Citrus fruits are some of the most widely consumed fruits in the world and contain significant levels of flavonoids, a category of plant secondary metabolites which control taste, color, plant defense, and overall marketability. In citrus and other plants, flavonoids are found in their glucosylated form. Glucosyltransferases (GT’s) are enzymes that add glucose to secondary metabolites like flavonoids. They make up a diverse class of enzymes ubiquitous throughout the plant and animal kingdoms. While many GT’s have been identified, they vary greatly in their structural identity, and their chemical properties make it such that only a small percentage of existing GT’s have been functionally characterized. Research on GT structure function relationships strengthens the reliability of genomic databases and makes significant contributions to the field of enzyme biotechnology. Bioenergy research and custom enzyme synthesis rely on GT structural data, making this research critical to the success of many promising current and future projects. A GT was isolated from grapefruit and was shown to glucosylate the flavonol class of flavonoids at the 3-OH position, called CP3GT. Subsequent analysis showed there are specific arrangements of amino-acids inside the catalytic cleft of CP3GT that likely account for its specificity with flavonols. These interactions are not fully understood and make CP3GT an excellent model for elucidating unique structure function relationships of a GT enzyme. X-ray crystallography is one of the best methods for structure determination that allows a 3D image of the protein in question to be resolved at the molecular level. This method has vast potential for advancing plant enzymology, yet to date only 6 plant glucosyltransferases have had their crystal structures solved. The structural similarities and complementary specificities that CP3GT shares with these crystallized GT’s make CP3GT an excellent candidate for crystallization. This research hypothesizes that there are unique structural features that give CP3GT its specificity, and that these features can be elucidated using x-ray crystallography. Wild type CP3GT and 3 recently characterized mutants are being prepared for crystallization. The crystallization of 3 CP3GT mutants in addition to wild type will compliment structure/function analysis by providing insight into how structural modifications can alter enzyme function. It is recommended that protein be in its native form for crystallization, thus a thrombin-cleavage site was inserted into WT CP3GT and 3 mutants to remove tags following purification. Some studies have suggested that the presence of tags alters enzyme activity, thus this presented the opportunity to test the effect of tags by assaying both native and tagged enzyme. Initial results showed that WT CP3GT treated with thrombin retained 70 percent activity after a 2-hour treatment at 4o C. Additional assays will be conducted to fully determine tag effects and will run concurrently with crystallization experiments

No description available.

No description available.

This thesis presents two papers addressing important biochemical prediction challenges. The first paper focuses on accurate protein distance predictions and introduces updates to the ProSPr network. We evaluate its performance in the Critical Assessment of techniques for Protein Structure Prediction (CASP14) competition, investigating its accuracy dependence on sequence length and multiple sequence alignment depth. The ProSPr network, an ensemble of three convolutional neural networks (CNNs), demonstrates superior performance compared to individual networks. The second paper addresses the issue of accurate ligand ranking in virtual screening for drug discovery. We propose MILCDock, a machine learning consensus docking tool that leverages predictions from five traditional molecular docking tools. MILCDock, an ensemble of eight neural networks, outperforms single-network approaches and other consensus docking methods on the DUD-E dataset. However, we find that LIT-PCBA targets remain challenging for all methods tested. Furthermore, we explore the effectiveness of training machine learning tools on the biased DUD-E dataset, emphasizing the importance of mitigating its biases during training. Collectively, this work emphasizes the power of ensembling in deep learning-based biochemical prediction problems, highlighting improved performance through the combination of multiple models. Our findings contribute to the development of robust protein distance prediction tools and more accurate virtual screening methods for drug discovery.

The type VII secretion system is a protein export pathway linked to diverse phenotypes in both Actinobacteria and Firmicutes. The Actinobacterial subtype of the T7SS, referred to as T7SSa, has been shown to play a critical role in various aspects of Mycobacterial life including virulence, conjugation, and metal homeostasis. The T7SSb of Firmicutes bacteria on the other hand has similarly been shown to influence virulence but by the direct growth inhibition of competitor bacteria. Structure-function analyses of the T7SSa apparatus as well as various effectors and chaperones have begun to build a more mechanistic understanding of how T7SSa functions. In contrast, we know little of how the T7SSb functions despite its noted importance to both pathogens and environmental bacteria such as Bacillus, Staphylococcus, Enterococcus, and Streptococcus. During my thesis work, I have addressed several gaps in our understanding of T7SSb function. The three major questions that I have studied are: (1) how do T7SSb immunity proteins inhibit the toxicity of their cognate toxins, (2) how does the T7SSb export effectors through the thick Gram-positive cell wall, and (3) what is the role of chaperone proteins in facilitating T7SSb effector export? / Thesis / Doctor of Philosophy (PhD) / Bacteria require space and various nutrients to survive and grow and must therefore compete against other bacteria for access to these resources. To gain advantage over their competitors, many bacteria have developed molecular weapons that target and kill other closely related bacteria. Some of these weapons take the form of protein secretion machines that export antibacterial toxins. Gram-positive bacteria use the type VIIb secretion system (T7SSb) to inhibit the growth of other Gram-positive bacteria. In this work, I explore several aspects of T7SSb including: (1) how toxins are inhibited by immunity proteins, (2) how toxins are secreted through the cell envelope, and (3) how toxins are recognized by the secretion apparatus. The goal of this work is to better understand how T7SSb functions at the molecular level.

The random-coil model has been dominant for unfolded proteins since the 1950’s; however, some experiments showed that the unfolded proteins were biased toward specific conformations in conflict with the random-coil model. Recently, residual dipolar couplings (RDCs) and paramagnetic relaxation enhancement (PRE) were applied to obtain a large amount of structural information on unfolded proteins. Typically, these data were interpreted in a framework of random-coil ensembles with a good agreement between experimental data and theoretical predictions. In this thesis, it was tested whether locally organized nonrandom ensembles could describe this agreement equally as well. Using a complete set of RDC and PRE data for denatured ubiquitin, it was revealed that there was no distinguishable difference between random-coil ensembles and ensembles containing 50% native structure. Thus, while it is important to measure as many RDCs or PRE as possible, even the best datasets may be insensitive to local organization in unfolded proteins.

No description available.

No description available.

No description available.

No description available.