G-quadruplexes are non-canonical nucleic acid secondary structures of increasing biological and medicinal interest due to their proposed physiological functions in transcription, replication, translation and telomere biology. Aberrant G4 formation and stabilisation have been linked to genome instability, cancer and other diseases. However, the specific genes and pathways involved are largely unknown, and the work within this thesis aims to investigate this. Stabilisation of G4s by small molecules can perturb G4-mediated processes and initial studies suggest that this approach has chemotherapeutic potential. I therefore also aimed to identify cell genotypes sensitive to G4-ligand treatment that may offer further therapeutic opportunities. To address these aims, I present the first unbiased genome-wide genetic screen in cells where genes were silenced via short-hairpin RNAs (shRNAs) whilst being treated with either PDS or PhenDC3, two independent G4-stabilising small molecules. I explored gene deficiencies that enhance cell death (sensitisation) or provide a growth advantage (resistance) in the presence of these G4-ligands. Additionally, I present a validation screen, comprising hits uncovered via genome-wide screening, and also the use of this in another cell line of different origin. Sensitivities were enriched in DNA replication, cell cycle, DNA damage repair, splicing and ubiquitin-mediated proteolysis proteins and pathways. Ultimately, I uncovered four synthetic lethalities BRCA1, TOP1, DDX42, GAR1, independent of cell line and ligand. These were validated with three G4-stabilising ligands (PDS, PhenDC3 and CX-5461) using an independent siRNA approach. The latter siRNA methodology was used to screen 12 PDS derivatives with improved medicinal chemistry properties and ultimately identified SA-100-128, as a lead compound. The mechanism behind synthetic lethality with G4-stabilising ligands was explored further for DDX42, which I show has in vitro affinity for both RNA- and DNA-G4s and may represent a previously unknown G4-helicase. Also within this thesis, gene deficiencies that provided a growth advantage to PDS and/or PhenDC3 as uncovered by genome-wide and focused screening were explored. These showed enrichment in transcription, chromatin and lysosome-associated genes. The resistance phenotype of three gene deficiencies, TAF1, DDX39A and ZNF217 was further supported by additional siRNA experiments. Overall, I satisfied the primary aims and established many novel synthetic lethal and resistance interactions that may represent new therapeutic possibilities. Additionally, the results expand our knowledge of G4-biology by identifying genes, functions and subcellular locations previously not known to involve or regulate G4s.

To identify novel therapeutic targets for basal-like breast cancer (BLBC) subtype, we investigated several DNA repair mechanisms associated with maintenance of high genomic instability for cell survival in cancer cells. We identified that the mismatch repair proteins, MSH2 and MSH6 (referred to as MSH2/6 hereafter), are highly elevated across BLBC samples. High expression level of MSH2/6 in BLBC is associated with worse prognosis and survivability for patients. Therefore, we knocked out MSH2 in BLBC cell lines and performed in vivo xenograft and syngeneic mice model studies to find significant attenuation of tumor growth in MSH2 KO group. Also, MSH2-deficient BLBC cells have increased rate of new mutations. Additionally, we tested the efficacy of conventional chemotherapeutics and radiation treatment that would further tip the genomic instability in MSH2-deficient BLBC cells towards cell death, but found them to be ineffective. Next, we performed high-throughput screening of 1280 FDA-approved compounds to discover that calcium channel blockers preferentially kill MSH2-deficient BLBC cells. This was likely due to association of significantly mutated pathways that involved calcium ion binding and calmodulin binding sites. Here we provide evidence of an alternative therapeutic strategy targeting DNA repair genes in BLBC patients utilizing bioinformatics analysis, high-throughput drug screening, in vitro,and vivoexperimentalmodels.

During every cell cycle and upon exogenous stress, tumor suppression programs are engaged to ensure genomic stability. In response to replicative aging and oncogenic stimuli, the p53 and Rb pathways are activated to prevent the proliferation of damaged cells. Several lines of evidence suggest that escape from senescence is a crucial early step in oncogenic progression. A major challenge in the cancer field is to combine genomic information regarding cancer-associated genetic changes with high-throughput functional studies, in order to confirm genetic requirements and pinpoint biological roles of these perturbed genes in oncogenesis. Furthermore, a complete genetic understanding of replicative senescence, and how it might be bypassed, is lacking. We describe here two genome scale loss-of-function genetic screens that interrogate these tumor suppressor programs. We utilized a unique sensitization approach to isolate senescence pathways and unmask compensatory mechanisms that may have been difficult to identify in previous studies. These genetic screens have generated comprehensive and validated datasets of putative senescence and p53 pathway genes. We present this dataset as a high-quality resource for further investigation into these biological pathways. We have uncovered several genes in distinct biological pathways which have not been demonstrated to have a functional role in senescence, and which may be putative tumor suppressors. We have identified BRD7 and BAF180, two SWI/SNF components, as critical regulators of p53. BRD7 and BAF180 are required for p53 activity and p21 expression during replicative and oncogene-induced senescence, and evidence suggests that they are inactivated in human cancer. In addition, we have uncovered a role for the deubiquitinating enzyme USP28 in the regulation of p53 accumulation during senescence, such that loss of USP28 results in bypass of the senescence program. We have also investigated several other novel senescence genes including SEMA6A, SEMA3b, and TMEM154. We have found that the expression of these genes is highly regulated during senescence by distinct means, including both p53-dependent and p53-independent mechanisms. These results demonstrate the efficacy of our sensitized screening approach, and also highlight the emerging view that the senescence program requires the combined action of multiple biological pathways for its execution.

The human ribosomal RNA genes are critically important for cell metabolism and viability. They code for the catalytic RNAs which, encased in a housing of more than 80 ribosomal proteins, link together amino acids by peptide bonds to generate all cellular proteins. Because the RNAs are not repeatedly translated, as is the case with messenger RNAs, multiple copies are required. The genes which code for the human ribosomal RNAs (rRNAs) are arranged as clusters of tandemly repeated sequences. Three of four catalytic RNAs are spliced from a single transcript. The genes are located on the short arms of the five acrocentric chromosomes (13, 14, 15, 21, and 22). The genes for the fourth rRNA are on chromosome 1q42, also arranged as a cluster of tandem repeats. The repeats are extremely similar in sequence, which makes them ideal for misalignment, non‐allelic homologous recombination (NAHR), and genomic destabilization during meiosis , replication, and damage repair. In this dissertation, I have used pulse‐field gel electrophoresis and in‐blot Southern hybridization to explore the physical structure of the human rRNA genes and determine their stability and heritability in normal, healthy individuals. I have also compared their structure in solid tumors compared to normal, healthy tissue from the same patient to determine whether dysregulated homologous recombination is an important means of genomic destabilization in cancer progression. Finally, I used the NCI‐60 panel of human cancer cell lines to compare the results from the pulsed‐field analysis, now called the gene cluster instability (GCI) assay, to two other indicators of homologous‐recombination-mediated genomic instability: sister chromatid exchange, and 5‐hydroxymethyl‐2’deoxyuridine sensitivity.

Genomic instability driven by non-allelic homologous recombination (NAHR) provides a realistic mechanism that could account for the numerous chromosomal abnormalities that are hallmarks of cancer. We recently demonstrated that this type of instability could be assayed by analyzing the copy number variation of the human ribosomal RNA gene clusters (rDNA). Further, we found that gene cluster instability (GCI) was present in greater than 50% of the human cancer samples that were tested. Here, data is presented that confirms this phenomenon in the human GAGE gene cluster of those cancer patients. This adds credence to the hypothesis that NAHR could be a driving force for carcinogenesis. This data is followed by experimental results that demonstrate the same gene cluster instability in cultured cells that are deficient for the human BLM protein. Bloom’s Syndrome (BS) results from a genetic mutation that results in the abolition of BLM protein, one of human RecQ helicase. Studies of Bloom’s Syndrome have reported a 10-fold increase in sister chromatid exchanges during mitosis which has primarily been attributed to dysregulated homologous recombination. BS also has a strong predisposition to a broad spectrum of malignancies. Biochemical studies have determined that the BLM protein works in conjunction with TOPOIIIα and RMI1/RMI2 to function as a Holliday Junction dissolvase that suppress inadvertent crossover formation in mitotic cells. Because of the similarities in their biochemical activities it was suggested that another DNA helicase found in E. coli, the RecG DNA translocase, is the functional analog of BLM. RecG shares no sequence homology with BLM but it can complement both the sister chromatid exchange elevation and the gene- cluster instability phenotype caused by BLM deficiency. This indicates that the physiological function of BLM that is responsible for these phenotypes rests somewhere in the shared biochemical activities of these two proteins. These data taken together give new insights into the physiological mechanism of BLM protein and the use of Bloom’s Syndrome as a model for carcinogenesis.

Diss. (sammanfattning) Stockholm : Karol. inst., 2004. / Härtill 3 uppsatser.

<p>Like most Archaea, the hypersaline-adapted organism Halobacterium salinarum exhibits characteristics from all three domains of life, including a eukaryotic histone protein, a universal propensity to genetic rearrangements, and homologs of bacterial cell division proteins. Here we investigate the ancestral function of histone protein in the Archaea. Transcriptomics, proteomics, and phenotypic assays of histone mutants determine that histone regulates gene expression and cell shape but not genome compaction in H. salinarum. We further explore the regulation of gene expression on a genome-wide scale through the study of genomic instability. Genomic deletions and duplications are detected through the meta-analysis of 1154 previously published gene expression arrays and 48 chromatin immunoprecipitation arrays. We discover that a 90 kb duplication event in the megaplasmid pNRC100 directly leads to increased gene expression, and find evidence that the chromosome is far more unstable than previously assumed. These events are all linked with the presence of mobile insertion elements. Finally, in response to questions generated by these experiments, we develop a novel time-lapse protocol for H. salinarum and ask basic questions about single cell dynamics during division. Fluorescent labeling of homologs to bacterial cell division proteins confirms their involvement in cell division but localization dynamics contradict the basic bacterial model. The discovery of unusual facets of morphology during cell division is consistent with these novel protein dynamics and opens up new avenues of inquiry into archaeal cell division.</p> / Dissertation

Common fragile sites are a set of genomic locations with a propensity to form lesions, breaks and gaps on mitotic chromosomes upon induction of replication stress. While the exact reasons for their fragility are unknown, CFS display instability in a cell-type specific manner, suggesting a substantial contribution from an epigenetic component. CFSs also overlap with sites of increased breakage and deletions in tumour cells, as well as evolutionary breakpoints, implying that their features shape genome stability in vivo. Previously, factors such as delays in replication timing, low origin density and transcription of long genes have been implicated in instability at CFS locations but comprehensive molecular studies are lacking. Chromatin structure, an important factor that fits the profile of cell-type specific contributor, has also not been investigated yet. Throughout their efforts to determine the factors that lead to the appearance of CFS lesions, investigators have focused on a single component at a time, potentially missing out complex interactions between cellular processes that could underlie fragility. Additional difficulties come from the cell-type specificity of CFS breakage: it indicates that only cell type-matched data would be informative, limiting the scope for studies using publicly available data. To perform a comprehensive study defining the role of different factors in determining CFS fragility, I explored replication timing, transcriptional landscapes and chromatin environment across a number of CFSs in two cell types exhibiting differential CFS breakage. Initially, I characterised the patterns of CFS fragility in the two cell types on both the cytogenetic and the molecular level. I then used a FISH-based technique to investigate the process of mitotic compaction at active CFS sites and found that the cytogenetically fragile core of these sites sits within larger regions which display a tendency to mis-fold in mitosis. The aberrant compaction of these regions could be observed on cytogenetically normal metaphase chromosomes, suggesting that finer scale abnormalities in chromosome structure underlie the cytogenetically visible breaks at fragile sites. I also investigated the links between transcription of long genes and CFS fragility using two approaches: I quantified levels of expression across all fragile sites using RNA-seq and modified transcription at a single active CFS using the CRISPR genome engineering methodology. My results indicate a complex interplay between transcription and CFS fragility: no simple linear correlation can be observed, but an increase of transcriptional levels at the active CFS led to a corresponding increase in fragility. To investigate the influence of the cell type specific replication programme and replication stress on CFS instability, I mapped replication timing genome-wide in unperturbed cells and under conditions of replication stress in both cell types. I found that replication stress induces bi-directional changes in replication timing throughout the genome as well as at CFS regions. Surprisingly, the genomic regions showing the most extreme replication timing alterations under replication stress do not overlap with CFS, implying that CFS instability is not fully explained by replication delays as previously suggested. Instead, I observed a range of replication-stress induced timing changes across CFS regions: while some CFSs appear under-replicated, others display switches to both earlier and later replication as well as differential recruitment of both early and late origins, implying that dis-regulation of replication timing and origin firing, rather than simply delays, underlie the sensitivity to CFS regions to replication stress. Finally, I investigated large-scale chromatin states at two active CFSs throughout S phase and into G2, the cell cycle stages most relevant stage for CFS breakage. I found that changes in large-scale chromatin architecture accompany the replication timing shifts triggered by replication stress, raising the possibility that such alterations contribute to instability. In conclusion, I assessed the influence of multiple relevant factors on CFS fragility. I found that bi-directional replication timing changes and alterations in interphase chromatin structure are likely to play a role, converging to promote mitotic folding problems which ultimately result in the well-described cytogenetic lesions on metaphase chromosomes and genomic instability.

Mon travail de thèse s'intéresse au rôle du récepteur à l'immunité innée NLRP3, composant essentiel de l'inflammasome, dans le développement tumoral pulmonaire. Nos résultats révèlent que les cellules épithéliales pulmonaires immortalisées expriment un inflammasome NLRP3 fonctionnel. De façon inattendue, nous montrons que l'expression du récepteur NLRP3 est fortement diminuée, voire perdue des lignées tumorales de CBNPC et dans des tumeurs de patients, comparé au tissu sain adjacent. Nous montrons que NLRP3, de façon totalement indépendante de l'inflammasome, est impliquée dans la régulation transcriptionnelle de H2AFX, le gène codant pour le variant d'histone H2AX, élément clé de la signalisation des dommages à l'ADN. L'absence de NLRP3 dans les cellules HBEC altère l'amplification et la transmission du signal en réponse à des cassures double brin, résultant in fine à moins de réparation. Ce défaut de réparation des cassures se traduit par une instabilité génomique, qui est en effet plus forte dans les adénocarcinomes pulmonaires exprimant de faible niveau de NLRP3. Mon travail de thèse identifie donc le récepteur NLRP3 comme un facteur clé de la réponse aux dommages à l'ADN et du maintien de l'intégrité génomique en promouvant la transcription de H2AFX dans les cellules épithéliales pulmonaires. Ce nouveau rôle de NLRP3, associé à sa perte dans les tumeurs de CBPNC en font un potentiel suppresseur de tumeur / During my PhD, I have been interested in the role of the innate immune receptor NLRP3, a key component of the inflammasome, in lung cancer development. Our results show the presence of a functional NLRP3 inflammasome in normal human bronchial epithelial cells (HBEC). Surprisingly, NLRP3 expression is strongly down-regulated in a large panel of NSCLC cell lines and patient tumors compared to healthy tissue. Moreover, we unravel that NLRP3 contributes to the transcription of H2AFX, the coding gene for the histone variant H2AX, in an inflammasome independent-manner. The deletion of NLRP3 in HBEC impairs double strand break signal amplification and transduction, resulting in a decrease in DNA repair. This repair defect leads to genomic instability, which is increased in lung adenocarcinomas expressing low levels of NLRP3. My PhD work identifies NLRP3 as a key factor of the DNA damage response and genomic integrity maintenance by regulating the transcription of H2AFX. This new role for NLRP3, together with its loss in NSCLC, makes it as a potential tumor suppressor

L’instabilité génomique est une caractéristique majeure des cellules cancéreuses. Dans les premières étapes du développement du cancer, l’activation des oncogènes induit du stress réplicatif à l’origine de cette instabilité. Le mécanisme par lequel la dérégulation des oncogènes induit un blocage des fourches de réplication et du gammaH2AX sur la chromatine reste peu compris. Ainsi déterminer l'origine de ce stress réplicatif dans les cellules précancéreuses est donc essentiel afin de mieux comprendre les premières étapes de la tumorigenèse. Il a été montré dans notre laboratoire que la déplétion dans des cellules humaines de la topoisomérase 1 ou du facteur d’épissage ASF/SF2 perturbe la progression des fourches de réplication, active le checkpoint de phase S et induit des cassures chromosomiques (Tuduri et al., 2009). Puisque les dommages à l’ADN et les défauts de réplication sont corrigés par la RNase H1, il est possible que ce stress réplicatif soit dû à la formation de R-loops. Ces hybrides ADN/ARN se forment au cours de la transcription lorsque l’ARN naissant revient s’hybrider avec sa matrice d’ADN laissant le brin non-transcrit sous forme simple brin. Les R-loops se formant dans des sites spécifiques du génome, nous avons alors cartographié leurs distributions et l’avons comparé à celle de marqueurs de stress réplicatif et de cassure double brin (CDB) de l’ADN dans nos cellules shASF et shTop1. Nous avons donc combiné différentes approches génomiques comme le DRIP-seq (R-loops), le ChIP-seq (pRPA et gammaH2AX) et le BLESS (CDB ; Crosetto et al., Nature Methods, 2013). Nos données montrent une corrélation importante entre les régions formant du stress réplicatif et la formation de R-loops appuyant l’idée que l’interférence entre réplication et transcription augmente l’instabilité génomique dans les cellules humaines. Toutefois puisque les R-loops ont de multiples rôles physiologiques, toutes régions qui en forment ne corrèlent pas avec l’induction de stress réplicatif. Ce projet devrait nous aider à déterminer dans quelles conditions les R-loops représentent une menace pour l’intégrité du génome. Par microscopie confocale à fluorescence, nous avons confirmé que les R-loops s’accumulaient dans nos lignées HeLa déplétées pour ASF et Top1. A noter que les R-loops s’accumulent également dans des fibroblastes immortalisées exprimant la forme oncogénique de Ras et dans des préplasmablastes, une étape du développement plasmocytaire particulièrement à risque pour le développement de myélome multiple. Ensemble, ces données indiquent que les R-loops pourraient être une source du stress réplicatif induit par les oncogènes. / Genome instability is a hallmark of cancer cells. It has been proposed that at early stages of the cancer process, genomic instability is caused by oncogene-induced replication stress, a poorly-understood process characterized with the accumulation of stalled replication forks and gammaH2AX on chromatin. Understanding the origin of chronic replication stress represents a major challenge in cancer biology. We have previously shown that depletion of DNA Topoisomerase 1 or the splicing factor ASF/SF2 in mammalian cells interferes with replication fork progression, activating the DNA damage response and inducing chromosome breaks (Tuduri et al., 2009). Since DNA damage and replication fork stalling are relieved by RNaseH1, an attractive hypothesis could be that replication stress is caused by R-loops. These RNA-DNA hybrid structures form when nascent RNA re-anneals to the template DNA strand, leaving the non-template strand unpaired. Using immunofluorescence confocal microscopy with the S9.6 antibody that recognizes RNA-DNA hybrids, we confirmed that R-loops accumulate in ASF/SF2 and Top1-depleted HeLa cells. Since R-loops are enriched at specific sites in the human genome, wecombined different genomic approaches, including DRIP-seq (R-loops), ChIP-seq (gammaH2AX, pRPA) and BLESS (DSBs; Crosetto et al., Nature Methods, 2013) to monitor their distribution relative to replication stress markers and DNA double-strand breaks (DSBs) in the absence of Top1 or ASF/SF2. . Our data reveal a significant correlation between replication stress and cotranscriptional R-loops, supporting the view that the interference between replication and transcription promotes genomic instability in human cells. However, not all R-loops forming regions colocalize with replication stress since these structures have multiple physiological roles. This approach allowed us to determine the conditions in which R-loops may represent a threat to genome integrity. Moreover, we also observed the accumulation of R-loops in immortalized fibroblasts expressing an oncogenic form of Ras and in preplasmablast during plasma cell differentiation, a crucial process during which multiple myeloma may evolve. In clonclusion, our data indicate that R-loops may represent an important source of oncogene-induced replication stress.