Virtual screening is a fast, low cost method to identify potential small molecule therapeutics from large chemical databases for the vast amount of target proteins emerging from the life sciences and bioinformatics. In this work, we applied several conventional and newly developed virtual screening approaches to identify novel non-steroidal ligands for the human and zebrafish sex hormone binding globulin (SHBG). The ‘benchmark set of steroids’ is a set of steroids with known affinities for human SHBG that has been widely used for validation in the development of different virtual screening methods. We have updated this data set by including additional steroidal SHBG ligands and by modifying the predicted binding orientations of several benchmark steroids in the SHBG binding site based on the use of an improved docking protocol and information from recent crystallographic data. The new steroid binding orientations and the expanded version of the benchmark set was then used to create new in silico models which were applied in virtual screening to identify high-affinity non-steroidal human SHBG ligands from a large chemical database. Anthropogenic compounds with the capacity to interact with the steroid-binding site of SHBG pose health risks to humans and other vertebrates including fish. We constructed a homology model of SHBG from zebrafish and applied virtual screening to identify ligands for zebrafish SHBG from a set of 80 000 existing commercial substances, many of which can be exposed to the aquatic environment. Six hits from this in silico screen were tested experimentally for zebrafish SHBG binding and three of them, hexestrol, 4-tert-octylcatechol, dihydrobenzo(a)pyren-7(8H)-one demonstrated micromolar binding affinity for the zebrafish SHBG. These findings demonstrate the feasibility of using virtual screening to identify anthropogenic compounds that may disrupt or highjack functionally important protein:ligand interactions. Studies applying this new computational toxicology method could increase the awareness of hazards posed by existing commercial chemicals at relatively low cost. / Science, Faculty of / Graduate

Syfte – Rapportens syfte är att undersöka förbättringsmöjligheter av materialhantering mellan in-och utlastning på en cross-docking terminal. Baserat på syftet har två frågeställningar formulerats. Vilka aktiviteter ingår i materialhantering? Hur kan materialhanteringen effektiviseras? Metod – Studien har genomfört en förstudie på fallföretaget för att definiera ett problemområde. Utifrån problemområdet har syfte och frågeställningar formulerats. Fallstudien och litteraturgenomgång har genomförts parallellt för att skapa en abduktiv ansats. De inhämtade teorier från litteraturstudierna ligger stöd för det teoretiska ramverket.  Resultat – Flera slöserier identifierades under studien vilket påverkar den interna materialhanteringen. Inom de primära cross-docking aktiviteterna existerar slöserier vilket behöver elimineras för en effektivisering ska genomföras. Studien lyfter även vikten av ett fungerande informationsflöde för en effektiv materialhantering.  Implikationer – Eftersom problemområdet redan har etablerade teorier har ingen ny forskning genomförts. Däremot har studien inriktat sig på en annan synvinkel, där förhållandet mellan informationsflöde och materialflöde studerats.  Begränsningar – Begreppet materialhantering är brett och innefattar även materialhantering utanför terminalen. Rapporten har begränsat sig till enbart den interna materialhanteringen. Studien begränsar sig till enbart ett fallföretag vilket påverkar generaliserbarheten.  Nyckelord – Cross-docking, Materialhantering, Materialflöde, Informationsflöde.

Fragment-based virtual screening is an in-silico method that potentially identifies new startingpoints for drug molecules and provides an inexpensive and fast exploration of the relevantchemical space compared to its experimental counterpart. It focuses on docking small potentialbinding fragments to a binding pocket and is used to design improved binders by growing thefragments or joining fragments using suitable linkers. In this project, a fragment-based virtualscreening was evaluated by docking 21 fragments that are obtained from 4 different drugs. Here,the fragments were evaluated using SP score in place and SP and XP flexible docking methodsand were compared to the results of the two decoy fragment datasets. Three of the investigatedfragments are positioned at the top and docked with the correct poses and pockets when comparedto the corresponding substructure in the crystal structure and thus could be considered a successfulfragment starting points. Out of the two flexible docking methods used, the SP method providedadditional correct poses and pockets than XP in this limited dataset.

The goal of the ENCODE project has been to characterize regulatory elements in the human genome, such as regions bound by transcription factors (TFs), regions of open chromatin and regions with altered histone modifications. The ENCODE consortium has performed a large number of whole-genome experiments to measure TF binding, chromatin accessibility, gene expression and histone modifications, on a multitude of cell types and conditions in both human and mouse. In this dissertation I describe the analysis of numerous datasets comprising 66 epigenomes, chromatin accessibility and expression data across twelve tissues and seven time points, during mouse embryonic development. We defined chromatin states using histone modification data and performed integrative analysis on the states. We observed coordinated changes of histone mark signals at enhancers and promoters with gene expression. We detected evolutionary conserved bivalent promoters, selectively silencing ~3,400 genes, including hundreds of TFs regulating embryonic development. Second, I present a supervised method to predict TF binding across cell types, with features based on DNA sequence and patterns in DNase I cleavage data. We found that sequence and DNase read counts can outperform other features as well as state-of-the-art methods. I also describe our contribution to the ENCODE TF Binding DREAM challenge where we developed a method, using multiscale features and Extreme Boosting. Third, I describe methods, tools, and computational infrastructure that we have developed to handle large amounts of experimental data and metadata. These tools are fundamental to the selection and integration of large experimental datasets and are at the core of our pipelines, which are described in this dissertation. Finally, I present the protein docking server I developed, as well as algorithms and routines for post-processing predictions and protein structures. Collectively, this body of work encompasses computational approaches to the analyses of chromatin states, gene regulation, and the integration of large experimental datasets. / 2021-08-31T00:00:00Z

This dissertation presents a collection of biological simulations and predictions in collaboration with experiment to support and elucidate the trends observed in various protein–ligand systems. Within the model systems, there is a strong focus on the support for the development of peptidomimetic inhibitors for post-translational reader proteins (CBX proteins). The systems studied throughout this document each present their own unique challenges but fall under the general theme of protein flexibility and the difficulties of sampling such systems. As part of this work, methodological advances were made to address the challenges of structural prediction on flexible proteins and ultimately form the method Selective Ligand-Induced Conformational Ensemble (SLICE). The development, validation, and future directions of the SLICE method are also discussed. Ultimately, the collaborative efforts presented in this dissertation bring forward a greater understanding of the drug design challenges on the CBX proteins as well a new methodology in the field of structure-based drug design. / Graduate

京都大学 / 0048 / 新制・課程博士 / 博士(農学) / 甲第20065号 / 農博第2194号 / 新制||農||1045(附属図書館) / 学位論文||H28||N5021(農学部図書室) / 京都大学大学院農学研究科地域環境科学専攻 / (主査)准教授 赤松 美紀, 教授 植田 和光, 教授 宮川 恒 / 学位規則第4条第1項該当 / Doctor of Agricultural Science / Kyoto University / DFAM

Disease caused by antibiotic resistant (ABR) bacteria has become a widespread global public health issue as humanity's existing collection of effective antibiotics dwindles. ABR bacteria are responsible for approximately 5 million deaths worldwide annually, which is predicted to reach 10 million yearly by 2050. Antivirulence therapeutics have been explored in recent times as another approach to tackling the global ABR pandemic by disrupting the function of virulence factors that promote disease development. The bacterial type IV pilus (T4P) is a prevalent virulence factor in many ABR pathogens, contributing to bacterial pathogenesis by facilitating cell motility, surface adhesion, and biofilm formation. Critically, the T4P facilitates early stages of disease, providing a means to invade and colonize a host. T4P assembly is driven by the PilB/PilF motor ATPase that localizes to the cytoplasmic face of the inner membrane to drive pilus biogenesis by ATP hydrolysis. The thesis work here explores computational and experimental methods for the discovery of antivirulence therapeutics targeting the T4P assembly ATPase PilB. A computational model of Chloracidobacterium thermophilum PilB was generated by homology modeling and molecular docking was performed to analyze the binding characteristics of six anti-PilB inhibitory compounds identified in previous studies. Computational docking aligns with the existing body of work and reveals important protein-ligand interactions and characteristics, particularly involving the ATP binding domain of PilB. This work supports the use of PilB in structure-based virtual screening to identify novel compounds targeting PilB. Additionally, through heterologous expression and chromatography methods, the ATPase core of Neisseria gonorrhoeae PilF was successfully expressed and purified as an active ATPase. This work optimized conditions for its ATPase activity in vitro. Additionally, this thesis documents the experimental attempt to express and purify Clostridioides difficile PilB as an active ATPase. Two of the seven C. difficile PilB variant proteins expressed led to soluble protein while one construct remains to be explored. The results of these studies provide insight for future methodology design for antivirulence therapeutic research targeting the T4P assembly ATPase using both in silico and in vitro methods. / Master of Science / Antibiotic resistant bacterial infections are responsible for nearly 5 million deaths worldwide every year. These infections are becoming increasingly more difficult to treat as bacterial pathogens acquire greater means to overcome our dwindling antibiotic repertoire. This has prompted researchers to explore alternative therapeutic strategies, including the antivirulence approach that aims to disable the function or production of bacterial virulence factors. Virulence factors serve as arms and armor that help bacteria cause disease, but they may be disrupted in such a way that renders potentially pathogenic bacteria harmless to humans. One major virulence factor in many antibiotic resistant bacteria is the type IV pilus (T4P), which is important in the early stages of host invasion by mediating adhesion and biofilm formation. This work explores both computational and experimental strategies to antivirulence drug discovery targeting the T4P, specifically the primary motor protein PilB/PilF. Newly identified PilB inhibitors were evaluated by molecular docking and molecular dynamics simulation to assess the use of PilB for drug discovery via virtual screening in silico. This revealed key characteristics and protein-ligand interactions that contribute to successful PilB inhibition and supports the use of CtPilB for structure-based virtual screening. Additionally, the PilF motor protein from Neisseria gonorrhoeae was successfully purified and demonstrated to be active for inhibitor discovery in the future. This work also covers efforts to establish Clostridioides difficile PilB as potential model enzyme for inhibitor discovery in the future.

Antibiotics have transformed modern medicine in manifold ways. However, the misuse and over-consumption of antibiotics or antimicrobials have led to the rise in antimicrobial resistance (AMR). Unfortunately, robust tools or techniques for the detection of potential loci responsible for AMR before it happens are lacking. The emergence of resistance even when a strain lacks known AMR genes has puzzled researchers for a long time. Clearly, there is a critical need for the development of novel approaches for uncovering yet unknown resistance elements in pathogens and advancing our understanding of emerging resistance mechanisms. To aid in the development of new tools for deciphering AMR, here we propose a machine learning (ML) based framework that provides ML models trained and tested on (1) genotypic AMR and phenotypic antimicrobial susceptibility testing (AST) data, which can predict novel resistance factors in bacterial strains that lack already implicated resistance genes; and (2) complete gene set and AST phenotypic data, which can predict the most important genetic loci involved in resistance to specific antibiotics in bacterial strains. The validation of resistance loci prioritized by our ML pipeline was performed using homology modeling and in silico molecular docking.

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.

No description available.