Inheritance of one mutant von Hippel-Lindau (VHL) allele gives rise to the development of the autosomal dominant VHL disease, which affects approximately 1 in 36 000 individuals. The VHL tumour suppressor protein plays a critical role in the E3 ubiquitin ligase-mediated destruction of hypoxia-inducible factor (HIF) and the promotion of fibronectin extracellular matrix assembly. A failure in either process is associated with oncogenic progression. Work included in this thesis provides evidence that these tumour suppressor functions are mutually exclusive. Additionally, post-translational modification of VHL by NEDD8 is shown to act as a ‘molecular switch’, altering VHL protein associations and providing a mechanism of pathway segregation. As a result of HIF stabilization, the expression of a homophilic adhesion molecule E-cadherin is significantly down-regulated in primary renal clear-cell carcinoma (RCC) upon VHL loss. E-cadherin down-regulation is shown to increase the invasive potential and is of prognostic value in RCC. Finally, VHL and SOCS1 are shown to dimerize and negatively regulate the JAK2-STAT5 signalling cascade. Defects in this dimerization are shown to underlie Chuvash polycythemia and provide a molecular understanding of the phenotypic observations associated with VHL-related polycythemias.

MicroRNA is a single-stranded RNA molecule of about 22 nucleotides in length and is expressed endogenously. It functions as a gene regulator by pairing imperfectly with 3’ untranslated region (3’UTR) of target mRNAs, leading to translational inhibition. MicroRNA is implicated in many regulatory pathways and hence affects various cellular activities. In the development of cancer, genetic alterations occurred at miRNA locus and its expression level is dysregulated in various cancers versus normal tissue counterparts. It is thus important to find the targets of dysregulated microRNAs contributing to progression of cancer. To facilitate long term functional studies, a microRNA expression construct with unique futures was generated. Stable expression of miR-378 enhanced cell survival, reduced caspase-3 activity, and promoted tumor growth and angiogenesis. By algorithmic predictions and proteomic analysis, two tumor suppressors, SuFu and Fus-1, were found to be translationally regulated by miR-378. Target validation was confirmed by co-transfection experiments and luciferase activity assays, reassuring its oncogenic role by regulating two tumor suppressor genes simultaneously. Conversely, microRNA can also function as a tumor suppressor by modulating expression of Versican, an extracellular matrix protein known to facilitate tumorigenesis and angiogenesis. By a novel PCR method, more than one microRNA were found to bind to Versican 3’UTR. iii Among these microRNAs, targeting of Versican and Fibronectin by miR199a-3p was validated. Expression of a fragment of Versican 3’UTR was expected to antagonize the function of miR-199a-3p. Stable expression of Versican 3’UTR resulted change in cell morphology and increased cell-cell adhesion. Analysis of primary tissues from transgenic mice expressing versican 3’UTR showed an increase expression of Versican and Fibronectin, and organ adhesion was found between liver and its surrounding tissues. In addition, 3’UTR also modulated the level of miR-199a-3p and miR-136, alleviating translation of negative cell cycle regulators, PTEN and Rb1. This resulted in reduced cell proliferation and hence diminished tumor growth. These findings suggest a role of microRNA in tumor growth, providing a valuable target for therapeutic intervention.

The CTCF protein is involved in several important aspects of gene regulation including transcriptional activation, transcriptional repression and insulator ability. It is also involved in the regulation of epigenetic processes including X-chromosome inactivation and the maintenance of genomic imprinting. CTCF has been shown to bind to approximately 15 000 sites in the mammalian genome and has been implicated in nuclear organization. The CTCF protein mediates long-range chromatin interactions and is believed to form DNA loops. It also acts to block the communication of an enhancer with a promoter by acting as an insulator. Despite its importance in gene regulation, the molecular mechanisms that govern CTCF’s ability to perform its myriad functions remain enigmatic. In this thesis, I add insight into our understanding of the mechanisms behind CTCF’s function. I show that CTCF is post-translationally modified by the Small Ubiquitin-like Modifier proteins and that this post-translational modification contributes to its repressive ability at the c-myc P2 promoter. I also show that CTCF is localized to the sub-nuclear compartment called the Polycomb bodies. The Polycomb protein Pc2 acts as an E3 ligase to enhance the SUMOylation of CTCF by SUMOs 2 and 3. These findings help to explain CTCF’s ability to act as a transcriptional repressor. I also report biochemical evidence to support the role for CTCF in forming an unusual DNA structure, possibly a loop. I hypothesize that a single CTCF binding site is able to form DNA loops. These findings suggest mechanisms by which CTCF is able to organize the mammalian genome and to function as an insulator protein. In addition to these findings I have also purified CTCF interacting proteins through the use of the tandem affinity purification technique. The interacting proteins contain many chromatin and DNA binding proteins further suggesting a role for CTCF in chromatin organization. The results in this thesis enhance our knowledge of the molecular mechanisms of CTCF function and provide a basis for the improved understanding of CTCF mediated gene expression.

The aims of this thesis were to characterize the molecular ecology of the hyporheic zone – between dissolved organic matter (DOM) and microbes – and to test whether seasonal and spatial patterns existed in correlation with seasonal ecosystem processes. The hyporheic zone is an area of vertical integration between groundwater and surface water, and lateral integration between terrestrial and stream ecosystems. Colonization corers were used to collect in situ DOM and bacterial communities from the hyporheic sediments of two streams that varied in hydroperiod (i.e., permanent vs. intermittent). DOM was collected using passive samplers and analyzed using 1H NMR and fluorescence spectroscopy; bacteria were characterized using terminal-restriction fragment length polymorphism. At the permanent site, bacteria correlated significantly with seasonal environmental factors including: fall communities with DOM concentration; spring and winter communities with nitrate concentrations; and summer communities with temperature. Bacterial communities at the intermittent site were significantly correlated with flooding as a function of hydrologic connectivity. Sediment communities were discriminated between hyporheic sediments and interstitial porewaters, and shared several operational taxonomic units (OTUs). Sediment communities were more distinct when hydrologic connectivity was low, and porewater communities changed dramatically upon flooding. Fifteen out of 259 OTUs were shared across aquatic sediments, interstitial porewater and watershed soil samples. DOM was spatially and seasonally dynamic in both sites. Five key DOM groups described using 1H NMR spectroscopy revealed spatial differences between the permanent and intermittent sites. EEM-PARAFAC models confirmed that despite significantly different molecular components, the relative sources of DOM at both sites were similar, including humic-like terrestrial sources and tyrosine (microbial) sources. This study provides new knowledge on both organic matter dynamics and bacterial communities in a dynamic aquatic ecotone, and also confirmed the hypothesis that bacterial communities correlated significantly with ecosystem processes within a watershed.

To date, most genome-scale approaches designed to explore kinase pathways have been targeted towards substrate identification for individual kinases but provide little functional information and only a limited view of the interplay of kinases and their targets in key biological processes. I attempted to tackle the complexity of kinase networks using an unbiased integrated global analysis in budding yeast. I used functional genomics screens to study the yeast kinome using combinatorial genetic perturbations. I first assessed the effects of gene overexpression on the fitness of non-essential kinase deletion mutants to generate a comprehensive view of Synthetic Dosage Lethal (SDL) interactions involving yeast kinases. These data were complemented by assessing genome-wide Synthetic Lethal (SL) interactions for kinases that gave SDL interactions. By measuring >600,000 potential interactions between kinase-gene pairs, I produced a meta-network of ~1300 dosage lethal interactions and 7500 negative and positive genetic interactions. I reasoned that by combining two complementary genetic datasets for kinases, I could: 1) characterize the unexpected phenotypic outcomes that result from the interplay of gain-of-function and loss-of-function phenotypes; 2) better comprehend the complexity of kinase signaling networks and; 3) predict the function of novel genes that arise from the combined genetic network and a gold standard list of known kinase-substrate pairs in the literature. The SDL network alone was enriched for pathways known to be regulated by cognate kinases including phosphoproteins, and kinase targets and kinases that yielded informative SDL interactions were largely those with cell polarity roles. Condition-specific screens and analysis of kinase double mutants suggested that the apparent resistance of many kinases to genetic perturbation cannot be solely attributed to kinase redundancy but most likely reflects the requirement for many kinases in certain activating conditions. Next, I created the first systematic gold standard for kinase-substrate pairs and generated the first kinase interaction database, specifically curated for experiments pertaining to kinase-substrate relationships, in order to analyze the SDL network for identification of kinase targets. Also, using a novel approach that combines the gold standard and the integrated SDL-genetic interaction meta-network, I found that the integrated network is more functionally informative than either SGA or SDL networks alone. Additional integration of known kinase-substrate relationships extracted from the biochemical literature into this network was key to the identification of recurring motifs that enable accurate prediction of single mutant phenotypes. I have also identified ~2000 triplet regulatory network motifs, unraveling novel pathways regulated by kinases. I have tested several motifs using phenotypic and biochemical assays and have identified a novel gene involved in the regulation of cell wall integrity pathway, and found a new regulatory mechanism involving the mitotic exit network machinery. My study provides a general framework for predicting phenotypic outcomes from different combinations of genetic mutations, but also delineates the complexity of signaling pathways involving phosphorylation. The identified network motifs belie the simplistic notion of linear kinase cascades, and support a complex network model where multi-dimensional signaling waves dictate a phenotypic outcome. My data suggest that unraveling the complexity of network biology demands an unbiased global analysis using an integrated set of functional genomics approaches with a high quality gold standard.

The Transforming Growth Factor beta (TGFb) superfamily of ligands are highly versatile, functioning throughout development and in adult organisms as diverse as worms and humans to regulate a myriad of biological activities. TGFb family members signal through their cognate serine/threonine kinase receptors to mediate the phosphorylation and activation of receptor-regulated Smads (R-Smads), that then complex with the common Smad (co-Smad/Smad4) to transduce TGFb signals from the membrane into the nucleus. This thesis recounts the first identification of a mammalian Smad-interacting transcription factor, Foxh1. Investigation of the Smad/Foxh1 DNA-binding complex, which mediates TGFb-dependent regulation of transcription from a Gsc enhancer, determined that both Smad and Foxh1 binding sites are required. These studies also defined the first known biological difference between the highly related R-Smads, Smad2 and Smad3. Specifically, it was shown that while both can similarly participate in Smad/Foxh1 DNA-binding complexes, Smad2 activates and Smad3 represses Foxh1-mediated TGFb-dependent transcription. A detailed analysis of the Gsc enhancer element subsequently defined the sequence req irements for a functional Smad/Foxh1 enhancer (SFE). This information was utilized to direct in silico, genome wide searches for genes harbouring evolutionarily conserved SFEs, which successfully expanded the repertoire of Smad/Foxh1 targets. Analysis of these targets revealed novel roles for Smad/Foxh1 signalling in forebrain development and retinoic acid production. Finally, the importance of Foxh1 to heart development was examined. The interaction between Foxh1 and the heart specific factor Nkx2-5 was characterized with respect to TGFb-dependent regulation of Mef2c expression via a compound Foxh/Nkx2-5 enhancer (FNE). Genome-wide searches for similar FNEs identified many potential Foxh1/Nkx2-5 targets, further analysis of which will provide greater insights into how Foxh1 functions in heart development. In summary, the work presented herein expands our understanding of the role of TGFb in development through the identification and characterization of Foxh1 and its genomic targets downstream of TGFb signalling.

Protein phosphorylation is a central regulatory mechanism in countless cellular processes. Deletion of the PP1 serine/threonine phosphatase gene Ppp1cc in mice results in male infertility due to a severe impairment in spermatogenesis. This disruption in spermatogenesis is hypothesized to arise due to a deficiency of the testis specific Ppp1cc isoform PPP1CC2. To learn more about the function of PPP1CC2 in spermatogenesis, we have employed several proteomic approaches aimed at identifying both regulatory proteins and substrates that interact with PPP1CC2 in the testis. First, we created transgenic mouse embryonic stem cell lines expressing a tandem affinity tagged version of PPP1CC2. Tandem affinity purification using these cell lines identified a number of known PP1 interacting proteins, and one novel interactor DDOST (dolichyl-di-phosphooligosaccharide-protein glycotransferase) which we hypothesize to have a role in spermatogenesis. In a second approach, we conducted GST pull down assays from mouse testis lysates to identify PPP1CC2 interacting proteins. TSSK1 (testis-specific serine kinase 1) was identified as a novel PPP1CC2 interacting protein. We then demonstrated that TSSK1 interacts with PPP1CC2 in an indirect manner via a common interacting protein TSKS (testis-specific serine kinase substrate). Binding of TSKS to PPP1CC2 is regulated via phosphorylation of a PP1 docking motif on the TSKS surface, and localization of TSSK1 and TSKS in the testis is disrupted in Ppp1cc mutants. Finally, to identify candidate substrates of PPP1CC2 in the testis, we conducted a comparative phosphoproteomic analysis and identified 33 different peptides that were hyperphosphorylated in the testis of 3 week old Ppp1cc knockout mice. Amongst these candidate substrates are several proteins essential for mouse spermatogenesis—HMGA1 (high mobility group AT-hook 1), HSPA4 (heat shock protein 4), YBX2 (Y box protein 2) and SYCP2 (synaptonemal complex protein 2). Taken together, our results suggest that PPP1CC2 interacts with a number of different proteins in the testis, and is likely to play a role at several different stages of spermatogenesis, in both meiotic and post-meiotic spermatogenic cells.

APLF is a forkhead associated (FHA) domain-containing protein with unique poly(ADP)-ribose (PAR)-binding zinc finger (PBZ) domains that are involved in the DNA damage response. The interaction of the APLF PBZ domains with PAR is essential for the rapid recruitment of APLF to sites of DNA double strand breaks (DSBs), while the FHA domain facilitates non-homologous end-joining. In response to ionizing radiation (IR), APLF is phosphorylated at Serine-116 (APLFS116), although the function of this post-translational modification has yet to be defined. Here we provide a detailed characterization of the IR-induced and ATM- dependent phosphorylation of endogenous APLF at Serine-116 in the context of DNA damage. We additionally examine a novel APLF FHA-dependent interaction with 53BP1 (p53 Binding protein 1). Together, we illustrate that APLFS116 phosphorylation is dependent upon both the tandem PBZ domains, as well as the FHA-domain, and that the depletion of either PARP3 or 53BP1, similarly affected APLFS116 phosphorylation. Furthermore, we show that DSB-repair was compromised in cells expressing the APLFS116A mutation. Collectively, our findings provide a detailed understanding of the molecular pathway that leads to the phosphorylation of APLF following DNA damage and suggest that APLFS116 phosphorylation facilitates APLF-dependent DSB repair.

The yeast Saccharomyces cerevisiae is a prevalent system for studying gene regulation because of the ease of experimental methods and the simplicity of its gene structure. Here, I describe my work that aims to identify the sequences and factors responsible for demarcating genes within the genome sequence. With comparative genomics and RNA-Seq, we are quite adept at identifying gene structure. However, the cell does not have access to this kind of information. Instead, it uses the specificities of DNA- and RNA-binding proteins to read and interpret the sequence of the genome; it is this process that I have studied in my thesis. In the first chapter, I describe my work collecting yeast transcription factor specificities. I evaluated these specificities using available confirmatory data to determine which one best represents the transcription factor; this gave me a high-confidence description of what DNA sequences yeast transcription factors recognize. Next, I look for over- and under-represented DNA words within and surrounding gene structures and attempt to explain these in terms of the specificities of known factors or other known biological phenomena. I found that the sequences in the 5' and 3' gene ends are very similar and can often be explained by similar phenomena. I also provide evidence that several factors may be involved in regulating transcription in non-canonical ways. In the final chapter, I describe my efforts to build a model that uses my collection of transcription factor specificities as well as DNA structural features to identify gene structure as we think the cell would. This model is comprised of two classifiers that identify mRNA initiation and termination sites, and these are used to provide evidence to a hidden Markov model that predicts gene structure. I test that the predicted determinants of promoter structure are sufficient to initiate transcription, and that the transcription arising from randomly-generated DNA is correctly predicted. Overall, my work demonstrates that the sequence elements demarcating yeast genes are relatively simple in nature, which has implications for how transcription is regulated and how genes evolve.

APLF is a forkhead associated (FHA) domain-containing protein with unique poly(ADP)-ribose (PAR)-binding zinc finger (PBZ) domains that are involved in the DNA damage response. The interaction of the APLF PBZ domains with PAR is essential for the rapid recruitment of APLF to sites of DNA double strand breaks (DSBs), while the FHA domain facilitates non-homologous end-joining. In response to ionizing radiation (IR), APLF is phosphorylated at Serine-116 (APLFS116), although the function of this post-translational modification has yet to be defined. Here we provide a detailed characterization of the IR-induced and ATM- dependent phosphorylation of endogenous APLF at Serine-116 in the context of DNA damage. We additionally examine a novel APLF FHA-dependent interaction with 53BP1 (p53 Binding protein 1). Together, we illustrate that APLFS116 phosphorylation is dependent upon both the tandem PBZ domains, as well as the FHA-domain, and that the depletion of either PARP3 or 53BP1, similarly affected APLFS116 phosphorylation. Furthermore, we show that DSB-repair was compromised in cells expressing the APLFS116A mutation. Collectively, our findings provide a detailed understanding of the molecular pathway that leads to the phosphorylation of APLF following DNA damage and suggest that APLFS116 phosphorylation facilitates APLF-dependent DSB repair.