Dokument: Approximate and efficient prediction of entropy changes upon complex formation

Titel:	Approximate and efficient prediction of entropy changes upon complex formation
URL für Lesezeichen:	https://docserv.uni-duesseldorf.de/servlets/DocumentServlet?id=39485
URN (NBN):	urn:nbn:de:hbz:061-20160914-100500-3
Kollektion:	Dissertationen
Sprache:	Englisch
Dokumententyp:	Wissenschaftliche Abschlussarbeiten » Dissertation
Medientyp:	Text
Autor:	MSc Ben-Shalom, Ido [Autor]
Dateien:	[Dateien anzeigen] Adobe PDF [Details] 34,11 MB in einer Datei [ZIP-Datei erzeugen] Dateien vom 06.09.2016 / geändert 06.09.2016
Beitragende:	Prof. Dr. Gohlke, Holger [Betreuer/Doktorvater] Jun.-Prof. Dr. Schröder, Gunnar [Gutachter]
Dewey Dezimal-Klassifikation:	500 Naturwissenschaften und Mathematik » 540 Chemie
Beschreibung:	All biological processes depend on the specific molecular recognition and binding affinity between ligand molecules and their macromolecular targets. Thus, the rational design of potent ligands requires an accurate knowledge of the binding free energy, which is composed of enthalpic (ΔH) and entropic (ΔS) contributions. However, ΔS is difficult to predict as it requires a full understanding of all possible states of the system and is thus often neglected. In my thesis, I developed the BEERT (Binding Entropy Estimation for Rotation and Translation) method for fast predictions of translational and rotational entropy contributions of the ligand (ΔSconfig.) to the binding free energy. BEERT estimates ΔSconfig. by modeling the naturally occurring reduction of accessible microstates that leads to changes in translational and rotational entropy. For validation, I fitted predicted ΔSconfig. and MM-PBSA effective energies (ΔG eff.) to experimental binding affinities of HIV-1 protease, Factor Xa (FXa), and Heat shock protein 90 (Hsp90) inhibitors using multiple linear regression. For all datasets, the obtained correlations are highly significant, markedly improved compared to MM-PBSA results alone, and robust in a leave-one-out cross-validation. In the second project, I calculated changes in the vibrational entropy (ΔSvib.) upon ligand binding to proteins. ΔSvib. originate from the varying vibrational degrees of freedom between a protein-ligand complex and the unbound partners. To this end, our working group introduced a computationally highly efficient approximation of ∆Svib. upon binding to biomolecules based on rigidity theory. In constraint network representations of the binding partners, ∆Svib. is estimated from changes in the variation of the number of low (i.e., zero) frequency modes with respect to variations in the networks’ coordination number. Compared to ∆Svib. computed by normal mode analysis as a gold standard, our approach yields significant and good to fair correlations for datasets of protein-protein complexes, alanine scanning, and two protein-small molecule datasets: FXa and trypsin. For the HIV-1 protease and Hsp90 datasets the correlations were poor and caused by the width of the distribution of ∆Svib. calculated using NMA, which is very similar in magnitude to the average standard deviation of the computed ∆Svib.. Therefore, the maximum possible squared Pearson correlation coefficient on these datasets vanishes. Taken together, both methods allow calculating the configurational entropy change upon ligand binding in a highly efficient manner, thus complementing existing scoring functions or free energy calculation. To the best of our knowledge, this is the first time that a prediction of configurational entropy was successfully implemented in a way that allows large scale virtual screening, making our approach a valuable tool for identifying, understanding, and optimizing drug molecules and molecular interactions in general.
Lizenz:	Urheberrechtsschutz
Fachbereich / Einrichtung:	Mathematisch- Naturwissenschaftliche Fakultät » WE Pharmazie » Pharmazeutische und Medizinische Chemie
Dokument erstellt am:	14.09.2016
Dateien geändert am:	14.09.2016
Datum der Promotion:	29.08.2016