Les relations entre les histones qui composent les nucléosomes et le processus de transcription des gènes codants, sont à la fois multiples et extrêmement complexes. Au cours de ma thèse, je me suis intéressé à deux de ces relations. Tout d’abord, une première étude a été réalisée en collaboration avec les laboratoires de Franck Holstege et de Catherine Dargemont. Ce travail permet de préciser clairement l’effet des différentes formes de méthylation de la lysine 4 de l’histone H3 sur l’activité transcriptionnelle. Dans cette étude nous démontrons que la méthylation de H3K4 n’influence la transcription que d’un nombre très limité de gènes. Concernant ces gènes, un profil non conventionnel de distribution des formes de méthylation de H3K4 a été identifié par la présence inhabituelle d’un enrichissement en 3’ de ces gène des formes di- et triméthylées de H3K4.L’effet majoritaire de cette marque est d’induire une répression transcriptionnelle selon au moins deux mécanismes distincts. L’enrichissement atypique de la triméthylation de H3K4 influence négativement l’expression des gènes via la production d’ARN non codant anti-sens. Concernant l’effet répressif associé à la diméthylation de H3K4, la quantité d’ARN anti-sens ainsi que sa production ne sont pas impliquées.Dans une seconde étude réalisée en collaboration avec les laboratoires de Sebastian Chavez etd’Akash Gunjan, nous nous sommes intéressés au complexe FACT qui est impliqué dans l’assemblage et le désassemblage des nucléosomes lors du passage de l’ARN polymérase II. Jusqu’alors, un défaut de croissance chez les mutants thermosensibles du complexe FACT avait pu être observé. Dans notre étude, nous montrons que l’altération de FACT conduit, lors de la transcription, à l’éviction d’histones normalement incorporées à la chromatine. L’accumulation de ces histones libres à fort potentiel toxique, induit une répression spécifique de CLN3 qui code pour la première cycline dephase G1. Pour la première fois, nous mettons en évidence dans cette étude l’existence d’un mécanisme de surveillance moléculaire du cycle cellulaire induit par l’excès d’histones non incorporées à la chromatine / Relationships between histones, components of nucleosomes, and the transcription process of coding genes are both multiple and extremely complex. During my Thesis, I looked at twoof these relationships. First, we performed a study in collaboration with the Franck Holstedge and Catherine Dargemont labs. This work has allowed us to clearly define the effect of various methylation forms of the lysine 4 of the histone 3 on gene transcription. In this study we have shownthat H3K4 methylation influences the transcription of only a very limited number of genes. For these genes, a non conventional distribution profile of H3K4 methylation forms has been identified by the presence of an unusual enrichment in di- and trimethylated H3K4 in the 3’ of these genes. The principal effect of this mark is to promote transcriptional repression by at least two distinct mechanisms. The atypical enrichment of H3K4 trimethylation negatively influences gene expressionvia the production of non coding antisense RNA. For the repressive effect associated with dimethylH3K4, the quantity of antisense RNA as well as its production are not involved. We propose severalh ypotheses that link our results to the data known on this subject. In a second study performed incollaboration with the Sebastian Chavez and Akash Gunjan labs, we concentrated on the FACT complex that is involved in the assembly and disassembly of nucleosomes as RNA polymerase IImoves past. Previously, a growth defect in thermosensitive mutants of the FACT complex had been observed. In our study, we show that FACT deterioration leads to the eviction of histones that arenormally incorporated into chromatin during transcription. The accumulation of these free histones,which have a high toxic potential, induces the specific repression of CLN3 which encodes for the firstcyclin of G1 phase. For the first time, we show in this study the existence of a cell cycle molecular surveillance mechanism that is induced by an excess of free histones

Cellular checkpoints are typically considered to both facilitate the ordered execution of the cell cycle and to act as a barrier to oncogene driven cell cycles and the transmission of unresolved genetic lesions from one phase to the next. Furthermore, these mechanisms are also believed to underpin the responses of cells, both in normal and cancerous tissues, to those therapies that either directly or indirectly generate DNA damage. In recent studies however, it has become clear these checkpoints permit the passage of significant genomic aberrations into subsequent cell cycle phases and even descendant cells, and that heterogeneous responses are apparent amongst genetically identical cells. The consequences of this checkpoint ‘negligence’ remain relatively uncharacterised despite the importance of checkpoints in current models for how genomic instability is avoided in the face of ubiquitous DNA damage. Unresolved DNA damage is presumably inherited by subsequent cell cycle phases and descendant cells yet characterisation of the consequences of this has been relatively limited to date. I therefore utilised microscopy-based lineage tracing of cells expressing genetically encoded fluorescent sensors, particularly the Fluorescent Ubiquitination-based Cell Cycle Indicator (FUCCI) probes (Sakaue-Sawano et al., 2008), with semi-automated image analysis to characterise the response of single cells and their descendants to DNA lesions across multiple cell cycle generations. This approach, complemented by generational tracing by flow cytometry, permitted me to characterise the timing of cell fate determination in treated and descendant cells, the non-genetic heterogeneity in checkpoint responses and overall lineage behaviour, correlations between cells (similarly to Sandler et al., 2015) and cell cycle timing dependencies in the response to DNA damaging agents. With these single cell analytical approaches I show that the consequences of DNA damage on descendant cell fate is dramatic, suggesting checkpoint mechanisms may have consequences and even cooperate across phases and generations. U2OS cell lineages traced for three generations following the induction of DNA damage in the form of strand breaks showed greatly induced cell death in the daughters and granddaughters of DNA damaged cells, termed delayed death. Furthermore, lineage behaviour was characterised as highly heterogeneous in when and whether cell death occurred. Complementary flow cytometric approaches validated the findings in U2OS cells and suggested HeLa cells may show similar behaviour. These findings indicate that checkpoint models need to incorporate multigenerational behaviour in order to better describe the response of cells to DNA damage. Understanding the processes governing cell fate determination in descendant cells will impact upon our understanding of the development of genomic instability during carcinogenesis and how DNA-damaging chemotherapeutics drive cells to ‘quit’ the cell cycle.