Search

This thesis details a tripartite approach to understanding genetic and epigenetic modifications in human leukaemia. First, an in vitro mouse model of precursor B-cell leukaemia was established by retroviral transduction of primary B-cell precursors with the oncogenes BCR-ABL1 or MYC. The model was used to study DNA damage incurred during the initial phase of oncogene activation i.e. oncogenic stress. Analysis of γH2AX (a mark of DNA damage) and the H3K27 acetylation histone modification (a mark of active enhancers and promoters) by chromatin immunoprecipitation sequencing (ChIP-seq) and mRNA expression (RNA-seq) were combined to establish that lineage specific transcription predisposes lineage specific genes to DNA damage. This was functionally validated with a lineage-switch model and technically validated using FISH. Second, the same cells were maintained during their recovery from the period of acute oncogenic stress until stably transformed. Re-analysis of the transformed populations using the same H3K27Ac ChIP-seq and RNA-seq techniques showed that the stable transformation of these cells is paralleled by activation of a subset of intergenic enhancers which induce an aberrant transcriptome enriched for cell activation and metabolic pathways. Finally, the techniques used in these in vitro experiments and the knowledge of oncogenic enhancers gained from them were then applied to the clinical setting. EVI1, a key haematopoietic transcription factor that regulates differentiation, is aberrantly expressed in a subset of myeloid leukaemia where it is universally associated with a poor prognosis. A patient sample biobank was established and suitable samples were filtered using a combination of RT-qPCR and FISH. Thus, H3K27AC ChIP-seq and newly optimised chromosome conformation capture sequencing (3C-seq or 4C) was used to understand the epigenetic interactions which result in aberrant expression. This has revealed several interactions between the EVI1 promoter with up- and down-stream intergenic regions which must be further investigated.

DNA double strand breaks (DSBs) are the most lethal radiation-induced lesions in response to which cells employ either the error free homologous recombination (HR) repair pathway or error prone mechanisms, such as non homologous end joining (NHEJ) and microhomology- mediated end joining (MMEJ). While MMEJ is suppressed by C-NHEJ, the relationship between HR and MMEJ is less clear. In this thesis, I have exploited the human endogenous HPRT gene to develop a novel genetic technology to detect mutation frequencies and signatures in response to DNA DSBs in different genetic backgrounds. Using a sensitive HPRT assay we describe a role for HR genes in suppressing MMEJ in human cells. By monitoring DSB mis-repair, we found that depletion of HR proteins, including RAD51, BRCA2, BRCA1 or SETD2, resulted in a distinct mutational signature associated with significant increases in break-induced mutation frequencies, deletion lengths and the annealing of short regions of microhomology (2 - 6 bp) across the break-site. This signature was dependent on CtIP, MRE11, POLQ and PARP, and thus indicative of MMEJ. In contrast to CtIP or MRE11, depletion of BRCA1 resulted in increased partial resection and MMEJ, thus revealing a functional distinction between these early acting HR factors. Together these findings indicate that HR factors suppress mutagenic MMEJ following DSB resection. Further, I have defined a role for the SETD2 histone methyltransferase in suppressing break-induced mutations, and have shown that CRISPR/Cas9 and ISceI- induced DSBs resulted in different repair profiles.

β-thalassemia and sickle cell disease, which are caused by mutations in the β-globin gene, are two of the most common single gene disorders worldwide and the only available cure is allogeneic bone marrow transplantation that is limited by donor availability. Gene therapy, by delivery of a β-globin expression vector into autologous haematopoietic stem cells, is a valuable alternative but the technique is affected by unpredictable protein expression levels and, more significantly, by random integration of the vector and the risk of insertional oncogenesis. Gene correction by Homologous Recombination (HR) with a DNA repair template would avoid the above-mentioned issues, becoming ideal therapeutic approach. Although spontaneous HR events are very rare, Double Strand Breaks (DSBs) at the target locus can greatly stimulate them. Therefore, the development of Zinc Finger Nucleases (ZFNs), which are customised endonucleases capable of cleaving any desired DNA sequence, has created the opportunity to design molecular tools to treat many rare monogenic disorders. The aim of this thesis has been to develop a ZFN-based gene targeting at the β-globin locus as a basis for β-thalassaemia gene-correction therapy. Initially, six ZF domains specific for the β-globin gene were designed, assembled and tested using publicly available reagents, but these were found to have a low binding efficiencies. Therefore, a commercial approach was used to obtain a functional ZFN (ZFN4) and this was shown to produce DSBs at the β-globin gene in ≈ 1% of the transfected human erythroleukemia cells without detectably cleaving the most closely related sequence in the human genome (the δ-globin gene). Using a reporter assay, ZFN4 was also shown to promote gene correction through intrachromosomal HR but was found to be 20 times less efficient than the homing endonuclease I-SceI. Ultimately, ZFN4-stimulated targeted integration of a drug resistance marker at the endogenous β -globin gene locus was tested: 95% of the drugresistant cells were targeted while the absolute frequency compared to the whole cell population resulted to be 0.1% The collected data show that ZFN4-mediated gene targeting of the β-globin locus is possible but further studies are required in order address the discrepancy between cutting and targeting efficiencies and to increase the absolute frequency of gene targeted cells.

Mitotic homologous recombination (HR) stabilises the genome by repairing harmful double stranded DNA (dsDNA) breaks. Rad52 promotes the annealing of complementary DNA strands and binds to Rad51 suggesting that Rad52 acts in both strand invasion (SI) and single strand annealing (SSA) HR pathways. Overexpression of Rad52 in mammalian cells can greatly stimulate HR, but the degree of stimulation varies widely between studies and inhibitory effects have also been described. The in vivo roles for Rad52 in mammalian cells are thus poorly understood as is the potential of using Rad52 as a tool to promote genome engineering methods [i.e. gene targeting (GT)]. Here I have systematically compared the effects of overexpressing wild-type (wt) and mutant Rad52 proteins on cell viability and various HR assays in human cells. Mutants were designed to test the potential involvement of defined domains/residues in any such effects. Human Rad52 (hRad52) and its derivatives negatively affected cell viability and proliferation which correlated with the presence of C-terminal residues 331-418, rather than, as expected, the Rad51 binding domain which limited inhibitory effects. Negative effects with yeast Rad52 (scRad52) were not observed. Consistent with previous findings, hRad52 inhibited GT, however, this was converted to a stimulatory effect when residues 331-418 were removed. When single stranded oligonucleotide (ssO) templates were used for GT, both hRad52 and its derivatives were stimulatory. These results are consistent with SI and SSA models for dsDNA- and ssO-mediated GT, respectively, and suggest that residues 331-418 cause a dominant-negative effect on SI and their removal promotes the strand annealing activity of the N-terminal domain. ScRad52 stimulated and inhibited GT with ds and ss templates, respectively. Altogether, these results provide the first evidence that truncated forms of hRad52 may serve as useful tools for promoting GT using both ds and ss DNA templates in human cells.

In a mouse model of radiation-induced acute myeloid leukaemia (rAML), the CBA/H strain, copy loss of Sfpi1 is found in up to 90 % of cases. The gene codes for the transcription factor PU.1, an important regulator of haematopoiesis, which acts as a tumour suppressor. Point mutations specifically in the DNA sequence encoding the binding domain of the protein are found at a high frequency, suggesting that the transcription factor has a reduced function in leukaemic blasts. Transgenic CBA/H mice expressing green fluorescent protein (GFP) under the control of the Sfpi1 promoter allowed the use of flow cytometry to, for the first time, study copy loss of Sfpi1 in cases of rAML by analysing the expression levels of GFP. Analysis of mRNA levels of Sfpi1 showed there probably is a different mechanism for leukaemogenesis in cases with point mutations in the sequence encoding the DNA binding domain compared to cases with a wild type Sfpi1, the former potentially through lack of function in the transcription factor PU.1 and the latter through suppression of Sfpi1 gene expression. Fluorescent activated cell sorting was used to isolate immature bone marrow cells (BMC) containing Sfpi1 deletions from previously irradiated mice. The isolated cells were transplanted into myeloablated hosts with some success, although this new approach to study radiation-induced leukaemogenesis needs refinement. A population of immature BMCs was found to be less radiosensitive in the CBA/H strain compared to the C57BL/6 strain, which is not susceptible to rAML. This gave an indication of where the target cell population for the initial leukaemic events may occur. The work in this thesis presents novel ways of studying early events in rAML induction in the CBA/H mouse and gives new insights into the mechanisms of leukaemogenesis in the model.

BCL6 is a zinc finger transcriptional repressor, which is highly expressed in germinal centre B-cells, and is essential for germinal centre formation and T-dependent antibody responses. Deregulated BCL6 expression is associated with certain non- Hodgkin’s lymphomas. High expression is observed in breast cancer. Tight lineage and temporal regulation of BCL6 is, therefore, required for normal immunity and abnormal regulation occurs in cancer. Regulatory mechanisms have been analysed in two settings. Firstly, BCL6 is strongly induced by the tyrosine kinase inhibitor, Imatinib, in chronic myeloid leukaemia lymphoid blast crisis cell lines, and this effect was used in order to study the effects of phospho-protein signalling on BCL6 expression and a major finding is that p38 MAPK induced BCL6. Also, p38 is, at least in part, responsible for BCL6 expression in basal conditions in the germinal centre representative Burkitt’s lymphoma cell lines and that qualitatively different CD40 stimuli can either induce or repress BCL6 expression. Luciferase assays showed that p38 acts at a 300bp sequence immediately 5’ of exon 1, and probably also at more distal sequences. Overall it appears that the balance between positive and negative regulatory controls BCL6 expression with inhibitory signalling pathways being predominant in most circumstances. Focusing on BCL6 exon 1, a binding site for the multifunctional regulator CTCF was identified. CTCF interacts in vitro and in vivo with this sequence. Reduced expression of CTCF in germinal centre cells caused a moderate reduction of BCL6 expression. Finally, although no clear differences were observed in the methylation status of the CTCF binding site on exon 1, a significant enrichment of active histone modifications at this site was observed in BCL6 expressing cells, suggesting that CTCF may have a role in the epigenetic regulation of BCL6.

B cell lymphoma 6 (Bcl6) is a proto-oncogene functioning as a transcriptional repressor at the germinal centre (GC) stage of B cell development, regulating proliferation, differentiation and survival. Bcl6 is commonly dysregulated in GCderived lymphomas and hence a promising therapeutic target. Although a number of agents targeting Bcl6 are being developed, there is currently no good model available for functional analysis and hence drug target validation. To address this issue, I have established a conditional BCL6 knockout cell line for in-depth functional analyses of Bcl6. I have used homologous recombination to disrupt one BCL6 allele in the human Burkitt’s lymphoma cell line DG75 with a zeocin resistance cassette and have shown this line to be capable of supporting tetracycline (tet)-regulated Bcl6 expression from an integrated expression construct. The zeocin cassette used to disrupt the first BCL6 allele was subsequently deleted by Cre recombination. This allowed me to reuse the same targeting construct to disrupt the second BCL6allele. I subsequently set out to characterise the resulting conditional knockout cell line and found that depletion of Bcl6 leads to less proliferation, characterised by a prolonged G0/G1cell cycle phase. I went on to further characterise the effect of Bcl6 depletion in my system by cDNA microarray analysis, which revealed interesting insights into functions of Bcl6 such as the modulation of B cell receptor signalling and calcium signalling. The system I have developed is the first precisely regulated genetic model for studying Bcl6 function and will be of great benefit for research aiming to develop Bcl6 targeted drug therapy.