Typewritten copy / 1 v. / Title page, contents and abstract only. The complete thesis in print form is available from the University Library. / Thesis (Ph.D.)--University of Adelaide, 1953

Stroke induces multiple pathological sequelae directly affecting neuronal survival and eliciting short and long-term deficits in behavioral outcome. Most of stroke models utilized to investigate these pathological consequences are based on pure cerebral ischemia models. However, human thromboembolic stroke is characterized by a complex multifactorial response that involves the activation of the cerebral microcirculation by the occluding thrombus. Here, we have characterized a novel mouse model of tromboembolic stroke that mimics most of the clinical aspects of the human pathology. The common carotid artery thrombosis (CCAT) model produces consistent and reproducible infarcts and triggers an inflammatory response comparable to other well established models of stroke. Several of the pathological consequences of cerebral ischemia are triggered by focal inflammatory processes that occur early after the ischemic event. Cerebral inflammation is initiated by an early release of pro-inflammatory cytokines. These active cytokines promote the recruitment of inflammatory cells from the blood cerebral circulation into the brain parenchyma and subsequent release of additional amounts of inflammatory cytokines. This exacerbated cytokine response result in further irreversible neuronal and histopathological damage. Cytokines interleukyne-1 beta (IL-1 beta) and interleukyne-18 (IL-18) maturation requires the presence of active caspase-1. Activation of caspase-1 in the peripheral immune response involves the recruitment of several caspase-1 molecules into a macromolecular complex termed the inflammasome. Cerebral ischemia triggers the synthesis and activation of caspase-1. However, the cellular mechanisms associated to the activation of caspase-1 in the ischemic brain remain to be elucidated. In this study, we demonstrate that the NLRP1-inflammasome composed by capase-1, ASC (apoptosis-associated speck-like protein containing a caspase-activating recruitment domain) and NLRP1 (NLR (nucleotide binding, leucine-rich repeat) is assembled following the ischemic event. Moreover, we have characterized the cellular distribution of the inflammasome proteins in the normal and the ischemic brain. Data from this investigation suggest that six to twenty four hours following CCAT the inflammasome complex is assembled in neurons while microglia, macrophages and astrocytes form this complex at 7 days following cerebral ischemia. On the basis of these findings we next investigated whether inhibition of the inflammasome complex reduces the inflammatory response after ischemia. Neutralization of NLRP1 utilizing a specific antibody, revealed decreased activation of caspase-1 and IL-1 beta and reduced histopathological damage within the ischemic brain. Thus, the inflammasome complex is a major contributor of the inflammatory response following cerebral ischemia and inhibition of this complex may be a novel therapeutic target for reducing the pathological consequences of stroke.

An arginine (R) to a glycine (G) mutation at position 145 in the highly reserved inhibitory domain of cardiac troponin I (cTnI) is associated with hypertrophic cardiomyopathy (HCM), an autosomal dominant disease characterized by left ventricular hypertrophy. An arginine (R) to tryptophan (W) mutation at the same position in cTnI is associated with restrictive cardiomyopathy (RCM), a disease characterized by diastolic dysfunction with normal left ventricular size and normal systolic function. In this study we addressed the functional consequences of the human cardiac troponin I (hcTnI) HCM R145G mutation and hcTnI RCM R145W mutation in transgenic mice. Simultaneous measurements of the ATPase activity and force in skinned papillary fibers from hcTnI R145G transgenic mice (Tg-R145G) versus hcTnI wild type transgenic mice (Tg-WT) showed a significant decrease in the maximal Ca2+ activated force without changes in the maximal ATPase activity and an increase in the Ca2+ sensitivity by both ATPase activity and force development. No difference in the cross-bridge turnover rate was observed at the same level of cross-bridge attachment (activation state) showing that changes in Ca2+ sensitivity were not due to changes in cross-bridge kinetics. Energy cost calculations demonstrated higher energy consumption in Tg-R145G fibers compared to Tg-WT fibers. The addition of 3mM BDM at pCa 9.0 showed that there was approximately 2~4 percent of force generating cross-bridges attached in Tg-R145G fibers compared to less than 1.0 percent in Tg-WT fibers, suggesting the mutation impairs the ability of the cardiac troponin complex to fully inhibit cross-bridge attachment under relaxing conditions. Prolonged force and intracellular [Ca2+] transients in electrically stimulated intact papillary muscles were observed in Tg-R145G compared to Tg-WT. These results suggest that the phenotype of HCM is most likely caused by the compensatory mechanisms in the cardiovascular system which are activated by: 1) higher energy cost in the heart resulting from a significant decrease in average force per cross-bridge; 2) incomplete relaxation (diastolic dysfunction) caused by prolonged [Ca2+] and force transients; and 3) an inability of the cardiac TnI to completely inhibit activation at low levels of diastolic Ca2+ in Tg-R145G. Simultaneous measurements of the ATPase activity and force in transgenic skinned papillary fibers from hcTnI R145W transgenic mice (Tg-R145W) versus Tg-WT showed that there was a ~13 to ~16 percent increase in the maximal Ca2+ activated force and ATPase activity, respectively. The rate of dissociation of force generating cross-bridges (g) and energy cost (ATPase/force) was the same in all groups of fibers. These results suggest that the increase in force and ATPase activity is associated with an increase in the number of force generating cross-bridges attached at all activation levels. Additionally, there was a large increase in the Ca2+ sensitivity of force development and ATPase activity. In intact fibers, the mutation caused prolonged force and intracellular [Ca2+] transients, as expected due to the increased Ca2+ sensitivity (slower dissociation rate of Ca2+ from cTnC). The above cited results suggest that: 1) there would be an increase in resistance to ventricular filling during diastole resulting from the prolonged force and Ca2+ transients, especially at high heart rates; 2) there would be a decrease in ventricular filling (diastolic dysfunction); and 3) an increase in contractility during systole that would off-set the negative effect of a decrease in diastolic filling on ventricle stroke volume thus allowing the heart to maintain normal stroke volume despite the compromise in RCM (Tg-R145W) heart.

Induction of type I interferons by the transcription factors IRF3 and IRF7 is essential in the initiation of antiviral innate immunity. Activation of IRF3/7 requires C-terminal phosphorylation by the upstream kinases TBK1/IKKi, where IRF3/7 phosphorylation promotes dimerization, and subsequent nuclear translocation to the IFN-beta promoter. Recent studies have described the ubiquitin-editing enzyme A20 as a negative regulator of IRF3 signaling by associating with TBK1/IKKi, however the regulatory mechanism of A20 inhibition remains unclear. Here we describe the adaptor protein, TAX1BP1, as a key regulator of A20 function in terminating signaling to IRF3. Murine embryonic fibroblasts (MEFs) deficient in TAX1BP1 displayed increased amounts of IFN-beta production upon viral challenge compared to WT MEFs. TAX1BP1 inhibited virus-mediated activation of IRF3 at the level of TBK1/IKKi. TAX1BP1 and A20 blocked antiviral signaling by disrupting K63-linked polyubiquitination of TBK1/IKKi independently of the A20 deubiquitination (DUB) domain. Furthermore, TAX1BP1 was required for A20 effector function as A20 was defective for the targeting and inactivation of TBK1 and IKKi in Tax1bp1–/– MEFs. Additionally, we found the E3 ubiquitin ligase TRAF3 to play a critical role in promoting TBK1/IKKi ubiquitination. Collectively, our results demonstrate TBK1/IKKi to be novel substrates for A20 and further identifies a novel mechanism whereby A20 and TAX1BP1 restrict antiviral signaling by disrupting a TRAF3/TBK1/IKKi signaling complex. Several viruses utilize a number of strategies to evade the host innate immune response by inhibiting the production of type I interferons. The Human T-cell leukemia virus type 1 (HTLV-1) has been shown to block interferon signaling, however the mechanism of inhibition is poorly understood. We show here that the HTLV-1 encoded protein, Tax plays a critical role in blunting the activation of type I interferons. Tax expression rendered MEFs hyper-permissive in supporting virus replication. Correspondingly, Tax blocked the production of IFN-beta. Interestingly, Tax did not require NEMO interaction to inhibit antiviral signaling to IRF3/7. Instead, Tax targeted RIP1 and further blocked IRF7 K63-linked polyubiquitination. Altogether, we show that Tax inhibits IFN activation by disrupting the ubiquitin dependent activation of IRF7 mediated by RIP1.

<p>Neural networks are mathematical formulations that can be "trained" to perform certain functions. One particular application of these networks of interest in this thesis is to "model" a physical system using only input-output information. The physical system and the neural network are subjected to the same inputs. The neural network is then trained to produce an output which is the same as the physical system for any input. This neural network model so created is essentially a "blackbox" representation of the physical system. This approach has been used at the University of Saskatchewan to model a load sensing pump (a component which is used to create a constant flow rate independent of variations in pressure downstream of the pump). These studies have shown the versatility of neural networks for modeling dynamic and non-linear systems; however, these studies also indicated challenges associated with the morphology of neural networks and the algorithms to train them. These challenges were the motivation for this particular research.</p> <p>Within the Fluid Power Research group at the University of Saskatchewan, a "global" objective of research in the area of load sensing pumps has been to apply dynamic neural networks (DNN) in the modeling of loads sensing systems.. To fulfill the global objective, recurrent generalized neural network (RGNN) morphology along with a non-gradient based training approach called the complex algorithm (CA) were chosen to train a load sensing pump neural network model. However, preliminary studies indicated that the combination of recurrent generalized neural networks and complex training proved ineffective for even second order single-input single-output (SISO) systems when the initial synaptic weights of the neural network were chosen at random.</p> <p>Because of initial findings the focus of this research and its objectives shifted towards understanding the capabilities and limitations of recurrent generalized neural networks and non-gradient training (specifically the complex algorithm). To do so a second-order transfer function was considered from which an approximate recurrent generalized neural network representation was obtained. The network was tested under a variety of initial weight intervals and the number of weights being optimized. A definite trend was noted in that as the initial values of the synaptic weights were set closer to the "exact" values calculated for the system, the robustness of the network and the chance of finding an acceptable solution increased. Two types of training signals were used in the study; step response and frequency based training. It was found that when step response and frequency based training were compared, step response training was shown to produce a more generalized network.</p> <p>Another objective of this study was to compare the use of the CA to a proven non-gradient training method; the method chosen was genetic algorithm (GA) training. For the purposes of the studies conducted two modifications were done to the GA found in the literature. The most significant change was the assurance that the error would never increase during the training of RGNNs using the GA. This led to a collapse of the population around a specific point and limited its ability to obtain an accurate RGNN.</p> <p>The results of the research performed produced four conclusions. First, the robustness of training RGNNs using the CA is dependent upon the initial population of weights. Second, when using GAs a specific algorithm must be chosen which will allow the calculation of new population weights to move freely but at the same time ensure a stable output from the RGNN. Third, when the GA used was compared to the CA, the CA produced more generalized RGNNs. And the fourth is based upon the results of training RGNNs using the CA and GA when step response and frequency based training data sets were used, networks trained using step response are more generalized in the majority of cases.</p>

The course of pathogenic bacterial infection is dependent on the interactions between the host immune response and the bacterial virulence mechanisms. Our lab previously discovered that the Survival of Motor Neuron (SMN) protein complex undergoes a change in subcellular localization during infection with invasive Shigella bacteria, forming novel cytoplasmic aggregates called "U bodies". Similar results were obtained with other intracellular bacterial pathogens suggesting that these U bodies are a fundamental entity in microbial pathogenesis. Notably, the SMN complex normally plays a key role in the assembly of the spliceosomal U snRNA. We have shown during infection that there are changes in U snRNA maturation and splicing patterns. Importantly, we have found that U bodies are downstream of a stress pathway involving the stress-inducible ATF3 protein. Altogether, intracellular bacterial infection induces novel cellular stress pathways that disrupt normal SMN complex function and leads to changes in U snRNA associated functions.

The bounded Fatou components for certain quadratic polynomials are attached to each other at the boundary and form chain-like structures called ``bubble rays". In the context of mating quadratic polynomials, these bubble rays can serve as a replacement for external rays. The main objective of this thesis is to apply this idea to the mating of starlike quadratics.

The bounded Fatou components for certain quadratic polynomials are attached to each other at the boundary and form chain-like structures called ``bubble rays". In the context of mating quadratic polynomials, these bubble rays can serve as a replacement for external rays. The main objective of this thesis is to apply this idea to the mating of starlike quadratics.

<p>Neural networks are mathematical formulations that can be "trained" to perform certain functions. One particular application of these networks of interest in this thesis is to "model" a physical system using only input-output information. The physical system and the neural network are subjected to the same inputs. The neural network is then trained to produce an output which is the same as the physical system for any input. This neural network model so created is essentially a "blackbox" representation of the physical system. This approach has been used at the University of Saskatchewan to model a load sensing pump (a component which is used to create a constant flow rate independent of variations in pressure downstream of the pump). These studies have shown the versatility of neural networks for modeling dynamic and non-linear systems; however, these studies also indicated challenges associated with the morphology of neural networks and the algorithms to train them. These challenges were the motivation for this particular research.</p> <p>Within the Fluid Power Research group at the University of Saskatchewan, a "global" objective of research in the area of load sensing pumps has been to apply dynamic neural networks (DNN) in the modeling of loads sensing systems.. To fulfill the global objective, recurrent generalized neural network (RGNN) morphology along with a non-gradient based training approach called the complex algorithm (CA) were chosen to train a load sensing pump neural network model. However, preliminary studies indicated that the combination of recurrent generalized neural networks and complex training proved ineffective for even second order single-input single-output (SISO) systems when the initial synaptic weights of the neural network were chosen at random.</p> <p>Because of initial findings the focus of this research and its objectives shifted towards understanding the capabilities and limitations of recurrent generalized neural networks and non-gradient training (specifically the complex algorithm). To do so a second-order transfer function was considered from which an approximate recurrent generalized neural network representation was obtained. The network was tested under a variety of initial weight intervals and the number of weights being optimized. A definite trend was noted in that as the initial values of the synaptic weights were set closer to the "exact" values calculated for the system, the robustness of the network and the chance of finding an acceptable solution increased. Two types of training signals were used in the study; step response and frequency based training. It was found that when step response and frequency based training were compared, step response training was shown to produce a more generalized network.</p> <p>Another objective of this study was to compare the use of the CA to a proven non-gradient training method; the method chosen was genetic algorithm (GA) training. For the purposes of the studies conducted two modifications were done to the GA found in the literature. The most significant change was the assurance that the error would never increase during the training of RGNNs using the GA. This led to a collapse of the population around a specific point and limited its ability to obtain an accurate RGNN.</p> <p>The results of the research performed produced four conclusions. First, the robustness of training RGNNs using the CA is dependent upon the initial population of weights. Second, when using GAs a specific algorithm must be chosen which will allow the calculation of new population weights to move freely but at the same time ensure a stable output from the RGNN. Third, when the GA used was compared to the CA, the CA produced more generalized RGNNs. And the fourth is based upon the results of training RGNNs using the CA and GA when step response and frequency based training data sets were used, networks trained using step response are more generalized in the majority of cases.</p>

The goal of this project was to identify and characterize polymorphic markers spanning regions of the bovine major histocompatibility complex (BoLA) to analyze patterns of genetic variation and haplotype structure across diverse cattle breeds with various breed histories and selection pressures. Genetic markers that demonstrated sufficient levels of polymorphism, locus specificity, Mendelian inheritance, and the accurate typing of alleles across diverse haplotypes were chosen to define separate haplotype structures for the BoLA IIb and BoLA IIa-III-I regions and to evaluate breakpoints in linkage disequilibrium within the regions surrounding BoLA IIa-III-I. A total of 23 microsatellites, two SNPSTRs, 62 SNPs, and the alleles of three class IIa genes were selected for use in this study. These markers revealed eleven recombination events, low levels of recombination in BoLA IIa-III-I, a sharp break in haplotype structure in the region centromeric to class IIa, prolonged linkage disequilibrium in the extended class I region, strong conservation of BoLA IIa-III-I haplotype structure, BoLA IIa-III-I homozygous haplotype identity across seven different breeds of cattle, and a small number of common BoLA IIa-III-I haplotypes within the Angus and Holstein breeds. This work demonstrated that 52 SNPs from the Illumina 50K SNPchip could accurately predict BoLA IIa-III-I haplotypes. These 52 SNPs represent tagSNPs that can predict BoLA IIa-III-I genetic variation and could offer a cost-effective means for screening large sample sizes for haplotype/disease association studies in the future.