Abstract The aim of this PhD-study was to address some of the overlapping bottlenecks in protein engineering and metagenomics by developing or applying new tools which are useful for both disciplines. Two enzymes were studied as an example: Triosephosphate Isomerase (TIM) and Uridine Phosphorylase (UP). TIM is an important enzyme of the glycolysis pathway and has been investigated via means of protein engineering, while UP is a key enzyme in the pyrimidine-salvage pathway. In this thesis TIM was used to address protein engineering aspects, while UP was used in regards to some metagenomic and bioinformatic aspects. The aspects of a structural driven rational design approach and its implications for further engineering of monomeric TIM variants are discussed. Process development based on a new technology, EnBase®, addresses the relative instability of new variants, compared to its ancestors, for further studies. EnBase® is then applied for the production of 15N isotope labeling of a monomeric TIM variant, A-TIM. Systematical function- and engineering studies on dimeric TIM and monomeric TIM in regards to the hinges of the catalytic loop-6 were conducted to investigate enzyme activity and stability. Both the A178L and P168A were proposed to induce loop-6 closure, a wanted feature for A-TIM variants. The P168A mutants are hardly active, but gave great insight into the catalytic machinery, while the A178L mutants did induce partial loop-6 closure, however in addition, monomeric A178L was destabilized. Homology driven genome mining and subsequent isolation- high throughput (HTP) overexpression of a thermostable UP from the Archaea Aeopyrum pernix was carried out as an example for the production of recombinant proteins. In addition an alternative kinetic method to study the kinetics of UP by means of NMR directly from cell lysate is discussed. The combination of expression libraries and EnBase® in a HTP manner may relieve up the gene-to-product bottleneck. The structural aspects of A. pernix UP are explored by means of simple bioinformatic tools in the last section of this thesis. A thermostable, truncated version of UP was created and its use for protein engineering in the future is explored. The long N-terminal and C-terminal ends of A. pernix UP seem to be involved in stabilizing the dimeric and hexameric structures of UP. However, deletion of the N-terminal end of A. pernix UP yielded a thermostable protein. Overall, the finding in regards to process optimization and HTP expression and optimization and the underlying methods used in the TIM studies and the UP studies are interchangeable.

Glioblastoma Multiforme (GBM) is a primary malignant brain cancer, affecting children and adults and has a very poor prognosis. Up to 30% of the tumour mass consists of host derived immune cells, and a better understanding of how these cells affect tumour behaviour is required. These cells, called ‘Tumour Associated Macrophages’ (TAM) have been shown to occur in peripheral solid organ cancers, where they can cause local immune suppression, increase invasiveness and aid tumour growth. In the brain however, TAMs can consist of centrally derived microglia and peripherally derived macrophages, and although these cells could be exerting different effects on the tumour, there is currently no reliable way of distinguishing between the two. Cancer Stem Cells (CSC) are a subpopulation of cells within the tumour mass with stem-like features, are capable of self-renewal, and can recapitulate a tumour in an immunocompromised mouse host. It is thought that these cells play a role in disease recurrence and hence represent a potential target for anti-GBM therapies. In the first project we investigate the interaction between Cancer Stem Cells and TAMs. We first describe a method of culturing CSCs and TAMs from a single human patient sample, followed by an investigation into the functional properties of these cell types. We found functional differences between established cell line pairings of U87-MG and CHME3 versus primary patient derived CSCs and TAM cell line pairings. Polarisation of microglia/TAMs with lipopolysaccharide caused significant decrease in proliferative capacity of the GBM cell lines. Next we used a Non-Myeloablative Conditioning System (NMCS) to selectively replace the peripheral bone marrow compartment of wild type animals with labelled bone marrow cells, without disturbing brain homeostasis. We demonstrate that peripheral cells home exclusively to the orthotopically implanted tumour, and that some of these cells are CD11b+ and Gr1+, suggestive of myeloid derived suppressor cells (MDSC). We evaluate current CD45 based gating strategies for distinguishing peripheral and central cells and show them to be inadequate. Finally we compared the chemosensitivity profiles of different patient derived CSC lines using high throughput content screening (HTCS), against currently approved chemotherapeutic drugs and rank these drugs in a response space, based on HTCS parameters including 2D and 3D culture with and without irradiation. Differential chemosensitivities were noted between different patient derived cell lines. Drugs not currently used in the treatment of GBM such as Actinomycin D and Mitoxantrone were also identified using HTCS, suggesting the potential utility of such an approach to personalised treatments in GBM.

Many systems for field-based, high-throughput phenotyping (FB-HTP) quantify and characterize the reflected radiation from the crop canopy to derive phenotypes, as well as infer plant function and health status. However, given the technology's nascent status, it remains unknown how biophysical and physiological properties of the plant canopy impact downstream interpretation and application of canopy reflectance data. In that light, we assessed relationships between leaf thickness and several canopy-associated traits, including normalized difference vegetation index (NDVI), which was collected via active reflectance sensors carried on a mobile FB-HTP system, carbon isotope discrimination (CID), and chlorophyll content. To investigate the relationships among traits, two distinct cotton populations, an upland (Gossypium hirsutum L.) recombinant inbred line (RIL) population of 95 lines and a Pima (G, barbaderise L.) population composed of 25 diverse cultivars, were evaluated under contrasting irrigation regimes, water-limited (WL) and well-watered pm conditions, across 3 years. We detected four quantitative trait loci (QTL) and significant variation in both populations for leaf thickness among genotypes as well as high estimates of broad-sense heritability (on average, above 0.7 for both populations), indicating a strong genetic basis for leaf thickness. Strong phenotypic correlations (maximum r = -0.73) were observed between leaf thickness and NDVI in the Pima population, but not the RIL population. Additionally, estimated genotypic correlations within the RIL population for leaf thickness with CID, chlorophyll content, and nitrogen discrimination (r(gij) = -0.32, 0.48, and 0.40, respectively) were all significant under WW but not WL conditions. Economically important fiber quality traits did not exhibit significant phenotypic or genotypic correlations with canopy traits. Overall, our results support considering variation in leaf thickness as a potential contributing factor to variation in NDVI or other canopy traits measured via proximal sensing, and as a trait that impacts fundamental physiological responses of plants.

Doctor of Philosophy / Department of Chemistry / Stefan Bossmann / Numerous Proteases are implicated in cancer initiation, survival, and progression. Therefore, it is important to diagnose the levels of protease expression by tumors and surrounding tissues, which are reflected in blood and tissue samples. Nanoplatforms for Cathepsin(CTS) B and L, matrix metalloproteinases(MMP) 1, 2, 3, 7, 9, 13 and urokinase plasminogen activator(uPA) detection have been synthesized. Nanoplatforms feature a central dopamine-coated core/shell Fe/Fe₃O₄ nanoparticle. Cyanine 5.5 is permanently tethered to the dopamine ligands via amide bonds. Tetrakis(4-carboxy-phenyl)porphyrin (TCPP) is co-tethered to Fe/Fe₃O₄/dopamine by means of protease consensus sequences. In the presence of a relevant protease sequence, it is cleaved, releasing TCPP from the nanoplatform. In contrast, Cy 5.5 will remain permanently tethered to the nanoparticle. Therefore, an extensive increase of emission intensity of the fluorescence signal from TCPP is observed. This permits the detection of the activity of proteases at femtomolar levels in biospecimens by fluorescence spectroscopy. 46 breast cancer and 20 healthy human blood serum samples were analyzed. Based on the expression pattern of analyzed enzymes, human breast cancer can be detected at stage I. By monitoring CTS B and L stage 0 detection may be achieved. This study demonstrates the feasibility of minimally invasive successful early cancer diagnosis. Immunosuppression is one of the hallmarks of aggressive cancers. Arginase is overexpressed in cancer patients, resulting in systemic immunosuppression. Two nanoplatforms for arginase detection have been synthesized. Both feature a central dopamine-coated core/shell Fe/Fe₃O₄ nanoparticle to which cyanine 7.0 or cyanine 7.5 is tethered via amide bonds. In both nanoplatforms, cyanine 5.5 is linked to the N-terminal of the peptide sequence GRRRRRRRG. Arginine (R) reacts to ornithine (O) in the presence of arginase. According to our results obtained from fluorescence spectroscopy, the oligopeptides GRRRRRRRG and GOOOOOOOG differ in their chain dynamics. In the presence of arginase, and dependent on arginase activity, fluorescence increase of both nanoplatforms is observed, which is an indication that proton-transfer quenching decreases when arginine gets converted to ornithine. The novel assays permit the detection of active arginase within an hour. Additionally, Förster Resonance Energy Transfer (FRET) is observed in nanoplatforms featuring cy 5.5/7.0 pairs, resulting in picomolar detection limits. This is the first example of a “post-translational” enzyme sensor, in which the tether is subjected to chemical transformations of the aminoacid side chains and not cleaved by an enzyme, resulting in the modified mobility of the tether. The nanoplatforms do not show a fluorescence increase when incubated with NO-reductase, an enzyme indicative of immunoactivation, which also uses arginase as substrate. Copper dependent inhibitory activity of 10000 compound library has been studied against of Staphylococcus aureus. 53 copper- dependent hit molecules were recognized featuring extended thiourea core structure with NNSN motif. NMR titrations, UV/Vis studies have been performed for characterization of metal complexation and structure modeling. Chemoinformatic meta-analysis of the ChEMBL chemical database confirmed the NNSNs as an unrecognized staphylococcal inhibitor, in spite of other compound groups in chemical screening libraries. This will lead to the development of novel class of antibacterial agents against Staphylococcus aureus.

Fueled by increasing human appetite for high computing performance, semiconductor technology has now marched into the deep sub-micron era. As transistor size keeps shrinking, more and more transistors are integrated into a single chip. This has increased tremendously the power consumption and heat generation of IC chips. The rapidly growing heat dissipation greatly increases the packaging/cooling costs, and adversely affects the performance and reliability of a computing system. In addition, it also reduces the processor's life span and may even crash the entire computing system. Therefore, dynamic thermal management (DTM) is becoming a critical problem in modern computer system design. Extensive theoretical research has been conducted to study the DTM problem. However, most of them are based on theoretically idealized assumptions or simplified models. While these models and assumptions help to greatly simplify a complex problem and make it theoretically manageable, practical computer systems and applications must deal with many practical factors and details beyond these models or assumptions. The goal of our research was to develop a test platform that can be used to validate theoretical results on DTM under well-controlled conditions, to identify the limitations of existing theoretical results, and also to develop new and practical DTM techniques. This dissertation details the background and our research efforts in this endeavor. Specifically, in our research, we first developed a customized test platform based on an Intel desktop. We then tested a number of related theoretical works and examined their limitations under the practical hardware environment. With these limitations in mind, we developed a new reactive thermal management algorithm for single-core computing systems to optimize the throughput under a peak temperature constraint. We further extended our research to a multicore platform and developed an effective proactive DTM technique for throughput maximization on multicore processor based on task migration and dynamic voltage frequency scaling technique. The significance of our research lies in the fact that our research complements the current extensive theoretical research in dealing with increasingly critical thermal problems and enabling the continuous evolution of high performance computing systems.

A major driving force for development of new catalyst systems is the need for more efficient synthesis of chemical compounds essential to modern life. Catalysts having superior performance offer significant environmental and economic advantages, but their discovery is not trivial. Well-defined, homogeneous catalysts can offer unparalleled understanding of ligand effects, which proves invaluable in directing redesign strategies. This thesis work focuses on the design of ruthenium complexes for applications in dinitrogen activation and olefin metathesis. The complexes developed create new directions in small-molecule activation and asymmetric catalysis by late-metal complexes. Also examined are the dual challenges, ubiquitous in catalysis, of adequate interrogation of catalyst structure and performance. Insight into both is essential to enable correlation of ligand properties with catalyst activity and/or selectivity. Improved methods for accelerated assessment of catalyst performance are described, which expand high-throughput catalyst screening to encompass parallel acquisition of kinetic data. A final aspect focuses on direct examination of metal complexes, both as isolated species, and under catalytic conditions. Applications of charge-transfer MALDI mass spectrometry to structural elucidation in organometallic chemistry is described, and the technique is employed to gain insight into catalyst decomposition pathways under operating conditions.

This work explores the use of computational methods to aid in the design of Metal Organic Frameworks (MOFs) for use as CO2 scrubbers in carbon capture and storage applications. One of the main challenges in this field is in identifying important MOF design characteristics which optimize the complex interactions governing surface adsorption. We approach this in a high-throughput manner, determining properties important to CO2 adsorption from generating and sampling a large materials search space. The utilization of MOFs as potential carbon scrubbing agents is a recent phenomenon, as such, many of the computational tools necessary to perform high-throughput screening of MOFs and subsequent analysis are either underdeveloped or non-existent. A large portion of this work therefore involved the development of novel tools designed specifically for this task. The chapters in this work are contiguous with the goal of designing MOFs for CO¬2 capture, and somewhat chronological in order and complexity, meaning as time and expertise progressed, more advanced tools were developed and utilized for the purposes of computational MOF discovery. Initial work towards MOF design involved the detailed analysis of two experimental structures; CALF-15 and CALF-16 using classical molecular dynamics, grand canonical Monte Carlo simulations, and DFT to determine the structural features which promote CO2 adsorption. An unprecedented level of agreement was found between theory and experiment, as we are able to capture, with simulation, the X-ray resolved binding sites of CO2 in the confined pores of CALF-15. Molecular simulation was then used to provide a detailed breakdown of the energy contributions from nearby functional groups in both CALF-15 and CALF-16. A large database of hypothetical MOF structures is constructed for the purposes of screening for CO2 adsorption. The database contains 1.3 million hypothetical structures, generated with an algorithm which snaps together rigid molecular building blocks extracted from existing MOF crystal structures. The algorithm for constructing the hypothetical MOFs and the building blocks themselves were all developed in-house to form the resulting database. The topological, chemical, and physical features of these MOFs are compared to recently developed materials databases to demonstrate the larger structural and chemical space sampled by our database. In order to rapidly and accurately describe the electrostatic interactions of CO2 in the hypothetical database of MOFs, parameters were developed for use with the charge equilibration method. This method assigns partial charges on the framework atoms based on a set of parameters assigned to each atom type. An evolutionary algorithm was used to optimize the charge equilibration parameters on a set of 543 hypothetical MOFs such that the partial charges generated would reproduce each MOFs DFT-derived electrostatic potential. Validation of these parameters was performed by comparing the CO2 adsorption from the charge equilibration method vs DFT-derived charges on a separate set of 693 MOFs. Our parameter set were found to reproduce DFT-derived CO2 adsorption extremely well using only a fraction of the time, making this method ideal for rapid and accurate high-throughput MOF screening. A database of 325,000 MOFs was then screened for CO2 capture and storage applications. From this study we identify important binding pockets for CO2 in MOFs using a binding site analysis tool. This tool uses a pattern recognition method to compare the 3-D configurations of thousands of pore structures surrounding strong CO2 adsorption sites, and present common features found amongst them. For the purposes of developing larger databases which sample a more diverse materials space, a novel MOF construction tool is devloped which builds MOFs based on abstract graphs. The graph theoretical foundations of this method are discussed and several examples of MOF construction are presented to demonstrate its use. Notably, not only can it build existing MOFs with complicated geometries, but it can sample a wide range of unique structures not yet discovered by experimental means.

Reducing anthropogenic carbon dioxide emissions from coal-fired power plants is an important step in mitigating climate change. To implement carbon dioxide capture technologies, materials capable of removing carbon dioxide efficiently are required. Currently, liquid amine technology is used for carbon dioxide capture. However, the mechanism for carbon dioxide removal in liquid amine requires extraordinary amounts of energy input. Alternatively, solid sorbents such as metal-organic frameworks (MOFs) show promising potentials as a type of material for carbon dioxide capture. Due their varying structural properties, MOFs can be configured for specific purposes. Certain MOFs carry a net charge on their frameworks, which may allow for increased interactions with carbon dioxide molecules. In this work, charged MOFs were studied for their potential in carbon dioxide capture. Due to the massive number of MOFs available, computational methods were employed for the study. This project includes three major components: (1) the development of novel computational methods to simulate the gas adsorption properties in charged materials, (2) a diverse database of 47,244 hypothetical charged MOFs was constructed to represent the capabilities of charged MOFs, and (3) screening of high performing charged MOFs for carbon capture application by combining the previous two portions of the project. The methods developed in this work include fitting intermolecular interaction parameters to quantum mechanical calculations in periodic systems with net charges. No methods have been reported in literature for such parameter fittings, even in well studied materials such as zeolites. Therefore, the gas adsorption estimation method for charged materials developed in this work is proprietary. Also, databases of hypothetical MOFs with framework net charges have never been reported previously in literature. By screening the charged MOFs in the database with the methods developed, gas adsorption capabilities were evaluated. The adsorption properties of a neutral group of hypothetical MOFs were also obtained for a baseline comparison. Between the two groups of MOFs, charged MOFs were found to outperform neutral MOFs in three key aspects. Firstly, charged MOFs were able to adsorb an average of three times as much carbon dioxide than the neutral group. Secondly, charged MOFs were capable of removing twice the amount of carbon dioxide per adsorption/desorption cycle than the neutral MOFs. Lastly, charged MOFs were able to selectively adsorb much more carbon dioxide over other gasses present in the carbon dioxide capture situations. Specific structural features that resulted in the selectiveness of adsorption in charged MOFs were identified. Also, positive correlations were found between the adsorption of carbon dioxide and the charge present in the MOFs. As seen in the results, charges present in MOFs can greatly increase their ability to remove carbon dioxide. Charged MOFs in the hypothetical database not only outperformed neutral MOFs, certain top performers were also found to exceed the requirements for post-combustion carbon capture application. Therefore, charged MOFs were shown to be a possible material for future carbon dioxide capture. The proprietary methods developed in this work can not only be used to simulate gas adsorptions in charged MOFs, but also for other porous materials, regardless of net charges presented in their systems. Also, the database constructed in this work can be utilized in multiple ways. Aside from carbon dioxide capture capabilities, the charged MOFs in the database can be screened for other gas separations and catalysis via high throughput screening. The database and the computational methods developed in this work pave the way for discovering the capabilities of charged materials.

Soluble dietary fibers (SDF) are fermented more than insoluble dietary fibers (IDF), but their effect on colonic bacterial community structure and function remains unclear. Thus, bacterial community composition and function in the colon of BALB/c mice (n = 7) fed with a high level (approximately 20%) of typical SDF, oat-derived beta-glucan (G), microcrystalline cellulose (M) as IDF, or their mixture (GM), were compared. Mice in group G showed a lowest average feed intake (p < 0.05) but no change on the average body weight gain (p > 0.05) compared to other groups, which may be associated with the highest concentration of colonic propionate (p < 0.05) in these mice. The bacterial alpha-diversity of group G was significantly lower than other groups (p < 0.01). In group G, the relative abundance of bacteria belonging to the phylum Bacteroidetes was significantly increased, whereas bacteria from the phylum Firmicutes were significantly decreased (p < 0.01). The core bacteria for different treatments showed distinct differences. Bacteroides, Dehalobacterium, and Prevotella, including known acetogens and carbohydrate fermenting organisms, were significantly increased in relative abundance in group G. In contrast, Adlercreutzia, Odoribacter, and Coprococcus were significantly more abundant in group M, whereas Oscillospira, Desulfovibrio, and Ruminoccaceae, typical hydrogenotrophs equipped with multiple carbohydrate active enzymes, were remarkably enriched in group GM (p < 0.05). The relative abundance of bacteria from the three classes of Proteobacteria, Betaproteobacteria, Gammaproteobacteria (including Enterobacteriaceae) and Deltaproteobacteria, were significantly more abundant in group G, indicating a higher ratio of conditional pathogenic bacteria in mice fed dietary beta-glucan in current study. The predicted colonic microbial function showed an enrichment of "Energy metabolism" and "Carbohydrate metabolism" pathways in mice from group G and M, suggesting that the altered bacterial community in the colon of mice with the two dietary fibers probably resulted in a more efficient degradation of dietary polysaccharides. Our result suggests that the influence of dietary beta-glucan (SDF) on colonic bacterial community of mice was more extensively than MCC (IDF). Co-supplementation of the two fibers may help to increase the bacterial diversity and reduce the conditional pathogens in the colon of mice.

Mathematical modelling is playing an increasingly important role in helping us to understand biological systems. The construction of biological models typically requires the use of experimentally-measured parameter values. However, varying degrees of uncertainty surround virtually all parameters in these models. Sensitivity analysis is one of the most important tools for the analysis of models, and shows how the outputs of a model, such as concentrations and reaction fluxes, are dependent on the parameters which make up the input. Unfortunately, small changes in parameter values can lead to the results of a sensitivity analysis changing significantly. The results of such analyses must therefore be interpreted with caution, particularly if a high degree of uncertainty surrounds the parameter values. Global sensitivity analysis methods can help in such situations by allowing sensitivities to be calculated over a range of possible parameter values. However, these techniques are computationally expensive, particularly for larger, more detailed models. Software was developed to enable a number of computationally-intensive modelling tasks, including two global sensitivity analysis methods, to be run in parallel in a high-throughput computing environment. The use of high-throughput computing enabled the run time of these analyses to be drastically reduced, allowing models to be analysed to a degree that would otherwise be impractical or impossible. Global sensitivity analysis using high-throughput computing was performed on a selection of both theoretical and physiologically-based models. Varying degrees of parameter uncertainty were considered. These analyses revealed instances in which the results of a sensitivity analysis were valid, even under large degrees of parameter variation. Other cases were found for which only a slight change in parameter values could completely change the results of the analysis. Parameter uncertainties are a real problem in biological systems modelling. This work shows how, with the help of high-throughput computing, global sensitivity analysis can become a practical part of the modelling process.