Search

BACKGROUND: Alzheimer’s disease (AD) is a progressive neurodegenerative disease that is characterized by two neuropathological hallmarks: neurofibrillary tangles composed of hyperphosphorylated tau protein and extracellular plaques composed of amyloid-β peptides (Aβ). Recent GWAS have linked PICALM and BIN1 to AD risk, and several studies have implicated PICALM and BIN1 in Aβ production and clearance. However, so far their roles in modulating Aβ42 toxicity remain unclear. AIMS: This study aimed 1.) to test whether the fly orthologues of PICALM and BIN1, lap and Amph, suppress Aβ42 toxicity and to elucidate the molecular mechanisms at work. 2.) to identify novel suppressors of Aβ42 toxicity by genetic screens. RESULTS: 1.) I showed that over-expression of the Drosophila PICALM orthologue, lap, ameliorated Aβ42 toxicity in an adult onset AD model without affecting Aβ levels. I then performed a small-scale targeted genetic screen of endocytic-exocytic genes and identified Rab5 as a suppressor of Aβ42 toxicity. Aβ in flies led to the accumulation of pre-synaptic vesicular glutamate transporter (VGlut). Lap directly interacted with VGlut and reduced its accumulation back to control levels. Consistent with this, Aβ caused an increase in extracellular glutamate levels, while lap expression reduced the Aβ-induced increase in extracellular glutamate levels, suggesting lap might play a role in modulating glutamatergic transmission. Amph, the homologue of another AD risk factor, BIN1, modulated post-synaptic glutamate receptor (GluRII) localization. Amph localisation was disrupted upon Aβ expression and restored upon lap over-expression. 2.) The rough eye phenotype (REP) of Drosophila allows for high-throughput screening of enhancers and suppressors of therapeutic targets in neurodegenerative diseases. Using the REP, my collaborators and I performed a genome-wide genetic screen to identify novel modifiers of Aβ toxicity. The primary hits were further validated using negative geotaxis and lifespan assays. I found 4 gene-knockdown suppressors of Aβ42 toxicity. 3 of these suppressors did not reduce Aβ42 levels. CONCLUSIONS: 1.) I propose a model where lap and Amph collaborate to modulate glutamatergic transmission and so affect disease development. 2.) I find Rab5, EndoA and Snap25 are potential gene-overexpression suppressors of Aβ42 toxicity, and CG15011, Larp, Pxn, and Sodh2 are potent gene-knockdown suppressors of Aβ42 toxicity.

Large-scale photovoltaic power plants are becoming popular worldwide owing to the cost reduction in all parts of the solar power system, conversion efficiency improvement and the efforts from both governments and commercial companies to seek for new, clean, and sustainable energy sources. Inverters are important parts that connect the photovoltaic modules to the grid. Compared to the conventional two-level inverters, multilevel inverters feature improved harmonic performance with a reduced switching loss. Among all multilevel conversion topologies, the cascaded H-bridge converter is considered as one of the most promising candidates for the next-generation large-scale photovoltaic inverters. However, the solar irradiance and temperature conditions may vary significantly in a large-scale solar power plant. As a result, the bridges connected to different photovoltaic arrays are expected to generate different amounts of power. The thesis deals with the challenge to provide three-phase balanced grid currents when the power imbalance problem occurs. To begin with, several new zero-sequence injection methods are presented, featuring superior power balance capabilities to guarantee balanced currents under severe power imbalance. The optimal zero-sequence injection, in terms of the power balance capability, is also presented and proved to be the optimal solution of the star-connected H-bridge converter. Furthermore, the power balance capabilities of the presented zero-sequence injection methods are evaluated from both theoretical and practical perspectives. Additionally, the fault-tolerant scheme allowing the cascaded H-bridge to continue the operation after one or more bridges are bypassed owing to faults is presented. Different to the fault-tolerant scheme of the same converter with other applications, the unique problem of power imbalance in the photovoltaic application is taken into consideration. Finally, the delta-connected cascaded H-bridge converter is presented as an alternative to the star-connected converter. The converter overratings (both voltage and current), required by the star and delta-connected cascaded H-bridge converters to cope with all theoretically possible power imbalance cases, are compared to demonstrate that the delta-connected converter features significantly less converter overrating. All the findings are analyzed theoretically and verified experimentally with a converter prototype built in the laboratory.

The terrestrial biosphere is responsible for removing roughly one quarter of the total anthropogenic emissions of carbon dioxide from the atmosphere.This process is dominated by the forests. The terrestrial portion of the global carbon cycle is also the most uncertain. Growing forests are a major sink of CO2 while deforestation contributes as a major source of CO2. While the locations of deforestation can be tracked with remote sensing data relatively well, the amount of carbon removed is not well known due to the lack of knowledge on the biomass density of the forests that were disturbed. Spatially explicit distribution of carbon stocks in global forests can greatly reduce this uncertainty by improving estimates of emissions from land use activities. It will also help with green house gas inventory at regional and national scales.Increasing amounts of remote sensing based forest studies are being conducted, however, there is currently no global spatially explicit map of forest biomass/carbon. This dissertation aims to address that by producing the first global spatially explicit map of live biomass and carbon stock in forests and other living woody vegetation at ~1 km resolution for circa year 2005. A combination of remote sensing data from the Quick Scatterometer (QuikSCAT), Moderate Resolution Imaging Spectroradiometer (MODIS), and Shuttle Radar Topography Mission (SRTM) are used to provide wall-to-wall spatial information. Ground inventory and LiDAR from the Geosciences Laser Altimeter System (GLAS) onboard the Ice, Cloud, and land Elevation Satellite (ICESat) are used to create a global network of aboveground biomass (AGB) samples. Landcover map based on the GlobCover product from the European Space Agency (ESA) is used as ancillary data. These are combined in a statistical model using Maximum Entropy to create the first global map of AGB and total carbon stock (above + below ground) in living biomass. Uncertainty at the pixel level is also produced alongside the above products. The total global carbon in live woody vegetation is estimated at 337 PgC, with 311.4 PgC being in the forests (92% of total) for circa year 2005, compared with 302.8 PgC in total global live forest biomass as reported in the United Nations Food and Agriculture Organization's 2010 Forest Resources Assessment report (FRA 2010). Shrublands and savannas account for the other 25.6 PgC (8% of total) globally. National carbon stocks are calculated and reported for all the countries using several different thresholds for forest cover.