosomaux et des

protéines ribosomales serait un moyen extrêmement efficace pour assurer la corégulation

de ces deux systèmes transcriptionnels dans la biogenèse des ribosomes. L'étude des

phénotypes des souris UBF-/- permettra vraisemblablement de déterminer les fonctions in

vivo d'UBF. La comparaison du phénotype de ces souris avec celui des souris « Knock-

Out » pour Rrn3, TAF168 et TAFi95 permettra de savoir si la fonction d'UBF est spécifique

à la transcription ribosomale ou si cette protéine possède aussi d'autres fonctions. De plus,

l'étude des lignées cellulaires UBF-/- complémentées par un transgène inductible (chapitre

5) permettra d'étudier indépendamment les fonctions in vivo d'UBFl et d'UBFZ

6.2. Le rôle de la méthylation de l'ADN dans l'établissement et le maintien des gènes ribosomaux silencieux.

6.2. 1. La méthylation de l'ADN est essentielle pour la biogenèse des ribosomes.

Le principal sujet de ma thèse concerne la régulation du nombre de gènes

ribosomaux transcrits. Pour l'étude de ce niveau de régulation, je me suis intéressée aux

conséquences endogènes de la perte de la méthylation de l'ADN sur le nombre de gènes

ribosomaux transcrits ainsi que sur la transcription ribosomale et la croissance cellulaire

(chapitre 4). J'ai montré que l'inactivation génique des enzymes Dnmtl et Dnmt3b (cellule

DKO) entraîne une perte totale de la méthylation de l'ADN sur les gènes ribosomaux et une

219

augmentation de 2,5 fois du nombre de gènes transcrits. Par contre, cette augmentation du

nombre de gènes actifs ne confère pas un avantage sélectif à la cellule DKO. En effet,

l'hypométhylation des promoteurs entraîne plutôt une diminution de la transcription

ribosomale, causée par une réduction de la vitesse d'élongation de la polymérase, et une

diminution de la vitesse de croissance. De plus, cette hypométhylation cause des problèmes

dans la transformation de TARN ribosomal précurseur en ARNr matures. Des analyses

d'immunofluorescence ont aussi montré que la structure nucléolaire est radicalement

modifiée dans les cellules DKO. Finalement, une des conséquences majeures de la perte de

la méthylation est la transcription des gènes ribosomaux par TARN polymérase II. Ainsi,

les gènes ribosomaux, dans les cellules DKO, sont transcrits en même temps par Poil et

PolII ce qui entraîne vraisemblablement les défauts observés dans la transformation de

l'ARNr précurseur et dans l'assemblage co-transcriptionnel des sous-unités ribosomales.

Nous suggérons que la méthylation de l'ADN est nécessaire pour maintenir une haute

densité de complexes Poli sur les gènes ribosomaux de façon à exclure PolII de ces gènes

et à assurer une biogenèse des ribosomes optimale. Nous suggérons aussi que la

conservation des gènes ribosomaux silencieux est indispensable à la bonne fonction des

gènes ribosomaux.

La provenance des gènes nouvellement transcrits, c'est-à-dire s'ils sont localisés sur

des NORs actifs ou inactifs n'est pas connue. Il est possible que la perte de méthylation de

l'ADN entraîne la transcription de gènes normalement silencieux localisés sur les NORs

actifs et inactifs. Le pourcentage de gènes transcrits dans les cellules DKO serait alors

déterminé par la disponibilité des différents facteurs nécessaires à la transcription par

l'ARN polymérase I. Il est aussi possible que les gènes ribosomaux des NORs inactifs

soient dans un tel état de compaction que la perte de la méthylation seule ne soit pas

suffisante pour permettre leur transcription. Les gènes ribosomaux nouvellement transcrits

proviendraient alors des NORs actifs et le pourcentage de gènes transcrits dans les cellules

DKO serait déterminé par l'addition de ces gènes normalement silencieux à ceux

normalement transcrits. Une analyse FISH, pour « Fluorescence in situ hybridisation », sur

les chromosomes des cellules DKO et, à titre comparatif sur ceux des cellules parentales,

permettrait de savoir de quel type de NOR proviennent les gènes ribosomaux nouvellement

220

transcrits. Bien que ces gènes puissent provenir de NORs actifs ou inactifs, nous avons

montré que la perte de la méthylation de l'ADN entraîne une augmentation du nombre de

gènes transcrits qui pourrait probablement causer une augmentation la longueur de la fibre

ribosomale et conséquemment la formation d'une structure nucléosomale aberrante. Cette

augmentation du nombre de gènes transcrits aurait comme effet direct de diminuer le

nombre d'ARN polymérase I engagés sur chaque gène ribosomal. Cette réduction dans la

densité des complexes permettrait l'interaction de l'ARN polymérase II avec les gènes

ribosomaux. Les études faites chez la levure ont montré que l'ARN polymérase II peut

transcrire les gènes ribosomaux. C'est donc son exclusion du nucléole qui l'empêche de

transcrire ces gènes. Le maintien d'une haute densité de Poli sur les gènes ribosomaux est

vraisemblablement le mécanisme le plus efficace d'exclusion de PolII. Cette hypothèse est

supportée par des études montrant qu'il existe une relation inversement proportionnelle

entre la densité de Poli et la présence de PolII sur les gènes ribosomaux (voir la section

1.3.1 de l'introduction).

6.2.2. La méthylation et la réplication de l'ADN. Une des caractéristiques de la méthylation de l'ADN est qu'elle n'est pas effacée

lors de la réplication. Comme les gènes PolII, les gènes ribosomaux transcrits sont

répliqués plus tôt que les gènes ribosomaux silencieux. Toutefois, comme ces gènes sont

répétés en tandem, il est raisonnable de penser que l'entièreté d'un NOR actif, y compris

ses gènes silencieux, est répliquée au cours de la même période de la phase S. Je suggère

que la méthylation de l'ADN est importante pour définir la frontière entre les gènes

transcrits et les gènes silencieux d'un même NOR actif au cours de sa réplication et que

cette définition est indépendante de NoRC. Cette marque serait ensuite reconnue et

induirait l'inactivation de ces gènes, vraisemblablement par l'activité de NoRC.

L'importance de la méthylation sur les gènes silencieux des NORs actifs, suite à la

réplication, expliquerait pourquoi ces gènes ne sont pas maintenus silencieux dans les

cellules où la méthylation de l'ADN est absente, cellules DKO. Ces gènes ne seraient tout

simplement pas reconnus comme devant être inactivés. Des résultats expérimentaux

supportent cette idée. Li et collègues, ont montré que les gènes transcrits précocement ne

221

sont pas associés à NoRC et qu'une partie de ces gènes est résistante à une enzyme de

restriction dont l'activité est bloquée par la méthylation (Li et al., 2005). Cette enzyme de

restriction, Hpa II, est la même que celle utilisée pour l'étude de l'effet de la 5-aza-

2'dexoxycytidine (section 4.9.4, figure 4.9). Ainsi, une partie des gènes ribosomaux

répliquée de façon précoce, gènes qui seront théoriquement tous transcrits, est méthylée.

Ces résultats suggèrent que ces gènes répliqués de façon précoce seront par la suite

inactivés puisque les gènes transcrits sont dépourvus de méthylation. De plus, cette même

étude (Li et al., 2005) montre que la méthylation de l'ADN, associée aux gènes qui

répliquent précocement, est conservée au cours des divisions cellulaires ce qui suggère que

ces gènes seront réinactivés après chaque réplication de l'ADN. En ce qui concerne les

NORs inactifs, la méthylation de l'ADN n'est peut-être pas essentielle puisque le moment

de la réplication, de façon tardive, et les protéines qui leur sont associées sont probablement

suffisantes, comme c'est le cas pour les gènes PolII, pour les replacer dans une structure

hétérochromatinienne. En accord avec cette hypothèse, des résultats montrent l'association

du complexe de remodelage NoRC avec les gènes ribosomaux répliqués de façon tardive et

suggèrent que NoRC est essentiel pour le maintien, à travers les cycles cellulaires, de ces

gènes dans un état silencieux chez la souris (Li et al., 2005). Pour l'instant, le rôle de la

méthylation de l'ADN au cours de la réplication des gènes ribosomaux n'a pas été testé. Il

serait aussi important d'étudier le rôle du complexe de remodelage de la chromatine NoRC

dans les cellules humaines car rien n'a été fait à ce niveau.

6.2.3. L'hypométhylation du promoteur ribosomal et la tumorogenèse? Entre le début et la fin de mon doctorat, deux études ont été publiées par le Dr Jacob

sur l'effet de la méthylation de l'ADN sur la transcription ribosomale dans les cellules

humaines. La première montre que les promoteurs ribosomaux sont légèrement

hypométhylés dans des cellules de carcinomes du foie comparativement aux tissus sains

(Ghoshal et al., 2004). Cependant, contrairement à notre étude, aucun des promoteurs

étudiés n'est totalement dépourvu de méthylation. Ces comparaisons de patron de

méthylation, entre les cellules saines et les cellules de carcinome, ont été faites par

séquençage suite à un traitement au sodium bisulfite. Ce traitement transforme toutes les

222

cytosines non méthylées en uracils qui sont transcrites en thymines lors de l'amplification

par PCR. Les cytosines méthylées sont quant à elles protégées (figure 6.1 A et B). Les

auteurs suggèrent, à partir de cette seule comparaison de séquence, que l'hypométhylation

des promoteurs est la cause de l'augmentation de la transcription ribosomale dans les

cellules cancéreuses. Il est impossible de confirmer avec certitude ces résultats et ce pour

trois raisons majeures (voir 6.2.3.1-6.1.2.3).

6.2.3.1. La conception des oligonucléotides favorise l'amplification de l'ADN méthylé.

Premièrement, la séquence de deux des quatre oligonucléotides utilisés pour

l'amplification par PCR est introuvable dans la séquence du locus ribosomal humain. Les

deux autres oligonucléotides ressemblent, à quelques bases près, à la séquence ribosomale

publiée. Par contre, la séquence de ces oligonucléotides contient un dinucléotide CpG qui

va préférentiellement s'hybrider avec les promoteurs méthylés, qui seront donc

préférentiellement amplifiés (figure 6.2). Lors de la conception des oligonucléotides, il faut

tenir compte de la conversion des cytosines (figure 6.1B). Il est donc important de ne pas

favoriser l'hybridation de l'oligonucléotide avec l'un ou l'autre des types de promoteurs,

méthylé ou non (figure 6.2). Ainsi, il ne faut pas inclure un G, appariement avec la cytosine

méthylée, ou à un A, appariement avec la cytosine non méthylée, dans la séquence de

l'oligonucléotide à la position de la base qui s'hybridera avec le dinucléotide CpG (figure

6.2). Il faut donc placer un C ou un T pour faire un mauvais appariement avec les deux

types d'ADN (figure 6.2). La séquence d'une partie d'un des oligonucléotides utilisés par

Ghoshal et collègues (Ghoshal et al., 2004) est montrée en exemple à la figure 6.2. Comme

le montre cette figure, la séquence de leur oligonucléotide est spécifique pour le promoteur

méthylé et l'amplification par PCR est donc biaisée. Malgré le fait que je n'ai pas pu

identifier les autres oligonucléotides utilisés par Ghoshal et collègues, leur séquence

contient toujours un ou deux dinucléotides CpG ce qui suppose une amplification

préférentielle de l'ADN méthylé.

6.2.3.2. Sous-clonage préférentiel de l'ADN méthylé. Deuxièmement, Ghoshal et collègues affirment vérifier la transformation des

cytosines non méthylées en digérant l'ADN amplifié par PCR avec l'enzyme de restriction

223

Taql dont le site, selon eux, est créé uniquement lors de la complète conversion de ces

cytosines non méthylées (Ghoshal et al., 2004). Or, le site Taql, dont la séquence de

reconnaissance est TCGA, contient un dinucléotide CpG et est donc problématique pour ce

type d'analyse (figure 6.3). De plus, comme le site TCGA n'existe pas dans la séquence

non transformée, la séquence originale est CCGA. La séquence de ce site de clivage

contient donc potentiellement deux cytosines non méthylées. L'apparition du site de

clivage de Taql, TCGA, confirme la conversion de la première cytosine en thymine

puisque le site TCGA est créé suite à cette conversion (figure 6.3). Cependant, si la

deuxième cytosine, celle contenue dans le CpG, n'est pas méthylée, la séquence

transformée sera TTGA, et non pas TCGA, et le site de clivage de Taql ne sera pas créé

(figure 6.3). Ghoshal et collègues n'ont sous-cloné que les fragments PCR complètement

convertis, c'est-à-dire que tous les sites Taql ont été créés. Comme la séquence amplifiée

par PCR contient 5 sites potentiellement Taql, séquence CCGA, et que les auteurs n'ont

sous-cloné que les fragments PCR complètement convertis, donc complètement digérés, ils

ont sélectionné les fragments PCR contenant des cytosines méthylées à chacun de ces cinq

sites. L'impartialité de la méthode et la quantification qui en découle est donc tout à fait

faussée. Leurs analyses comparatives portent sur la séquence entre -200 et -9, par rapport

au site d'initiation +1. Cette séquence ne contient qu'un site Taql à la position -30. De

façon surprenante, leurs analyses montrent que le dinucléotide CpG contenu dans ce site de

clivage est fréquemment hypométhylé. Si les auteurs n'ont sous-cloné que les fragments

complètements transformés, séquence TCGA au site Taql, comment la cytosine du CpG

peut ne pas être méthylée? Si c'était le cas, la séquence transformée serait TTGA et non

TCGA et donc il n'y aurait pas eu de site Taql et théoriquement pas de sous-clonage de ce

fragment PCR. Malgré le fait qu'ils ne discutent pas de ce dinucléotide CpG

spécifiquement, leurs techniques expérimentales et leurs analyses sont vraisemblablement

inexactes ou à tout le moins contestables.

6.2.3.3. L'hypométhylation du promoteur ribosomal et son implication suggérée dans la tumorogenèse n'est pas démontrée.

Troisièmement, toutes les études publiées à ce jour montrent que les gènes actifs

sont dépourvus de méthylation contrairement aux gènes silencieux qui sont méthylés. Or,

224

tous les promoteurs séquences dans l'article de Ghoshal et collègues sont, à différents

niveaux, méthylés (Ghoshal et al., 2004). Est-ce parce qu'ils ne détectent pas les gènes

ribosomaux transcrits? Ce point n'est pas abordé dans leur discussion. De plus, ils n'ont pas

vérifié si l'hypométhylation des promoteurs, dans les cellules tumorales, augmentait le

nombre de gènes transcrits. Ils n'ont pas non plus montré que la transcription ribosomale

était effectivement augmentée dans les cellules tumorales comparativement aux cellules

saines qu'ils ont utilisées. Bien que ce soit probablement le cas, il aurait fallu le démontrer.

Je ne sais toutefois pas si ces expériences sont possibles dans des lignées cellulaires

primaires.

B)

ADN de départ

s'ACMGTACCATTCGCGAATCMGa' + sodium JL bisulfite ▼

s'ACGTATTATTTGTGAATCGs-

ADN traité

Figure 6.1 : Séquençage suite à un traitement de l'ADN avec du sodium bisulfite. A) Réaction chimique de la transformation des cytosines non méthylées en uraciles. B) Exemple de transformation de la séquence de l'ADN traité au sodium bisulfite. Les cytosines méthylées sont en rouge et sont protégées de la transformation. Les cytosines non méthylées sont en vert et sont amplifiées comme thymines suite à la réaction PCR.

À ma connaissance, cet article est le seul suggérant une corrélation entre

l'hypométhylation des promoteurs et l'augmentation de la transcription ribosomale dans les

cellules tumorales. Comme cette conclusion a été fréquemment référée depuis la

publication de cet article, il m'apparaît très important de relativiser la justesse de

l'argumentation en soulignant les faiblesses de l'expérimentation et des analyses. Le reste

225

de l'article présente des résultats, provenant d'un système de gènes rapporteurs luciférase

transcrits par Poli, qui sont en accord avec le modèle actuel du rôle de la méthylation du

promoteur dans l'établissement et le maintien des gènes ribosomaux silencieux. Les auteurs

montrent que la méthylation de la région promotrice inhibe la transcription ribosomale et

que cette inhibition est sensible à la densité de la méthylation. De plus, ils montrent que

MBD2, une « Methyl Binding Protein », serait importante pour l'inhibition de la

transcription.

A)

B)

ADN non méthylé

5TCCCCGT3-

5TTTTTGT3 3'AAAAACAs-

+ sodium bisulfite

— Oligo —~ spécifiques

ADN méthylé

STCCCCMGTS

5TTTTCGT3 3AAAAGCA5

I Séquence contenue dans l'oligo de Ghoshal et al.

C) 5 I I I I I GÏ3-3 ATTTT/CCAs — Oligo —

non spécifiques

sTTTTCGTs' sATTTT/CCAs

Figure 6.2 : Conception des oligonucléotides pour la réaction d'amplification par PCR de l'ADN traité au sodium bisulfite. A) ADN de départ non méthylé, à gauche, et méthylé, à droite. Les cytosines non méthylées qui seront converties en thymines sont en vert et celles protégées sont suivies d'un M et sont en rouge. B) Séquence résultant du traitement avec du sodium bisulfite de l'ADN présenté en A. La séquence des oligonucléotides spécifiques à chaque type d'ADN, méthylé ou non, est indiquée sous la séquence de l'ADN modifié. La séquence utilisée par Ghoshal et collègues est indiquée par une flèche et est spécifique pour l'ADN méthylé. Les cytosines qui étaient non méthylées ont été converties en thymines et sont en vert. Celles qui étaient méthylées et ont été protégées de la conversion sont en rouge. C) La séquence des oligonucléotides non spécifiques est présentée sous la séquence de l'ADN modifié. Le code de couleur est le même qu'en B et les nucléotides non spécifiques sont indiqués en bleu. Le fragment de séquence présenté dans cette figure provient de la séquence promotrice ribosomale utilisée dans un des oligonucléotides de Ghoshal et collègues.

226

6.2.3.4. Rôle des Dnmts dans la régulation de la transcription ribosomale Le même groupe a publié un second article concernant le rôle des Dnmts dans la

régulation de la transcription ribosomale (Majumder et al., 2006). Certaines expériences ont

été faites dans les mêmes cellules que celles que j 'ai utilisées dans le chapitre 4 de cette

thèse. Quelques-uns de leurs résultats sont en accord avec les nôtres, malgré l'utilisation de

techniques différentes. Par exemple, ils montrent que Dnmt 1 et 3b sont importantes pour la

méthylation des gènes ribosomaux. Ils montrent aussi que le niveau d'ARN précurseur,

mesuré par des analyses de type Northern et par RT-PCR, est équivalent dans les cellules

parentales et Double Knock-out.

ADN non méthylé ADN méthylé

S'NNNCCGANNN3- 5'NNNCQ<GANNN3'

+ sodium bisulfite W

+ Taq1 (site de clivage: TCGA)

5'NNNTCGANNN3'

digestion

partielle 3ins de 5 syes)

totale (5 sites)

souarclona^e et sous-clonage et j^fuençageNes séquençage des ?lones obtenus clones obtenus

Figure 6.3: Vérification de la conversion des cytosines non méthylées par digestion avec l'enzyme de restriction Taq 1. À gauche, la conversion des deux cytosines non méthylée, en vert, mènent à la séquence TTGA suite à la conversion de ces deux cytosines après le traitement avec le sodium bisulfite. À droite, la conversion de la cytosine non méthylé, en vert, mène à la création du site de reconnaissance de l'enzyme de restriction Taql, TCGA. La cytosine du dinucléotide CpG est méthylée et protégée de la conversion par le sodium bisulfite. Cette cytosine est suivie d'un M et est en rouge. Par la suite, Ghoshal et collègues, ont sous clone seulement les fragments amplifiés par PCR qui étaient complètement convertis et qui contenaient donc cinq sites Taql nouvellement créés.

227

Les autres résultats qu'ils ont obtenu dans les lignées cellulaires HCT 116 sont

différents des nôtres. Selon eux, le pourcentage de promoteurs non méthylés est de 28%

dans les cellules parentales et de 76% dans les cellules DKO. Selon mes résultats, 50% des

promoteurs sont non méthylés dans les cellules parentales et 100% dans les cellules DKO.

Cette différence est probablement due à la technique expérimentale utilisée. J'ai obtenu ces

résultats par séquençage suite à un traitement au sodium bisulfite. Leurs résultats

proviennent de la comparaison de l'amplification par PCR, suite à la digestion de l'ADN

génomique par l'enzyme de restriction Hpall décrite précédemment, avec celle de

l'amplification de l'ADN non digéré. Leur pourcentage de promoteurs méthylés correspond

donc au ratio de l'amplification suite à la digestion Hpall sur l'amplification de l'ADN non

digéré. Comme l'activité de Hpall est bloquée par la méthylation, les fragments PCR

amplifiés sur l'ADN digéré par Hpall proviennent de promoteurs méthylés seulement. Le

24% de promoteurs méthylés qu'ils détectent dans les cellules DKO est peut-être dû à une

amplification non spécifique ou à une digestion partielle.

De façon surprenante, malgré le fait que le niveau d'ARN précurseur soit constant

dans les cellules parentales et DKO, ils mesurent une augmentation de cet ARNr dans les

simples mutants Dnmtl-/- et Dnmt 3b-/-. Ces résultats sont en contradiction avec les miens

et je ne peux pas expliquer pourquoi. J'ai mesuré quatre Northern différents et trois d'entre

eux contenaient de l'ARN déposé en duplicata. De plus, chacun de ces Northern a été

hybride avec trois sondes différentes et je n'ai jamais vu d'augmentation d'ARNr

précurseur dans les simples mutants, figure 4.8. Le reste des expériences présentées dans

cet article ont été faites en utilisant un système de gènes rapporteurs transfecté dans des

cellules HeLa, humaines (Majumder et al., 2006). Ils montrent que l'activité

méthyltransférase de Dnmtl est essentielle pour l'inhibition de la transcription ribosomale

d'un promoteur non méthylé, mais est non essentielle pour l'inhibition d'un promoteur

méthylé. L'inhibition de la transcription ribosomale par Dnmt3b serait indépendante de

l'état de méthylation du promoteur et de l'activité de son domaine catalytique, mais serait

dépendante de ses domaines d'interactions avec l'ADN et avec ses partenaires protéiques.

L'inhibition de la transcription ribosomale par ces deux Dnmts seraient augmentée par

l'activité de HDACs dont HDAC2. Finalement, l'inhibition de la transcription de

228

promoteurs ribosomaux non méthylés et méthylés répondrait différemment à l'action des

Dnmts et des HDACs.

6.2.4 La méthylation des gènes ribosomaux et la croissance cellulaire L'inactivation épigénique de gènes suppresseurs de tumeur est maintenant connue

comme étant impliquée dans le développement d'une large variété de cancers humains

(Esteller, 2002; Jones et Baylin, 2002; Robertson, 2001; Robertson et Wolffe, 2000).

Plusieurs études ont été faites pour tenter de mieux comprendre le rôle individuel des

Dnmts dans ce processus d'inactivation épigénique. L'implication probable des Dnmts dans

cette inactivation épigénique a soulevé beaucoup d'intérêt et la drogue 5-aza-

2'dexoxycytidine, un inhibiteur de la méthylation de l'ADN utilisé dans la section 4.9.4, est

déjà en phase clinique 2. Ces traitements seront efficaces s'ils permettent la réactivation des

gènes suppresseurs de tumeur, mais surtout, s'ils entraînent une réduction de la vitesse de

croissance des cellules tumorales traitées. La seule réactivation de gènes suppresseurs de

tumeur, si elle n'est pas couplée à un ralentissement de la croissance des cellules tumorales,

n'aura pas d'effets bénéfiques sur les patients de ces traitements. De plus, ces drogues ne

sont pas spécifiques aux gènes suppresseurs de tumeur et les effets secondaires d'une perte

totale de méthylation de l'ADN pourraient s'avérer à risque pour les patients à long terme.

Une des lignées de cellules tumorales humaines utilisées dans les études sur

l'inactivation épigénétique est la lignée cellulaire HCT 116 (chapitre 4; Rhee et al., 2002;

Rhee et al., 2000). Les cellules HCT 116 sont des cellules de cancers du côlon humain qui

possèdent un allèle muté de pl6INK4A qui est exprimé et un allèle sauvage qui est inactivé

par l'hyperméthylation de son promoteur. Les deux études citées dans l'introduction et

faites par le groupe du Dr Vogelstein (Rhee et al., 2002; Rhee et al., 2000) ont montré que

dans ces cellules, la réactivation de l'allèle sauvage de pl61NK4A, CDKN2A, dépend de

l'inactivation génique combinée de Dnmtl et Dnmt3b. Ainsi, ces deux Dnmts seraient

importantes pour l'inactivation épigénétique de ce gène. La croissance de ces cellules

« Double Knockout », DKO, est diminuée de près de trois fois suite à la double inactivation

de Dnmtl et 3b (Rhee et al., 2002). Ces résultats suggèrent que l'inhibition de ces protéines

229

est efficace pour contrer l'avantage sélectif des cellules tumorales au niveau de leur

croissance cellulaire.

Toutefois, le même groupe a aussi montré que dans les cellules DKO de passage 22

l'allèle sauvage de pl6INK4A est de nouveau inactivé, mais cette fois par un mécanisme

indépendant de la méthylation de son promoteur (Bachman et al., 2003). Le défaut de

croissance qui est toujours observé dans ces cellules est donc indépendant de la réactivation

de l'allèle sauvage de pl6INK4A. Les passages que nous avons utilisés pour les cellules DKO

dans notre étude présentée au chapitre 4 sont de passage 13 à 25 pour les cellules avec peu

de passages, E pour « early » et de passages 27 à 40 pour les cellules avec plus de passages,

L pour « late ». Il est donc raisonnable de penser que les cellules DKO L et probablement

aussi une partie des cellules DKO E expriment seulement l'allèle muté de pl6

Comme le montre la figure 4.4, les cellules DKO E et L présentent un défaut de croissance

qui est vraisemblablement indépendant de l'activité de pl6INK4A ce qui expliquerait

pourquoi ces cellules ne bloquent pas en phase Gl du cycle cellulaire comme le prévoit

l'action de pl6INK4A (figure 4.4). L'ensemble des résultats présentés au chapitre 4 suggère

que ce sont les défauts dans la biogenèse des ribosomes et non la réactivation de gènes

suppresseurs de tumeur qui sont responsables des défauts de croissance résultant de la perte

de la méthylation de l'ADN. Il est donc probable que les effets des inhibiteurs de la

méthylation de l'ADN sur la croissance soient causés par leur interférence avec la

biogenèse normale des ribosomes et non parce qu'ils réactivent des gènes suppresseurs de

tumeur. Il serait intéressant de développer une drogue qui ciblerait de façon spécifique la

transcription ribosomale et qui pourrait diminuer la biogenèse des ribosomes et donc la

croissance des cellules tumorales.

6.2.5. Perspectives L'étude de la perte de la méthylation de l'ADN a fourni des résultats très

intéressants qui lient encore une fois, comme les résultats présentés aux chapitres 2 et 3, la

transcription ribosomale à la croissance cellulaire. Peu d'études ont été faites sur les

mécanismes d'inactivation et de maintien de gènes ribosomaux silencieux dans les cellules

humaines. Il serait intéressant de vérifier, par FISH, la provenance des gènes nouvellement

230

transcrit pour déterminer s'ils sont sur des NORs actifs ou inactifs. À ce sujet, il serait

intéressant, comme déjà mentionné, de caractériser les protéines associées aux gènes actifs

et silencieux. Il serait aussi important de confirmer l'existence de TiP5 et de NoRC dans les

cellules humaines et leur rôle dans l'inactivation des gènes ribosomaux. Puisque la perte de

la méthylation de l'ADN cause une augmentation du nombre de gènes transcrits, il serait

intéressant de comparer, par ChIP, la distribution de ces deux protéines dans les cellules

HCT116 parentales et DKO.

6.3. Une haute densité de complexes d'ARN polymérase I engagés sur les gènes ribosomaux est essentielle pour la biogenèse normale des ribosomes.

Les résultats obtenus suite à la perte de la méthylation de l'ADN montrent à quel

point il est important de maintenir un haut niveau de densité de complexes d'ARN

polymérases I sur les gènes ribosomaux de façon à exclure l'ARN polymérase II du locus

ribosomal et à assurer une biogenèse efficace des ribosomes. Un autre niveau de régulation,

présenté dans cette thèse, assure aussi le maintien d'une haute densité de complexes Poil

tout en permettant un ajustement de la transcription aux stimuli cellulaires. Il s'agit du

contrôle de la vitesse de l'élongation de l'ARN polymérase I. Si la régulation n'avait lieu

qu'au niveau de l'initiation de la transcription, alors des changements dans le taux

d'initiation feraient varier la densité des complexes engagés dans la transcription. Un faible

taux d'initiation, pour répondre à une faible demande en ARN ribosomaux précurseurs,

entraînerait une faible densité de complexes sur les gènes alors qu'une forte demande

entraînerait une haute densité de complexes. Le nombre d'ARN polymérase I par gène

varierait donc en fonction de la demande transcriptionnelle. La faible densité de complexes

lors d'une faible demande transcriptionnelle entraînerait vraisemblablement les mêmes

problèmes que ceux causés par la perte de la méthylation de l'ADN soit une transcription

ribosomale diminuée, une mauvaise transformation de l'ARN ribosomal précurseur, une

incursion de PolII dans le locus et un défaut de croissance. Une régulation au niveau de la

vitesse d'élongation de l'ARN polymérase I permet, au contraire, de conserver une densité

élevée de Poli sur les gènes ribosomaux tout en permettant d'ajuster la synthèse de l'ARN

ribosomal précurseur en augmentant ou en diminuant cette vitesse d'élongation.

231

Le maintien d'un certain nombre de gènes ribosomaux silencieux ainsi que

l'existence d'un mécanisme de régulation de Pélongation suggèrent que l'exclusion de

l'ARN polymérase II des gènes ribosomaux, par une haute densité de complexes de

transcription, est absolument essentielle pour la biogenèse normale des ribosomes et pour la

croissance cellulaire. Bien que tout ce qui est étudié en biologie cellulaire et moléculaire est

complexe, il semble que la cellule ait trouvé un moyen simple pour s'assurer d'une

biogenèse efficace des ribosomes: elle protège la transcription ribosomale par la

transcription ribosomale.

Bibliographie Aapola, U., Kawasaki, K., Scott, H. S., Ollila, J., Vihinen, M., Heino, M., Shintani, A., Kawasaki, K., Minoshima, S., Krohn, K., et al. (2000). Isolation and initial characterization of a novel zinc finger genc, DNMT3L, on 21q22.3, related to the cytosine-5-methyltransferase 3 gène family. Genomics 65, 293-298.

Albert, A. C, Denton, M., Kermekchiev, M., and Pikaard, C. S. (1999). Histone acetyltransferase and protein kinase activities copurify with a putative Xenopus RNA polymerase I holoenzyme self-sufficient for promoter-dependent transcription. Mol Cell Biol 19, 796-806.

Albert, A. L., Lavoie, S. B., and Vincent, M. (2004). Multisite phosphorylation of Pinl-associated mitotic phosphoproteins revealed by monoclonal antibodies MPM-2 and CC-3. BMC Cell Biol 5, 22.

Aoki, A., Suetake, I., Miyagawa, J., Fujio, T., Chijiwa, T., Sasaki, H., and Tajima, S. (2001). Enzymatic properties of de novo-type mouse DNA (cytosine-5) methyltransferases. Nucleic Acids Res 29, 3506-3512.

Bachman, K. E., Park, B. H., Rhee, I., Rajagopalan, H., Herman, J. G., Baylin, S. B., Kinzler, K. W., and Vogelstein, B. (2003). Histone modifications and silencing prior to DNA methylation of a tumor suppressor gène. Cancer Cell 3, 89-95.

Bachman, K. E., Rountree, M. R., and Baylin, S. B. (2001). Dnmt3a and Dnmt3b are transcriptional repressors that exhibit unique localization properties to heterochromatin. J Biol Chem 276, 32282-32287.

Bacolla, A., Pradhan, S., Roberts, R. J., and Wells, R. D. (1999). Recombinant human DNA (cytosine-5) methyltransferase. II. Steady-state kinetics reveal allosteric activation by methylated dna. J Biol Chem 274, 33011-33019.

Basi, G., and Draetta, G. (1995). pl3sucl of Schizosaccharomyces pombe régulâtes two distinct forms of the mitotic cdc2 kinase. Mol Cell Biol 15, 2028-2036.

Bazett-Jones, D. P., Leblanc, B., Herfort, M., and Moss, T. (1994). Short-range DNA looping by the Xenopus HMG-box transcription factor, xUBF. Science 264, 1134-1137.

Beard, C, Li, E., and Jaenisch, R. (1995). Loss of methylation activâtes Xist in somatic but not in embryonic cells. Gènes Dev 9, 2325-2334.

Bell, P., Mais, C , McStay, B., and Scheer, U. (1997). Association of the nucleolar transcription factor UBF with the transcriptionally inactive rRNA gènes of pronuclei and early Xenopus embryos. J Cell Sci 110 (Pt 17), 2053-2063.

Benguria, A., Hernandez, P., Krimer, D. B., and Schvartzman, J. B. (2003). Sir2p suppresses recombination of replication forks stalled at the replication fork barrier of ribosomal DNA in Saccharomyces cerevisiae. Nucleic Acids Res 31, 893-898.

Bestor, T. H. (1988). Cloning of a mammalian DNA methyltransferase. Gène 74, 9-12.

Bird, A. (2002). DNA methylation patterns and epigenetic memory. Gènes Dev 16, 6-21.

Bodem, J., Dobreva, G., Hoffmann-Rohrer, U., Iben, S., Zentgraf, H., Delius, H., Vingron, M., and Grummt, I. (2000). TIF-IA, the factor mediating growth-dependent control of ribosomal RNA synthesis, is the mammalian homolog of yeast Rrn3p. EMBO Rep /, 171-175.

Bonfils, C , Beaulieu, N., Chan, E., Cotton-Montpetit, J., and MacLeod, A. R. (2000). Characterization of the human DNA methyltransferase splice variant Dnmtlb. J Biol Chem 275, 10754-10760.

Boukhgalter, B., Liu, M., Guo, A., Tripp, M., Tran, K., Huynh, C, and Pape, L. (2002). Characterization of a fission yeast subunit of an RNA polymerase I essential transcription initiation factor, SpRrn7h/TAF(I)68, that bridges yeast and mammals: association with SpRrnl lh and the core ribosomal RNA gène promoter. Gène 291, 187-201.

Bourc'his, D., Xu, G. L., Lin, C. S., Bollman, B., and Bestor, T. H. (2001). Dnmt3L and the establishment of maternai genomic imprints. Science 294, 2536-2539.

Brock, G. J., and Bird, A. (1997). Mosaic methylation of the repeat unit of the human ribosomal RNA gènes. Hum Mol Genêt 6, 451-456.

Brown, J. D. (2001). Ribosome biogenesis: stripping for AAAction? Curr Biol 11, R710-712.

Buck, S. W., Sandmeier, J. J., and Smith, J. S. (2002). RNA polymerase I propagates unidirectional spreading of rDNA silent chromatin. Cell 111, 1003-1014.

Busch, H., Muramatsu, M., Adams, H., Steele, W. J., Liau, M. C, and Smetana, K. (1963). Isolation of Nucleoli. Exp Cell Res 24, SUPPL9:150-163.

Carlson, L. L., Page, A. W., and Bestor, T. H. (1992). Properties and localization of DNA methyltransferase in preimplantation mouse embryos: implications for genomic imprinting. Gènes Dev 6,2536-2541.

Chedin, F., Lieber, M. R., and Hsieh, C. L. (2002). The DNA methyltransferase-like protein DNMT3L stimulâtes de novo methylation by Dnmt3a. Proc Natl Acad Sci U S A 99, 16916-16921.

Chen, D., and Huang, S. (2001). Nucleolar components involved in ribosome biogenesis cycle between the nucleolus and nucleoplasm in interphase cells. J Cell Biol 153, 169-176. Chen, T., Ueda, Y., Xie, S., and Li, E. (2002). A novel Dnmt3a isoform produced from an alternative promoter localizes to euchromatin and its expression correlates with active de novo methylation. J Biol Chem 277, 38746-38754.

Chen, Z. J., Cornai, L., and Pikaard, C. S. (1998). Gène dosage and stochastic effects détermine the severity and direction of uniparental ribosomal RNA gène silencing (nucleolar dominance) in Arabidopsis allopolyploids. Proc Natl Acad Sci U S A 95, 14891-14896.

Chen, Z. J., and Pikaard, C. S. (1997). Epigenetic silencing of RNA polymerase I transcription: a rôle for DNA methylation and histone modification in nucleolar dominance. Gènes Dev 11, 2124-2136.

Chow, J. C, and Brown, C. J. (2003). Forming facultative heterochromatin: silencing of an X chromosome in mammalian females. Cell Mol Life Sci 60, 2586-2603.

Cioci, F., Vu, L., Eliason, K., Oakes, M., Siddiqi, I. N., and Nomura, M. (2003). Silencing in yeast rDNA chromatin: reciprocal relationship in gène expression between RNA polymerase I and II. Mol Cell 12, 135-145.

Cmarko, D., Verschure, P. J., Martin, T. E., Dahmus, M. E., Krause, S., Fu, X. D., van Driel, R., and Fakan, S. (1999). Ultrastructural analysis of transcription and splicing in the cell nucleus after bromo-UTP microinjection. Mol Biol Cell 10, 211-223.

Cornai, L., Tanese, N., and Tjian, R. (1992). The TATA-binding protein and associated factors are intégral components of the RNA polymerase I transcription factor, SL1. Cell 68, 965-976.

Conconi, A., Widmer, R. M., Koller, T., and Sogo, J. M. (1989). Two différent chromatin structures coexist in ribosomal RNA gènes throughout the cell cycle. Cell 57, 753-761.

Dechampesme, A. M., Koroleva, O., Leger-Silvestre, L, Gas, N., and Camier, S. (1999). Assembly of 5S ribosomal RNA is required at a spécifie step of the pre-rRNA processing pathway. J Cell Biol 145, 1369-1380.

Dennis, K., Fan, T., Geiman, T., Yan, Q., and Muegge, K. (2001). Lsh, a member of the SNF2 family, is required for genome-wide methylation. Gènes Dev 15, 2940-2944.

Derenzini, M. (2000). The AgNORs. Micron 31, 117-120.

Derenzini, M., Trere, D., Pession, A., Govoni, M., Sirri, V., and Chieco, P. (2000). Nucleolar size indicates the rapidity of cell prolifération in cancer tissues. J Pathol 191, 181-186.

Dick, F. A., Eisinger, D. P., and Trumpower, B. L. (1997). Exchangeability of Qsrlp, a large ribosomal subunit protein required for subunit joining, suggests a novel translational regulatory mechanism. FEBS Lett 419, 1-3.

Dimario, P. J. (2004). Cell and molecular biology of nucleolar assembly and disassembly. IntRevCytol 239, 99-178.

Dousset, T., Wang, C, Verheggen, C, Chen, D., Hernandez-Verdun, D., and Huang, S. (2000). Initiation of nucleolar assembly is independent of RNA polymerasc I transcription. Mol Biol Cell 77,2705-2717.

Dragon, F., Gallagher, J. E., Compagnone-Post, P. A., Mitchell, B. M., Porwancher, K. A., Wehner, K. A., Wormsley, S., Settlage, R. E., Shabanowitz, J., Osheim, Y., étal. (2002). A large nucleolar U3 ribonucleoprotein required for 18S ribosomal RNA biogenesis. Nature 477,967-970.

Dundr, M., Hoffmann-Rohrer, U., Hu, Q., Grummt, I., Rothblum, L. L, Phair, R. D., and Misteli, T. (2002). A kinetic framework for a mammalian RNA polymerase in vivo. Science 298, 1623-1626.

Dundr, M., and Misteli, T. (2001). Functional architecture in the cell nucleus. Biochem J 556,297-310.

Dundr, M., Misteli, T., and Oison, M. O. (2000). The dynamics of postmitotic reassembly of the nucleolus. J Cell Biol 750, 433-446.

Dundr, M., and Raska, I. (1993). Nonisotopic ultrastructural mapping of transcription sites within the nucleolus. Exp Cell Res 208, 275-281.

Eberhard, D., Tora, L., Egly, J. M., and Grummt, I. (1993). A TBP-containing multiprotein complex (TIF-IB) médiates transcription specificity of murine RNA polymerase 1. Nucleic Acids Res 27,4180-4186.

Esteller, M. (2002). CpG island hypermethylation and tumor suppressor gènes: a booming présent, a brighter future. Oncogene 21, 5427-5440.

Esteller, M., and Herman, J. G. (2002). Cancer as an epigenetic disease: DNA methylation and chromatin altérations in human tumours. J Pathol 196, 1-7.

Fatemi, M., Hermann, A., Pradhan, S., and Jeltsch, A. (2001). The activity of the murine DNA methyltransferase Dnmtl is controlled by interaction of the catalytic domain with the N-terminal part of the enzyme leading to an allosteric activation of the enzyme after binding to methylated DNA. J Mol Biol 309, 1189-1199.

Fath, S., Milkereit, P., Peyroche, G., Riva, M., Caries, C, and Tschochner, H. (2001). Differential rôles of phosphorylation in the formation of transcriptional active RNA polymerase I. Proc Natl Acad Sci U S A 98, 14334-14339.

Fatica, A., and Tollervey, D. (2002). Making ribosomes. Curr Opin Cell Biol 14, 313-318. Fcinberg, A. P., and Tycko, B. (2004). The history of cancer epigenetics. Nat Rev Cancer 4, 143-153.

Franceschini, P., Martino, S., Ciocchini, M., Ciuti, E., Vardeu, M. P., Guala, A., Signorile, F., Camerano, P., Franceschini, D., and Tovo, P. A. (1995). Variability of clinical and immunological phenotype in immunodeficiency-centromeric instability-facial anomalies syndrome. Report of two new patients and review of the literature. Eur J Pediatr 154, 840-846.

French, S. L., Osheim, Y. N., Cioci, F., Nomura, M., and Beyer, A. L. (2003). In exponentially growing Saccharomyces cerevisiae cells, rRNA synthesis is determined by the summed RNA polymerase I loading rate rather than by the number of active gènes. Mol Cell Biol 23, 1558-1568.

Frieman, M., Chen, Z. J., Saez-Vasquez, J., Shen, L. A., and Pikaard, C. S. (1999). RNA polymerase I transcription in a Brassica interspecific hybrid and its progenitors: Tests of transcription factor involvement in nucleolar dominance. Genetics 152, 451-460.

Fromont-Racine, M., Senger, B., Saveanu, C, and Fasiolo, F. (2003). Ribosome assembly in eukaryotes. Gène 313, 17-42.

Fuks, F., Burgers, W. A., Godin, N., Kasai, M., and Kouzarides, T. (2001). Dnmt3a binds deacetylases and is recruited by a sequence-specific repressor to silence transcription. Embo J 20, 2536-2544.

Fuks, F., Hurd, P. J., Déplus, R., and Kouzarides, T. (2003). The DNA methyltransferases associate with HP1 and the SUV39H1 histone methyltransferase. Nucleic Acids Res 31, 2305-2312.

Gadal, O., Labarre, S., Boschiero, C, and Thuriaux, P. (2002). Hmol, an HMG-box protein, belongs to the yeast ribosomal DNA transcription system. Embo J 21, 5498-5507. Gadal, O., Strauss, D., Kessl, J., Trumpower, B., Tollervey, D., and Hurt, E. (2001). Nuclear export of 60s ribosomal subunits dépends on Xpolp and requires a nuclear export sequence-containing factor, Nmd3p, that associâtes with the large subunit protein RpllOp. Mol Cell Biol 21, 3405-3415.

Gallagher, J. E., Dunbar, D. A., Granneman, S., Mitchell, B. M., Osheim, Y., Beyer, A. L., and Baserga, S. J. (2004). RNA polymerase I transcription and pre-rRNA processing are linked by spécifie SSU processome components. Gènes Dev 18, 2506-2517.

Gebrane-Younes, J., Fomproix, N., and Hernandez-Verdun, D. (1997). When rDNA transcription is arrested during mitosis, UBF is still associated with non-condensed rDNA. J Cell Sci 110 (Pt 19), 2429-2440.

Gerbi, S. A., Borovjagin, A. V., and Lange, T. S. (2003). The nucleolus: a site of ribonucleoprotein maturation. Curr Opin Cell Biol 15, 318-325.

Ghidelli, S., Donze, D., Dhillon, N., and Kamakaka, R. T. (2001). Sir2p exists in two nucleosome-binding complexes with distinct deacetylase activities. Embo J 20, 4522-4535. Ghoshal, K., Majumder, S., Datta, J., Motiwala, T., Bai, S., Sharma, S. M., Frankel, W., and Jacob, S. T. (2004). Rôle of human ribosomal RNA (rRNA) promoter methylation and of methyl-CpG-binding protein MBD2 in the suppression of rRNA gène expression. J Biol Chem 279, 6783-6793.

Gibbons, R. J., McDowell, T. L., Raman, S., O'Rourke, D. M., Garrick, D., Ayyub, H., and Higgs, D. R. (2000). Mutations in ATRX, encoding a SWI/SNF-like protein, cause diverse changes in the pattern of DNA methylation. Nat Genêt 24, 368-371.

Goll, M. G., and Bestor, T. H. (2005). Eukaryotic cytosine methyltransferases. Annu Rev Biochem 74, 481-514.

Gowher, H., and Jeltsch, A. (2002). Molecular enzymology of the catalytic domains of the Dnmt3a and Dnmt3b DNA methyltransferases. J Biol Chem 277, 20409-20414.

Grandi, P., Rybin, V., Bassler, J., Petfalski, E., Strauss, D., Marzioch, M., Schafer, T., Kuster, B., Tschochner, H., Tollervey, D., et al. (2002). 90S pre-ribosomes include the 35S pre-rRNA, the U3 snoRNP, and 40S subunit processing factors but predominantly lack 60S synthesis factors. Mol Cell 10, 105-115.

Gruenbaum, Y., Cedar, H., and Razin, A. (1982). Substrate and séquence specificity of a eukaryotic DNA methylase. Nature 295, 620-622.

Grummt, I. (2003). Life on a planet of its own: régulation of RNA polymerase I transcription in the nucleolus. Gènes Dev 17, 1691-1702.

Grummt, L, and Pikaard, C. S. (2003). Epigenetic silencing of RNA polymerase I transcription. Nat Rev Mol Cell Biol 4, 641-649.

Guarente, L. (2000). Sir2 links chromatin silencing, metabolism, and aging. Gènes Dev 14, 1021-1026.

Guarente, L., and Kenyon, C. (2000). Genetic pathways that regulate ageing in model organisms. Nature 408, 255-262.

Guillen, A. K., Hirai, Y., Tanoue, T., and Hirai, H. (2004). Transcriptional repression mechanisms of nucleolus organizer régions (NORs) in humans and chimpanzees. Chromosome Res 12, 225-237.

Hall, D. B., Wade, J. T., and Struhl, K. (2006). An HMG protein, Hmol, associâtes with promoters of many ribosomal protein gènes and throughout the rRNA gène locus in Saccharomyces cerevisiae. Mol Cell Biol 26, 3672-3679.

Hanada, K., Song, C. Z., Yamamoto, K., Yano, K., Maeda, Y., Yamaguchi, K., and Muramatsu, M. (1996). RNA polymerase I associated factor 53 binds to the nucleolar transcription factor UBF and functions in spécifie rDNA transcription. Embo J 75, 2217-2226.

Hannan, K. M., Brandenburger, Y., Jenkins, A., Sharkey, K., Cavanaugh, A., Rothblum, L., Moss, T., Poortinga, G., McArthur, G. A., Pearson, R. B., and Hannan, R. D. (2003). mTOR-dependent régulation of ribosomal gène transcription requires S6K1 and is mediated by phosphorylation of the carboxy-terminal activation domain of the nucleolar transcription factor UBF. Mol Cell Biol 23, 8862-8877.

Hannan, K. M., Hannan, R. D., Smith, S. D., Jefferson, L. S., Lun, M., and Rothblum, L. I. (2000). Rb and pi30 regulate RNA polymerase I transcription: Rb disrupts the interaction between UBF and SL-1. Oncogene 19, 4988-4999.

Hannan, R. D., Cavanaugh, A., Hempel, W. M., Moss, T., and Rothblum, L. (1999). Identification of a mammalian RNA polymerase I holoenzyme containing components of the DNA repair/replication System. Nucleic Acids Res 27, 3720-3727.

Hâta, K., Okano, M., Lei, H., and Li, E. (2002). Dnmt3L coopérâtes with the Dnmt3 family of de novo DNA methyltransferases to establish maternai imprints in mice. Development 129, 1983-1993.

Heix, J., and Grummt, I. (1995). Species specificity of transcription by RNA polymerase I. Curr Opin Genêt Dev 5, 652-656.

Heix, J., Vente, A., Voit, R., Budde, A., Michaelidis, T. M., and Grummt, I. (1998). Mitotic silencing of human rRNA synthesis: inactivation of the promoter selectivity factor SL1 by cdc2/cyclin B-mediated phosphorylation. Embo J 17, 7373-7381.

Heix, J., Zomerdijk, J. C, Ravanpay, A., Tjian, R., and Grummt, I. (1997). Cloning of murine RNA polymerase I-specific TAF factors: conserved interactions between the subunits of the species-specific transcription initiation factor TIF-IB/SL1. Proc Natl Acad SciUSAP4, 1733-1738.

Heliot, L., Kaplan, H., Lucas, L., Klein, C , Beorchia, A., Doco-Fenzy, M., Ménager, M., Thiry, M., O'Donohue, M. F., and Ploton, D. (1997). Electron tomography of metaphase

nucleolar organizer régions: évidence for a twisted-loop organization. Mol Biol Cell 8, 2199-2216.

Hempel, W. M., Cavanaugh, A. H., Hannan, R. D., Taylor, L., and Rothblum, L. I. (1996). The species-specific RNA polymerase 1 transcription factor SL-1 binds to upstream binding factor. Mol Cell Biol 16, 557-563.

Hermann, A., Gowher, H., and Jeltsch, A. (2004). Biochemistry and biology of mammalian DNA methyltransferases. Cell Mol Life Sci 61, 2571-2587.

Hermann, A., Schmitt, S., and Jeltsch, A. (2003). The human Dnmt2 has residual DNA-(cytosine-C5) methyltransferase activity. J Biol Chem 278, 31717-31721.

Hernandez-Verdun, D., Roussel, P., and Gebrane-Younes, J. (2002). Emerging concepts of nucleolar assembly. J Cell Sci 115, 2265-2270.

Hirschler-Laszkiewicz, L, Cavanaugh, A., Hu, Q., Catania, J., Avantaggiati, M. L., and Rothblum, L. I. (2001). The rôle of acetylation in rDNA transcription. Nucleic Acids Res 29,4114-4124.

Hirschler-Laszkiewicz, L, Cavanaugh, A. H., Mirza, A., Lun, M., Hu, Q., Smink, T., and Rothblum, L. I. (2003). Rrn3 becomes inactivated in the process of ribosomal DNA transcription. J Biol Chem 278, 18953-18959.

Hisatake, K., Nishimura, T., Maeda, Y., Hanada, K., Song, C. Z., and Muramatsu, M. (1991). Cloning and structural analysis of cDNA and the gène for mouse transcription factor UBF. Nucleic Acids Res 19, 4631-4637.

Ho, J. H., Kallstrom, G., and Johnson, A. W. (2000). Nmd3p is a Crmlp-dependent adapter protein for nuclear export of the large ribosomal subunit. J Cell Biol 151, 1057-1066.

Howell, C. Y., Bestor, T. H., Ding, F., Latham, K. E., Mertineit, C, Trasler, J. M., and Chaillet, J. R. (2001). Genomic imprinting disrupted by a maternai effect mutation in the Dnmtl gène. Cell 104, 829-838.

Hozak, P., Cook, P. R., Schofer, C, Mosgoller, W., and Wachtler, F. (1994). Site of transcription of ribosomal RNA and intranucleolar structure in HeLa cells. J Cell Sci 107 ( Pt 2), 639-648.

Jackson, J. P., Lindroth, A. M., Cao, X., and Jacobsen, S. E. (2002). Control of CpNpG DNA methylation by the KRYPTONITE histone H3 methyltransferase. Nature 416, 556-560.

Jantzen, H. M., Chow, A. M., King, D. S., and Tjian, R. (1992). Multiple domains of the RNA polymerase I activator hUBF interact with the TATA-binding protein complex hSLl to médiate transcription. Gènes Dev 6, 1950-1963.

Johnson, A. W., Lund, E., and Dahlberg, J. (2002). Nuclear export of ribosomal subunits. Trends Biochem Sci 27, 580-585.

Jones, J. C, Phatnani, H. P., Haystead, T. A., MacDonald, J. A., Alam, S. M., and Greenleaf, A. L. (2004). C-terminal repeat domain kinase I phosphorylates Ser2 and Ser5 of RNA polymerase II C-terminal domain repeats. J Biol Chem 279, 24957-24964.

Jones, P. A., and Baylin, S. B. (2002). The fondamental rôle of epigenetic events in cancer. NatRev Genêt 3, 415-428.

Jones, P. L., Veenstra, G. J., Wade, P. A., Vermaak, D., Kass, S. U., Landsberger, N., Strouboulis, J., and Wolffe, A. P. (1998). Methylated DNA and MeCP2 recruit histone deacetylase to repress transcription. Nat Genêt 19, 187-191.

Juttermann, R., Li, E., and Jaenisch, R. (1994). Toxicity of 5-aza-2'-deoxycytidine to mammalian cells is mediated primarily by covalent trapping of DNA methyltransferase rather than DNA demethylation. Proc Natl Acad Sci U S A 91, 11797-11801.

Kaiser, B. K., Zimmerman, Z. A., Charbonneau, H., and Jackson, P. K. (2002). Disruption of centrosome structure, chromosome ségrégation, and cytokinesis by misexpression of human Cdcl4A phosphatase. Mol Biol Cell 13, 2289-2300.

Kihm, A. J., Hershey, J. C, Haystead, T. A., Madsen, C. S., and Owens, G. K. (1998). Phosphorylation of the rRNA transcription factor upstream binding factor promûtes its association with TATA binding protein. Proc Natl Acad Sci U S A 95, 14816-14820.

Kim, G. D., Ni, J., Kelesoglu, N., Roberts, R. J., and Pradhan, S. (2002). Co-operation and communication between the human maintenance and de novo DNA (cytosine-5) methyltransferases. Embo J 21, 4183-4195.

Klein, J., and Grummt, I. (1999). Cell cycle-dependent régulation of RNA polymerase I transcription: the nucleolar transcription factor UBF is inactive in mitosis and early Gl. Proc Natl Acad Sci U S A 96, 6096-6101.

Klose, R. J., and Bird, A. P. (2006). Genomic DNA methylation: the mark and its mediators. Trends Biochem Sci 31, 89-97.

Kressler, D., Linder, P., and de La Cruz, J. (1999). Protein trans-acting factors involved in ribosome biogenesis in Saccharomyces cerevisiae. Mol Cell Biol 19, 7897-7912.

Kuhn, A., and Grummt, I. (1992). Dual rôle of the nucleolar transcription factor UBF: trans-activator and antirepressor. Proc Natl Acad Sci U S A 89, 7340-7344.

Kuhn, A., Stefanovsky, V., and Grummt, I. (1993). The nucleolar transcription activator UBF relieves Ku antigen-mediated repression of mouse ribosomal gène transcription. Nucleic Acids Res 21, 2057-2063.

Kuhn, A., Voit, R., Stefanovsky, V., Evers, R., Bianchi, M., and Grummt, I. (1994). Functional différences between the two splice variants of the nucleolar transcription factor UBF: the second I1MG box détermines specificity of DNA binding and transcriptional activity. Embo J 13, 416-424.

Kunert, N., Marhold, J., Stanke, J., Stach, D., and Lyko, F. (2003). A Dnmt2-like protein médiates DNA methylation in Drosophila. Development 130, 5083-5090.

Laferte, A., Favry, E., Sentenac, A., Riva, M., Caries, C., Chédin, S. (2006). The transcriptional activity of RNA polymerase I is a key déterminant for the level of ail ribosome components. Gènes Dev 20, 2030-2040.

Langst, G., Becker, P. B., and Grummt, I. (1998). TTF-I détermines the chromatin architecture of the active rDNA promoter. Embo J 17, 3135-3145.

Langst, G., Blank, T. A., Becker, P. B., and Grummt, I. (1997). RNA polymerase I transcription on nucleosomal templates: the transcription termination factor TTF-I induces chromatin remodeling and relieves transcriptional repression. Embo J 16, 760-768.

Leonhardt, H., Page, A. W., Weier, H. U., and Bestor, T. II. (1992). A targeting séquence directs DNA methyltransferase to sites of DNA replication in mammalian nuclei. Cell 71, 865-873.

Leung, A. K., Trinkle-Mulcahy, L., Lam, Y. W., Andersen, J. S., Mann, M., and Lamond, A. I. (2006). NOPdb: Nucleolar Proteome Database. Nucleic Acids Res 34, D218-220.

Li, E. (2002). Chromatin modification and epigenetic reprogramming in mammalian development. Nat Rev Genêt 3, 662-673.

Li, E., Bestor, T. H., and Jaenisch, R. (1992). Targeted mutation of the DNA methyltransferase gène results in embryonic lethality. Cell 69, 915-926.

Li, J., Santoro, R., Koberna, K., and Grummt, I. (2005). The chromatin remodeling complex NoRC controls replication timing of rRNA gènes. Embo J 24, 120-127.

Liang, G., Chan, M. F., Tomigahara, Y., Tsai, Y. C, Gonzales, F. A., Li, E., Laird, P. W., and Jones, P. A. (2002). Cooperativity between DNA methyltransferases in the maintenance methylation of répétitive éléments. Mol Cell Biol 22, 480-491.

Liu, C. J., Wang, H., and Lengyel, P. (1999). The interferon-inducible nucleolar p204 protein binds the ribosomal RNA-specific UBF1 transcription factor and inhibits ribosomal RNA transcription. Embo J 18, 2845-2854.

Liu, K., Wang, Y. F., Cantemir, C , and Muller, M. T. (2003). Endogenous assays of DNA methyltransferases: Evidence for differential activities of DNMT1, DNMT2, and DNMT3 in mammalian cells in vivo. Mol Cell Biol 23, 2709-2719.

Liu, Y., Oakeley, E. J., Sun, L., and Jost, J. P. (1998). Multiple domains are involved in the targeting of the mouse DNA methyltransferase to the DNA replication lbci. Nucleic Acids Res 26, 1038-1045.

Mailand, N., Lukas, C, Kaiser, B. K., Jackson, P. K., Bartek, J., and Lukas, J. (2002). Deregulated human Cdcl4A phosphatase disrupts centrosome séparation and chromosome ségrégation. Nat Cell Biol 4, 317-322.

Majumder, S., Ghoshal, K., Datta, J., Smith, D. S., Bai, S., and Jacob, S. T. (2006). Rôle of DNA methyltransferases in régulation of human ribosomal RNA gène transcription. J Biol Chem 281, 22062-22072.

Marilley, M., and Pasero, P. (1996). Common DNA structural features exhibited by eukaryotic ribosomal genc promoters. Nucleic Acids Res 24, 2204-2211.

Marilley, M., Radebaugh, C. A., Geiss, G. K., Laybourn, P. J., and Paule, M. R. (2002). DNA structural variation affects complex formation and promoter melting in ribosomal RNA transcription. Mol Genêt Genomics 267, 781-791.

McStay, B. (2006). Nucleolar dominance: a model for rRNA gène silencing. Gènes Dev 20, 1207-1214.

Mertineit, C., Yoder, J. A., Taketo, T., Laird, D. W., Trasler, J. M., and Bestor, T. H. (1998). Sex-specific exons control DNA methyltransferase in mammalian germ cells. Development 125, 889-897.

Milkereit, P., and Tschochner, H. (1998). A specialized form of RNA polymerase I, essential for initiation and growth-dependent régulation of rRNA synthesis, is disrupted during transcription. Embo J 17, 3692-3703.

Miller, G., Panov, K. L, Friedrich, J. K., Trinkle-Mulcahy, L., Lamond, A. L, and Zomerdijk, J. C. (2001). hRRN3 is essential in the SLl-mediated recruitment of RNA Polymerase I to rRNA gène promoters. Embo J 20, 1373-1382.

Moorefield, B., Greene, E. A., and Reeder, R. H. (2000). RNA polymerase I transcription factor Rrn3 is functionally conserved between yeast and human. Proc Natl Acad Sci U S A P7, 4724-4729.

Moss, T. (2004). At the crossroads of growth control; making ribosomal RNA. Curr Opin Genêt Dev 14, 210-217.

Moss, T., Langlois, F., Gagnon-Kugler, T., and Stefanovsky, V. (2006a). A housekeeper with power of attorney: the rRNA gènes in ribosome biogenesis. Cell Mol Life Sci.

Moss, T., Stefanovsky, V., Langlois, F., and Gagnon-Kuglcr, T. (2006b). A new paradigm for the régulation of the mammalian ribosomal RNA gènes. Biochem Soc Trans 34, 1079-1081.

Moss, T., and Stefanovsky, V. Y. (2002). At the center of eukaryotic life. Cell 109, 545-548. Mougey, E. B., O'Reilly, M., Osheim, Y., Miller, O. L., Jr., Beyer, A., and Sollner-Webb, B. (1993). The terminal balls characteristic of eukaryotic rRNA transcription units in chromatin spreads are rRNA processing complexes. Gènes Dev 7, 1609-1619.

Moy, T. I., and Silver, P. A. (1999). Nuclear export of the small ribosomal subunit requires the ran-GTPase cycle and certain nucleoporins. Gènes Dev 13, 2118-2133.

Murphy, S. K., and Jirtle, R. L. (2003). Imprinting évolution and the price of silence. Bioessays 25, 577-588.

Muth, V., Nadaud, S., Grummt, L, and Voit, R. (2001). Acetylation of TAF(I)68, a subunit of TIF-IB/SL1, activâtes RNA polymerase I transcription, Embo J 20, 1353-1362.

Nan, X., Ng, H. H., Johnson, C. A., Laherty, C. D., Turner, B. M., Eisenman, R. N., and Bird, A. (1998). Transcriptional repression by the methyl-CpG-binding protein MeCP2 involves a histone deacetylase complex. Nature 393, 386-389.

O'Mahony, D. J., Smith, S. D., Xie, W., and Rothblum, L. I. (1992). Analysis of the phosphorylation, DNA-binding and dimerization properties of the RNA polymerase I transcription factors UBF1 and UBF2. Nucleic Acids Res 20, 1301-1308.

O'Sullivan, A. C., Sullivan, G. J., and McStay, B. (2002). UBF binding in vivo is not restricted to regulatory séquences within the vertebrate ribosomal DNA repeat. Mol Cell Biol 22, 657-668.

Oakes, M., Siddiqi, L, Vu, L., Aris, J., and Nomura, M. (1999). Transcription factor UAF, expansion and contraction of ribosomal DNA (rDNA) repeats, and RNA polymerase switch in transcription of yeast rDNA. Mol Cell Biol 19, 8559-8569.

Oka, M., Meacham, A. M., Hamazaki, T., Rodic, N., Chang, L. J., and Terada, N. (2005). De novo DNA methyltransferases Dnmt3a and Dnmt3b primarily médiate the cytotoxic effect of 5-aza-2'-deoxycytidine. Oncogene 24, 3091-3099.

Okano, M., Bell, D. W., Haber, D. A., and Li, E. (1999). DNA methyltransferases Dnmt3a and Dnmt3b are essential for de novo methylation and mammalian development. Cell 99, 247-257.

Panov, K. L, Friedrich, J. K., and Zomerdijk, J. C. (2001). A step subséquent to preinitiation complex assembly at the ribosomal RNA gène promoter is rate limiting for human RNA polymerase I-dependent transcription. Mol Cell Biol 21, 2641-2649.

Pelletier, G., Stefanovsky, V. Y., Faubladier, M., Hirschler-Laszkiewicz, I., Savard, J., Rothblum, L. I., Cote, J., and Moss, T. (2000). Compétitive recruitment of CBP and Rb-HDAC régulâtes UBF acctylation and ribosomal transcription. Mol Cell 6, 1059-1066.

Pikaard, C. S. (2000). The cpigenetics of nucleolar dominance. Trends Genêt 16, 495-500. Pradhan, S., Bacolla, A., Wells, R. D., and Roberts, R. J. (1999). 'Recombinant human DNA (cytosine-5) methyltransferase. I. Expression, purification, and comparison of de novo and maintenance methylation. J Biol Chem 274, 33002-33010.

Pradhan, S., and Esteve, P. O. (2003). Allosteric activator domain of maintenance human DNA (cytosine-5) methyltransferase and its rôle in methylation spreading. Biochemistry 42, 5321-5332.

Pradhan, S., Talbot, D., Sha, M., Benner, J., Hornstra, L., Li, E., Jaenisch, R., and Roberts, R. J. (1997). Baculovirus-mediated expression and characterization of the full-length murine DNA methyltransferase. Nucleic Acids Res 25, 4666-4673.

Preiss, T., and Hentze, W. H. (2003). Starting the protein synthesis machine: eukaryotic translation initiation. Bioessays 25, 1201-1211.

Prives, C, and Hall, P. A. (1999). The p53 pathway. J Pathol 187, 112-126. Puvion-Dutilleul, F., Puvion, E., and Bachellerie, J. P. (1997). Early stages of pre-rRNA formation within the nucleolar ultrastructure of mouse cells studied by in situ hybridization with a 5'ETS leader probe. Chromosoma 105, 496-505.

Raska, I. (2003). Oldies but goldies: searching for Christmas trees within the nucleolar architecture. Trends Cell Biol 13, 517-525.

Raska, L, Koberna, K., Malinsky, J., Fidlerova, H., and Masata, M. (2004). The nucleolus and transcription of ribosomal gènes. Biol Cell 96, 579-594.

Rhee, I., Bachman, K. E., Park, B. H., Jair, K. W., Yen, R. W., Schuebel, K. E., Cui, H., Feinberg, A. P., Lengauer, C, Kinzler, K. W., et al. (2002). DNMTl and DNMT3b cooperate to silence gènes in human cancer cells. Nature 416, 552-556.

Rhee, L, Jair, K. W., Yen, R. W., Lengauer, C , Herman, J. G., Kinzler, K. W., Vogelstein, B., Baylin, S. B., and Schuebel, K. E. (2000). CpG methylation is maintained in human cancer cells lacking DNMTl. Nature 404, 1003-1007.

Robert, M. F., Morin, S., Beaulieu, N., Gauthier, F., Chute, I. C, Barsalou, A., and MacLeod, A. R. (2003). DNMTl is required to maintain CpG methylation and aberrant gène silencing in human cancer cells. Nat Genêt 33, 61-65.

Robertson, K. D. (2001). DNA methylation, methyltransferases, and cancer. Oncogene 20, 3139-3155.

Robertson, K. D. (2002). DNA methylation and chromatin - unraveling the tangled web. Oncogene 27, 5361-5379.

Robertson, K. D., Uzvolgyi, E., Liang, G., Talmadge, C, Sumegi, J., Gonzales, F. A., and Jones, P. A. (1999). The human DNA methyltransferases (DNMTs) 1, 3a and 3b: coordinate mRNA expression in normal tissues and overexpression in tumors. Nucleic AcidsRes 27, 2291-2298.

Robertson, K. D., and Wolffe, A. P. (2000). DNA methylation in health and disease. Nat Rev Genêt/, 11-19.

Rose, K. M., Szopa, J., Han, F. S., Cheng, Y. C., Richter, A., and Scheer, U. (1988). Association of DNA topoisomerase I and RNA polymerase I: a possible rôle for topoisomerasc I in ribosomal gène transcription. Chromosoma 96, 411-416.

Rouquette, J., Choesmel, V., and Gleizes, P. E. (2005). Nuclear export and cytoplasmic processing of precursors to the 40S ribosomal subunits in mammalian cells. Embo J 24, 2862-2872.

Roussel, P., André, C., Cornai, L., and Hernandez-Verdun, D. (1996). The rDNA transcription machinery is assembled during mitosis in active NORs and absent in inactive NORs. J Cell Biol 133, 235-246.

Roussel, P., and Hernandez-Verdun, D. (1994). Identification of Ag-NOR proteins, markers of prolifération related to ribosomal gène activity. Exp Cell Res 214, 465-472.

Rubbi, C. P., and Milner, J. (2003). Disruption of the nucleolus médiates stabilization of p53 in response to DNA damage and other stresses. Embo J 22, 6068-6077.

Ruggero, D., and Pandolfi, P. P. (2003). Does the ribosome translate cancer? Nat Rev Cancer 3, 179-192.

Saez-Vasquez, J., and Pikaard, C. S. (1997). Extensive purification of a putative RNA polymerase I holoenzyme from plants that accurately initiâtes rRNA gène transcription in vitro. Proc Natl Acad Sci U S A 94, 11869-11874.

Santoro, R. (2005). The silence of the ribosomal RNA gènes. Cell Mol Life Sci 62, 2067-2079.

Santoro, R., and Grummt, I. (2001). Molecular mechanisms mediating methylation-dependent silencing of ribosomal gène transcription. Mol Cell 8, 719-725.

Santoro, R., Li, J., and Grummt, I. (2002). The nucleolar remodeling complex NoRC médiates heterochromatin formation and silencing of ribosomal gène transcription. Nat Genêt 32, 393-396.

Savino, T. M., Gebrane-Younes, J., De Mey, J., Sibarita, J. B., and Hernandez-Verdun, D. (2001). Nucleolar assembly of the rRNA processing machinery in living cells. J Cell Biol 153, 1097-1110.

Schnapp, A., and Grummt, I. (1991). Transcription complex formation at the mouse rDNA promoter involves the stepwise association of four transcription factors and RNA polymerase I. J Biol Chem 266, 24588-24595.

Schnapp, A., Schnapp, G., Erny, B., and Grummt, I. (1993). Function of the growth-regulated transcription initiation factor TIF-IA in initiation complex formation at the murine ribosomal gène promoter. Mol Cell Biol 13, 6723-6732.

Schneider, D. A., French, S. L., Osheim, Y. N., Bailey, A. O., Vu, L., Dodd, J., Yates, J. R., Beyer, A. L., and Nomura, M. (2006). RNA polymerase II elongation factors Spt4p and Spt5p play rôles in transcription elongation by RNA polymerase I and rRNA processing. Proc Natl Acad Sci U S A 103, 12707-12712.

Schneider-Stock, R., Diab-Assef, M., Rohrbeck, A., Foltzer-Jourdainne, C., Boltze, C, Hartig, R., Schonfeld, P., Roessner, A., and Gali-Muhtasib, H. (2005). 5-Aza-cytidine is a potent inhibitor of DNA methyltransferase 3a and induces apoptosis in HCT-116 colon cancer cells via Gadd45- and p53-dependent mechanisms. J Pharmacol Exp Ther 312, 525-536.

Seither, P., Iben, S., and Grummt, I. (1998). Mammalian RNA polymerase I exists as a holoenzyme with associated basai transcription factors. J Mol Biol 275, 43-53.

Seither, P., Iben, S., Thiry, M., and Grummt, I. (2001). PAF67, a novel protein that is associated with the initiation-compétent form of RNA polymerase I. Biol Chem 382, 1163-1170.

Seither, P., Zatsepina, O., Hoffmann, M., and Grummt, I. (1997). Constitutive and strong association of PAF53 with RNA polymerase I. Chromosoma 106, 216-225.

Shou, W., Sakamoto, K. M., Keener, J., Morimoto, K. W., Traverso, E. E., Azzam, R., Hoppe, G. J., Feldman, R. M., DeModena, J., Moazed, D., et al. (2001). Netl stimulâtes RNA polymerase I transcription and régulâtes nucleolar structure independently of controlling mitotic exit. Mol Cell 8, 45-55.

Shou, W., Seol, J. H., Shevchenko, A., Baskerville, C, Moazed, D., Chen, Z. W., Jang, J., Shevchenko, A., Charbonneau, H., and Deshaies, R. J. (1999). Exit from mitosis is triggered by Teml-dépendent release of the protein phosphatase Cdcl4 from nucleolar RENT complex. Cell 97, 233-244.

Sims, R. J., 3rd, Belotserkovskaya, R., and Reinberg, D. (2004). Elongation by RNA polymerase II: the short and long of it. Gènes Dev 18, 2437-2468.

Sinclair, D. A., and Guarente, L. (1997). Extrachromosomal rDNA circles--a cause of aging in yeast. Cell 91, 1033-1042. Sirri, V., Roussel, P., and Hernandez-Verdun, D. (1999). The mitotically phosphorylated form of the transcription termination factor TTF-1 is associated with the repressed rDNA transcription machinery. J Cell Sci 112 (Pi 19), 3259-3268.

Stage-Zimmermann, T., Schmidt, U., and Silver, P. A. (2000). Factors affecting nuclear export of the 60S ribosomal subunit in vivo. Mol Biol Cell 11, 3777-3789.

Stancheva, I., Lucchini, R., Koller, T., and Sogo, J. M. (1997). Chromatin structure and methylation of rat rRNA gènes studied by formaldehyde fixation and psoralen cross-linking. Nucleic Acids Res 25, 1727-1735.

Stefanovsky, V., Langlois, F., Gagnon-Kugler, T., Rothblum, L. L, and Moss, T. (2006a). Growth factor signaling régulâtes elongation of RNA polymerase I transcription in mammals via UBF phosphorylation and r-chromatin remodeling. Mol Cell 21, 629-639.

Stefanovsky, V., and Moss, T. (2006). Régulation of rRNA synthesis in human and mousc cells is not determined by changes in active gène count. Cell Cycle 5, 735-739.

Stefanovsky, V. Y., Bazett-Jones, D. P., Pelletier, G., and Moss, T. (1996). The DNA supercoiling architecture induced by the transcription factor xUBF requires three of its five HMG-boxes. Nucleic Acids Res 24, 3208-3215.

Stefanovsky, V. Y., Langlois, F., Bazett-Jones, D., Pelletier, G., and Moss, T. (2006b). ERK modulâtes DNA bending and enhancesome structure by phosphorylating HMG1-boxes 1 and 2 of the RNA polymerase I transcription factor UBF. Biochemistry 45, 3626-3634.

Stefanovsky, V. Y., Pelletier, G., Bazett-Jones, D. P., Crane-Robinson, C , and Moss, T. (2001a). DNA looping in the RNA polymerase I enhancesome is the resuit of non-cooperative in-phase bending by two UBF molécules. Nucleic Acids Res 29, 3241-3247.

Stefanovsky, V. Y., Pelletier, G., Hannan, R., Gagnon-Kugler, T., Rothblum, L. L, and Moss, T. (2001b). An immédiate response of ribosomal transcription to growth factor stimulation in mammals is mediated by ERK phosphorylation of UBF. Mol Cell 8, 1063-1073.

Straight, A. F., Shou, W., Dowd, G. J., Turck, C. W., Deshaies, R. J., Johnson, A. D., and Moazed, D. (1999). Netl, a Sir2-associated nucleolar protein required for rDNA silencing and nucleolar integrity. Cell 97, 245-256.

Strohner, R., Nemeth, A., Jansa, P., Hofmann-Rohrer, U., Santoro, R., Langst, G., and Grummt, I. (2001). NoRC—a novel member of mammalian ISWI-containing chromatin remodeling machines. Embo J 20, 4892-4900.

Suja, J. A., Gebrane-Younes, J., Geraud, G., and Hernandez-Verdun, D. (1997). Relative distribution of rDNA and proteins of the RNA polymcrase I transcription machinery at chromosomal NORs. Chromosoma 105, 459-469.

Sullivan, G. J., and McStay, B. (1998). Dimerization and HMG box domains 1-3 présent in Xenopus UBF are sufficient for its rôle in transcriptional enhancement. Nucleic Acids Res 26,3555-3561.

Tamaru, IL, and Selker, E. U. (2001). A histone H3 methyltransferase controls DNA methylation in Neurospora crassa. Nature 414, 277-283.

Tang, L. Y., Reddy, M. N., Rasheva, V., Lee, T. L., Lin, M. J., Hung, M. S., and Shen, C. K. (2003). The eukaryotic DNMT2 gènes encode a new class of cytosine-5 DNA methyltransferases. J Biol Chem 278, 33613-33616.

Thompson, M., Haeusler, R. A., Good, P. D., and Engelke, D. R. (2003). Nucleolar clustering of dispersed tRNA gènes. Science 302, 1399-1401.

Tschochner, H., and Hurt, E. (2003). Pre-ribosomes on the road from the nucleolus to the cytoplasm. Trends Cell Biol 13, 255-263.

Tuan, J. C., Zhai, W., and Cornai, L. (1999). Recruitment of TATA-binding protein-TAFI complex SL1 to the human ribosomal DNA promoter is mediated by the carboxy-terminal activation domain of upstream binding factor (UBF) and is regulated by UBF phosphorylation. Mol Cell Biol 19, 2872-2879.

Turek-Plewa, J., and Jagodzinski, P. P. (2005). The rôle of mammalian DNA methyltransferases in the régulation of gène expression. Cell Mol Biol Lett 10, 631-647.

Valdez, B. C, Henning, D., Le, T. V., and Busch, H. (1995). Spécifie aspartic acid-rich séquences are responsible for silver staining of nucleolar proteins. Biochem Biophys Res Commun 207, 485-491.

Visintin, R., Craig, K., Hwang, E. S., Prinz, S., Tyers, M., and Amon, A. (1998). The phosphatase Cdcl4 triggers mitotic exit by reversai of Cdk-dependent phosphorylation. Mol Cell 2, 709-718.

Voit, R., Hoffmann, M., and Grummt, I. (1999). Phosphorylation by Gl-spécifie cdk-cyclin complexes activâtes the nucleolar transcription factor UBF. Embo J 18, 1891-1899.

Voit, R., Schnapp, A., Kuhn, A., Rosenbauer, H., Hirschmann, P., Stunnenberg, H. G., and Grummt, I. (1992). The nucleolar transcription factor mUBF is phosphorylated by casein kinase II in the C-terminal hyperacidic tail which is essential for transactivation. Embo J 77,2211-2218.

Vu, L., Siddiqi, I., Lee, B. S., Josaitis, C. A., and Nomura, M. (1999). RNA polymerase switch in transcription of yeast rDNA: rôle of transcription factor UAF (upstream activation factor) in silencing rDNA transcription by RNA polymerase II. Proc Natl Acad Sci U S A 96, 4390-4395.

Walsh, C. P., Chaillet, J. R., and Bestor, T. H. (1998). Transcription of IAP endogenous retroviruses is constrained by cytosine methylation. Nat Genêt 20, 116-117.

Warner, J. R. (1999). The économies of ribosome biosynthesis in yeast. Trends Biochem Sci 24, 437-440.

Weber, J. D., Taylor, L. J., Roussel, M. F., Sherr, C. J., and Bar-Sagi, D. (1999). Nucleolar Arf sequesters Mdm2 and activâtes p53. Nat Cell Biol /, 20-26.

Weisenberger, D. J., Velicescu, M., Preciado-Lopez, M. A., Gonzales, F. A., Tsai, Y. C., Liang, G., and Jones, P. A. (2002). Identification and characterization of alternatively spliced variants of DNA methyltransferase 3a in mammalian cells. Gène 298, 91-99.

White, R. J. (2005). RNA polymerases I and III, growth control and cancer. Nat Rev Mol Cell Biol 6, 69-78.

Workman, J. L. (2006). Nucleosome displacement in transcription. Gènes Dev 20, 2009-2017.

Yoder, J. A., Walsh, C. P., and Bestor, T. H. (1997). Cytosine methylation and the ecology of intragenomic parasites. Trends Genêt 13, 335-340.

Yuan, X., Zhao, J., Zentgraf, H., Hoffmann-Rohrer, U., and Grummt, I. (2002). Multiple interactions between RNA polymerase I, TIF-IA and TAF(I) subunits regulate preinitiation complex assembly at the ribosomal gène promoter. EMBO Rep 3, 1082-1087.

Yuan, X., Zhou, Y., Casanova, E., Chai, M., Kiss, E., Grone, H. J., Schutz, G., and Grummt, I. (2005). Genetic inactivation of the transcription factor TIF-IA leads to nucleolar disruption, cell cycle arrest, and p53-mediated apoptosis. Mol Cell 19, 77-87.

Zachariae, W., and Nasmyth, K. (1999). Whose end is destruction: cell division and the anaphase-promoting complex. Gènes Dev 13, 2039-2058.

Zhai, W., and Cornai, L. (2000). Repression of RNA polymerase I transcription by the tumor suppressor p53. Mol Cell Biol 20, 5930-5938.

Zhao, J., Yuan, X., Frodin, M., and Grummt, I. (2003). ERK-dependent phosphorylation of the transcription initiation factor TIF-IA is required for RNA polymerase I transcription and cell growth. Mol Cell 11, 405-413.

Zhou, Y., Santoro, R., and Grummt, I. (2002). The chromatin remodeling complex NoRC targets HDACl to the ribosomal gène promoter and represses RNA polymerase I transcription. Embo J 21, 4632-4640.

Annexe 1 : L'importance des gènes ribosomaux dans la biogenèse des ribosomes;

« A housekeeper with power of attorney, the rRNA gènes in ribosome biogenesis»

Tom Moss, Frédéric Langlois, Thérèse Gagnon-Kugler and Victor Stefanovsky

Cancer Research Centre and Department of Médical Biology, Laval University, CHUQ-

HDQ, Pavillon St Patrick, 9 rue McMahon, Québec, G1R 3S3, Canada.

Cet article a été publié en janvier 2007 dans «Cellular and Molecuiar Life Science», vol 64

(1): 29-49 et il est reproduit avec la permission de l'éditeur Birkhâuser Verlag.

Cell. Mol. Life Sci. 64 (2007) 2SM9 1420-682X/07/010029-21 DOI 10.1007/sOOO 18-006-6278-1 © BirkhâuserVerlag, Basel, 2007

I Cellular and Molecular Life Sciences

Review

A housekeeper with power of attorney: the rRNA gènes in ribosome biogenesis T. Moss*, F. Langlois,T. Gagnon-Kugler and V. Stefanovsky

Cancer Research Centre and Department of Médical Biology, Laval University, CHUQ-HDQ, Pavillon St Patrick, 9 nie McMahon, Québec, G1R 3S3 (Canada), Fax: +1 418 6915439, e-mail: [email protected]

Received 12 June 2006; received after revision 16 August 2006; accepted 5 October 2006 Online First 14 December 2006

Abstract. Ribosome biogenesis centres both physically and functionally on the activity of the ribosomal RNA (rRNA) gènes. Ribosome assembly occurs co-transcrip-tionally on thèse gènes, requires the coordinated expres­sion and assembly of many hundreds of proteins and is finely tuned to cell and organism growth. This review présents contemporary understanding of the mode and

the means of rRNA gène transcription and how growth factors, oncogenes and tumour suppressors regulate this transcription. It is argued that transcription elongation is a key mechanism regulating rRNA gène transcription. This unorthodox view provides a logical framework to explain the co-transcriptional phase of ribosome biogenesis.

Keywords. Ribosome biogenesis, growth régulation, gène régulation, RNA polymerase I, RPI, transcription, elonga­tion, transcription-coupled ribosome assembly.

A senior professor once gave some stern advice: 'Keep well clear of muscle and ribosomes,' he said, 'they've been done to death.' Though this advice was followed assidu-ously during his postgraduate years, T. M. failed miserably as; a postdoctoral fellow, becoming fascinated by the prob-lem of how a few hundred ribosomal RNA (rRNA) gènes are able to produce 80% of the total cellular RNA. Since that time, he has become ever more convinced that under­standing the rôle thèse gènes play in regulating cell growth is; one of the most important challenges facing modem bi­ology. Yet it is clearly also one of the most neglected chal­lenges. Hère, we attempt to review the knowledge of how, and indeed why, the rRNA gènes are transcribed and sum-marise what is known of the mechanisms used to coordi-nate their output with cell growth.

" Corresponding author.

Why are the rRNA gènes important?

The rRNA gènes encode the major RNA components of the ribosome, the most ancient and most complex of ail molecular machines. In eukaryotes, the synthetic activity of thèse gènes générâtes the largest sub-nuclear structure, the nucleolus, and it is hère that ribosomes are assembled. Given the overwhelming emphasis paid to protein-coding gènes, it is a sobering thought that each organism must provide around 10 ribosomes for every messenger RNA it synthesizes. Ribosome bio­genesis - the process of ribosome synthesis - therefore occupies a very large fraction of the metabolic effort of a cell. But does it also control cell growth, prolif­ération and perhaps differentiation, or is it just another 'housekeeping' function? Several récent studies give little alternative but to consider ribosome biogenesis as a def ining élément in the control of cellular and or­ganism growth. Hère we will review current understand­ing of the growth-related aspects of rRNA gène regul-

30 T. Moss et al. Ribosomal RNA Gènes

ation, emphasising their relevance to mammalian Sys­tems.

Assembling ribosomes

The ribosome, the factory of protein synthesis, is probably descended from a primitive catalytic RNA. The existence in modem organisms of ribozymes and the démonstration that peptide bond formation is catalysed predominantly by the rRNAs strongly suggest that when proteins finally arrived in the primaeval RNA world they were synfhe-sized on an 'all-RNA' ribosome [see e.g. refs. 1, 2]. This central rôle in the development of life forms appears to hâve been carried over into modem organisms. In récent years we hâve corne to understand that the ability to syn-thesise ribosomes détermines growth and cell division rates and we now know that many oncoproteins and tu-mour suppressor proteins modulate ribosome biogenesis [3,4]. More surprisingly, ribosome biogenesis appears in turn to be a regulator of several tumour suppressors, in-cludingp53 [5,6]. The mammalian ribosome is a 4-MDa structure made up of two-thirds RNA and one-third protein and assembled into two distinct units referred to as the large, or 60S, and the small, or 40S, subunits. The large 60S subunit contains three RNA species, the 28S, 5.8S and 5S rRNAs, and -49 ribosomal proteins (r-proteins), while the small 40S subunit contains a single RNA, the 18S rRNA, and -33 r-proteins. However, several hundred other proteins hâve been implicated in the process of ribosome biogene­sis [7-10].The 18S, 5.8S and28S rRNAs are synthesised, processed and assembled into ribosomes in the larg-est sub-nuclear structure, the nucleolus (Fig. 1). Thèse three rRNAs are transcribed by a dedicated polymerase, RNA polymerase I (RPI, also called Poli), from a set of repeated gènes, the rRNA gènes or rDNA, as part of a single precursor, which in mammals is referred to as the 45 or 47S pre-rRNA (Fig. 2A). The short 5S rRNA is in-dependently synthesised by RNA polymerase III (RPIII, also called PolIII), and since its régulation is beyond the scope of this review, the reader is referred to other review articles [11]. Initial assembly of the ribosome occurs co-transcriptionally with 47S pre-rRNA synthesis leading to a 90S precursor particle, a process elegantly visualized in the 'Miller spread' électron micrographs (Fig. 1 A, B) [re-viewed in réf. 12]. Structural studies of the ribosome sug­gest that this co-transcriptional assembly process is im­portant in establishing the complex folding of the mature rRNAs and in positioning the r-proteins [13]. Hence, this co-transcriptional phase of assembly is probably a key factor in the fidelity of ribosome biogenesis [14]. Soon after its synthesis, the pre-rRNA is cleaved in a number of distinct steps, first to yield 40S and 60S subunit precur­sor complexes and finally the mature ribosomal subunits

Figure 1. The cytological and low-resolution macromolecuiar structures of the nucleolus and active ribosomal gènes. FC, fibrillar centre; DFC, dense fibrillar centre; GC, granular centre. (A) Elec­tron micrograph ofa thin section through the nucleus and nucleolus ofa bovine endothelial cell. (B) A 'Miller spread' from a mouse Ltk~ cell in culture showing closely packed transcribing polymerases. (C) Electron micrograph ofa thin section through the nucleus and nucleolus of a Saccharomyces cerevisiae cell. (B) 'Miller spread' from an S. cerevisiae cell. A and C kindly provided by Dr N. Gas; B by Prof. U Scheer; D by Dr Y. Osheim and Prof. A. Beyer.

Cell. Mol. Life Sci. Vol. 64, 2007 Review Article 31

[reviewed in refs. 15-17]. Not only is ribosomc biogen­esis thc most complcx undcrtaking of prolifcrating cclls, it is also a major metabolic task. Ribosomcs account for around 80% of total ccllular RNA. In ycast, ribosomc bio­genesis accounts for >75% of ail nuclear transcription, -60% cngaged in thc production of thc rRNAs them-selvcs and -15% in transcribing the 78 yeast ribosomal protcin gencs [18]. In prolifcrating mammalian cclls, around 35% of nuclcar transcription is dcdicated to the production of thc rRNAs, while a significant proportion of total mRNA gcne transcription is rcquircd to produce the proteins needed for ribosomc assembly. Intcrestingly, thc rate of genome-wide transcription has been shown to bc coordinated with ribosome biogenesis. Rcgardlcss of growth rate, yeast maintains a constant ratio of about 10 ribosomcs per mRNA, though how this is accomplished remains a mystery [19].

Synthesis of the rRNA prccursor is thc central focus of ribosomc biogenesis. The nucleolus forms in the nuclcus wherever thc rRNA gencs are transcribed [20]. Thus, the existence of the ccllular ribosomc factory is thc consé­quence of rRNA gcne activity. Consistent with this, thc nucleolus shows a distinct sub-structurc, thc so-called fibrillar centres and associated dense fibrillar centres, which are the centres of pre-rRNA synthesis and co-tran-scriptional assembly, and the outer granular centres, thc arca wherc the large and small ribosomal subunits are indcpendently matured (Fig. 1) [21-25]. The 5S rRNA is transcribed indcpendently of the 18S, 5.8S and 28S rRNAs and is importcd into thc nucleolus, as are thc r-proteins. The only known exception among eukaryotes is Saccharomyces cerevisiae, in which thc 5S gènes are linked to thc rRNA genes and hence must neccssarily bc transcribed in the nucleolus (Fig. 2A). In mammals, however, transcription of thc unlinkcd 5S and even tRNA

A) Mammals: 1289- ■ I G S -

-5kb—

ETS

Xenopus: r*-

47S

J85 | I5%l4

JH-5.8S

40S

ifcid 333

S. cerevisiae: .*. ^ 37S

^HEUIj B) X.laevis:

rat/mouse:

T2 Spacer Promoters (SpPr) y«

-illMllllllllïlIllllllir I U U 3 C . y

Enhancers Enhancers

Spjr,-». Enhancers

D. melanogaster: Directional Enhancers T

•lllllllllllllllllllïl T1 T2 T3AT3B TP - ,

■ i I 11 lf£ S. cerevisae:

"Enhancer"

Figure 2. (A) Organisation of thc rRNA genes in mammals, amphibia and ycast. (li) Organisation of the inter-genic spacer (IGS) in rat/ moust: [221, 222] [36, 37, 60, 61 ], Xenopus laevis, Drosophila melanogaster and S. eerevisiae [reviewed in rcfs. 60, 61 ; reviews also avail-able on request froni the author], Tcrmination sites in S. cerevisae are taken from Van der Sandc et al. [62], but some questions remain as to the functions of thèse sites in vivo [sec c.g. refs. 223, 224].

32 T. Moss et al. Ribosomal RNA Gencs

gènes occurs at the nucleolar periphery [26]. During the co-transcriptional phase of ribosome assembly, the rRNA is subjected to extensive, séquence spécifie modifica­tion. In vertebrates, around 115 residues of the rRNAs are 2'-0-methylated and about 95 uridines are converted to pseudo-uridine [27]. Thèse modifications, which also occur on tRNAs and the 5 S rRNAs, are dépendent on several hundred complementary small nucleolar RNAs (snoRNAs) [27-30]. As can be seen, assembly and matu­ration of the ribosome subunits is a complex process and will not be discussed further hère; the reader is referred to several specialised reviews [16, 17, 31]. For the purposes of the présent review, it suff ices to say that the transcrip-tional activity of the rRNA gènes concentrâtes ail thèse functions in the nucleolus.

The mode of rRNA gène transcription.

In trying to understand the mechanisms that underlie rRNA gène transcription and its régulation, we hâve generally made the assumption that, irrespective of the eukaryotic System studied, ribosome biogenesis and its régulation will be fundamentally the same. Given the pri-maeval origins of the ribosome and its conservation both structurally and functionally, this seems the rnost reason-able starting point, at least until solid évidence exists to the contrary. Thus, it is our contention that one should seek out the parallels between higher and lower eukary-otes rather than emphasise their apparent différences. As will be seen below, it could be argued from présent in­complète knowledge that the promotion of rRNA gène transcription in yeast is quite différent from that in mam-mals. But, I would suggest that this apparent différence is more than likely due to our présent incomplète state of knowledge of the mammalian System rather than to a fundamental différence in molecular mechanism. When the existence of active promoters within the intergenic spacer of the rRNA gènes was first demonstrated [32], this was seen as a peculiarity of the amphibian Xenopus. However, within a few months, Drosophila species were found to hâve such promoters [33-35], and a few years later mammals also underwent a rapid évolution [36-38]. Thus, until our knowledge of différent Systems is suffi-cient to demonstrate a clear lack of mechanistic conserva­tion, it may be more profitable to seek out the commonal-ities rather than to stress the apparent différences. In this vein, let us first consider the fundamental mechanisms of rRNA gène transcription. The 200 to 300 ribosomal gènes per haploid mammalian génome exist as direct repeats at the secondary constric-tions of acrocentric chromosomes, five in humans [39] and probably five in mouse [40]. Each of thèse sites has the potential to form a nucleolus and is hence referred to as a nucleolar organiser or nor, a term that predates

knowledge of the gènes thèse sites encompass [41, 42]. Each rRNA gène repeat is made up of an intergenic spacer (IGS), originally referred to as the non-transcribed spacer (NTS), but now known to be partly transcribed into non-coding, non-structural RNA. Early studies revealed that the DNA séquences surrounding the site coding the pre-rRNA 5' terminus, and now known to promote its tran­scription, were also found repeated one or more times within the IGS (Fig. 2B) [32, 43-45]. Thèse 'spacer pro­moters' were shown to direct transcription of the IGS and to be required for efficient pre-rRNA synthesis, though they could not direct this synthesis themselves [32, 46]. The ability of the spacer promoters to enhance pre-rRNA transcription was shown to requirc the adjacent short re­peated séquences referred to as enhancers [32, 47, 48]. However, we showed that thèse séquences do not func-tion to increase overall transcription but rather to greatly increase the likelihood of transcription from the linked gène [32,48], and in this sensé they are functionally simi-lar to the so-called yeast enhancer [49, 50]. Later work has demonstrated the veracity of thèse findings [51-53]. Several explanations for IGS transcription hâve been sug­gested. The spacer promoters could represent a means of trapping polymerase and supplying it to the 45 S pre-rRNA promoter [32, 54]. Alternatively, spacer transcrip­tion could represent a mechanism for activating or main-taining the active state of a gène, perhaps by maintaining chromatin of the IGS in some as yet undefined 'open' state or, as recently shown in yeast, for regulating recom-bination of the rRNA gènes [55]. Most recently, it has been suggested that spacer transcription is required for rDNA silencing [56], a subject treated in depth elsewhere [4, 57]. Such a silencing function for the IGS transcripts is presently difficult to reconcile with the copious data showing an in cis positive selector function for the spacer promoters. However, it is consistent with the rôle of mi-cro-RNAs in centromere silencing [58, 59] and almost certainly provides an important link in the complex chain of events leading to rRNA gène silencing. In most Systems, though perhaps not ail, IGS transcrip­tion is terminated immediately preceding the true pre-rRNA promoter (Fig. 2B) [32,60,61 ]. A similar termina-tion site also occurs in yeast, despite the apparent absence of any IGS transcription directed towards it [62]. Thèse promoter-proximal terminators hâve been ascribed the functions of (i) preventing promoter occlusion, that is the inactivation of the promoter by displacement of essential factors [63] or (ii) recycling polymerase molécules deliv-ered there by the spacer promoters or by 'read-through' from the upstream gène [64, 65]. More recently, the pro­moter proximal terminator in mammals has been sug­gested to médiate gène silencing [57], though again this is difficult to reconcile with the transcription enhancing functions of the terminator.

Cell. Mol. Life Sci. Vol. 64, 2007 Review Article 33

The nu ans of rRNA gène transcription

A specialised set of proteins has evolved uniquely to transcribe thc rRNA gènes. With rare exceptions [66], thc a-amanitin-resistant DNA-dcpendent RNA RPI is dedicated solcly to thc transcription of thc rRNA gènes, synthesising both the non-coding spacer transcripts and thc productive prc-rRNA transcripts. In yeast, RPI is an enzyme of 14 subunits, half of which arc shared with onc or both of thc other two cukaryotic RNA polymerases, RPI1 and RPIII (Table I). In vertebrates and yeasts, the RPI promotcr is a séquence of 140-160 bp encompass-ing at least thc first four nucleotides downstream of thc mapped pre-rRNA initiation site (Fig. 3). The very poor séquence conservation of RPI promoters is consistent with thc extrême species specificity of thc RPI transcrip­tion system and thc rapid évolution of the rRNA genc

spacer. Despite this, functional studics have shown lhat most RPI promoters contain two distinct séquence clé­ments, a corc promotcr (Corc) séquence and an upstrcam promoter élément (UPE or UE), originally callcd thc up­strcam control séquence (UCE) in humans. The spacing of thèse promotcr cléments is crucial to in vivo function, but often the Core promoter clément is sufficient to spec-ify correct transcription initiation in vitro (Fig. 3). Short promoter séquences, similar in length to thesc Core pro­moters appear to be the norm in plants and some singlc-ccll organisms (Fig. 3) [67-69]. Thcrc is gênerai agreement that thc formation of an RPI pre-initiation complex requircs the TATA box-binding protein (TBP) and a group of RPI-specific TBP-associ-ated factors (TAF,s) that form onc or two complexes ablc to recognize the promoter (Table 2). But herc thc similari­t é s appear to end. In human and mouse, formation of thc

- 1 5 0 - 1 0 0

Ho CCCGGGCC ïcTCTGG CTQCGATQGT «JCOTT^ffl ET CCPTGTCGCG CGTCGCCTGG GCCGGCG . GC GTGGTCGGTG ACGCGACCTC ^ ■ ■ M G I - a f l M ^ •^■■fcs i - aoxWÊÊ^^-

Mm GTTGTTCCTT TGAGGTCCGG TTCTTTTCOT TATGQGGTCA TTTTTOOOCC ACÇTCCAO^. GTAT^^TTCCAGGTATICT CTGXGGCCTG TCACTTTCCT

Rn TGTTCCTTTG CGGTCCGGTT CTCTTTCTACATW3GQACCT CtTOoSSSÂc AÇQTCACCOA ACATGACTTC ÇAGACGTTCC GTGTGGCCTG TÇATGTTTAT

XI CCGGGGCCCT CCCGCOOAQQ CCCCGATOAQ QACGGATTCG CCCGGCCCGC CCCGGCCGGA GTTCCGGGAG CCCGGGGAGA GGAGCCGGCG GCCCGGCCTC Dm AGGGGAAA AAATAATCAT ATAATATATA AGAGAATAGC CGCTATGTGG Dv CCGGCA TAAGTCAATT ATGTTTATAA AAGGAGAATA ATGAAGTTAT AAAAGTGTAT TATTAAATTA GTACATGAAG ACATTAAGGT

Ao TCATGAAAAA AAACACGTTT GCTTGGGGGC TTGCTCTAGG GACTTTGCTG CTGCAAGGTG TCTCGGCCGG GCCGGTGGCC GGAAAAATCC CGG™ Tt AAAACTGAAA AATTTACAAG GGATTGAAAA TTTTGGCAGA GTCTTTTTTT TGGCAAAAAA AAAAACAAAA ATAGTAAACC TTCCGAACTT TTTTGACTTT At ATTATGTAAA TTTACCAGAA AATAGGATTT AGTATCCTTA TGATGCATGC TAAAAAGAAT TTTCAAATTC CAAGTATTTC TTTTTTTTTG GCACCGGTGT Se AGAATAOCTT AAATTGAAGT TTTTCTCGOC GAGAAATACG TAGTTAAOOC AOAGCGACAG AGAOGGCAAA KGAAAATAAA AGTAAQATTT TAGTTTGTAA Sp GAA GAAGTAGTTT TGTGGTGGTA GTAGTAGGAC TTTTGGTGGA GAGAGGAGGG ATATGGAAGA AAGGATAGCA AATAGGCACC OAAATGOACG

bO I

Hs CCGG.CCCCG GGGAG8TATA TCTTTCSCTC

Mm 3CCTGTCTCT TTTATGCTTG TOATCTTTTC

-1 +1 I

'ATT GCTGACACG I

GCGACCTGTC GCTGGAGAGG

T GGAGATAOQG ACTOACACO CTGTCCTTTC CCTATTAACA CTAAAGGACA

A.:

2CTGTGTÇT TTTACACTTT TCATCTTTGC TATCTOTCCT TATÏOTACCT OOAQATATAT OCTOACACG CTGTCCTTTT GACTCTTTTT GTCATTAAAG

XI TCGGGCCCCC CGCACGACGC C GÇATOJOCOO OCACQ^flOT AQQ^AAiiA CCGGCC :G oj^ArGOG CGCCCGAAAA

Dm GGTGGTAAAT GGAATTOAAA ATACCCGCTT TQAGGACAGC Dv GAATGGTAGC ATTTGAAAAA AATATCGCCA TTATAGATGA

CAAACCGGC^GCXBCACTTT TCTOOCACCT Tt GAGAAAAATT CTTTGGCAAA AAAAATAAAA ATAATATCAG At CTCCTCAQAC ATTTCAATOT CTGTTGGTGC CAAGAGGGAA Se TGGGAGGGGG OOTTTAOTCA TOOAOTACAA GTGTGAGGAA Sp AAATTCACAC AACCACAAAT GTTTCTATAA AATCGAGOAA

OOOTTCAAAA A. .CTACTAT TGTGTCAAAA AACCTATTC.

:CGTCC<3JTAAGTÂTATÂT GGGGGTAAAA ATGCATATTT AAGGG.CTAT TAAOCTATAT AAOTAOTTOE GAGGTACTTC AAATAOOTCC AAGQAACTAT

A.OOTAGGC AGTGGTTGCC GACCTCGCAT TGTTCGAAAT ATGGTGAGC AGTGTGTGCT CATCACATTA CGCTGAAAGC

AAAOOOACG AAGAAGGGG AOOOaOGTG ATOCOAAAG GTAAAAGGA

GGTCCGGCCG AAACATCTCC GGTGTTGAGG CAGTTGAAGA GTGGTCGTTG

GAAAAAAAAA GGATCAAAAA GAGTCTGGGC CAAGTTCGAA AAAGGAGAAA

ATACGCCCAT TAAAATATCA AGTCCGTGGG AAGAGTTTGG AOAAAAAAGA

TBP; X-link TBPi-complex; Footprint TBPi-complex; X-link

^ B | ^ HMGl -boxas of UBF; Footprint

Figure 3. Structure of RPI promoters. Promoter séquences are shown aligned to the mapped initiation site (+1) with some small adjust-mcnls of alignmcnt to emphasisc the limited homologies. Mapped functional séquence cléments arc shown in bluc and the régions dem-onstiated by footprinting or crosslinking to be physically contacted by the TBP, complex (SL-I, TIF-1B), by UBF or by TBP arc indicatcd graphically. [The data for human (Hs) were taken from refs. 74, 80, 225-228, for mouse (Mm) from refs. 229, 230, for rat (Rn) from refs. 231-233, for Xenopus laevix (XI) from refs. 81, 234-238, for Drosophiia melanogaster (Dm) from réf. 239; thc Drosophila virilis (Dv) promotcr séquence was dedueed from comparisons of répétitive IGS séquences in réf. 35; the data for Acanthamoeba castallanii (Ac) was taken from refs. 118, 119, 240, for Arabidopsis thaliana (At) from réf. 241, for Saccharomyces cerevisiae (Se) from refs. 101, 242 and for Schizosaccharomycespombe from réf. 69.J

34 T. Moss et al. Ribosomal RNA Ocncs

pre-initiation complex requircs SL1 [for selcctivity factor I, rcfcrrcd to in mousc as transcription initiation factor IB (TIF-IB)], which has bcen shown to contain TBP and three TBP-associated factors, TAF,48, 63 and 94 (Ta­ble 2). A second non-specific DNA-binding protcin, up­stream binding factor (UBF), has bcen shown to enhanec RP1 transcription and will bc discusscd in greater détail

Table 1. List of RP1 subunits and associated factors in ycast and mammals.

Yeast RPI Yeast genc Yeast null Human subunits and mutant and associated mousc proteins

A190 RPAI90 lethal RPA194 AI35 RPAI35 lethal RPA135 A49 RPA49 conditional PAF53/PAF51? N/l PAF49/Ase-1/CAST A43 RPA43 lethal RPA43=TWISTNB AC40 RPC40 lethal RPA40 A34.5 RPA34 viable not identified ABC27 RPB5 lethal RPB25 ABC23 RPB6 lethal RPBI4.4/RPB6 AC19 RPC19 lethal RPA16 ABCI4.5 RPB8 lethal RPB17 AI4 RPAI4 viable not identified A12.2 RPA12 conditional RPA12 ABCIOa RPC10 lethal RPBlOa ABClO/j RPB10 lethal RPBIO/i Rrn3p RRN3 lethal Rrn3(TIF-IA)

A, B andC referto the three RNA polymerase forms RPI, RPI I, and RPIII, e.g. in ycast 'ABC indicates a common subunit. In mam­mals, the shared subunits were named after the polymerase with which they were first associated. The data arc mainly taken from Caries and Riva [219] combincd with searches of the BioBasc Pro-teome Library and the récent identification of PAF49 [220].

Table 2. List of ycast corc factor (CF) and upstream activating fac­tor (UAF) and mammalian SLI (TIF-IB) subunits and theirpossible correspondences.

Yeast CF Yeast UAF Yeast null Mammalian SLI mutant (TIF-IB)

Rrn6p >? lethal TA F, 110/95 (102 kDa) (95/92 kDa) Rrn7p lethal TAF,63/68 (60 kDa) (68 kDa) Rml lp lethal TAF,48 (59 kDa) (53 kDa) TBP = > < = TBP lethal TBP

Rrn5p viable Rrn9p viable RrnlOp viable UAF30 H3 114

viable

below. On the other hand, in the ycast S. cerevisiae, two factors are required for pre-initiation complex formation, the upstream activating factor (UAF) and the core factor (CF). The isolation of factors from Schizosaccharomyces pombe has suggested that parallcls cxist bctween the CF components of the yeast and thosc of mammals (Table 2) [69]. Howcvcr, thèse parallcls still rcmain tentative and no mammalian équivalents to the ycast UAF are known.

Initiation complex formation

The mechanisms of initiation of RPI transcription have bcen studied in yeast, Acanthamoeba castellani, human, rat and mousc. In each case, reconstruction of the pre-initiation and initiation complexes has been studied in solution using ccll-free extracts and purified factors. A summary of the known interactions of the various factors with their cognatc promoters is given in Figure 3.

1. Mammals The earliest data came from studios of the human and mousc factors and established a paradigm that is only

A) M * .

S U '

In the cases of the TAF,s, the original protcin names arc indicated for human and mouse, rcspcctivcly, followcd by the calculated mo-lecular wcight(s) roundcd to the nearest kDa. Sec text for the origins of the data.

Figure 4. Assembly of the pre-initiation complex and the initiation cycle in mammals. DNA is shown in red and promoter éléments in ycllow and chequered ycllow. The possibility of further SLI sub­units (?) and the possible implication of TFIIH arc indicated. Prob­able corrélations between SLI subunits and ycast CF subunits is indicated by colouring (compare with Fig. 6).

Cell. Mol. Life Sci. Vol. 64, 2007 Review Article 35

A)

mUBF HMG1 -boxes 2. 3

&SSS-

v Core UBF

B)

Enhancesome Superhelical model for RPI promoter

Figure 5. (A) The domain structure ofmammalian UBF. Eaeh HMG box is indicatcd by the homologous fold of HMG-D [243] and in-ter-box peptidcs as arrows of approximately correct length. The acidic residuos of the C-terminal tail arc indicatcd by a wavy line and the approximate positions of the blocks of serine residuos arc in yellow. (li) Low-resolution structure of the enhancesome. Lcft, the low-resolution structure of a single enhancesome as determined by électron spectroscopic imaging [83, 84, 186]; right, possible folding of the RPI promoter by two adjacent enhancesomes induced by UBF binding. Promoter séquences are indicatcd as in Figure 4. Only the Core UBF région is shown in li and inter-HMGl box linkers are shown generically. UBF is in bluc and DNA in red, and * indicates sites of phosphorylation by ERK.

now being questioned. Esscntially, this paradigm is dc-scribed in the cartoons depicting the steps of human ini­tiation complex asscmbly in Figure 4 [sec refs. 4, 70-73 for reviews]. The non-specific DNA-binding HMG 1-box protein UBF was shown to bind the upstream and/or core promoter régions, creating a situation propitious for the SL1 complex to bind and form a 'stable' pre-initiation complex [74]. This complex is ablc to recruit RPI and, in the présence of nuclcotidc triphosphates, initiate tran­scription. How UBF is able to aid in RPI initiation rcmains somewhat of a mystery. If this is truly its function, several observations suggest how it might occur. The C-terminal domain of UBF is made up almost exclusively of blocks of aspartic and glutamic acid residues, cach terminating in blocks of serine residues (Fig. 5A). This domain can bind and recruit SL1, binding being enhaneed by phosphory­

lation of the serine motifs, probably by casein kinase II [75-78], However, carly studies also suggested that UBF could more weakly enhance initiation complex asscmbly even in the absence of this C-terminal domain [79, 80]. When it was discovered that the N-tcrminal half of UBF (core UBF) could form the 'enhancesome', a nuclcopro-tein structure that somewhat resemblcs the nucleosome (Fig. 5B), the finding immediatcly led to a spéculative cxplanation for the cooperativity between UBF and SL1 and the bimodal organisation of RPI promoters [81-84]. It was suggested that by binding two régions of the pro­moter, UBF would juxtapose key promoter séquences and thus prcsçnt the SL1-binding sites on the same surface of a DNA superhelix (Fig. 5B). Thèse interprétations werc based on the original observations in the human ccll-free system showing that UBF was required for SL1 recruit-

36 T. Moss et al.

ment and RP1 promotion. Howevcr, it is clear that UBF is not esscntial for promotion in the mousc and rat ccll-frcc Systems and, indccd, it is oftcn difficult to find conditions undcr which it has any positive effect at ail on RPI tran­scription initiation. In vitro, UBF has been found vari-ously to activate RPI transcription at pre-initiation [79, 80] or promoter releasc [73, 85], to relieve H1 repression [86] or simply to be unnecessary [87]. Thèse conflicting observations may bc thc rcsult of the polariscd basic-acidic nature of UBF, allowing it to compete non-specifi-cally for inhibitory DNA-binding activities in in vitro as-says. But they may also be relatcd to thc low DNA-bind­ing constant of UBF combincd with its ability to interact non-spccifically with almost ail DNA séquences [81, 83, 88]. Most recently, it has been argued that human SL1 is ablc to functionally bind thc promoter in thc absence of UBF, something alrcady known for rodent SL1, and that UBF binding is dépendent on SL1 rather than the converse [89]. Howevcr, given the rapid off-ratc of UBF [90, 91], it may be difficult to détermine the truc order of binding of thèse factors to thc DNA. SL1 binds to both thc promoter DNA and to UBF. SL1 would then naturally reducc the off-ratc of UBF by its coopérative interaction with both components of thc UBF-DNA complex. Hcncc, wc do not believe that the data to date provide définitive information on thc order of promoter association of UBF and SU, and may, rather, reflect thc changes in DNA-protein complex stabilitics. Thc observation that UBF is not restricted to thc RPI promoter but is also found to bind throughout thc rDNA, further complicates thc issue of whether or not UBF has a spécifie function in tran­scription initiation [92]. Rather, UBF appears to form a nuclcolar or rRNA genc chromatin and this may rcgulate multiple aspects of transcription, including rRNA genc accessibility, much as nuclcar histone chromatin does for thc rest of thc génome. Clearly, this rRNA gene chroma­tin is able to recruit SL1 and RPI regardlcss of the under-lying DNA [93-95]. Consistent with the rôle of UBF in the formation of an rRNA genc chromatin, thc most ré­cent data show that growth factor-dcpcndcnt remodelling of this chromatin controls rRNA synthesis by rcgulating RPI clongation [96] (sec below).

2. Yeast RPI initiation complex formation in yeast is in broad terms similar to that in mammals. Howevcr, thc numbcr of factors dircctly involved appears to be significantly greater, suggesting that much may still nced to be learnt about thc mammalian situation. The stcps of in vitro as-sembly of a yeast initiation complex in yeast arc shown in thc cartoons of Figure 6 [20, 97]. After establishment of thc UAF-UE complex, TBP is cither alrcady présent or is recruitcd along with the CF [98-100]. Efficient promo­tion requircs the UAF complex, though low-levcl spécifie

Ribosomal RNA Gènes

Upslream Activating Factor

Figure 6. Cartoon of the assembly of a pre-initiation complex and thc initiation cycle in the yeast S. cerevisae, bascd on Nomura [20] and Aprikian et al. [97], DNA is shown in red and promoter clé­ments in yellow. Colouiïng of the différent factors indicates prob­able équivalence with thc mammalian factors shown in Figure 4.

transcription initiation in vitro docs not require this com­plex, nor does it rcquire TBP or the UE of thc promoter [99, 101, 102]. This is reminiscent of the mammalian, plant and single-cell organism data, wherc in gênerai only a corc promoter and a single TBP complex arc abso-lutcly required in vitro (Fig. 3). A significant inercase in the TBP concentration in thc absence of functional UAF enhances RPI transcription from thc ycast core promoter in vitro, but in vivo, over-expression of TBP is unablc to rcscue thc Ioss of UAF [103, 104]. UAF subunits arc not strictly required for ycast survival. Howevcr, viability then dépends on polymerase switching, a phenomenon in which ycast uses RPII to producc functional levcls of rRNA and which requircs an inercase in rDNA copy num­bcr [105]. Thus, UAFisin faet esscntial for thc functional synthesis of rRNA by RPI. Once thc ycast pre-initiation complex has been assem-blcd, recruitment of the polymerase permits transcription

Cell. Mol. Life Sci. Vol. 64, 2007 Review Article 37

initiation and the complex is released into the elongation phase (Fig. 6). The CF complex may be disrupted and reibrmed at each new round of initiation [97], though it is unclear whether or not this occurs in vivo and it is certainly not obligatory for re-initiation [106]. Hmolp, a UBF-like protein, is not essential for yeast viability, but its loss induces a slow-growth phenotype and is lethal in combination with inactivation of non-essential RPI sub-un its [ 107]. Récent work has shown that like UBF, Hmo 1 p is found bound throughout the rDNA and hence may also de fine an rDNA chromatin [108]. However, Hmolp may also bc required for r-protein gène transcription [108].

3. Protozoa In vitro RPI promotion in cell-free extracts from A. castel-lanii resembles basai in vitro transcription in rodent and yeast extracts. A single TBP complex, TIF-IB, is required for in vitro RPI transcription from a short, core-like, promoter (Fig. 3). TIF-IB is purified as a five-subunit complex that includes TBP and its high-affinity bind-ing within the promoter région has been mapped by both footprinting and protein-DNA crosslinking (Fig. 3). The data from Acanthamoeba is probably the best in terms of the détails of the early steps in initiation [109-117]. The System has shown that the exact site of RPI initiation is determined by the positioning of the TBP complex and is relatively séquence insensitive [110]. Acanthamoeba TBP was shown to contact the DNA at around -45 bp, and RPI in the pre-initiation complexes sits across the initiation site, protecting the DNA from the downstream edge of the TBP complex to around +20 (Fig. 3) [118, 119].

Transcription initiation and the transition to elongation

There is gênerai agreement that unlike RPII, but like bacterial polymerases, RPI initiation does not require tri-phosphate hydrolysis. This was initially demonstrated in cell-free rat extracts [120], but has also been shown in the mouse and Acanthamoeba Systems [121, 122]. Initia­tion in yeast and mammals requires the RPI-associated factor Rrn3. Rrn3p was initially identified in yeast as an essential factor for RPI transcription (Fig. 6) [123]. It was shown to associate with a fraction of the RPI, and to be essential for functional recruitment of the polymerase to the pre-initiation complex [101,124]. Though RPI can be recruited to the yeast promoter in the absence of Rrn3p, initiation does not occur [97]. Rrn3p is normally found associated with a small fraction of RPI, and in yeast this association requires RPI phosphorylation [125]. Rrn3p is also phosphorylated, but in yeast this does not appear to be required for initiation [125]. Soon after the polymerase

initiâtes transcription, it releases Rrn3p, somewhat in the vein of the release of bacterial a factor and the RPII fac­tor TFIIF [124, 126]. This release is not obligatory for normal elongation, as fusion of Rrn3p to the RPI subunit A43 has no effect on viability or growth rate in yeast but does prevent normal down-regulation of RPI transcrip­tion [106]. In this context, it is worth noting that a factor is often maintained throughout elongation of the bacterial rRNA gènes, its stochastic release being only necessary to allow a more rapid reprogramming of transcription levels [127,128]. This said, dephosphorylation of RPI by Fcplp does appear to enhance the early phase of RPI elongation in vitro [129]. The mammalian Rrn3 homologue was first identified in human [130], where it has been shown to be required for RPI recruitment [131, 132]. Recycling of mammalian Rrn3 (TIF-IA) requires a post-translational modification that is lost during initiation [133] and in contrast to yeast Rrn3p, phosphorylation does indeed ap­pear to play a rôle in its activity [134]. Depending on the promoter context, bacterial RNA poly­merases and eukaryotic RPII may pass through a phase called promoter escape, during which the polymerase re-peatedly aborts synthesis and re-initiates, producing many short transcripts of 1O-20 bases. Eventually, the poly­merase escapes the influence of the promoter and makes the transition to a highly processive elongation phase [see e.g. réf. 135]. Despite the potential importance of pro­moter escape as a means of regulating transcription, little is known about RPI in this respect, and no published data on abortive initiation exist. Studies hâve inferred from the measurement of in vitro transcription kinetics that RPI passes through a rate-limiting post-initiation step, con­sistent with promoter escape [85, 136]. However, while studying the transition of RPI from initiation to elonga­tion, we were unable to detect the production of abortive transcripts, despite a highly sensitive single-round initia­tion assay [réf. 96 and unpublished data]. Thèse findings are consistent with the data for the rRNA gène promût-ers of Escherichia coli, which do not generate abortive transcripts and are not limited by promoter escape [137]. Thus, if promoter release of RPI is rate limiting, the un-derlying enzymatic mechanism must be distinctly différ­ent from that of either E. coli RNA polymerase or RPII. Clearly, the rôle of promoter escape in RPI initiation war­rants more detailed investigation. Once RPI has made the transition from initiation to pro­cessive elongation it may encounter various impediments to continued rRNA synthesis, such as chromatin. As men-tioned above, UBF is a major component of the rRNA gène chromatin. Surprisingly, we found that rather than aiding RPI elongation, in its unmodified state, UBF very effectively blocks elongation. Hence, UBF is a potential modulator of RPI elongation and, as will be seen below, this property is growth regulated. Thèse data may also help to explain the very varied properties that hâve been

38 T. Moss et al. Ribosomal RNA Gènes

ERK

figure 7. Summary of the regulatory targets of growth signalling, oncogenes and tumour suppressors within the RP1 transcription ma-chinery.

attributed to UBF in the past. Givcn the complexitics of RPII elongation that have been rcvcalcd over the last few years [see e.g. réf. 138], it is highly likely that yet more rcgulators of RPI elongation will be discovered.

In vivo régulation of rRNA gène expression

Without new ribosomes, a ccll cannot make protein and hence cannot grow and proliferate. Thus, an inercased rate of ribosomc biogenesis is a fundamcntal factor in hy­pertrophie discase. But is ribosomc biogenesis a control-ling factor or simply a housekeeping function? The most probable answer is: a bit of both. The inability to make ribosomcs quickly enough will, without doubt, limit ccll growth and slow prolifération [3,19,139,140]. The many signalling pathways, tumour suppressors and oncogenes that impinge on mammalian ribosome biogenesis would suggest that it is the cell and its environment that con­trol ribosome biogenesis (see Fig. 7). Howcvcr, the rate at which ribosomcs are made in turn détermines whether a ccll will enter S phase and commit to ccll growth and division [3, 140]. Thus, wc must consider ribosome bio­genesis as one component of a communication network controlling growth and prolifération. In bacteria, growth is associated with a high Icvcl of ri­bosome synthesis, while severc nutrient deprivation is associated with a rapid shutdown of this synthesis. In cubacteria, the shutdown is known as the 'stringent re-sponse' and is related to the production of (p)ppGpp by the idling ribosomc [137, 141]. The key éléments in this form of growth régulation werc shown to be the proximal promoters of the rRNA gencs. Thcsc rRNA genc promot-ers are regulatcd by nutrient availability while the weaker distal promoters and the r-protein gene promoters arc not. R-protcin expression is believed to be kept in check by an autoregulatory loop, frec r-protcin lcvcls ncgativcly regulating their corrcsponding gencs. Thus, in bacteria, ribosomc biogenesis appears to be regulated at the Icvcl of rRNA synthesis, this in turn regulating r-protcin con­

centrations and hence their synthesis rates by driving ri­bosomc assembly [142-144]. In eukaryotes the situation seems to be more complex, both r-protcin and rRNA genes being growth regulatcd. For cxample, yeast mRNA and r-protein lcvcls arc both directly regulated in responsc to nutrient availability [145] and even in the absence of rRNA synthesis, HcLa cclls continue to make r-proteins [146]. Thus, wc must assume that the mechanisms coordinating ribosome bio­genesis do not simply rcly on rRNA levcls controlling r-protcin synthesis via an autoregulatory loop as oc-curs in bacteria. Howcvcr, a very récent study in yeast in which the factor Rrn3p was fused to the A43 subunit of RPI (Table 1, Fig. 6) demonstrated clearly that ex­pression of the r-protein gencs dépends on the Icvcl of rRNA synthesis [106]. Thus, as in bacteria, transcription of the yeast rRNA gencs appears to détermine r-protein expression levcls. The régulation of ribosome biogenesis in eukaryotes, then, probably involves spécifie signalling and feedback networks to coordinatc r-protein and rRNA synthesis precisely. At Icast some of thèse signais may rcly on detecting functional 60S ribosome subunit levcls, since inhibition of large subunit nuclcar export lcads to a coordinatc down-regulation of rRNA and r-protein syn­thesis [147]. Strangely, this is not the case for the small ribosomal subunit. Despite the différences between pro- and eukaryotes, a stringent-like responsc has not only been identified in cukaryotic micro-organisms such as yeast [148], but also much more rcccntly in mammalian cclls [149]. In the late 1960s, ribosome production in mammalian ccll cultures was shown to bc down-rcgulated in conditions of amino acid starvation [150, 151], and a few years later this effect was shown to be due at least in part to a down-regulation of rRNA gene transcription [152]. Glucocorticoid was also shown to down-rcgulate the mammalian rRNA genes, as was the global arrest of translation by cyclohcximidc [153-156] and sérum withdrawal [157]. Encystment of Acanthamoeba was shown to cause a complète shutdown of rRNA transcription but also of ail other nuclcar tran­scription [158]. However, the responscs of mammalian

Cell. Mol. Life Sci. Vol. 64, 2007 Review Article 39

ce Ils to nutrient withdrawal, hormones and drugs were generally considered to be slow; for example, a 16-h treat-ment was used to observe repression of transcription with glucocorticoid [155], and cycloheximide required 2 h to reach maximal effect [153], though histidine withdrawal gave a relatively rapid response, transcriptional activity dropping by 50% in 60 min [152, 159]. Not until much later did studies of the response of mammalian cells to growth factor and MAP kinase activation reveal an im­médiate effect on rRNA gène transcription in vivo [149]. Indeed, thèse studies showed that a réversible response to growth factor (EGF) stimulation or direct MAP-kinase (ERK.) activation occurred within 10 min, identifying a stringent-like response in mammals. Subséquent studies hâve confirmed thèse observations and extended them to include sérum, fibroblast growth factor (FGF), insu-lin and insulin-like growth factor (IGF) responses [134, 160, 161]. However, before considering the mechanisms underlying thèse regulatory responses, we should first consider the potential cellular responses leading to rRNA gène régulation.

Growth response or stress response?

To corne to terms with various, often conflicting data it may be necessary to consider at least two distinct rRNA gène regulatory modes. The first is a true growth régula­tion, in which rRNA synthesis and ribosome biogenesis are modulated to meet spécifie growth rate requirements. The second is a 'stress' response, in which the cell estab-lishes a protective mode as an antécédent to the resump-tion of growth or to cell death. UV or other DNA damage and possibly cycloheximide or long-term withdrawal of nutrients may resuit in a stress response in which most if not ail ribosome biogenesis is shut down. On the other haiid, a short-term réduction in nutrients, changes in growth factors, or cell differentiation would be expected to lead to the adaptation of ribosome biogenesis to the new growth conditions, more akin to letting the engine 'tick over' rather than stalling it. Unfortunately, it is not always possible to détermine which experiments fall into which of thèse catégories, and this should be borne in mind when attempting to reconcile conflicting data.

Mechanisms of growth régulation

Unlike protein-coding gènes, only four distinct possi­bilités exist to regulate the rRNA gènes: gène activation, productive transcription initiation, transcription elonga-tion and rRNA dégradation, Since the last of thèse does not appear to be a significant factor, it will not be con­sidered further.

1. rRNA gène activation Since the ribosomal gènes are présent in several hundred copies and only a proportion appear to be transcribed, modulating the number of actively transcribed gènes could in principle regulate rRNA synthesis. Growth of yeast into stationary phase leads to a réduction in the numbers of active ribosomal gènes [162, 163]. However, growth stimulation of mammalian cell cultures does not detectably change the active gène number, despite a sev-eral-fold increase in rRNA synthesis rates [164]. This said, we hâve recently found that artificial changes in the heterochromatic state of the rRNA gencs can lead to an increase in the active gène number in mammalian cells, though this does not induce a corresponding increase in transcription [T. Gagnon-Kugler and T. Moss unpublished data]. Furthermore, the number of active rRNA gènes has not to our knowledge been determined under conditions of stress in mammals. Hence, though gène activation does not explain the rapid growth factor-mediated régulation of rRNA synthesis in mammalian cells in culture, its im­portance in vivo cannot be excluded.

2. Regulating initiation of rRNA transcription Situations in which RPI transcription is very strongly repressed, for example long-term sérum withdrawal, cycloheximide treatment and encystment of Acantham-oeba, hâve been used as the basis for attempts to identify mechanisms of rRNA gène régulation. Factors from ac­tive and inactive cells hâve been isolated and their abili-ties to support spécifie RPI transcription initiation in vi­tro investigated. Such studies led to the identification of active and inactive RPI fractions. Thèse fractions of poly-merase are equally capable of DNA-templated nucleotide polymérisation, but only one form retains the capacity to specifically initiate transcription from the RPI promoter [ 156,165]. The ability to initiate transcription was associ-ated with polymerase phosphorylation [124, 165] and/or with soluble factors TIF-IA, TIFIC or factor C [ 156,157, 166-169]. More récent data show that TIF-IA and prob-ably factor C correspond to the yeast and human Rrn3 (Figs. 4, 6) [131], but that TIFIC may be a distinct ac­tivity [133]. Later work demonstrated that mammalian Rrn3 (TIF-IA) phosphorylation changes with cell treat­ment and correlates with in vivo rRNA gène transcrip­tion levels [133, 134]. Phosphorylation of mammalian Rrn3 at several sites has been demonstrated to be due to a combination of RSK and ERK activities, and mutation of thèse sites in mouse Rrn3 suppresses transcription in transfected mouse and human cells (Fig. 7) [134]. Mouse Rrn3 was also shown to be regulated via the mTOR nutri-ent-sensing pathway and most recently via Jun N-terminal kinase (JNK) during a true stress response [170, 171]. A model emerged from thèse data of rRNA gène régulation occurring exclusively at the level of transcription initia-

40 T. Moss et al. Ribosomal RNA Gènes

tion via activation and inactivation of Rrn3/TIF-IA [71]. This model is consistent with stationary phase régulation and TOR inhibition in yeast [163], however, as will be seen below, it is not consistent with our knowledge of growth régulation in mammalian Systems. Despite intense study of RPI and Rrn3, the first molecu-lar pathway from growth signalling to rRNA gène tran­scription actually came from a study of UBF [149]. Re-sponse to stimulation of human and mouse cells by EGF and by direct activation of the Raf-MEK-ERK pathway was shown to require phosphorylation of UBF. The two N-terminal HMG boxes of UBF display on their DNA-binding surfaces consensus ERK. phosphorylation sites. Phosphorylation of thèse sites was shown to be required for stimulation of rRNA gène transcription. Récent data hâve shown that this phosphorylation régulâtes RPI elon-gation rates (see below). UBF has been implicated in the largest number of rRNA gène regulatory responses (Fig. 7), and it is even a direct target of the FGF2 growth factor [172,173]. SL1 function (Figs. 4, 6), is regulated by PCAF acetyla-tion of its TAF,68 subunit [174]. SL1 is also inactivated by CDK1 phosphorylation [175]. Furthermore, the SL1 complex is disrupted in cells overexpressing the phospha-tase PTEN [176]. At first sight, this would suggest that gène activation should be regulated, leading to a change in the number of active rRNA gènes. However, as we hâve seen, this does not usually occur [164]. Rather, the data suggest that the level of active SL1 does not define the number of active gènes. Such an interprétation is consis­tent with observations in yeast showing that CF, the SL1 homologue, is released at each new round of initiation and hence forms part of a catalytic cycle [97]. It may, then, be wrong to consider the formation of an SLl-pro-moter complex as a gène activation step, and it should rather be perceived of as a catalytic event much like initiation itself. The observation of 'holo-polymerase' complexes that include SL1 further supports this conten­tion [177-181]. We may then find that mammalian SL1 performs a growth-regulated function that is unrelated to rRNA gène activation.

3. Regulating the rRNA transcription elongation rate Despite the emphasis that until recently was placed on ini­tiation as a means of regulating rRNA gène transcription, several early studies suggested that the capacity of RPI to initiate does not explain growth régulation in vivo [154, 182-184]. Most telling, however, are the Miller spread observations of tightly packed transcription complexes along the coding régions of the rRNA gènes (Fig. 1). The dense packing of polymerases that thèse pictures reveal clearly excludes any large increase in polymerase load-ing. Yet, similar observations of tight polymerase packing hâve been made in animal cells exhibiting a wide range

of rRNA synthesis rates [see réf. 60 for a discussion], suggesting that in vivo elongation must be regulated. A notable exception to this occurs in yeast, where the sta­tionary phase shutdown of transcription correlates with a réduction in the density of transcribing polymerases [162, 163,185]. However, even in this extrême case of régula­tion, yeast makes an attempt to maintain a high density of transcription by also reducing the number of active rRNA gène repeats. The obvious inconsistency between mammalian models that invoke régulation solely at the level of initiation [see e.g. réf. 71], and the observations of near saturating levels of transcription complexes in Miller spreads, led us to ask whether changes in the RPI loading do in fact oc­cur during growth factor induction of the rRNA gènes [96, 164]. What we found was fully consistent with the électron microscope data. Using both nuclear run-on ànd ChIP approaches, we established that growth factor and MAP kinase activation of rRNA synthesis in human and mouse cells does not correlate with a significant increase in the total number of RPI transcription complexes re­quired by models of régulation at initiation. The obvious explanation was that RPI transcription elongation rates were also modulated. This we demonstrated to be the case in vivo by directly measuring RPI elongation under dif­férent conditions of growth stimulation [96]. To achieve near constant RPI loadings over a wide range of rRNA synthesis rates, either (i) elongation and initia­tion must be coordinately regulated or (ii) elongation must limit the rate of initiation of new transcripts. The latter would appear much the simpler explanation mechanisti-cally, since limiting elongation rates would naturally limit initiation. Several observations suggested that régulation of elongation was dominant over initiation [96]. More récent work has also shown that the capacity of nuclear extracts to specifically initiate RPI transcription in vitro does not change when cells are stimulated by growth fac­tor (Fig. 8A). Thus, fivefold or greater changes in rRNA synthesis can occur in the absence of changes in the com­pétence of SL1, RPI and Rrn3 to initiate new transcripts. We hâve further shown that the growth rates of human colon cancer cells are also quantitatively explained by changes in the RPI elongation rate [T. Gagnon-Kugler, unpublished data], suggesting that régulation of elonga­tion is a gênerai phenomenon in mammalian cells. How is RPI elongation regulated? As mentioned above, some years ago we demonstrated that the ability of UBF to enhanced RPI transcription in vivo dépends on a révers­ible phosphorylation of its two N-terminal HMG1 boxes by ERK1/2, and this phosphorylation was necessary for EGF to stimulate rRNA gène transcription (Fig. 5A) [149]. Each of the three most N-terminal HMG boxes of UBF is able to bend DNA and a dimer of UBF induces a 360° loop in the DNA template [83, 84,186]. The result-ing enhancesome structure resembles the nucleosome of

Cell. Mol. Life Sci. Vol. 64, 2007 Review Article 41

Relative 4SS transcription ( H labellng)

FigureS. (A) The rapid changes of in vivo rRNA synthesis rates upon growth factor stimulation arc not reflected in a change in the compétence of RP1 to initiate. Upper panel, RPl-speciftc in vitro transcription in nuclcar extracts from N1H3T3 cclls treated for 30 min with the MEK inhibitor PD98059 (PD) or with EGF [96, 149]. Exactly equal but increasing amounts of nuclcar proteins were used in otherwise identical in vitro transcription reactions per-formed in parallei. Cntrl, in vitro transcription of thc RPI template by a highly active nuclear extract from mouse LI210 cclls. Lower panel, parallei mcasurement of m vivo 45S rRNA transcription rates determined in a 30-min ['H]-uridine puise labelling [96, 149]. (Il) Régulation of RPI clongation by nuclcolar chromatin remodelling. The rRNA gènes arc shown foldcd into a séries of enhancesomes by the binding of consécutive UBF dimers. Only thc Core of UBF consisting of the dimerisation domains and HMG boxes 1-3 are indicated (Fig. 5). ERK. interaction with the first two I IMG boxes and their subséquent phosphorylation leads to an unfolding of the enhancesome, permitting passage of thc RPI clongation complex. The interaction of CBP with thc samc région of UBF may lcad to a coopérative elîect on transcription and may also be required in vivo. Rb is ablc to compote for CBP and perhaps ERK binding and hence would inhibit clongation. As in Figures 4 and 5, UBF is shown in dark blue and the DNA in red.

chromatin in protcin-DNA composition but contains only a single loop of DNA (Fig. 5B). We showed that ERK phosphorylation of UBF remodcls thc enhancesome and in this way détermines thc rate of clongation of the RPI transcription complexes through nucleolar chromatin (Fig. 8B) [96, 187]. This réaction is probably mediated by a direct interaction between ERK. and the HMG boxes it phosphorylates [149]. The démonstration that thc RPI clongation rate is a key growth rcgulator of rRNA synthesis also suggests a mcans to coordinate this synthesis with pre-ribosome as-scmbly. During the co-transcriptional phase of pre-ribo­some assembly, processing factors and r-proteins must be assembled in thc correct order on the nascent rRNA. A feedback mechanism allowing thc régulation of RPI clon­gation dépendent on correct ribonucleoprotein (RNP) assembly could provide an important means of 'proof-reading' pre-ribosome assembly. Indeed, recent studies of thc processome, thc carlicst visible co-transcriptional RNP structure, suggest that its assembly on thc prc-rRNA régulâtes the rate of rRNA synthesis in ycast [ 188].

The rôle of oncogenes and tumour suppressors

Tumour suppressors Rb and p53 hâve been implicated in limiting rRNA synthesis and affect thc interaction between UBF and SL1 (Fig. 7) [189-192]. The acetyltransferase CBP has been shown to enhanec rRNA gène transcrip­tion by competing for the Rb-binding site on UBF. CBP binding to HMG boxes 1 and 2 causes UBF acetylation, while Rb displaces CBP from its binding site and recruits HDAC1 to catalyse the deacetylation of UBF. A tantaliz-ing corrélation exists between this Rb-CBP compétition and ERK phosphorylation of UBF. Ail threc proteins bind to the same or adjacent sites on UBF [149, 193]. Further-more, CBP is known to bc bound and activated by ERK 1194, 195]. Thus, ERK and CBP may act cooperatively in growth factor activation of thc rRNA gcncs by targeting the same site on UBF. In support of this, we hâve shown that activation of thc rRNA gcncs with the histone deacet-ylase (HDAC) inhibitor trichostatin A (TSA) dépends on a functional ERK signalling pathway [ 164, 193]. UBF has also been rcportcd to bc acctylatcd by p300, PCAF and thc isolated Tip6() acetyltransferase subunit [196, 197]. However, thc functional significanec of UBF acetylation is still poorly understood. Thc ARF tumour suppressor was shown to regulate rRNA processing independently of p53 by catalysing the dégra­dation of B23/nuclcophosmin, a key protein in ribosome assembly [198-200]. However, it has also been shown to inhibit rRNA synthesis by an unknown mechanism [199]. Most recently, c-Myc was shown to enhance ribosome biogencsis, though the mechanism for its action is still

42 T. Mossetal. Ribosomal RNA Gènes

far from clear [201-204]. A scarch for c-Myc and N-Myc gène targets has consistently identified many gènes that are implicated directly in ribosome biogenesis, including many r-protein gènes and processing factors such as B23/ nucleophosmin as well as gènes that are indirectly impli­cated via cell cycle switches such as p27cip, CDK.4 and cyclin D2 [205-214]. Since it is évident that a coupling exists between r-protein and rRNA synthesis, it was not surprising to find that c-Myc lcvels modulated rRNA syn­thesis. The first study to demonstrate this directly showed that in a human B cell Une, rRNA processing was more efficient when c-Myc expression was induced [215]. However, this study detected no change in rRNA synthe­sis rates. A second study investigated c-Myc/MAD func-tion in mouse granulocytes and found that thèse proteins had reciprocal effects on rRNA synthesis rates [204]. It was argued that this was due to direct régulation of the UBF gène, enhanced UBF levels having previously been shown to drive rRNA synthesis [216-218]. Consistent with thèse data, a more récent study in Drosophila dem-onstrated that c-Myc drives rRNA synthesis indirectly by activating the gènes encoding the RPI machinery and the great majority of gènes required for ribosome assembly [201]. However, two further studies hâve shown that c-Myc and its co-factor TRRAP can also be found in the nucleolus and are able to interact directly with the rRNA gènes [202,203]. In one study, it was argued that nucleo-lar localisation and rRNA gène interaction of c-Myc was enhanced by inhibition of its proteasomal dégradation, while in the other study, proteasome inhibition was not found to be a factor. In contrast, both studies showed that when c-Myc levels were manipulated by small interfering RNA and inducible expression, thèse levels correlated positively with rRNA gène activity and with the level of histone H3 and H4 acetylation on thèse gènes. However, the mechanism of direct régulation of the rRNA gènes by c-Myc remains a mystery, even more so since nucleolar c-Myc levels appear to be exceedingly low. It should be noted that a common factor of many onco-genes and tumour suppressors is that they do not spe-cifically target the promoter of the rRNA gène. Rather, they act through proteins such as UBF or directly interact with DNA sites spread widely throughout the gènes. This suggests that they could be important for controlling the rate of transcription elongation or even pre-ribosome as­sembly.

In Minimal y

In the last few years, several reappraisals of the basic parameters of rRNA gène régulation hâve succeeded in pinpointing the levels at which régulation can and does occur. We can be certain that rRNA gène transcription is co-regulated with growth and, as we should expect,

changes in rRNA synthesis précède détectable changes in growth rate. To increase the rate of ribosome biogenesis, it is necessary to up-regulate the production of several hundred proteins in addition to the rRNAs. Who is driv-ing whom - rRNA or r-protein expression - and the rôle they play in growth régulation are still open questions. In contrast to prokaryotes, the answer for eukaryotes may be that neither rRNA nor r-protein is dominant and that a complex feedback network exists to coordinate their syn­thesis. Clearly, isolated mechanisms of controlling rRNA initiation rates via RPI activity cannot explain this coor­dination, though they likely play their part. Régulation at the level of transcription elongation provides a more sat-isfactory explanation since it allows for feedback mecha­nisms that could coordinate rRNA synthesis with pre-ri­bosome assembly. The challenge now is to find ways of identifying and piecing together the components of the network regulating ribosome biogenesis.

Acknowledgement. We hâve attempted to review work of direct relevance to the growth régulation of rRNA gène transcription. For this reason and due to constraints on space, much important work had to be omitted and we wish to apologise to those whose work has apparently been overlooked. We thank Drs N. Gas, A. Beyer, Y. Osheim and U. Scheer for kindly providing électron mi-crographs. We would also like to thank those who provided up­dates of their most récent work and J. Smith and M. Nomura who took the time to discuss various mechanistic scénarios. Thanks finally to B. Muntwyler for critical reading and the editorial staff of CMLS for their patience. This work was supported by an operat-ing grant from the Canadian Institutes of Health Research (CIHR) and a CIHR scholarship to T. G.-K. The University Laval Cancer Research Centre is supported by the Fonds de Recherche sur la Santé du Québec (FRSQ).

1 Poole, A. M., Jeffares, D. C. and Penny, D. (1998) The path from the RNA world. J. Mol. Evol. 46, 1-17.

2 Rodnina, M. V and Wintermeyer, W. (2003) Peptide bond formation on the ribosome: structure and mechanism. Curr. Opin. Struct. Biol. 13, 334-340.

3 Jorgensen, P., Rupes, L, Sharom, J. R., Schneper, L., Broach, J. R. and Tyers, M. (2004) A dynamic transcriptional network communicates growth potential to ribosome synthesis and critical cell size. Gènes Dev. 18,2491-2505.

4 Moss, T. (2004) At the crossroads of growth control: making ribosomal RNA. Curr. Opin. Genêt. Dev. 14,210-217.

5 Fontoura, B. M., Atienza, C. A., Sorokina, E. A., Morimoto, T. and Carroll, R. B. (1997) Cytoplasmic p53 polypeptide is associated with ribosomes. Mol. Cell. Biol. 17,3146-3154.

6 Rubbi, C. P. and Milner, J. (2003) Disruption of the nucleolus médiates stabilization of p53 in response to DNA damage and other stresses. EMBO J. 22,6068-6077.

7 Andersen, J. S., Lam, Y. W, Leung, A. K„ Ong, S. E., Lyon, C. E., Lamond, A. I. and Mann, M. (2005) Nucleolar pro-teome dynamics. Nature 433,77-83.

8 Coûte, Y., Burgess, J. A., Diaz, J. J., Chichester, C, Lisacek, F., Greco, A. and Sanchez, J. C. (2006) Deciphering the hu­man nucleolar proteome. Mass. Spectrom. Rev. 25,215-234.

9 Scherl, A., Coûte, Y, Deon, C, Calle, A., Kindbeiter, K., San­chez, J. C, Greco, A., Hochstrasser, D. and Diaz, J. J. (2002) Functional proteomic analysis of human nucleolus. Mol. Biol. Cell 13,4100-4109.

10 Andersen, J. S., Lyon, C. E., Fox, A. H., Leung, A. K., Lam, Y. W., Steen, H., Mann, M. and Lamond, A. I. (2002) Directed

Cell. Mol. Life Sci. Vol. 64, 2007 Review Article 43

proteomic analysis of the human nucleolus. Curr. Biol. 12, 1-11.

11 White, R. J. (2004) RNA polymerase 111 transcription and can­cer. Oncogene 23,3208-3216.

12 Scheer, U. (2005) Visualizing genetic activity of mammalian cells: looking back to the year 1972. Eur. X Cell Biol. 84, 91-96.

13 Klein, D. J., Moore, P. B. and Steitz, T. A. (2004) The rôles of ribosomal proteins in the structure assembly, and évolution of the large ribosomal subunit. X Mol. Biol. 340, 141-177.

14 Ferreira-Cerca, S., Poil, G., Gleizes, P.-E., Tschochner, H. and Milkereit, P. (2005) Rôles of eukaryotic ribosomal proteins in maturation and transport of pre-18S rRNA and ribosome function. Mol. Cell 20, 263-275.

15 Nazar, R. N. (2004) Ribosomal RNA processing and ribo­some biogenesis in eukaryotes. IUBMB. Life 56,457—465.

16 Tschochner, II. and Huit, E. (2003) Pre-ribosomes on the road from the nucleolus to the cytoplasm. Trends Cell Biol. 13,255-263.

17 Fromont-Racine, M., Senger, B., Saveanu, C. and Fasiolo, F. (2003) Ribosome assembly in eukaryotes. Gène 313, 17-42.

18 Warner, J. R. ( 1999) The économies of ribosome biosynthesis in yeast. Trends Biochem. Sci. 24,437-440.

19 Rudra, D., Zhao, Y. and Warner, J. R. (2005) Central rôle of Ifhlp-Fhllp interaction in the synthesis of yeast ribosomal proteins. EMBO J. 24, 533-542.

20 Notnura, M. (2001) Ribosomal RNA gènes, RNA polymer-ases, nucleolar structures, and synthesis of rRNA in the yeast Saccharomyces cerevisiae. Cold Spring Harb. Symp. Quant. Biol. 66, 555-565.

21 Shaw, P. and Doonan, J. (2005) The nucleolus: playing by dif­férent rules? Cell Cycle 4, 102-105.

22 Trumtel, S., Leger-Silvestre, I., Gleizes, P. E., Teulieres, F. and Gas, N. (2000) Assembly and functional organization of the nucleolus: ultrastructural analysis of Saccharomyces cerevi­siae mutants. Mol. Biol. Cell 11, 2175-2189.

23 Cheutin, T., O'Donohue, M. F., Beorchia, A., Vandelaer, M., Kaplan, H., Defever, B., Ploton, D. and Thiry, M. (2002) Three-dimensional organization of active rRNA gènes within the nucleolus. J. Cell Sci. 115, 3297-3307.

24 Scheer, U, Thiry, M. and Goessens, G. (1993) Structure, function and assembly of the nucleolus. Trends Cell Biol. 3, 236-241.

25 Thiry, M. and Goessens, G. (1992) Where, within the nucleo­lus, are the rRNA gènes located? Exp. Cell Res. 200, 1-4.

26 Thompson, M., Haeusler, R. A., Good, P. D. and Engelke, D. R. (2003) Nucleolar clustering of dispersed tRNA gènes. Science 302, 1399-1401.

27 Maden, B.E. H. and Hughes, J.M. X. (1997) Eukaryotic ribo­somal RNA: the récent excitement in the nucleotide modifica­tion problem. Chromosoma 105,391-400.

28 Bachellerie, J. P. and Cavaillé, J. (1997) Guiding ribose meth-ylation of rRNA. Trends Biochem. Sci. 22,257-261.

29 Cavaillé, X, Nicoloso, M. and Bachellerie, J. P. (1996) Tar-geted ribose methylation of RNA in vivo directed by tailored antisense RNA guides. Nature 383, 732-735.

30 Kiss, T. (2001) Small nucleolar RNA-guided post-transcrip-tional modification of cellular RNAs. EMBO J. 20, 3617-3622.

31 Fatica, A. and Tollervey, D. (2002) Making ribosomes. Curr. Opin. Cell Biol. 14,313-318.

32 Moss, T. (1983) A transcriptional function for the répétitive ribosomal spacer in Xenopus laevis. Nature 302,223-228.

33 Miller, J. R., Hayward, D. C. and Glover, D. M. (1983) Tran­scription of the non-transcribed spacer of Drosophila melano-gaster rDNA. Nucleic Acids Res. 11,11-19.

34 Coen, E. S. and Dover, G. A. (1983) Multiple polymerase I promoter séquences in rDNA of Drosophila melanogaster. Nucleic Acids Res. 10, 7017-7026.

35 Murtif, V L. and Rae, RM. M. (1985) In vivo transcription of rDNA spacers in Drosophila. Nucleic Acids Res. 13, 3221 -3240.

36 Cassidy, B. G., Yang-Yen, H. F. and Rothblum, L. I. (1987) Additional RNA polymerase I initiation site within the non-transcribed spacer région of the rat rRNA gène. Mol. Cell. Biol. 7,2388-2396.

37 Tower, X, Henderson, S. L., Dougherty, K. M., Wejksnora, P. X and Sollner-Webb, B. (1989) An RNA polymerase I pro­moter located in the CHO and mouse ribosomal DNA spac­ers: functional analysis and factor and séquence requirements. Mol. Cell. Biol. 9,1513-1525.

38 Kuhn, A. and Grummt, I. (1987) A novel promoter in the mouse rDNA spacer is active in vivo and in vitro. EMBO J. 6, 3487-3492.

39 Henderson, A. S., Warburton, D. and Atwood, K. C. (1972) Location of ribosomal DNA in the human chromosome com­plément. Proc. Natl. Acad. Sci. USA 69, 3394-3398.

40 Rowe, L. B., Janaswami, P. M., Barter, M. E. and Birkenmeier, E. H. (1996) Genetic mappingof 18S ribosomal RNA-related loci to mouse chromosomes 5, 6, 9, 12, 17, 18, 19, and X. Mamm. Génome 7, 886-889.

41 Heitz, E. (1931) Die Ursache der gesetzmâssigen Zahl, Lage, Form und Grosse pflanzlicher Nukleolen. Planta 12, 775-844.

42 McClintock, B. (1934) The relation of a particular chromo-somal élément to the development of the nucleoli in Zea mays. Z. Zellforsch. Mikroanat. 21,294-328.

43 Boseley, P., Moss, T., Machler, M., Portmann, R. and Birnstiel, M. (1979) Séquence organization of the spacer DNA in a ribo­somal gène unit of Xenopus laevis. Cell 17, 19-31.

44 Moss, T. and Birnstiel, M. L. (1979) The putative promoter of a Xenopus laevis ribosomal gène is reduplicated. Nucleic Acids Res. 6, 3733-3743.

45 Moss, T. (1982) Transcription of cloned Xenopus laevis ri­bosomal DNA microinjected into Xenopus oocytes, and the identification of an RNA polymerase I promoter. Cell 30, 835-842.

46 De Winter, R.F. J. and Moss, T. (1986) Spacer promoters are essential for efficient enhancement of X. laevis ribosomal transcription. Cell 44,313-318.

47 Labhart, P. and Reeder, R. H. (1984) Enhancer-like proper-ties of the 60/81 bp éléments in the ribosomal gène spacer of Xenopus laevis. Cell 37, 285-289.

48 De Winter, R.F. J. and Moss, T. ( 1987) A complex array of sé­quences enhances ribosomal transcription in Xenopus laevis. X Mol. Biol. 196,813-827.

49 Elion, E. A. and Warner, X R. ( 1984) The major promoter élé­ment of rRNA transcription in yeast lies 2kb upstream. Cell 39,663-673.

50 Wai, H., Johzuka, K., Vu, L., Eliason, K., Kobayashi, T., Horiuchi, T. and Nomura, M. (2001 ) Yeast RNA polymerase I enhancer is dispensable for transcription of the chromosomal rRNA gène and cell growth, and its apparent transcription enhancement from ectopic promoters requires Fobl protein. Mol. Cell. Biol. 21, 5541-5553.

51 Osheim, Y. N., Mougey, E. B., Windle, X, Anderson, M., O'Reilly, M., Miller, O. L. Jr, Beyer, A. and Sollner-Webb, B. (1996) Metazoan rDNA enhancer acts by making more gènes transcriptionally active. J. Cell Biol. 133,943-954.

52 Dunaway, M. and Drôge, P. (1989) Transactivation of the Xenopus rRNA gène promoter by its enhancer. Nature 341, 657-659.

53 Caudy, A. A. and Pikaard, C. S. (2002) Xenopus ribosomal RNA gène intergenic spacer éléments conferring transcrip­tional enhancement and nucleolar dominance-like compéti­tion in oocytes. X Biol. Chem. 277,31577-31584.

54 Grimaldi, G., Fiorentini, P. and Di Nocera, P. P. (1990) Spacer promoters are orientation-dependent activators of pre-rRNA

44 T. Moss et al. Ribosomal RNA Gènes

transcription in Drosophila melanogaster. Mol. Cell. Biol. 10, 4667^677.

55 Kobayashi, T. and Ganley, A. R. (2005) Recombination régu­lation by transcription-induced cohesin dissociation in rDNA repeats. Science 309, 1581-1584.

56 Mayer, C , Schmitz, K. M., Li, J., Grummt, I. and Santoro, R. (2006) Intergenic transcripts regulate the epigenetic state of rRNA gènes. Mol. Cell 22, 351-361.

57 Grummt, I. and Pikaard, C. S. (2003) Epigenetic silencing of RNA polymerase I transcription. Nat Rev Mol. Cell. Biol. 4, 641-649.

58 Almeida, R. and Allshire, R. C. (2005) RNA silencing and génome régulation. Trends Cell Biol. 15, 251-258.

59 Reinhart, B. J. and Bartel, D. P. (2002) Small RNAs corre­spond to centromere heterochromatic repeats. Science 297, 1831.

60 Moss, T., Mitchelson, K. and De Winter, R.F. J. (1985) The promotion of ribosomal transcription in eukaryotes. Oxford Surv. Eukaryot. Gènes 2, 207-250.

61 Moss, T. and Stefanovsky, V Y. (1995) Promotion and régula­tion of ribosomal transcription in eukaryotes by RNA poly­merase I. In: Progress in Nucleic Acids and Molecular Biol-ogy (Cohn, W. E. and Moldave, K, Eds), pp. 25-66. Académie Press, San Diego.

62 Van der Sande, C.A.F. M., Kulkens, T., Kramer, A. B., De Wijs, I. J., Van Heerikhuizen, H., KJootwijk, J. and Planta, R. J. (1989) Termination of transcription by yeast RNA poly­merase I. Nucleic Acids Res. 17, 9127-9146.

63 Henderson, S. L., Ryan, K. and Sollner-Webb, B. (1989) The promoter-proximal rDNA terminator augments initiation by preventing disruption of the stable transcription complex caused by polymerase read-in [published erratum appears in Gènes Dev. 1989, 3, 584]. Gènes Dev. 3, 212-223.

64 Mitchelson, K. and Moss, T. (1987) The enhancement of ri­bosomal transcription by the recycling of RNA polymerase I. Nucleic Acids Res. 15, 9577-9596.

65 Moss, T., Larose, A.-M., Mitchelson, K. and Leblanc, B. (1992) Readthrough enhancement and promoter occlusion on the ribosomal gènes ofXenopus laevis. Biochem. Cell Biol. 70,324-331.

66 Zomerdijk, J.C.B. M., Kieft, R. and Borst, P. (1991) Efficient production of functional mRNA mediated by RNA poly­merase I in Trypanosoma brucei. Nature 353, 772-775.

67 Paule, M. R. (1998) Promoter structure of class I gènes. In: Transcription of Ribosomal Gènes by Eukaryotic RNA Poly­merase I (Paule, M. R. Ed.), pp. 40-50. Landes Bioscience, Austin.

68 Hamada, M., Huang, Y, Lowe, T. M. and Maraia, R. J. (2001) Widespread use of TATA éléments in the core promoters for RNA polymerases III, II, and I in fission yeast. Mol. Cell. Biol. 21,6870-6881.

69 Boukhgalter, B., Liu, M., Guo, A., Tripp, M., Tran, K., Huynh, C. and Pape, L. (2002) Characterization of a fission yeast sub­unit of an RNA polymerase I essential transcription initiation factor, SpRrn7 h/TAF(I)68, that bridges yeast and mammals: association with SpRrn 11 h and the core ribosomal RNA gène promoter. Gène 291,187-201.

70 Moss, T. and Stefanovsky, V Y. (2002) At the center of eukary­otic life. Cell 109, 545-548.

71 Grummt, I. (2003) Life on a planet of its own: régulation of RNA polymerase I transcription in the nucleolus. Gènes Dev. 17,1691-1702.

72 Cavanaugh, A., Hirschler-Laszkiewicz, I. and Rothblum, L. I. (2004) Ribosomal DNA transcription in mammals. In: The Nucleolus (Oison, M. O. Ed.), pp. 90-129. Kluwer Académie, Dordrecht.

73 Russell, J. and Zomerdijk, J. C. (2005) RNA-polymerase-I-di-rected rDNA transcription, life and works. Trends Biochem. Sci. 30,87-96.

74 Bell, S. P., Learned, R. M., Jantzen, H. M. andTjian, R. (1988) Functional cooperativity between transcription factors UBF1 and SL1 médiates human ribosomal RNA synthesis. Science 241, 1192-1197.

75 Tuan, J. C , Zhai, W. G. and Cornai, L. (1999) Recruitment of TATA-binding protein-TAF, complex SL1 to the human ri­bosomal DNA promoter is mediated by the carboxy-terminal activation domain of upstream binding factor (UBF) and is regulated by UBF phosphorylation. Mol. Cell. Biol. 19, 2872-2879.

76 Kihm, A. J., Hershey, J. C , Haystead, TA. J., Madsen, C. S. and Owens, G. K.. (1998) Phosphorylation of the rRNA tran­scription factor upstream binding factor promûtes its associa­tion with TATA binding protein. Proc. Natl. Acad. Sci. USA 95,14816-14820.

77 Kwon, H. and Green, M. R. (1994) The RNA polymerase I transcription factor, upstream binding factor, interacts di-rectly with the TATA box-binding protein. J. Biol. Chem. 269, 30140-30146.

78 Voit, R., Schnapp,A., Kuhn,A., Rosenbauer, H., Hirschmann, P., Stunnenberg, H. G. and Grummt, I. (1992) The nucleolar transcription factor mUBF is phosphorylated by casein kinase II in the C-terminal hyperacidic tail which is essential for transactivation. EMBOJ. 11,2211-2218.

79 McStay, B., Frazier, M. W. and Reeder, R. H. (1991) xUBF contains a novel dimerization domain essential for RNA poly­merase I transcription. Gènes Dev. 5, 1957-1968.

80 Jantzen, H. M., Chow, A. M., King, D. S. and Tjian, R. ( 1992) Multiple domains of the RNA polymerase I activator hUBF interact with the TATA-binding protein complex hSLl to mé­diate transcription. Gènes Dev. 6, 1950-1963.

81 Leblanc, B., Read, C. and Moss, T. (1993) Récognition of the Xenopus ribosomal core promoter by the transcription factor xUBF involves multiple HMG box domains and leads to an xUBF interdomain interaction. EMBO J. 12, 513-525.

82 Moss, T., Bazett-Jones, D. P. and Leblanc, B. ( 1994) The struc­ture of the xUBF-DNA complex. J. Cell. Biochem. Suppl. 18CL505.

83 Stefanovsky, V Y, Bazett-Jones, D. P., Pelletier, G. and Moss, T. (1996) The DNA supercoiling architecture induced by the transcription factor xUBF requires three of its five HMG-boxes. Nucleic Acids Res. 24, 3208-3215.

84 Stefanovsky, V Y, Pelletier, G., Bazett-Jones, D. P., Crane-Robinson, C. and Moss, T. (2001) DNA looping in the RNA polymerase I enhancesome is the resuit of non-cooperative in-phase bending by two UBF molécules. Nucleic Acids Res. 29, 3241-3247.

85 Panov, K. L, Friedrich, J. K., Russell, J. and Zomerdijk, J. C. (2006) UBF activâtes RNA polymerase I transcription by stimulating promoter escape. EMBO J. 25, 3310-3322.

86 Kuhn, A. and Grummt, I. (1992) Dual rôle of the nucleolar transcription factor UBF: trans-activator and antirepressor. Proc. Natl. Acad. Sci. USA 89, 7340-7344.

87 Smith, S. D., O'Mahony, D. J., Kinsella, B. J. and Rothblum, L. I. (1993) Transcription from the rat 45S ribosomal DNA promoter does not require the factor UBF. Gène Express. 3, 229-236.

88 Copenhaver, G. P., Putnam, C. D., Denton, M. L. and Pi­kaard, C. S. ( 1994) The RNA polymerase I transcription fac­tor UBF is a sequence-tolerant HMG-box protein that can recognize structured nucleic acids. Nucleic Acids Res. 22, 2651-2657.

89 Friedrich, J. K, Panov, K. I., Cabart, P., Russell, J. and Zomerdijk, J. C. (2005) TBP-TAF complex SL1 directs RNA polymerase I pre-initiation complex formation and stabilizes upstream binding factor at the rDNA promoter. J. Biol. Chem. 280,29551-29558.

90 Chen, D. and Huang, S. (2001) Nucleolar components in-volved in ribosome biogenesis cycle between the nucleolus

Cell. Mol. Life Sci. Vol. 64,2007 Review Article 45

and nucleoplasm in interphase cells. J. Cell Biol. 153, 169-176.

91 Dundr, M., Hoffmann-Rohrer, U, Hu, Q., Grummt, L, Roth-blum, L. L, Phair, R. D. and Misteli, T. (2002) A kinetic frame-work for a mammalian RNA polymerase in vivo. Science 298, 1623-1626.

92 O'Sullivan, A. C , Sullivan, G. J. and McStay, B. (2002) UBF binding in vivo is not restricted to regulatory séquences within the vertebrate ribosomal DNA repeat. Mol. Cell. Biol. 22, 657-668.

93 Chen, D., Belmont, A. S. and Huang, S. (2004) Upstream bind­ing factor association induces large-scale chromatin decon-densation. Proc. Natl. Acad. Sci. USA 101, 15106-15111.

94 Jordan, P., Mannervik, M., Tora, L. and Carmo-Fonseca, M. (1996) In vivo évidence that TATA-binding protein/SLl colocalizes with UBF and RNA polymerase I when rRNA synthesis is either active or inactive. J. Cell Biol. 133, 225-234.

95 Mais, C , Wright, J. E., Prieto, J. L., Raggett, S. L. and Mc­Stay, B. (2005) UBF-binding site arrays form pseudo-NORs and sequester the RNA polymerase I transcription machinery. Gènes Dev. 19,50-64.

96 Stefanovsky, V Y., Langlois, F., Gagnon-Kugler, T., Rothblum, L. I. and Moss, T. (2006) Growth factor signaling régulâtes elongation of RNA polymerase I transcription in mammals via UBF phosphorylation and chromatin remodeling. Mol. Cell 21,629-^39.

97 Aprikian, P., Moorefield, B. and Reeder, R. H. (2001) New model for the yeast RNA polymerase I transcription cycle. Mol. Cell. Biol. 21,4847-4855.

98 Lin, C. W, Moorefield, B., Payne, J., Aprikian, P., Mitomo, K. and Reeder, R. H. (1996) A novel 66-kilodalton protein com­plexes with Rrn6, Rrn7, andTATA-binding protein to promote polymerase I transcription initiation in Sacchammyces cerevi­siae. Mol. Cell. Biol. 16,6436-6443.

99 Steffan, J. S., Keys, D. A., Dodd, J. A. and Nomura, M. (1996) The rôle of TBP in rDNA transcription by RNA polymerase I in Saccharomyces cerevisiae: TBP is required for upstream activation factor-dependent recruitment of core factor. Gènes Dev. 10,2551-2563.

100 Steffan, J. S., Keys, D. A., Vu, L. and Nomura, M. (1998) In­teraction of TATA-binding protein with upstream activation factor is required for activated transcription of ribosomal DNA by RNA polymerase I in Saccharomyces cerevisiae in vivo. Mot. Cell. Biol. 18,3752-3761.

101 Keener, J., Josaitis, C. A., Dodd, J. A. and Nomura, M. (1998) Reconstitution of yeast RNA polymerase 1 transcription in vi­tro from purif ied components - TATA binding protein is not required for basai transcription. J. Biol. Chem. 273, 33795-33802.

102 Keys, D. A., Lee, B. S., Dodd, J. A., Nguyen.T. T., Vu, L., Fan-tino, E., Burson, L. M., Nogi, Y. and Nomura, M. (1996) Mul-tiprotein transcription factor UAF interacts with the upstream élément of the yeast RNA polymerase I promoter and forms a stable preinitiation complex. Gènes Dev. 10,887-903.

103 Aprikian, P., Moorefield, B. and Reeder, R. H. (2000) TATA binding protein can stimulate core-directed transcription by yeast RNA polymerase I. Mol. Cell. Biol. 20, 5269-5275.

104 Siddiqi, I., Keener, J., Vu, L. and Nomura, M. (2001) Rôle of TATA binding protein (TBP) in yeast ribosomal DNA tran­scription by RNA polymerase I: defects in the dual functions of transcription factor UAF cannot be suppressed by TBP. Mol. Cell. Biol. 21,2292-2297.

105 Oakes, M., Siddiqi, I., Vu, L., Aris, J. and Nomura, M. (1999) Transcription factor UAF, expansion and contraction of ribo­somal DNA (rDNA) repeats, and RNA polymerase switch in transcription of yeast rDNA. Mol. Cell. Biol. 19, 8559-8569.

106 Laferte, A., Favry, E., Sentenac, A., Riva, M., Caries, C. and Chedin, S. (2006) The transcriptional activity of RNA poly­

merase I is a key déterminant for the level of ail ribosome components. Gènes Dev. 20, 2030-2040.

107 Gadal, O., Labarre, S., Boschiero, C. and Thuriaux, P. (2002) Hmol, an HMG-box protein, belongs to the yeast ribosomal DNA transcription System. EMBO J. 21, 5498-5507.

108 Hall, D. B., Wade, J. T. and Struhl, K. (2006) An HMG protein, Hmol, associâtes with promoters of many ribosomal protein gènes and throughout the rRNA gène locus in Saccharomyces cerevisiae. Mol. Cell. Biol. 26, 3672-3679.

109 Bateman, E. and Paule, M. R. ( 1986) Régulation of eukaryotic ribosomal RNA transcription by RNA polymerase modifica­tion. Cell 47,445-450.

110 Kownin, P., Bateman, E. and Paule, M. R. (1987) Eukaryotic RNA polymerase I promoter binding is directed by protein contacts with transcription initiation factor and is DNA sc-quence-independent. Cell 50, 693-699.

111 Bateman, E. and Paule, M. R. (1988) Events during eucaryotic rRNA transcription initiation and elongation: conversion from the closed to the open promoter complex requires nucleotide substrates. Mol. Cell. Biol. 8, 1940-1946.

112 Kownin, P., Bateman, E. and Paule, M. R. (1988) Effects of single-base substitutions within the Acanthamoeba castellanii rRNA promoter on transcription and on binding of transcrip­tion initiation factor and RNA polymerase I. Mol. Cell. Biol. 8, 747-753.

113 Radebaugh, C. A., Matthews, J. L., Geiss, G. K., Liu, F., Wong, J. M., Bateman, E., Camier, S., Sentenac, A. and Paule, M. R. (1994) TATA box-binding protein (TBP) is a constituent of the polymerase I-specific transcription ini­tiation factor TIF-IB (SL1) bound to the rRNA promoter and shows differential sensitivity to TBP-directed reagents in polymerase I, II, and III transcription factors. Mol. Cell. Biol. 14, 597-605.

114 Geiss, G. K., Radebaugh, C. A. and Paule, M. R. (1997) The fundamental ribosomal RNA transcription initiation factor-IB (TIF-IB, SL1, factor D) binds to the rRNA core promoter pri-marily by minor groove contacts. J. Biol. Chem. 272, 29243™ 29254.

115 Radebaugh, C. A., Kubaska, W. M., Hoffman, L. H., Stiffler, K. and Paule, M. R. (1998) A novel transcription initiation factor (TIF), TIF-IE, is required for homogeneous Acantham­oeba castellanii TIF-IB (SL1) to form a committed complex. J. Biol. Chem. 273,27708-27715.

116 Kahl, B. F., Li, H. and Paule, M. R. (2000) DNA melting and promoter clearance by eukaryotic RNA polymerase I. J. Mol. Biol. 299,75-89.

117 Al-Khouri, A. M. and Paule, M. R. (2002) A novel RNA polymerase I transcription initiation factor, TIF-IE, commits rRNA gènes by interaction with TLF-IB, not by DNA binding. Mol. Cell. Biol. 22, 750-761.

118 Bateman, E„ Iida, C. T., Kownin, P. and Paule, M. R. (1985) Footprinting of ribosomal RNA gènes by transcription initia­tion factor and RNA polymerase I. Proc. Natl. Acad. Sci. USA 82, 8004-8008.

119 Bric, A., Radebaugh, C. A. and Paule, M. R. (2004) Photo-cross-linking of the RNA polymerase I preinitiation and im­médiate postinitiation complexes: implications for promoter recruitment. J. Biol. Chem. 279,31259-31267.

120 Gokal, P. K., Mahajan, P. B. and Thompson, E. A. (1990) Hor­monal régulation of transcription of rDNA: formation of initi-ated complexes by RNA polymerase I in vitro. 1. Biol. Chem. 265,16234-16243.

121 Lofquist, A . K , Li, H., Imboden, M. A. and Paule, M. R. ( 1993) Promoter opening (melting) and transcription initiation by RNA polymerase I requires neither nucleotide beta.gamma hydrolysis nor protein phosphorylation. Nucleic Acids Res. 21, 3233-3238.

122 Schnapp, A. and Grummt, I. (1991) Transcription complex formation at the mouse rDNA promoter involves the stepwise

46 T. Moss et al. Ribosomal RNA Gènes

association of four transcription factors and RNA polymerase I. J. Biol. Chem. 266, 24588-24595.

123 Yamamoto, R. T.,Nogi,Y., Dodd, J. A. andNomura, M. (1996) RRN3 gène of Saccharomyces cerevisiae encodes an essential RNA polymerase I transcription factor which interacts with the polymerase independently of DNA template. EMBO J. 15, 3964-3973.

124 Milkereit, P. and Tschochner, 11. (1998) A specialized form of RNA polymerase I, essential for initiation and growth-depen-dent régulation of rRNA synthesis, is disrupted during tran­scription. EMBO J. 17, 3692-3703.

125 Fath, S., Milkereit, P., Peyroche, G., Riva, M„ Caries, C. and Tschochner, H. (2001) Differential rôles of phosphorylation in the formation of transcriptional active RNA polymerase I. Proc. Natl. Acad. Sci. USA 98, 14334-14339.

126 Bier, M., Fath, S. and Tschochner, H. (2004) The composition of the RNA polymerase I transcription machinery switches from initiation to elongation mode. FEBS. Lett. 564, 41-46.

127 Raffaelle, M., Kanin, E. L, Vogt, J., Burgess, R. R. and Ansari, A. Z. (2005) Holoenzyme switching and stochastic release of sigma factors from RNA polymerase in vivo. Mol. Cell 20, 357-366.

128 Kapanidis, A. N., Margeat, E., Laurence,T. A., Doose, S., Ho, S. O., Mukhopadhyay, J., Kortkhonjia, E., Mekler, V, Ebright, R. H. and Weiss, S. (2005) Rétention of transcription initia­tion factor sigma(70) in transcription elongation: single-mol-ecule analysis. Mol. Cell 20, 347-356.

129 Fath,S.,Kobor,M. S.,Philippi,A.,Greenblatt,J.andTschoch-ner, H. (2004) Dephosphorylation of RNA polymerase I by Fcplp is required for efficient rRNA synthesis. J. Biol. Chem. 279,25251-25259.

130 Moorefield, B., Greene, E. A. and Reeder, R. H. (2000) RNA polymerase I transcription factor Rrn3 is functionally con-served between yeast and human. Proc. Natl. Acad. Sci. USA 97,4724^1729.

131 Bodem, X, Dobreva, G., Hoffmann-Rohrer, U, Iben, S., Zent-graf, H., Delius, H., Vingron, M. and Grummt, I. (2000) TIF-IA, the factor mediating growth-dependent control of ribo­somal RNA synthesis, is the mammalian homolog of yeast Rrn3p. EMBO Rep. 1, 171-175.

132 Miller, G., Panov, K. I., Friedrich, J. K., Trinkle-Mulcahy, L., Lamond, A. I. and Zomerdijk, J. C. (2001) hRRN3 is essen­tial in the SL1 -mediated recruitment of RNA polymerase I to rRNA gène promoters. EMBO J. 20,1373-1382.

133 Cavanaugh, A. H., Hirschler-Laszkiewicz, I., Hu, Q., Dundr, M., Smink, T., Misteli, T. and Rothblum, L. I. (2002) Rrn3 phosphorylation is a regulatory checkpoint for ribosome bio-genesis. J. Biol. Chem. 277,27423-27432.

134 Zhao, J., Yuan, X., Frodin, M. and Grummt, I. (2003) ERK-de-pendent phosphorylation of the transcription initiation factor TIF-IA is required for RNA polymerase I transcription and cell growth. Mol. Cell 11,405-413.

13 5 Dvir, A. (2002) Promoter escape by RNA polymerase II. Bio-chim. Biophys. Acte 1577,208-223.

136 Panov, K. I., Friedrich, J. K. and Zomerdijk, J. C. (2001) A step subséquent to preinitiation complex assembly at the ri­bosomal RNA gène promoter is rate limiting for human RNA polymerase I-dependent transcription. Mol. Cell. Biol. 21, 2641-2649.

137 Paul, B. J., Ross, W., Gaal, T. and Gourse, R. L. (2004) rRNA transcription in Escherichia coll. Annu Rev Genêt 38, 749-770.

138 Sims, R. J., m, Belotserkovskaya, R. and Reinberg, D. (2004) Elongation by RNA polymerase II: the short and long of it. Gènes Dev. 18,2437-2468.

139 Rudra, D. and Warner, J. R. (2004) What better measure than ribosome synthesis? Gènes Dev. 18,2431-2436.

140 Jorgensen, P. and Tyers, M. (2004) How cells coordinate growth and division. Curr. Biol. 14, R1014-R1027.

141 Condon, C , Squires, C. and Squires, C. L. (1995) Control of rRNA transcription in Escherichia coli. Microbiol. Rev. 59, 623-645.

142 Miura, A., Krueger, J. H., Itoh, S., de Boer, H. A. and Nomura, M. (1981) Growth-rate-dependent régulation of ribosome synthesis in E. coli: expression of the lacZ and galK gènes fused to ribosomal promoters. Cell 25, 773-782.

143 Gourse, R. L., de Boer, H. A. and Nomura, M. (1986) DNA déterminants of rRNA synthesis in E. coli: growth rate dé­pendent régulation, feedback inhibition, upstream activation, antitermination. Cell 44, 197-205.

144 Sarmientos, P. and Cashel, M. (1983) Carbon storvation and growth rate-dependent régulation of the Escherichia coli ribo­somal RNA promoters: differential control of dual promoters. Proc. Natl. Acad. Sci. USA 80, 7010-7013.

145 Warner, J. R. (1989) Synthesis of ribosomes in Saccharomy­ces cerevisiae. Microbiol. Rev. 53, 256-271.

146 Warner, J. R. (1977) In the absence of ribosomal RNA syn­thesis, the ribosomal proteins of HeLa cells are synthesized normally and degraded rapidly. J. Mol. Biol. 115, 315-333.

147 Zhao, Y, Sohn, J. H. and Warner, J. R. (2003) Autoregulation in the biosynthesis of ribosomes. Mol. Cell. Biol. 23, 699-707.

148 Warner, J. R. and Gorenstein, C. (1978) Yeast lias a true strin-gent response. Nature 275, 338-339.

149 Stefanovsky, V Y, Pelletier, G., Hannan, R., Gagnon-Kugler, T., Rothblum, L. I. and Moss, T. (2001) An immédiate re­sponse of ribosomal transcription to growth factor stimulation in mammals is mediated by ERK phosphorylation of UBF. Mol. Cell 8, 1063-1073.

150 Maden, B. E., Vaughan, M. H., Warner, J. R. and Darnell, J. E. (1969) Effects of valine deprivation on ribosome formation in HeLa cells. J. Mol. Biol. 45,265-275.

151 Vaughan, M. H., Soeiro, R., Warner, J. R. and Darnell, J. E. (1967) The effects of methionine deprivation on ribosome synthesis in HeLa cells. Proc. Natl. Acad. Sci. USA 58,1527-1534.

152 Grummt, L, Smith, V A. and Grummt, F. (1976) Amino acid starvation affects the initiation frequency of nucleolar RNA polymerase. Cell 7,439-445.

153 Gokal, P. K., Cavanaugh, A. H. and Thompson, E. A. Jr (1986) The effects of cycloheximide upon transcription of rRNA, 5 S RNA, and tRNA gènes. J. Biol. Chem. 261,2536-2541.

154 Cavanaugh, A. H. and Thompson, E. A. Jr (1985) Hormonal régulation of transcription of rDNA: glucocorticoid effects upon initiation and elongation in vitro. Nucleic Acids Res. 13, 3357-3369.

155 Cavanaugh, A. H., Gokal, P. K., Lawther, R. P. and Thompson, E. A. Jr (1984) Glucocorticoid inhibition of initiation of tran­scription of the DNA encoding rRNA (rDNA) in lymphosar-coma PI798 cells. Proc. Natl. Acad. Sci. USA 81,718-721.

156 Tower, J. and Sollner-Webb, B. (1987) Transcription of mouse rDNA is regulated by an activated subform of RNA poly­merase I. Cell 50, 873-883.

157 Buttgereit, D , Pflugfelder, G. and Grummt, I. (1985) Growth-dependent régulation of rRNA synthesis is mediated by a transcription initiation factor (TIF-IA). Nucleic Acids Res. 13,8165-8180.

158 Stevens, A. R. and Pachler, P. F. (1973) RNA synthesis and turnover during density-inhibited growth and encystment of Acanthamoeba castellanii. J. Cell Biol. 57, 525-537.

159 Grummt, I. and Grummt, F. ( 1976) Control of nucleolar RNA synthesis by the intracellular pool sizes of ATP and GTP. Cell 7,447^153.

160 James, M. J. and Zomerdijk, J. C. (2004) Phosphatidylinositol 3-kinase and mTOR signaling pathways regulate RNA poly­merase I transcription in response to IGF-1 and nutrients. J. Biol. Chem. 279, 8911-8918.

Ce». Mol. Life Sci. Vol. 64, 2007 Review Article 47

161 Hannan, K. M., Rothblum, L. I. and Jefferson, L. S. (1998) Régulation of ribosomal DNA transcription by insulin. Am. J. Physiol. Cell Physiol. 275, C130-C138.

162 Sandmeier, J. J., French, S., Osheim, Y., Cheung, W. L., Gallo, C. M., Beyer, A. L. and Smith, J. S. (2002) RPD3 is required for the inactivation of yeast ribosomal DNA gènes in station-ary phase. EMBO J. 21,4959-4968.

163 Claypool, J. A., French, S. L., Johzuka, K., Eliason, K., Vu, L., Dodd, J. A., Beyer, A. L. and Nomura, M. (2004) Tor pathway régulâtes Rrn3p-dependent recruirment of yeast RNA poly­merase I to the promoter but does not participate in altération of the number of active gènes. Mol. Biol. Cell 15,946-956.

164 Stefanovsky, V Y. and Moss, T. (2006) Régulation of rRNA synthesis in human and mouse cells is not determined by changes in active gène count. Cell. Cycle 5, 735-739.

165 Paule, M. R., Iida, C. T., Perna, P. J., Harris, G. H., Knoll, D. A. and D'Alessio, J. M. (1984) In vitro évidence that eukaryotic ribosomal RNA transcription is regulated by modification of RNA polymerase I. Nucleic Acids Res. 12, 8161-8180.

166 Cavanaugh, A. II. and Thompson, E. A. Jr (1986) Hormonal régulation of transcription of rDNA: initiation of transcription by RNA polymerase I in vitro. J. Biol. Chem. 261, 12738-12744.

167 Mahajan, P. B., Gokal, P. K. and Thompson, E. A. (1990) Hor­monal régulation of transcription of rDNA: the rôle of TFIC in formation of initiation complexes. J. Biol. Chem. 265,16244-16247.

168 Mahajan, P. B. and Thompson, E. A. Jr (1992) Copurification of RNA polymerase I and the glucocorticoid-regulated tran­scription factor IC. Protein Express Purif. 3,410-416.

169 Mahajan, P. B. and Thompson, E. A. (1991) Glucocorticoid régulation of rRNA synthesis. Mol. Cell. Biochem. 104,195-200.

170 Mayer, C , Bierhoff, H. and Grummt, I. (2005) The nucleolus as a stress sensor: JNK2 inactivates the transcription factor TIF-IA and down-regulates rRNA synthesis. Gènes Dev. 19, 933-941.

171 Mayer, C , Zhao, J., Yuan, X. and Grummt, I. (2004) mTOR-dependent activation of the transcription factor TIF-IA links rRNA synthesis to nutrient availability. Gènes Dev. 18, 423-434.

172 Bouche, G., Gas, N., Prats, H., Baldin, V, Tauber, J.-R, Teissie, J. and Amalric, F. ( 1987) Basic fibroblast growth factor enters the nucleolus and stimulâtes the transcription of ribosomal gènes in ABAE cells undergoing Go-Gl transition. Proc. Natl. Acad. Sci. USA 84,6770-6774.

173 Sheng, Z., Liang, Y, Lin, C. Y, Comai, L. and Chirico, W. J. (2005) Direct régulation of rRNA transcription by fibroblast growth factor 2. Mol. Cell. Biol. 25,9419-9426.

174 Muth, V, Nadaud, S., Grummt, I. and Voit, R. (2001) Acety-lation of TAF(I)68, a subunit of TIF-IB/SL1, activâtes RNA polymerase I transcription. EMBO J. 20,1353-1362.

175 Heix, J., Vente, A., Voit, R., Budde, A., Michaelidis, T. M. and Grummt, I. (1998) Mitotic silencing of human rRNA synthesis: inactivation of the promoter selectivity factor SL1 by cdc2/cyclin B-mediated phosphorylation. EMBO J. 17, 7373-7381.

176 Zhang, C , Comai, L. and Johnson, D. L. (2005) PTEN re­presses RNA polymerase I transcription by disrupting the SL1 complex. Mol. Cell. Biol. 25, 6899-6911.

177 Hannan, R. D., Cavanaugh, A., Hempel, W. M., Moss, T. and Rothblum, L. (1999) Identification of a mammalian RNA polymerase I holoenzyme containing components of the DNA repair/replication System. Nucleic Acids Res. 27,3720-3727.

178 Hannan, R. D., Hempel, W. M., Cavanaugh, A., Arino, T., Dimitrov, S. I., Moss, T. and Rothblum, L. (1998) Aff inity pu­rification of mammalian RNA polymerase I - identification of an associated kinase. J. Biol. Chem. 273,1257-1267.

179 Saez-Vasquez, J., Albert, A. C , Earley, K. and Pikaard, C. S. (2003) Purification and transcriptional analysis of RNA polymerase I holoenzymes from broccoli (Brassica olera-ceà) and frog (Xenopus laevis). Methods Enzymol. 370, 121-138.

180 Seither, P., Iben, S. and Grummt, I. (1998) Mammalian RNA polymerase I exists as a holoenzyme with associated basai transcription factors. J. Mol. Biol. 275,43-53.

181 Fath, S., Milkereit, P., Podtelejnikov, A. V, Bischler, N., Schultz, P., Bier, M., Mann, M. and Tschochner, H. (2000) Association of yeast RNA polymerase 1 with a nucleolar sub­structure active in rRNA synthesis and processing. J. Cell Biol. 149, 575-589.

182 Fuhrman, S.A. and GUI, G. N. (1975) Adrenocorticotropic hormone stimulation of adrenal RNA polymerase I and III ac­tivités: nucleotide incorporation into internai positions and 3' chain termini. Biochemistry 14, 2925-2933.

183 Dauphinais, C. (1981) The control of ribosomal RNA tran­scription in lymphocytes: évidence that the rate of chain elon-gation is the limiting factor. Eur. J. Biochem. 114,487-492.

184 Dembinski, T. C. and Bell, P. A. ( 1984) Glucocorticoids mod-ify the rate of ribosomal RNA synthesis in rat thymus cells by regulating the polymerase elongation rate. J. Steroid Biochem. 21,497-504.

185 French, S. L., Osheim, Y. N., Cioci, F, Nomura, M. and Beyer, A. L. (2003) In exponentially growing Saccharomyces cerevi-siae cells, rRNA synthesis is determined by the summed RNA polymerase I loading rate rather than by the number of active gènes. Mol. Cell. Biol. 23, 1558-1568.

186 Bazett-Jones, D. P., Leblanc, B., Herfort, M. and Moss, T. (1994) Short-range DNA looping by the Xenopus HMG-box transcription factor, xUBF. Science 264, 1134-1137.

187 Stefanovsky, V Y, Pelletier, G., Bazett-Jones, D. P. and Moss, T. (2006) ERK modulâtes DNA bending and enhancesome structure by phosphorylating HMG1-boxes 1 and 2 of the RNA polymerase I transcription factor UBF. Biochemistry 45, 3626-3634.

188 Gallagher, J.E. G, Dunbar, D. A., Granneman, S., Mitchell, B. M., Osheim, Y, Beyer, A. L. and Baserga, S. J. (2004) RNA polymerase I transcription and pre-rRNA processing are linked by spécifie SSU processome components. Gènes Dev. 18,2506-2517.

189 Cavanaugh, A. H., Hempel, W. M., Taylor, L. J., Rogalsky, V, Todorov, G. and Rothblum, L. I. (1995) Activity of RNA polymerase I transcription factor UBF blocked by Rb gène product. Nature 374, 177-180.

190 Hannan, K. M., Hannan, R. D., Smith, S. D, Jefferson, L. S., Lun, M. and Rothblum, L. I. (2000) Rb and pi30 regulate RNA polymerase I transcription: Rb disrupts the interaction between UBF and SL-1. Oncogene 19,4988-4999.

191 Budde, A. and Grummt, I. (1999) p53 represses ribosomal gène transcription. Oncogene 18,1119-1124.

192 Zhai, W. and Comai, L. (2000) Repression of RNA polymerase I transcription by the tumor suppressor p53. Mol. Cell. Biol. 20,5930-5938.

193 Pelletier, G., Stefanovsky, V.Y., Faubladier, M., Hirschler-Laszkiewicz, I., Savard, J., Rothblum, L. L, Côté, J. and Moss, T. (2000) Compétitive recruirment of CBP and Rb-HDAC régulâtes UBF acetylation and ribosomal transcription. Mol. Cell 6,1059-1066.

194 Liu, Y. Z., Thomas, N.S. B. and Latchman, D. S. (1999) CBP associâtes with the p42/p44 MAPK enzymes and is phos-phorylated following NGF treatment. Neuroreport 10, 1239-1243.

195 Ait-Si-Ali, S., Carlisi, D., Ramirez, S., Upegui-Gonzalez, L. C , Duquet, A., Robin, P., Rudkin, B., Harel-Bellan, A. and Trouche, D. (1999) Phosphorylation by p44 MAP kinase/ ERK1 stimulâtes CBP histone acetyl transferase activity in vitro. Biochem. Biophys. Res. Commun. 262,157-162.

48 T. Moss et al. Ribosomal RNA Gènes

196 Hirschler-Laszkiewicz, I., Cavanaugh, A., Hu, Q., Catania, J., Avantaggiati, M. L. and Rothblum, L. I. (2001) The rôle of acetylation in rDNA transcription. Nucleic Acids Res. 29, 4114-4124.

197 Halkidou, K., Logan, I. R., Cook, S., Neal, D. E. and Robson, C. N. (2004) Putative involvement of the histone acetyltrans-ferase TipôO in ribosomal gène transcription. Nucleic Acids. Res. 32, 1654-1665.

198 Bertwistle, D., Sugimoto, M. and Sherr, C. J. (2004) Physical and functional interactions of the Arf tumor suppressor pro-tein with nucleophosmin/B23. Mol. Cell. Biol. 24,985-996.

199 Sugimoto, M., Kuo, M. L., Roussel, M. F. and Sherr, C. J. (2003) Nucleolar Arf tumor suppressor inhibits ribosomal RNA processing. Mol. Cell 11, 415—424.

200 Itahana, K., Bhat, K. P., Jin, A., Itahana, Y., Hawke, D„ Kobayashi, R. and Zhang, Y. (2003) Tumor suppressor ARF dégrades B23, a nucleolar protein involved in ribo-some biogenesis and cell prolifération. Mol. Cell 12, 1151-1164.

201 Grewal, S. S., Li, L., Orian, A., Eisenman, R. N., Edgar, B. A., Toshima, J., Toshima, J. Y, Martin, A. C. and Drubin, D. G. (2005) Myc-dependent régulation of ribosomal RNA syn-thesis during Drosophila development; phosphoregulation of Arp2/3-dependent actin assembly during receptor-mediated endocytosis. Nat. Cell Biol. 7, 295.

202 Arabi, A., Wu, S., Ridderstrale, K., Bierhoff, H., Shiue, C , Fatyol, K.., Fahlen, S., Hydbring, P., Soderberg, O., Grummt, I., Larsson, L.-G. and Wright, A.P. H. (2005) c-Myc associ­âtes with ribosomal DNA and activâtes RNA polymerase I transcription. Nat. Cell Biol. 7, 303-310.

203 Grandori, C , Gomez-Roman, N., Felton-Edkins, Z.A., Ngouenet, C , Galloway, D. A., Eisenman, R. N. and White, R. X (2005) c-Myc binds to human ribosomal DNA and stimu­lâtes transcription of rRNA gènes by RNA polymerase I. Nat. Cell Biol. 7, 311-318.

204 Poortinga, G., Hannan, K. M., Snelling, H., Walkley, C. R., Jenkins, A., Sharkey, K.., Wall, M., Brandenburger, Y, Palat-sides, M., Pearson, R. B., McArthur, G. A. and Hannan, R. D. (2004) MAD1 and c-MYC regulate UBF and rDNA transcrip­tion during granulocyte differentiation. EMBO X 23, 3325-3335.

205 Haggerty, T. J., Zeller, K. L, Osthus, R. C , Wonsey, D. R. and Dang, C. V (2003) A strategy for identifying transcription fac-tor binding sites reveals two classes of genomic c-Myc target sites. Proc. Natl. Acad. Sci. USA 100, 5313-5318.

206 Williams, N. S., Gaynor, R. B., Scoggin, S., Verma, U, Go-kaslan, T., Simmang, C , Fleming, X, Tavana, D , Frenkel, E. and Becerra, C. (2003) Identification and validation of gènes involved in the pathogenesis of colorectal cancer using cDNA microarrays and RNA interférence. Clin. Cancer Res. 9 ,931-946.

207 Watson, X D., Oster, S. K., Shago, M., Khosravi, F. and Penn, L. Z. (2002) Identifying gènes regulated in a Myc-dependent manner. J. Biol. Chem. 277,36921-36930.

208 Menssen, A. and Hermeking, H. (2002) Characterization of the c-MYC-regulated transcriptome by SAGE: identification and analysis of c-MYC target gènes. Proc. Natl. Acad. Sci. USA 99,6274-6279.

209 Zeller, K. I., Haggerty, T. X, Barrett, X F., Guo, Q., Wonsey, D. R. and Dang, C. V (2001) Characterization of nucleophos-min (B23) as a Myc target by scanning chromatin immunopre-cipitation. X Biol. Chem. 276,48285^18291.

210 Schuldiner, O. and Benvenisty, N. (2001) A DNA microarray screen for gènes involved in c-MYC and N-MYC oncogenesis in human tumors. Oncogene 20, 4984-4994.

211 Frank, S. R., Schroeder, M., Fernandez, P., Taubert, S. and Amati, B. (2001) Binding of c-Myc to chromatin médiates mitogen-induced acetylation of histone H4 and gène activa-tion. Gènes Dev. 15, 2069-2082.

212 Boon, K., Caron, H. N., van Asperen, R., Valentijn, L., Hermus, M. C , van Sluis, P., Roobeek, L, Weis, I., Voûte, P. A., Schwab, M. and Versteeg, R. (2001) N-myc enhances the expression of a large set of gènes functioning in ribo-some biogenesis and protein synthesis. EMBO X 20, 1383-1393.

213 Guo, Q. M., Malek, R. L„ Kim, S., Chiao, C , He, M., Ruffy, M., Sanka, K., Lee, N. H., Dang, C. V and Liu, E. T. (2000) Identification of c-myc responsive gènes using rat cDNA mi­croarray. Cancer Res. 60, 5922-5928.

214 Coller, H. A., Grandori, C , Tamayo, P., Colbert, T., Lander, E. S., Eisenman, R. N. and Golub, T. R. (2000) Expression analysis with oligonucleotide microarrays reveals that MYC régulâtes gènes involved in growth, cell cycle, signaling, and adhésion. Proc. Natl. Acad. Sci. USA 97, 3260-3265.

215 Schlosser, L, Holzel, M., Murnseer, M., Burtscher, H., Weidle, U. H. and Eick, D. (2003) A rôle for c-Myc in the régulation of ribosomal RNA processing. Nucleic Acids Res. 31, 6148 6156.

216 Brandenburger, Y, Jenkins, A., Autelitano, D. X and Hannan, R. D. (2001) Increased expression of UBF is a critical déter­minant for rRNA synthesis and hypertrophie growth of car-diac myocytes. FASEB X 15, NIL146-NIL171.

217 Hannan, R., Stefanovsky, V, Arino, T., Rothblum, L. and Moss, T. (1999) Cellular régulation of ribosomal DNA transcription: both rat and Xenopus UBFI stimulate rDNA transcription in 3T3 fibroblasts. Nucleic Acids Res. 27, 1205-1213.

218 Hannan, R. D., Stefanovsky, V, Taylor, L., Moss, T. and Roth­blum, L. I. (1996) Overexpression of the transcription factor UBFI is sufficient to increase ribosomal DNA transcription in neonatal cardiomyocytes: implications for cardiac hyper-trophy. Proc. Natl. Acad. Sci. USA 93, 8750-8755.

219 Caries, C. and Riva, M. (1998) Yeast RNA polymerase I sub-units and gènes. In: Transcription of Ribosomal Gènes by Eukaryotic RNA polymerase I (Paule, M. R. Ed.), pp. 9-38. Landes Bioscience, Austin.

220 Panov, K. L, Panova, T. B., Gadal, O., Nishiyama, K., Saito, T., Russell, X and Zomerdijk, X C. (2006) RNA polymerase I-specific subunit CAST/hPAF49 has a rôle in the activation of transcription by upstream binding factor. Mol. Cell. Biol. 26, 5436-5448.

221 ICuehn, M. and Arnheim, N. (1983) Nucleotide séquence of the geneticaly labile repeated éléments 5' to the origin of mouse rRNA transcription. Nucleic Acids Res. 11,211-224.

222 Chikaraishi, D. M., Buchanan, L., Danna, K. X and Har-rington, C. A. (1983) Genomic organisation of rat rDNA. Nucleic Acids Res. 11, 6439-6452.

223 Lang, W. H., Morrow, B. E., Ju, Q., Warner, X R. and Reeder, R. H. (1994) A model for transcription termination by RNA polymerase I. Cell 79, 527-534.

224 Melekhovets, Y. E, Shwed, P. S. and Nazar, R. N. (1997) In vivo analyses of RNA polymerase I termination in Schizosac-charomyces pombe. Nucleic Acids Res. 25,5103-5109.

225 Jones, M. H., Learned, R. M. and Tjian, R. (1988) Analysis of clustered point mutations in the human ribosomal RNA gène promoter by transient expression in vivo. Proc. Natl. Acad. Sci. USA 85,669-673.

226 Learned, R. M., Smale, S. T., Haltiner, M. M. and Tjian, R. (1983) Régulation of ribosomal RNA transcription. Proc. Natl. Acad. Sci. USA 80, 3558-3562.

227 Haltiner, M. M., Smale, S. T. and Tjian, R. (1986) Two dis­tinct promoter éléments in the human rRNA gène identified by linker scanning mutagenesis. Mol. Cell. Biol. 6,227-235.

228 Bell, S. P., Jantzen, H.-M. and Tjian, R. (1990) Assembly of alternative multiprotein complexes directs rRNA promoter se-lectivity. Gènes Dev. 4, 943-954.

229 Clos, X, Normann, A., Ohrlein, A. and Grummt, I. (1986) The core promoter of mouse rDNA consists of two functionally distinct domains. Nucleic Acids Res. 14, 7581-7595.

Coll. Mol. Life Sci. Vol. 64, 2007 Review Article 49

230 Henderson, S. L. and Sollner-Webb, B. (1990) The mouse ri-bosomal DNA promoter has more stringent requirements in vivo than in vitro. Mol. Cell. Biol. 10, 4970-4973.

231 Cassidy, B., Haglund, R. and Rothblum, L. I. (1987) Régions upstream from the core promoter of the rat ribosomal gène are required for the formation of a stable transcription initiation complex by RNA polymerase I in vitro. Biochim. Biophys. Acta 909, 133-144.

232 Xie, W., O'Mahony, D. J., Smith, S. D., Lowe, D. and Roth­blum, L. I. (1992) Analysis of the rat ribosomal DNA pro­moter: characterization of linker-scanning mutants and of the binding of UBF. Nucleic Acids Res. 20, 1587-1592.

2.33 Smith, S. D., Oriahi, E., Lowe, D., Yang-Yen, H.-F., O'Mahony, D., Rose, K., Chen, K. and Rothblum, L. I. (1990) Charac­terization of factors that direct transcription of rat ribosomal DNA. Mol. Cell. Biol. 10, 3105-3116.

234 Windle, J. J. and Sollner-Webb, B. ( 1986) Two distant and pre-cisely positioned domains promote transcription of Xenopus iaevis rRNA gènes: analysis with linker-scanning mutants. Mol. Cell. Biol. 6,4585-4593.

235 Windle, J. and Sollner-Webb, B. (1986) Upstream domains of the Xenopus Iaevis rDNA promoter are revealed in microin-jected oocytes. Mol. Cell. Biol. 6, 1228-1234.

236 Reeder, R. H., Pennock, D , McStay, B., Roan, J., Tolentino, E. and Walker, P. (1987) Linker scanner mutagenesis of the Xenopus Iaevis ribosomal gène promoter. Nucleic Acids Res. 15,7429-7441.

237 Firek, S., Read, C , Smith, D. R. and Moss, T. (1990) Point mutation analysis of the Xenopus Iaevis RNA polymerase 1 core promoter. Nucleic Acids Res. 18, 105-109.

238 Read, C , Larose, A. M., Leblanc, B., Bannister, A. J., Firek, S., Smith, D. R. and Moss, T. (1992) High resolution studies of the Xenopus Iaevis ribosomal gène promoter in vivo and in vitro. J. Biol. Chem. 267, 10961-10967.

239 Kohorn, B. D. and Rae, P. M. (1983) A component of Dro-sophila RNA polymerase I promoter lies within the rRNA transcription unit. Nature 304, 179-181.

240 Iida, C.T., Kownin, P. and Paule, M. R. (1985) Ribosomal RNA transcription: proteins and DNA séquences involved in preinitiation complex formation. Proc. Natl. Acad. Sci. USA 82, 1668-1672.

241 Doelling,J. H.andPikaard,C. S.(1995)Theminimalribosomal RNA gène promoter of Arabidopsis thaliana includes a critical élément at the transcription initiation site. Plant J. 8, 683-692.

242 Verbeet, M. P., Klootwijk, J., Van Heenkhuizen, H., Fontijn, R. D., Vreugdenhil, E. and Planta, R. (1984) A conserved sé­quence élément is présent around the transcription initiation site for RNA polymerase I in Saccharomycetoideae. Nucleic Acids Res. 12,1137-1148.

243 Murphy, F. V, Sweet, R. M. and Churchill, M. E. (1999) The structure of a chromosomal high mobility group protein-DNA complex reveals sequence-neutral mechanisms important for non-sequence-specific DNA récognition. EMBO J. 18,6610-6618.

Annexe 2 : Un nouveau paradigme pour la régulation de la transcription ribosomale chez les mammifères;

«A new paradigm for the régulation of the mammalian ribosomal RNA gènes»

Tom Moss, Victor Stefanovsky, Frédéric Langlois and Thérèse Gagnon-Kugler

Cancer Research Centre and Department of Médical Biology, Laval University, CHUQ-

HDQ, Pavillon St Patrick, 9 rue McMahon, Québec, G1R 3S3, Canada.

Cet article a été publié en décembre 2006 dans « Biochemical Society Transaction », vol 34

(6) : 1079-1081 et il est reproduit avec la permission de l'éditeur Portland Press. © the

Biochemical Society.

Molecular Basis of Transcription 1079

A new paradigm for the régulation of the mammalian ribosomal RNA gènes T. Moss1, V. Stefanovsky, F. Langlois and T. Gagnon-Kugler Cancer Research Centre and Department of Médical Biology, Laval University, CHUQ-HDQ, Pavillon St. Patrick, 9 rue McMahon, QC, Canada G1R 353

Abstract Ribosome assembly occurs co-transcriptionally on the rRNA gènes. This process requires the co-ordinated expression and assembly of many hundreds of proteins and is finely tuned to cell and organism growth. Co-ordinate régulation of the rRNA gènes and the ribosomal protein gènes is therefore essential for high-fidelity ribosome assembly. Récent work shows that rRNA gène transcription is regulated at the level of elongation via the mitogen-activated protein kinase pathway. We argue that this may provide an explanation for the high fidelity of ribosome assembly.

Introduction V/hen so much emphasis is being placed on the rôle of protein coding gènes, it is a sobering thought that a eukaryotic cell must provide on average ten ribosomes for every mRNA it synthesizes. Cytoplasmic ribosomes account for approx. 80% of total cellular RNA. Each ribosome is some 4 MDa in weight and consists of four RNAs, the 18 S, 5.8 S, 28 S and 5 S rRNAs, and 80 or so proteins. Hence, ribosome biogenesis is the major limit to growth and prolifération. In the présent paper, we will briefly review the mechanisms that co-ordinate production of the rRNAs with growth in mammalian cells.

The rRNA gènes encode the three largcst RNAs of the ribosome, as part of a single precursor, in mammals called the 47 S rRNA. Initial assembly of the ribosome occurs co-transcriptionally with 47 S rRNA synthesis, leading to a 90 S precursor particle, a process elegantly visualized in the 'Miller-Spread' électron micrographs [1]. Structural studies of the ribosome suggest that this 5' to 3' stepwise co-tran-scriptional assembly is important in establishing the complex secondary structure of the mature rRNAs and in positioning the r-proteins [2,3].

Transcription initiation on the rRNA gènes In eukaryotes, a dedicated set of proteins is used to transcribe the rRNA gènes, allowing the cell to regulate thèse gènes independently of the rest of the génome. The œ-amanatin-resistant DNA-dependent RPI (RNA polymerase I) is dedicated solely to the transcription of the rRNA gènes. In mammals, formation of an RPI pre-initiation complex requires SL1 (selectivity factor 1; or mTIF-IB) and an HMG1-box [DNA-binding motif homologous with those of HMG1 (high mobility group protein 1)] DNA-binding protein, UBF

Key woi . lv growth lactor, mammalian cell, ribosomal RNA gène, RNA polymerase I, rRNA,

transcription elongation.

Abbrevlatlons used: CREB, cAMP-response-element-binding protein, CBP, CREB-binding protein;

ERK, exttacellular-signal-regulated kinase; HMGi, high mobility group protein 1; HMGl-box, DNA-

binding motif homologous with those of HMG1; MAPK, mitogen-activated protein kinase; RPI,

RNA polymerase I; RNP, ribonucleoproteln; SU, selectivity factor 1; UBF, upstream binding lactor. 1 To whom correspondence should be addressed (email [email protected]).

(upstream binding factor), see Figure 1. Originally it was believed that UBF specifically bound the upstream and/or core promoter régions, creating a situation propitious for the SLl complex to bind and form a 'stable' pre-initiation com­plex [4]. In vitro, this complex is able to recruit RPI and, in the présence of nucleotide triphosphates, to initiate transcrip­tion. When it was discovered that the N-terminal half of UBF (core UBF) could form the 'enhancesome', a nucleoprotein structure that somewhat resembles the nucleosome, the finding led to a spéculative explanation for the co-operativity between UBF and SLl [5-8]. In this model, UBF juxtaposes the two promoter domains, presenting the SLl-binding sites on the same side of a DNA superhelix, see Figure 1. However, it was later found that UBF does not restrict itself to the RPI promoter but is found throughout the rDNA repeat [9]. This suggests that UBF defines a specialized rRNA gène chromatin and is consistent with its indiscriminate DNA-binding properties [7].

Growth factor régulation of the rRNA gènes Unlike protein coding gènes, expression of the rRNA gènes can only be regulated at four distinct levels: gène activation, transcription initiation, transcription elongation and rRNA dégradation. Since the rRNA gènes exist in tandem arrays of several hundred copies, the number of active gènes could in principle vary. However, in mammalian cell culture, several-fold changes in rRNA synthesis can be observed without a détectable change in active gène number [10]. Long-term sérum withdrawal and cycloheximide and glucocorticoid treatments hâve demonstrated that the ability of RPI to initiate transcription, as measured in vitro, is regulated via phosphorylation of mammalian Rrn3/TIF-IA, a protein essential for promoter-directed RPI initiation [11-14], see Figure 1. Thèse results led to claims that rRNA gène régul­ation occurred exclusively at the level of transcription initiation via régulation of Rrn3/TIF-IA [15].

Despite intense study of the régulation of RPI and Rrn3 activities, the first molecular pathway linking growth

O2006 Biochemical Society

1080 Biochemical Society Transactions (2006) Volume 34, paît 6

Figure 11 The rôle of UBF in rRNA gène transcription (A) A cartoon showing the now classic paradigm of RPI initiation in

mammals. Adapted from Cell, 109, I. Moss and V.Y. Slelanovsky,

'AI the center of eukaryolic life.', 545-548, © 200,', witli permission

from Elsevier. (B) A hypolhetical model of promoter structure showing

Iwo consécutive UBr DNA enhancesome complexes juxtaposing the

upstream control élément (UCF) and core promoter éléments. Data are

based on the low-resolulion enhancesome structure [6-8], Adapted from

(ell, 109, I. Moss and V.Y. Stefanovsky, 'At the center ofeukaryotic life.',

545-548, © 2002, wilh permission from Hsevier. (C) A new paradigm

lor RPI transcription régulation. Multiple adjacent enhancesomes arrest

elongating RPI. ERK and CBP combine to unlold the enhancesome,

allowing passage to the polymerase. Ihe DNA is shown in red, UBF

in dark blue, RPI and associated factors in light blue and S11 in orange.

TRP, TATA-box-binding protein; TAF, TBP-associated factor. Adapted from

Molecular Cell, 2 1 , V.Y. Stefanovsky, I. langlois, I. Gagnon-Kugler, L.l.

Rothblum and I. Moss, 'Growth factor signaling régulâtes elongation of

k'NA polymerase I transcription in mammals via Util phosphorylation

and r-chromatin remodeling.', 629-639, © 2006, wilh permission from

I Kevin

factor signalling to rRNA gène transcription came from a study of UBF [16], Responsc to stimulation of human and mouse cells by EGF (epidermal growth factor) and by direct activation of the Raf-MEK |MAPK (mitogen-activatcd protein kinase)/ERK (extracellular-signal-regulated kinasc) kinase]-ERK pathway was shown to require the phosphorylation of UBF. The two N-terminal HMG-boxes of UBF display consensus ERK phosphorylation sites on their DNA-binding surfaces. Phosphorylation of thèse sites was shown to bc required for growth factor stimulation of rRNA gène transcription, Given that UBF binds throughout the rRNA gènes, thèse results led us to spcculate that UBF might regulate transcription elongation [13,14,16].

Growth factors target RPI elongation If rRNA gène transcription is rcgulated solely via changes in the transcription initiation rate, the number of polymerase molécules engaged in transcribing the rRNA gènes should be directly proportional to the rate of rRNA synthesis. This led us to ask whether an increase in RPI loading on the rRNA gènes did in fact occur during growth factor induction of the rRNA gènes [10,17]. What wefound clearly contradicted a simple model of initiation régulation. Using both nuclear run-on and chromatin immunoprecipitation approaches, we established that growth factor and MAPK activation of rRNA synthesis in human and mouse cells does not correlate with a significant increase in the total number of RPI transcription complexes [10,17]. Indeed, it was difficult to detect any direct corrélation between rRNA synthesis levels and the RPI loadings on the gènes, strong suppression of synthesis often showing the highest loading levels. Clearly then, rRNA synthesis is not regulated simply by modulating the rate of transcription initiation. The obvious explanation was that RPI transcription elongation rates must also be regulated. This we confirmed using a novel approach to measure in vivo RPI elongation rates under différent conditions of growth stimulation [17]. We found that changes in RPI elongation rate quantitatively explain the full dynamic range of growth factor régulation.

To maintain the constant level of RPI engagement we observe over a wide range of rRNA synthesis rates, either (i) elongation and initiation must be exactly co-ordinatcly regulated or (ii) elongation must limit the rate of initiation of new transcripts. Of thèse two possibilities, the latter would appear much the simpler mechanistically, since by limiting elongation rates, initiation would be limited in exact proportion. Our observations that a block to MAPK activation gave an increase, albeit slight, in RPI loading while causing a drastic réduction in rRNA synthesis suggested that the reduced elongation rate was backing-up polymerases and preventing more rapid initiation 117]. Hence, it appeared from thèse results that elongation was dominant over initiation. In fact, we hâve now shown that the capacity of cell-free extracts to specifieally initiate UBF-independent RPI transcription in vitro does not change when the cells were stimulated by growth factor or MAPK activation was blocked

'• 2006 Biochemical Society

Molecular Basts of Transcription 1081

immediately prior to preparing the extracts (F. Langlois ind T. Moss, unpublished work). Thus, 5-6-fold changes in rRNA synthesis can occur without détectable changes in the activities of SL1, RPI or Rrn3. Under thèse conditions, elongation must then be the dominant factor in growth factor régulation of the rRNA gènes. We hâve now shown that growth rates of various human colon cancer ccll lines are also quantitatively explained by changes in RPI elongation rate (T. Gagnon-Kugler and T. Moss, unpublished work), suggest-ing that régulation of elongation is a gênerai phenomenon in unstressed mammalian cells.

The mechanism of growth factor régulation of transcription elongation Some years ago, we demonstrated that UBF enhances RPI transcription in vivo dépendent on a réversible ERK1/2 phosphorylation of its two N-terminal HMGl-boxes [16]. Given that UBF forms an rRNA gène chromatin through which RPI must transcribe, this suggested that UBF phosphorylation might regulate transcription elongation on thèse gènes. We had shown that each of the three N-terminal-most HMGl-boxes of UBF is able to bend D N A and that a dimer of UBF induces a 360" loop in the D N A template [6-8], sce Figure 1. The resulting enhancesome structure resembles the chromatin nucleosome in its protein/DNA ratio, but induces only a single gyre in the DNA, see Figure 1. In a récent study, we showed that ERK phosphorylation of UBF leads to an unwinding of the enhancesome [18]. Thus it seemed reasonable to hypothesize that this remodelling of the enhancesome could be the means by which RPI elongation is regulated. To our surprise, we found that when D N A templates were fully engaged by UBF, RPI elongation was strongly inhibited. ERK phosphorylation of UBF fully released this inhibition while at the same time unfolding the enhancesome [17,18].

A new paradigm for growth factor régulation The above results allowed us to establish a new paradigm for RPI transcription régulation. Active rRNA gènes are associ-ated with UBF and, hence, are predominantly in an enhance-somal conformation, see Figure 1(C). ERK phosphorylation of UBF causes local remodelling of the enhancesome and permits RPI to transcribe through the UBF-bound DNA. We hâve shown that ERK binds directly to the HMGl-boxes of UBF it phosphorylates, suggesting that it may be recruited to the rRNA gène chromatin directly. The process of re­modelling rRNA gène chromatin may need more than just ERK phosphorylation, since we hâve shown that acetylation of the rRNA gène chromatin by the histone acetyltransferase CBP [CREB (cAMP-response-element-binding protein)-binding protein] is also able to enhance RPI transcription [19]. Thèse results are ail the more tantalizing when one realizes that CBP is recruited to the same two HMG1 -boxes as is ERK

and that enhancement of RPI transcription by acetylation is completely dépendent on ERK activity [10,19].

Further implications of the new paradigm The démonstration that the RPI transcription elongation rate is a dominant regulator of rRNA synthesis suggests a mechanism for the co-ordinate régulation of pre-ribosome assembly. During co-transcriptional process, r-proteins and assembly factors must arrive on the nascent rRNA at exactly the right moment in order to be correctly positioned and ensure appropriate rRNA folding. If RPI elongation were to also dépend on correct RNP (ribonucleoprotein) assembly, this would provide a means of co-ordinating rRNA and r-protein synthesis. It would also provide an important means of 'proof-reading' pre-ribosome assembly. Consistent with thèse possibilities, studios ol the processome, the earliesl vis ible co-transcriptional RNP structure, suggest that its assem­bly is required for efficient rRNA synthesis in yeast [20].

This work was supported by an operating grant and a scholarship to T.G.-K. from the Canadian Institutes of Health Research. The University of Laval Cancer Research Centre is supported by the Fonds de Recherche sur la Santé du Québec.

Références 1 Scheer, II. (2005) Eur. J. Cell Biol. 84, 91-96 2 Klein, D.J., Moore, P.B. and Steitz, T.A. (2004) J. Mol. Biol. 340,141-177 3 Ferreira-Cerca, S., Poli, G., Gleizes, P.-E., Tschochner, H. and Milkereil, P.

(2005) Mol. Cell 20, 263-275 4 Bell, S.P., tearned, R.M., Jantzen, H.M. and Tjian, R. (1988) Science 241,

1192-1197 5 Leblanc, B., Read, C. and Moss, T. (1993) EMBOJ. 12, 513-525 6 Bazett-jones, O.P., Leblanc, B, Herfort, M. and Moss, T. (1994)

Science 264, 1134-1137 7 Stefanovsky, V.Y., Bazett-jones, O.P., Pelletier, G. and Moss, T. (1996)

Nucleic Acids Res. 24, 3208-3215 8 Stefanovsky, V.Y., Pelletier, G., Bazett-jones, O.P., Crane-Robinson, C. and

Moss, T. (2001) Nucleic Acids Res. 29, 3241-3247 9 O'Sullivan, A.C., Sullivan, G.J. and McStay, B. (2002) Mol. Cell. Biol. 22,

657-668 10 Stefanovsky, V.Y. and Moss, T. (2006) Cell Cycle 5, 735-739 11 Cavanaugh, A.H., Hirschler-laszkiewicz, I, Hu, Q., Dundr, M., Smink, 1.,

Misteli, T. and Rothblum, U. (2002) J. Biol. Chem. 277, 27423-27432 12 ZhaoJ., Yuan, X., Frodin, M. and Grummt, l. (2003) Mol. Cell 11, 405-413 13 Moss, T. (2004) Curr. Opin. Genêt. Oev. 14, 210-217 14 Moss, T. and Stefanovsky, V.Y. (2002) Cell 109, 545-548 15 Grummt, I. (2003) Gènes Oev. 17,1691-1702 16 Stefanovsky, V.Y., Pelletier, G., Hannan, R., Gagnon-Kugler, T., Rothblum,

L.l. and Moss, T. (2001) Mol. Cell 8, 1063-1073 17 Stefanovsky, V.Y., Langlois, F., Gagnon-Kugler, T., Rothblum, L.l. and

Moss, T. (2006) Mol. Cell 21 , 629-639 18 Stefanovsky, V.Y., Pelletier, G., Bazett-jones, O.P. and Moss, T. (2006)

Biochemistry 45, 3626-3634 19 Pelletier, G., Stefanovsky, V.Y., Faubladier, M., Hirschler-Laszkiewicz, L,

Savardj., Rothblum, L.L, CfltéJ. and Moss, T. (2000) Mol. Cell 6, 1059-1066

20 Gallagher, J.E.G., Dunbar, D.A., Granneman, S., Mitchell, B.M., Osheirn, Y., Beyer, A.L and Baserga, S.J. (2004) Gènes Dev. 18, 2506-2517

Received 21 June 2006

C2006 Biochemical Society