Identification of Ligands with Tailored Selectivity: Strategies & Application

In the field of computer-aided drug design, docking is a computational tool, often used to evaluate the sterical and chemical complementarity between two molecules. This technique can be used to estimate the binding or non-binding of a small molecule to a protein binding site. The classical applicat...

Full description

Saved in:
Bibliographic Details
Main Author: Schmidt, Denis
Contributors: Kolb, Peter (Dr.) (Thesis advisor)
Format: Doctoral Thesis
Language:English
Published: Philipps-Universität Marburg 2015
Subjects:
Online Access:PDF Full Text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In the field of computer-aided drug design, docking is a computational tool, often used to evaluate the sterical and chemical complementarity between two molecules. This technique can be used to estimate the binding or non-binding of a small molecule to a protein binding site. The classical application of docking is to find those molecules within a large set of molecules that bind a certain target protein and modulate its biological activity. This setup can be considered as established for a single target protein. In contrast to this, the docking to multiple target structures offers new possible applications. It can be used, for example, to assess the binding profile of a ligand against a number of proteins. In this work, the applicability of docking is assessed in such a scenario where multiple target structures are used. The corresponding proteins mostly belong to the family of G protein-coupled receptors. This protein family is very large and numerous GPCRs have been identified as potential drug targets, explaining the their relevance in pharmaceutical research. The protein structures used herein have different relationships and thus represent different application scenarios. The first case study uses two structures belonging to different proteins. These proteins are CXCR3 and CXCR4, a pair of chemokine GPCRs. In this chapter, new ligands are identified that bind to these proteins and modulate their biological activity. More importantly, for each of these newly identified ligands it could be predicted using docking, whether this ligand binds only to one of the two target proteins or to both. This study proves the applicability of docking to identify ligands with tailored selectivity. In addition, these ligands show excellent binding affinities to their respective target or targets. In the following two studies, the docking to different structures of the same target protein is investigated. The first application aims at identifying ligands selective for either one of two isoforms of the zebrafish CXC receptor 4. Subsequently, multiple conformations of the chemokine receptor CCR5 are used to show that different starting structures can identify different ligands. Next to the plain identification of chemically new ligands, experimental hurdles to prove the biological activity of these molecules in a functional assay is discussed. These difficulties are based on the fact that docking evaluates the structural complementarity between molecules and protein structures rather than predicting the effect of these molecules on the proteins. In addition, GPCRs form a challenging set of target proteins, since their ligands can induce a variety of different effects. Finally, the general applicability of multi-target docking to a very large number of structures is investigated. For this evaluation, kinases are used as protein family since many more structures have been experimentally determined for these proteins compared to GPCRs as membrane proteins. First, using published experimental data, a dataset is created consisting of several hundred kinase structures and a set of small-molecule kinase inhibitors. This dataset is characterised by the availability of experimental binding data for each single kinase-inhibitor combination. These experimental data were subsequently compared to the docking results of each ligand into each single kinase structure. The results indicate that a reliable selectivity prediction for a ligand is highly demanding in such a large-scale setup and beyond current possibilities. However, it can be shown that the prediction accuracy of docking can be improved by normalising the docking scores over multiple ligands and proteins. Based on these findings, the idea of "protein decoys" is developed, which might in the future allow more accurate predictions of selectivity profiles using docking.
DOI:10.17192/z2015.0588