Skip to main content

Crystallization of charged proteins

Final Activity Report Summary - PROTEIN-CRYSTALS (Crystallisation of charged proteins)

In the project "protein-crystals" we were studying the influence of electrostatics on the crystallisation process of proteins using computer simulations. Most proteins typically need salt to be stable in solution, and often carry charges themselves. These charges have to be compensated for by the salt; therefore the salt concentration differs between the inside and outside the crystals. This creates an electrostatic potential at the crystal surface, the so-called Donnan potential, which should influence the crystal formation process. The crystal surface is experimentally not accessible, therefore we employed computer simulations techniques to study the Donnan-effect.

Studying this effect required a simulation code that is on one hand efficient in treating the electrostatic interaction, and on the other hand employed state-of-the-art methods for sampling rare-events such nucleation, i.e. the onset of crystal growth. During the last two years, the necessary state-of-the-art algorithms were implemented into the freely available simulation package Espresso. While Espresso already contains efficient algorithms for treating charged systems, rare event sampling was not available and had to be developed. We decided to add the recently developed Forward Flux Sampling (FFS) method. Simulations on the Donnan effect have started as planned; however, due to the complexity of the problem, the results are not yet ready for publication.

With respect to transfer of knowledge, the project was extremely successful. The Espresso code, of which A. Arnold is one of the core developers, was established as a Molecular Dynamics simulation tool in the groups of D. Frenkel in AMOLF, Amsterdam and of B. Mulder at the University of Wageningen, and several ongoing projects perform their simulations using the package. In a joint project together with S. Jun from the Biophysics group of AMOLF, who recently joined Harvard University, and Bae-Yeun Ha from the University of Waterloo, we studied the dynamics of DNA in confinement using Espresso. We could show theoretically and by simulations that entropic effects alone can lead to the segregation of flexible polymers in confinement. This means, that for small cells such as bacteria, there is actually no need for a complex DNA segregation mechanism like it is found in all higher cells. This can not only explain, why despite many efforts no common mechanism for bacterial could be identified, but also, how early, primitive cells could replicate. Until now, this collaboration has led to three publications in international, peer-reviewed journals, but further projects on the dynamics of knots in DNA are in work.

In a second collaboration with S. Portegies Zwart and R. Belleman from the astrophysics and computer science departments of the University of Amsterdam, we studied the implementation of Molecular Dynamics (MD) simulations on recent graphics cards. With new technology introduced by NVidia at the end of the year 2006, we were able to perform MD simulations up to 80 times as fast as on a conventional processor. A quick adoption of this new technology brings the power of a supercomputer to the desk of every researcher. This should make it possible to tackle many problems that are inaccessible for computer simulations at the moment due to computation time requirements.