Download PDF - EPCC - University of Edinburgh

FOCUS: HARDWARESelf-assembly and hydration inbiological systems:lessons from smallmoleculesJason Crain, School of Physics, University of EdinburghUnderstanding the principles and processes by which proteinsin aqueous solutions assemble from amino acid chainsinto biologically functional three-dimensional structuresis recognised as one of the major challenges spanning thephysical and life sciences. In addition to fundamental interestin the factors surrounding this process, protein mis-foldingevents are also attributed to diseases such as cystic fibrosis,Creutzfeldt-Jakob disorder and Alzheimers. Therefore, thelong-term impact of advances in this area also reaches intomedicine and may have future ramifications for protein andpeptide design. The basic molecular-scale issues that underliefolding and other related ‘self-organisation’ phenomena arevery general and concern the following themes:1) Hydration structure: How are the water moleculesorganised around the polar, non-polar and peptide groups ofcomplex biomolecules? How is this organization influencedby externally-controllable conditions such as temperature andpressure?2) Molecular flexibility: How flexible are the bonds ofcomplex molecules that define overall macromolecular shape?And how is this flexibility affected by hydration structureand other environmental factors such as pressure andtemperature?3) Ionic (Hofmeister) Effects: How does the presence ofsimple inorganic ions influence hydration structure andflexibility?The size, complexity and diversity of real proteins andbiomolecules still renders a thorough molecular-levelinvestigation of these issues impractical for systems of directbiological relevance. A promising emerging strategy in thestudy of processes with inherently high levels of structuralcomplexity, multiple degrees of freedom and competinginteractions makes use of so-called model systems whichrepresent extreme approximations of ‘real’ macromolecularmaterials. Chosen carefully, and with appropriatecombinations of computational and experimental input,co-ordinated investigations on model systems can be veryA simulation of complexbiological molecules.powerful, leading to insight into muchmore complex systems. In the specificcase of the protein folding problem and self-organisationphenomena in biomolecules, there are potentially many viewsas to what level of abstraction relative to a real biomoleculeis ‘useful’ for a particular purpose. Ultimately, there mustbe some compromise found between the degree of ‘realism’of the system and the level of detail at which the system isdescribed. In those cases where the fundamental questionsare general rather than specific to particular biomolecules,extreme simplifications of real biological systems may bemade. The benefit of this approach is that very sophisticatednew computational (such as polarisable models and pathintegral molecular dynamics) and experimental tools(isotope-labelled neutron diffraction and empirical potentialstructure refinement) can be deployed together for the firsttime to build confidence in new models.With this strategy in mind, a major collaboration betweenthe University of Edinburgh and IBM Research has beenestablished to combine massively parallel supercomputingwith high-resolution neutron diffraction experiments atatomic resolution. To support the project IBM has recentlyannounced a major Shared University Research (SUR) awardto Edinburgh including leading-edge Power5 microprocessortechnology (16-way SMP, 1.9GHZ p570 processors with128GB RAM) and high performance graphics workstationsenhanced by world-class 9 million pixel LCD monitors forvisualization of complex structures. The pSeries system willbe housed at the ACF, where support for the machine will beprovided by EPCC. The project also makes extensive use ofthe Blue Gene systems in both the ACF and in the TJ WatsonCenter.9