Search

Apolipoproteins co-deposit with amyloids, yet apolipoprotein-amyloid interactions are enigmatic. To understand how apoE interacts with Alzheimer's amyloid-β (Aβ) peptide in fibrillary deposits, the NMR structure of full-length human apoE was docked to four structures of patient-derived Aβ 1-40 and Aβ 1-42 fibrils determined previously using cryo-electron microscopy or solid-state NMR. Similar docking was done using the NMR structure of human apoC-III. In all complexes, conformational changes in apolipoproteins were required to expose large hydrophobic faces of their amphipathic α-helices for sub-stoichiometric binding to hydrophobic surfaces on sides or ends of fibrils. Basic residues flanking the hydrophobic helical faces in apolipoproteins interacted favorably with acidic residue ladders in some amyloid polymorphs. Molecular dynamics simulations of selected apoE-fibril complexes confirmed their stability. Amyloid binding via cryptic sites, which became available upon opening of flexibly linked apolipoprotein α-helices, resembled apolipoprotein-lipid binding. This mechanism probably extends to other apolipoprotein-amyloid interactions. Apolipoprotein binding alongside fibrils could interfere with fibril fragmentation and secondary nucleation, while binding at the fibril ends could halt amyloid elongation and dissolution in a polymorph-specific manner. The proposed mechanism is supported by extensive prior experimental evidence and helps reconcile disparate reports on apoE's role in Aβ aggregation. Furthermore, apoE domain opening and direct interaction of Arg/Cys158 with amyloid potentially contributes to isoform-specific effects in Alzheimer's disease. In summary, current modeling supported by prior experimental studies suggests similar mechanisms for apolipoprotein-amyloid and apolipoprotein-lipid interactions; explains why apolipoproteins co-deposit with amyloids; and helps reconcile conflicting reports on the chaperone-like apoE action in Aβ aggregation., Competing Interests: Disclosure statement. All authors have no financial or any other conflicts of interests to disclose.

To understand the workings of a living cell, we need to identify its molecular components and determine how they associate with each other. To date, studies that identify new macromolecular assemblies have been mainly limited to low-throughput biochemistry assays. Structures of macromolecular assemblies also have been difficult to obtain, and are mostly available for a small subset of individual components or their assemblies amenable to conventional approaches, such as X-ray crystallography and nuclear magnetic resonance spectroscopy. In this dissertation, I describe my contributions to the development of four novel pipelines that utilize new technologies and datasets to facilitate the identification of macromolecular assemblies and the determination of their structures. First, we designed an algorithm to identify genes coding for biosynthetic macromolecular assemblies. Second, we developed a platform to identify previously unknown HIV-human protein assemblies based on affinity purification, mass spectrometry, and computational scoring. Third, to aid the structure determination of macromolecular assemblies that are challenging to isolate and purify, we proposed a new strategy based on in vivo measurements of genetic interaction and integrative modeling. Finally, to facilitate rational discovery of small molecule modulators of macromolecular assemblies and their components, we presented a new approach based on computational identification of putative ligand-binding sites that are not detectable in ligand-free structures, due to insufficient structure resolutions or flatness in the absence of a ligand.

Many proteins have small molecule-binding pockets that are not easily detectable in the ligand-free structures. These cryptic sites require a conformational change to become apparent; a cryptic site can therefore be defined as a site that forms a pocket in a holo structure, but not in the apo structure. Because many proteins appear to lack druggable pockets, understanding and accurately identifying cryptic sites could expand the set of drug targets. Previously, cryptic sites were identified experimentally by fragment-based ligand discovery, and computationally by long molecular dynamics simulations and fragment docking. Here, we begin by constructing a set of structurally defined apo-holo pairs with cryptic sites. Next, we comprehensively characterize the cryptic sites in terms of their sequence, structure, and dynamics attributes. We find that cryptic sites tend to be as conserved in evolution as traditional binding pockets, but are less hydrophobic and more flexible. Relying on this characterization, we use machine learning to predict cryptic sites with relatively high accuracy (for our benchmark, the true positive and false positive rates are 73% and 29%, respectively). We then predict cryptic sites in the entire structurally characterized human proteome (11,201 structures, covering 23% of all residues in the proteome). CryptoSite increases the size of the potentially “druggable” human proteome from ~40% to ~78% of disease-associated proteins. Finally, to demonstrate the utility of our approach in practice, we experimentally validate a cryptic site in protein tyrosine phosphatase 1B using a covalent ligand and NMR spectroscopy. The CryptoSite web server is available at http://salilab.org/cryptosite.

In der vorliegenden Arbeit wurden unterschiedliche Methoden des computergestützten strukturbasierten Wirkstoffdesigns in verschiedenen Projekten zur Anwendung gebracht. // Es wurden dabei fünf verschiedenen Zielproteine in drei Targetklassen untersucht. (1) Aus der Superfamilie der Aldo-Keto-Reduktasen wurden zwei Mitglieder der Familie 1 untersucht: Zum einen Aldosereduktase, die auf Grund ihrer Beteiligung an diabetischen Komplikationen im fortgeschrittenen Stadium dieser Erkrankung von Interesse ist, zum anderen das Mitglied AKR1B10 dieser Familie, dessen Überexpression mit verschiedenen Krebsarten assoziiert ist. (2) PRMT4 und PRMT6, die als epigenetische Zielproteine ein aktuelles und zukunftsträchtiges Themengebiet darstellen. (3) Ein bakterielles Target in dieser Arbeit stellte die tRNA-Guanin Transglykosylase (TGT) dar, die auf Grund ihrer initialen Funktion in der Aktivierungskaskade des pathogenen Phänotyps von Shigella flexneri ein interessantes Target für die Behandlung der durch Shigellen ausgelösten Bakterienruhr darstellt. //// Aldosereduktase // Die AR gehört zu der Superfamilie der Aldo-Keto-Reduktasen. Als NAD(P)(H)-abhängige Oxidoreduktase ist sie in erster Linie für die Umwandlung von Glukose zu Sorbitol im Polyolweg bekannt. Darüber hinaus leistet sie durch die Reduktion von giftigen Aldehyden und Carbonylen einen wichtigen Beitrag zur Entgiftung des Organismus. Die Aldosereduktase wurde in Kooperation mit der Arbeitsgruppe Diederich (Marburg) und GE Healthcare Bio-Sciences AB untersucht. Der Fokus dieses Projekts lag auf der Charakterisierung einer Ligandenserie und Untersuchung des Bindungsverhaltens dieser Liganden in Hinblick auf das Öffnungsverhalten der transienten Spezifitätstasche des Proteins. Der Beitrag dieser Dissertation war die Untersuchung der Proteindynamik im Bezug auf diese Tasche und die Analyse und Beschreibung eines möglichen Öffnungsmechanismus. Die Daten für diese Analysen wurden durch molekulardynamische Simulationen gewonnen. Ein weiterer Beitrag war die Durchführung von MM-GBSA-Berechnungen zur Evaluierung der Bindungspräferenz zweier ähnlicher Liganden in Bezug auf die Spezifitätstasche. // AKR1B10 // In diesem Kooperationsprodukt wurden vergleichende Simulationen eines Liganden (JF0064) durchgeführt, um (1) die röntgenkristallographisch beobachtete Protonierung des Liganden im Komplex mit dem Protein zu validieren und (2) strukturelle Einblicke in die Bindungskonformation des Liganden im Komplex mit dem Wildtyp-Protein zu bieten. Die molekulardynamischen Simulationen boten hier einen Einblick in die Struktur des Wildtyp:JF0064-Komplexes, dessen röntgenkristallographische Aufklärung nicht möglich war. // Protein-Arginin-Methyltransferasen // In diesem Kapitel wurden verschiedene Verfahren verwendet, um zwei Aufgabenstellungen umzusetzen. (1) In Kooperation mit der Arbeitsgruppe Bauer wurde ein Homologiemodell für die Gallus gallus PRMT4 erstellt. Bisher wurde in Vögeln, als einziger Klasse der Wirbeltiere, kein PRMT4-Homolog nachgewiesen. Hannah Berberich gelang erstmalig der Nachweis einer putativen Gallus gallus PRMT4 (ggPRMT4). Der Anteil dieser Dissertation war die Erstellung eines plausiblen in silico Modells der ggPRMT4. Dafür wurden verschiedene in silico Methoden miteinander kombiniert. Im Rahmen dieses Projekts wurde zur Erstellung des ggPRMT4-Modells eine Homologiemodellierung kombiniert mit einem Protein-Protein-Docking durchgeführt. Die Validierung und Untersuchung der Dynamik des modellierten Komplexes erfolgte durch ergänzende molekulardynamische Simulationen. (2) Im Rahmen eines Dockings von organischen Kleinmolekülen wurden die beiden Proteine PRMT4 und PRMT6 – ausgehend von bekannten Kristallstrukturen – als Targets adressiert. Diesen Dockingläufen schloss sich die visuelle Inspektion der durch Scoring-Funktionen vorselektierten Moleküle an. Anschließend wurden interessante Verbindungen kommerziell erworben und in Kooperation mit der Arbeitsgruppe Bauer von Antje Repenning in einem Assay getestet. Dabei konnten aktive Substanzen gefunden werden, die nun weiter untersucht werden und als Ausgangspunkte für rationales Wirkstoffdesign dienen können. // tRNA-Guanin Transglykosylase // Dieses Target wurde im Rahmen von zwei Studien untersucht. In (1) wurde zunächst die Dynamik einer bei röntgenkristallographischen Untersuchungen neu entdeckten Subtasche unter der beta1alpha1-Schleife untersucht. Dabei erfolgte die Untersuchung erst mit Hilfe molekulardynamischer Simulationen. Im zweiten Schritt wurde diese Tasche dann in einem Docking mit organischen Molekülen adressiert. In (2) wurde das Öffnungsereignis der transienten erweiterten Guanin/preQ1-Bindetasche durch molekulardynamische Simulationen analysiert., In this thesis different methods of computer-aided drug design have been applied to diverse problems in several projects. // Five different target proteins from three distinct target classes have been investigated. (1) From the super family of aldo-keto reductases, two members of the subfamily 1 have been examined. Firstly, aldose reductase which is of special interest due to its role during complications in advanced states of diabetes. Secondly, another member of the same family: AKR1B10 which is associated with various cancers. (2) Protein arginine methyltransferase 4 (PRMT4) and PRMT6 that represent an up-to-date and seminal topic due to their epigenetic function. (3) A bacterial target represented by tRNA-guanin transglycosylase (TGT). This protein fulfills an initial function in the activation cascade of the pathogenic phenotype of Shigella flexneri and, therefore, serves as an interesting target for the treatment of shigella induced bacillary dysentery. //// Aldose reductase // AR belongs to the super family of aldo-keto reductases. It is a NAD(P)(H) dependent oxidoreductase and, hence, mostly known for the conversion of glucose to sorbitol in the polyol pathway. Furthermore, it contributes to the detoxification of the organism by reducing toxic aldehydes and carbonyls. The Investigations on aldose reductase were conducted in cooperation with the Diederich group (Marburg) and GE Healthcare Bio-Sciences AB. The project was focused on characterizing a ligand series and examining the binding properties of these ligands with regard to opening events of a transient specificity pocket in this protein. This thesis contributed to this via an investigation of protein dynamics related to the pocket and an analysis and description of possible opening mechanisms. The necessary data for the analysis was obtained by conducting molecular dynamics simulations. Furthermore, MM-GBSA calculations were performed to evaluate binding preferences of two similar ligands to the specificity pocket. // AKR1B10 // In this cooperation project, comparative simulations of a ligand (JF0064) were executed to achieve two goals. (1) Validate the protonation state of the ligand in complex with the protein that has been observed in the x-ray structure. (2) Gaining structural insights on the binding conformation of the ligand to the wildtype protein. Since structure determination of the wildtype:JF0064 complex with x-ray crystallography was not possible, these molecular dynamic simulations allowed us to gain an insight on its structure. // Protein arginine methyltransferases // In this project several methods were used to accomplish two different tasks. (1) In cooperation with the Bauer group an homology model of Gallus gallus PRMT4 was built. To date, birds were the only class of vertebrates without experimental proof for an PRMT4 homologue. Hannah Berberich succeeded as the first to prove the existence of putative Gallus gallus PRMT4 (ggPRMT4). This thesis contributed by generating a feasible in silico model of ggPRMT4. Several in silico methods have been used in combination to accomplish this goal. To generate the ggPRMT4 model, homology modeling was combined with protein-protein docking. Furthermore, molecular dynamics simulations were used to validate the model and analyze the dynamics of the modeled complex further. (2) PRMT4 and PRMT6 were addressed as targets in a docking study using small organic molecules and known crystal structures of both proteins. All dockings were further evaluated by visual inspection subsequent to pre-selecting the molecules with a scoring function. Afterwards, interesting compounds were purchased and in cooperation with the BAUER group tested in a biological assay by Antje Repenning. As a result of this process, active compounds were identified. These compounds will now be further characterized and may be used as a starting point for rational drug development. // tRNA-guanin transglycosylase // This target was analyzed in two studies. (1) The dynamics of a, by x-ray crystallographic methods newly discovered, sub-pocket underneath the beta1alpha1-loop were analyzed. Therefore, firstly the pocket was examined via molecular dynamics methods. In a second step, the pocket was addressed by docking with organic molecules. (2) The opening event of the transient extended guanin/preQ1 binding pocket was analyzed with molecular dynamics simulations.

Molecular dynamics (MD) simulation is an emerging in silico technique with potential applications in diverse areas of pharmacology. Over the past three decades MD has evolved as an area of importance for understanding the atomic basis of complex phenomena such as molecular recognition, protein folding, and the transport of ions and small molecules across membranes. The application of MD simulations in isolation and in conjunction with experimental approaches have provided an increased understanding of protein structure-function relationships and demonstrated promise in drug discovery.

This review discusses the many roles atomistic computer simulations of macromolecular (for example, protein) receptors and their associated small-molecule ligands can play in drug discovery, including the identification of cryptic or allosteric binding sites, the enhancement of traditional virtual-screening methodologies, and the direct prediction of small-molecule binding energies. The limitations of current simulation methodologies, including the high computational costs and approximations of molecular forces required, are also discussed. With constant improvements in both computer power and algorithm design, the future of computer-aided drug design is promising; molecular dynamics simulations are likely to play an increasingly important role.