Modelling biomolecular interactions

Computational approaches for understanding and aiding in molecular biology has increased in significance over the last decades, where a wealth of biochemical experiments have provided a solid ground for developing computer models that can be used to predict unresolved issues within biology. Molecular dynamics (MD) is one of these methods, based on classical laws, and suitable for handling large macromolecules in their natural environment, water. A detailed picture at the atomic level can be obtained, and given that the formulation of the computer model is correct, new strategies for experiments and interpretation of real-world results are feasible.

An important target group for treating various diseases is found within the nuclear receptor super family, which is the largest known group of transcription regulators. They are ligand induced and promote gene regulation by recognising a specific DNA-sequence. This recognition is performed by a discrete functional DNA-binding domain (DBD), which consists of two perpendicularly packed amphipathic helix loop regions with eight out of nine invariant cysteine residues formed in two Cys4-zinc fingers.

This thesis is mainly based on molecular dynamics simulations performed on the glucocorticoid receptor DNA-binding domain (GR DBD) and follows two main topics: Develop and evaluate methods for describing metals within the framework of classical laws and resolve some of the biology behind recognition of protein-DNA assemblies.

Non-bonded models that only include the Coulomb electrostatic and van der Waals interactions between Zn to its ligands were in better agreement with experimental data. Being satisfied with the non-bonded metal description, MD was conducted variant structures of GR DBD. One ligand (C496) in the second zinc finger appeared less rigid than the other ligands in conjunction with conformational changes in this region. In some cases ligand exchange with water was observed, which led to the speculation that P493 acts as a regulator for the conformational switch of C476.

A virtual alanine-scan was done on variant DNA-DBD complexes and on the free GR DBD monomer for defining putative mutations that may be done in vivo and/or in vivo. In particular, P493 mutants appeared to stabilise the free protein monomer GR DBD as well as the dimer GR DBD associated with DNA. MD simulations on variant P493-mutants, suggest that P493 regulates the relative positions of secondary structure elements within the GR DBD, and that mutations involving side-chains with hydrogen bonding capabilities recovers the loss of steric isomerisation that is encountered in the wild type protein.

The methods and results presented in this study are readily applied to other DBDs free or associated with DNA.

List of scientific papers

I. Bredenber J, Nilsson L (2001). Modeling zinc sulfhydryl bonds in zinc fingers. International Journal of Quantum Chemistry. 83: 230-44.

II. Bredenberg J, Nilsson L (2002). Conformational states of the glucocorticoid receptor DNA-binding domain from molecular dynamics simulations. Proteins. 49(1): 24-36.
https://pubmed.ncbi.nlm.nih.gov/12211013

III. Bredenberg J, Mark P, Nilsson L (2003). Solvent effects in biomolecular dynamics simulations: A comparison of TIP3P, SPC and SPC/E acting on the glucocorticoid receptor DNA-binding domain. [Submitted]

IV. Bredenberg J, Nilsson L (2003). Computational alanine scanning on the glucocorticoid receptor DNA-binding domain reveals a biological function for Pro 493. [Manuscript]