Thesis (Ph. D.)--University of Hong Kong, 2006. / Title proper from title frame. Also available in printed format.

DNA is the central storage molecule for genetic information in the cell. Therefore, the DNA must be protected from damage that will otherwise be passed on to future generations as deleterious mutations. Although many different pathways have evolved for repairing different classes of damage there are certain features that are common to all repair pathways. Generically, for DNA damage to be repaired it must first be recognized, then excised and replaced with undamaged DNA. DNA damage recognition is highly varied since specific interactions are required between the protein and the damaged DNA. DNA damage repair, paradoxically, requires the action of highly processive nucleases. The nucleases may digest hundreds if not thousands of nucleotides, sometimes for the repair of a single mutant nucleotide. We have chosen to focus on Exonuclease VII (ExoVII), one of the processive nucleases that have been implicated in the process of Mismatch Repair (MMR). ExoVII is a hetero-pentameric enzyme composed of one large subunit (XseA) and four small subunits (XseB). It has been previously characterized as a processive, single-strand specific nuclease able to digest DNA in either the 5'->3' or 3'->5' direction by a metalindependent mechanism. Early studies have shown that although ExoVII is a hydrolytic nuclease it was completely active in the presence of large amounts of EDTA and was strongly stimulated by phosphate. This feature is unusual because hydrolytic DNA nucleases typically function by a mechanism that requires coordination of a divalent cation. To further our understanding of the mechanism ExoVII we have identified and characterized the ExoVII homolog from Thermotoga maritima (T. maritima, Tm), a hyperthermophilic bacterium. The genes responsible for Tm ExoVII (TM1768 and TM1769) were cloned into an overexpression construct and the resulting proteins were overexpressed, co-purified and characterized. Consistent with previous studies, we found that Tm ExoVII is a processive, single-strand specific nuclease. Surprisingly, unlike Ec ExoVII, the T. maritima homolog was found to have an absolute requirement for the divalent cation magnesium and was strongly inhibited by the presence of either phosphate or sulfate in the reaction buffer. Using multiple sequence alignments of the large subunit we have identified a conserved core present within the C-terminal ExoVII_Large domain. This conserved core, RGGGx27GHx2Dx4Dx9P, although unique among nucleases, is reminiscent of a metal-coordinating hydrolytic active site. We have tested this putative active site using site-directed mutagenesis to create the TmD235A/TmD240A double mutant. This mutant protein was purified and the resulting protein was found to be inactive. We propose that this conserved core represents the metal-coordinating active site of all ExoVII homologs and that the group of E. coli-like homologs are unique in their EDTA resistance and anion (phosphate and sulfate) stimulation. Since ExoVII is a bi-directional nuclease (both 5'->3' and 3'->5' activity), and MMR is a bi-directional process, our model was that ExoVII was the primary nuclease associated with MMR. To test this model and determine if, in fact, a minimal conserved MMR pathway can be defined, we performed an analysis of the genomic occurrence profiles for the genes involved in MMR. To do this we have developed a bioinformatic application, Magma, which assists in the creation of a searchable relational database. Using Magma we have found that MutH, the enzyme responsible for generating a nick that directs MMR to excise the newly synthesized DNA strand including a DNA mispair, is only present in E. coli and a subset of gamma-proteobacteria, suggesting that MutH is not a core component of MMR. Instead, most organisms employ a nicking activity found in the MutL subunit. We also show that, although four nucleases have been implicated as having "redundant" roles in bacterial mismatch repair, RecJ is the primary nuclease responsible for degrading the mutated DNA strand and that 5'->3' single-strand exonuclease activity is a core MMR component. From this analysis, it appears that prokaryotic mismatch repair is more similar to eukaryotic mismatch repair than was previously thought, from the genetic and biochemical work done in E. coli. We offer a model for a universal minimal MMR system.

In the fission yeast, Schizosacchromyces pombe, blocks to DNA replication elongation trigger the intra-S phase checkpoint that leads to the activation of the Cds1 kinase. Cds1 is required to both stabilize stalled replication forks and to prevent premature entry into mitosis. Interestingly, although Cds1 is essential to maintain the viability of mutants defective in DNA replication elongation, my study shows that mutants defective in DNA replication initiation require the Chk1 kinase, rather than Cds1. This suggests that failed initiation events can lead to activation of the DNA damage checkpoint independent of the intra-S phase checkpoint. This might result from reduced origin firing that leads to an increase in replication fork stalling or replication fork collapse that activates the G2 DNA damage checkpoint. I refer to the Chk1-dependent, Cds1-independent phenotype as the rid phenotype (for replication initiation defective). The data shows that Chk1 is active in rid mutants when grown under semi-permissive conditions, and rid mutant viability is dependent on the DNA damage checkpoint, and surprisingly Mrc1, an adaptor protein required for activation of Cds1. Mutations in Mrc1 that prevent activation of Cds1 have no effect on its ability to support rid mutant viability, suggesting that Mrc1 has a checkpoint-independent role in maintaining the viability of mutants defective in DNA replication initiation. Like Mrc1, Swi1 and Swi3 have been hypothesized as a part of the replication fork protection complex (RFPC). They are required for maintaining the viability of rid mutants, but are not essential for activation of Chk1 in response to failed initiation events. This suggests that Mrc1 in conjunction with Swi1 and Swi3 function in a similar pathway to alleviate replicative stress resulting from defects in DNA replication initiation. Using flow cytometry, I demonstrate that inhibition of DNA replication initiation has no significant impact on the duration of S phase, suggesting dormant origins might be activated in response to defects in DNA replication initiation. Fission yeast Rad22 is implicated in forming nuclear foci in response to damaged DNA. By tracking YFP-labeled Rad22, I screened for potential DNA damage in rid mutants grown at semi-permissive temperatures, and the results show that DNA damage occurs as the result of defects in DNA replication initiation. I also identified camptothecin, a DNA topoisomerase I inhibitor that can at low dose (2 µM) induce the rid phenotype, suggesting our assay (Chk1-dependent, Cds1-independent) can be used to screen small molecule inhibitors that interfere with the initiation step of DNA replication.

In order to maintain genome stability, DNA damage needs to be detected and repaired in a timely fashion. To cope with damaged DNA, cells have evolved mechanisms termed "checkpoints", where, upon damage, cells initiate a signal transduction cascade that results in the slowing or halting of the cell cycle, allowing efficient DNA repair. Defects in the DNA damage checkpoint result in an overall increase in genomic instability and are thought to fuel cancer progression. To facilitate our understanding of how DNA damage leads to cancer progression, it is crucial to fully comprehend how these signal transduction mechanisms function. In this work, we have characterized in great detail the mechanisms of regulation of Rad53 (a central regulator of the DNA damage response in Saccharomyces cerevisiae) at the genetic, biochemical and structural level. Firstly, we describe a complex biochemical two-step mode of activation of Rad53 by protein-protein interaction and multi-step phosphorylation. We also shed light onto the mechanisms by which Rad53 is turned off to allow the cell cycle to resume, a process termed DNA damage recovery and adaptation. We found that during adaptation, the polo-like kinase Cdc5 is required to attenuate Rad53 catalytic activity. Finally, the study of Rad53 at the molecular and atomic level revealed that in addition to being regulated through a complex network of protein-protein interactions, Rad53 autophosphorylation is orchestrated by a mechanism of dimerization, activation segment phosphorylation via A-loop exchange, as well as through an autoinhibition mechanism regulated by a specific alpha-helical region at the C-terminal extremity of its kinase domain. Such work is important in understanding the function of different proteins in DNA damage signaling. This knowledge will enhance our understanding of the progression of DNA damage related diseases such as cancer, and could eventually help in the long term the development of novel therapeutics as treatments against these conditions.

Neural precursor cells self-renew and differentiate throughout development and in response to neural injury in the adult brain. The DNA damage response in NPCs has yet to be characterized. Patients with defective nucleotide excision repair (NER) demonstrate neurodegeneration dismyelination, and microcephaly, suggesting a potential link to defective NPC function with accumulating DNA damage. We observed reduced self-renewal in Csbm/m and Xpam/m NPCs in response to UV damage. Serial passaging resulted in exhaustion of Csbm/m NPCs in the absence of exogenous DNA damage. In vitro neuronal differentiation resulted in abnormal neuritigenesis after UV DNA damage in Csbm/m NPCs, suggesting defects in the terminal differentiation process. Taken together, the results indicate that DNA damage can modulate the apoptotic, self-renewal and differentiation fates of NPCs.

Neural precursor cells self-renew and differentiate throughout development and in response to neural injury in the adult brain. The DNA damage response in NPCs has yet to be characterized. Patients with defective nucleotide excision repair (NER) demonstrate neurodegeneration dismyelination, and microcephaly, suggesting a potential link to defective NPC function with accumulating DNA damage. We observed reduced self-renewal in Csbm/m and Xpam/m NPCs in response to UV damage. Serial passaging resulted in exhaustion of Csbm/m NPCs in the absence of exogenous DNA damage. In vitro neuronal differentiation resulted in abnormal neuritigenesis after UV DNA damage in Csbm/m NPCs, suggesting defects in the terminal differentiation process. Taken together, the results indicate that DNA damage can modulate the apoptotic, self-renewal and differentiation fates of NPCs.

<i>Escherichia coli </i>RecA is a versatile protein that is involved in homologous recombination, and coordination of both the DNA damage response and translesion synthesis. Single-stranded DNA (ssDNA) that is generated at the site of double-stranded breaks serves as a signal to activate RecA. This allows RecA to form a long helical filament on the ssDNA, which is required in recombination, hydrolysis of ATP, and mediating the self-cleavage of some ser-lys dyad proteins such as the LexA repressor. In this thesis, the formation of the RecA/LexA complex did not require preactivation by ssDNA, instead a volume excluding agent in the presence of LexA was able to stimulate its formation. These preliminary results led to a hypothesis that the formation of the RecA/LexA complex is a thermodynamic process that involves three steps: (1) a change in RecAs conformation towards the active form, (2) a change in LexAs conformation towards the cleavable form (i.e. burial of the ser-lys dyad catalytic residues), and (3) the binding between the active form of RecA and the cleavable form of LexA. Evidence for this model was shown by the ability of either NaCl, LexA K156A, an ATP substrate, or a volume excluding agent to enhance the stability of the RecA/LexA complex, which was detected by both the ATPase and coprotease assays. Hyper-active RecA mutants, isolated form the yeast two-hybrid screen, were also tested, however they did not enhance the stability of the complex. Additionally, RecAs binding preference for the monomer or dimer form of LexA was examined, since it is unknown which species of LexA is able to enhance the stability of the complex. To generate the monomer form of LexA, single point mutations were introduced at the dimer interface of the protein such that its dimerization was disrupted by charge-charge repulsions. Based on the inhibition assay, RecA was found to bind preferentially to dimer form and not the monomer form of LexA, possible reasons for these results are discussed.

A crucial factor in determining the accuracy of DNA replication is maintenance of a balanced supply of deoxyribonucleoside triphosphates (dNTPs) at replication forks. Perturbation of dNTP biosynthesis can induce dNTP pool imbalance with deleterious genetic consequences, including increased mutagenesis, recombination, chromosomal abnormalities and cell death. Using the T4 bacteriophage system, I investigated the molecular basis of mutations induced by imbalanced dNTP pools in vivo. Two approaches were adopted to disturb dNTP biosynthesis: 1) using mutations which affect the deoxyribonucleotide biosynthesis pathway; 2) exogenously supplying mutagenic deoxyribonucleoside analogs which are then taken up by cells and are metabolized to dNTPs. The levels of dNTPs under different conditions were measured in crude extracts of phage-infected cells, while mutagenic effects were quantitated by analysis of certain rII mutations, thought to revert to wild type along either GC-to-AT or AT-to-GC transition pathways. The mutation pathways stimulated by dNTP pool perturbations were confirmed by direct DNA sequencing after amplification of template by the polymerase chain reaction (PCR). By replacing phage ribonucleotide (rNDP) reductase with the host, Escherichia coli, rNDP reductase, in phage-infected cells, I examined the mechanism of mutation induced by the thymidine analog 5- bromodeoxyuridine (BrdUrd) in vivo. Although both AT-to-GC and GC-to- AT transition mutations were stimulated many hundred-fold when cells were grown in medium containing 100 μM BrdUrd, GC-to-AT transitions were stimulated predominantly when T4 reductase was active, while ATto- GC transitions were stimulated more when E. coli reductase was active. By examining the control by dNTPs on CDP reduction, I found that the T4 rNDP reductase is substantially inhibited by either BrdUTP or dTTP in crude enzyme extracts. These experimental results are consistent with the hypothesis that mutagenic effects of BrdUrd are based on dNTP perturbations, supporting the model that rNDP reductase is a major determinant of BrdUrd mutagenesis. I also studied the mutator phenotype of one temperature-sensitive conditional lethal mutant, T4 ts LB3, which specifies a thermolabile T4 deoxycytidylate (dCMP) hydroxymethylase. At the sites of different rII mutations, I found 8- to 80-fold stimulation of GC-to-AT transitions induced by ts LB3 at a semipermissive temperature (34° C). Sequence analysis of revertants from the most sensitive gene marker, rII SN103, showed that either cytosine within the mutated triplet can undergo change to either thymidine or adenine, supporting a model in which mutagenesis induced by ts LB3 at a semipermissive temperature is based on dNTP pool perturbations. The putative depletion of hydroxymethyldeoxycytidine triphosphate (hm-dCTP) caused by the temperature-labile dCMP hydroxymethylase presumably enlarges effective dTTP/hm-dCTP and dATP/hm-dCTP pool ratios, resulting in the observed C-to-T transition and C-to-A transversion mutations. However, no significant dNTP pool abnormalities were observed in extracts from ts LB3 phageinfected cells even when cells were grown at the semi-permissive temperature, suggesting that imbalanced dNTP pools occurred only locally, close to replication forks. These results support a model of dNTP "functional compartmentation", in which DNA replication is fed by a small and rapidly depleted pool, with the bulk of measurable dNTP in a cell representing a replication-inactive pool. To further characterize the mutagenic specificity and DNA site specificity induced by T4 ts LB3, I developed a fast forward mutation approach using thymidine kinase as a marker gene. The studies confirmed that the principal mutagenic effect induced by ts LB3 is C-to- T transition, while C-to-A transversion mutagenesis also occurs. Analysis of DNA sequences around each mutation also suggests that local DNA context influences mutation frequency. / Graduation date: 1991

In order to maintain genome stability, DNA damage needs to be detected and repaired in a timely fashion. To cope with damaged DNA, cells have evolved mechanisms termed "checkpoints", where, upon damage, cells initiate a signal transduction cascade that results in the slowing or halting of the cell cycle, allowing efficient DNA repair. Defects in the DNA damage checkpoint result in an overall increase in genomic instability and are thought to fuel cancer progression. To facilitate our understanding of how DNA damage leads to cancer progression, it is crucial to fully comprehend how these signal transduction mechanisms function. In this work, we have characterized in great detail the mechanisms of regulation of Rad53 (a central regulator of the DNA damage response in Saccharomyces cerevisiae) at the genetic, biochemical and structural level. Firstly, we describe a complex biochemical two-step mode of activation of Rad53 by protein-protein interaction and multi-step phosphorylation. We also shed light onto the mechanisms by which Rad53 is turned off to allow the cell cycle to resume, a process termed DNA damage recovery and adaptation. We found that during adaptation, the polo-like kinase Cdc5 is required to attenuate Rad53 catalytic activity. Finally, the study of Rad53 at the molecular and atomic level revealed that in addition to being regulated through a complex network of protein-protein interactions, Rad53 autophosphorylation is orchestrated by a mechanism of dimerization, activation segment phosphorylation via A-loop exchange, as well as through an autoinhibition mechanism regulated by a specific alpha-helical region at the C-terminal extremity of its kinase domain. Such work is important in understanding the function of different proteins in DNA damage signaling. This knowledge will enhance our understanding of the progression of DNA damage related diseases such as cancer, and could eventually help in the long term the development of novel therapeutics as treatments against these conditions.

<i>Escherichia coli </i>RecA is a versatile protein that is involved in homologous recombination, and coordination of both the DNA damage response and translesion synthesis. Single-stranded DNA (ssDNA) that is generated at the site of double-stranded breaks serves as a signal to activate RecA. This allows RecA to form a long helical filament on the ssDNA, which is required in recombination, hydrolysis of ATP, and mediating the self-cleavage of some ser-lys dyad proteins such as the LexA repressor. In this thesis, the formation of the RecA/LexA complex did not require preactivation by ssDNA, instead a volume excluding agent in the presence of LexA was able to stimulate its formation. These preliminary results led to a hypothesis that the formation of the RecA/LexA complex is a thermodynamic process that involves three steps: (1) a change in RecAs conformation towards the active form, (2) a change in LexAs conformation towards the cleavable form (i.e. burial of the ser-lys dyad catalytic residues), and (3) the binding between the active form of RecA and the cleavable form of LexA. Evidence for this model was shown by the ability of either NaCl, LexA K156A, an ATP substrate, or a volume excluding agent to enhance the stability of the RecA/LexA complex, which was detected by both the ATPase and coprotease assays. Hyper-active RecA mutants, isolated form the yeast two-hybrid screen, were also tested, however they did not enhance the stability of the complex. Additionally, RecAs binding preference for the monomer or dimer form of LexA was examined, since it is unknown which species of LexA is able to enhance the stability of the complex. To generate the monomer form of LexA, single point mutations were introduced at the dimer interface of the protein such that its dimerization was disrupted by charge-charge repulsions. Based on the inhibition assay, RecA was found to bind preferentially to dimer form and not the monomer form of LexA, possible reasons for these results are discussed.