Modelling protein-DNA interactions: bend and twist until it fits

Abstract

DNA-interacting proteins fulfil a vital role in the living cell. They allow the cell to quickly respond to changes in its environment by tuning the level of transcription. Furthermore, sophisticated protein machineries guard the DNA against hazardous influences that may damage it. A disruption of this delicate balance of regulation can lead to severe consequences for the cell ranging from the inability to perform certain functions to cancer or cell death. Detailed knowledge of the structure of protein-DNA complexes at atomic resolution provides valuable insights into their function and impacts on many different fields such as medical sciences, molecular biology and pharmacology. Nuclear Magnetic Resonance (NMR) and X-ray crystallography are the classical experimental methods used to solve the three-dimensional structure of these complexes. Although they are readily able to solve the structures of individual proteins in their free, unbound form, they often experience difficulties in solving the complex. The transient and reversible nature of protein-DNA associations, together with the difficulty of solving the structure of the DNA, makes these types of complexes a challenge to solve. This thesis deals with computational docking as alternative method to model the complex starting from the individual proteins and the DNA in their unbound form. Chapter 2 provides and overview of the young, but growing protein-DNA docking field. Its describes that the development of effective docking methods is slowed down by two main challenges; dealing with the conformational changes in both the protein and the DNA upon complex formation, and the efficiency of finding the correct interaction interface. In chapter 3, a unique two-stage protein-DNA docking method implemented in HADDOCK is described, that aims to deal with these challenges. This method uses additional experimental information about the complex to drive the docking, enabling better reconstruction of the interface. The use of explicit flexibility during the docking together with a DNA modelling stage enabled the method to deal with the conformational changes in protein and DNA upon complex formation. Chapter 4 describes the DNA analysis and modelling tool 3D-DART, used in the modelling stage of the docking protocol. This tool, available as web server, can be used to model DNA structures with custom conformations. Chapter 5 describes a dedicated protein-DNA docking benchmark, providing a representative selection of challenging protein-DNA complexes useful for validation, development and comparison of protein-DNA docking methods. Finally, the two-stage protein-DNA docking method is challenged using this benchmark. The method was successful in accurately modelling the variety of DNA conformational changes present in the benchmark and to assemble the protein-DNA interfaces using available experimental information. Despite the excellent results, there are still a number of test cases that pose a considerable challenge to the method. However, with the methods and tools described, the stage is set for further developments, stimulating others to take the challenges and leverage protein-DNA docking to the next level of accuracy.