Computatonal Modelling & Informatics
We use high-performance computer modelling methods to study the structure, function, and recognition properties of biomolecules (proteins, nucleic acids, and biologically active small molecules). The majority of these studies are closely integrated with experimental projects within the division, more widely across the university, or as part of collaborative ventures with other institutions.
Treating Biomolecular Flexibility in Drug Design
It is widely recognised that one of the stumbling blocks in the application of structural information (e.g. X-ray or NMR structures of proteins) to drug design is that the target is not rigid, but may adapt its shape to optimise interactions with potential ligands.
We have developed a new modelling method, "Active Site Pressurization" (ASP) that simulates the process of pumping Lennard-Jones particles into a protein's ligand binding cavity. As the particles fill the cavity, it distorts in the most energetically favourable way to accommodate them.
The process can be stopped at any point, when the L-J particles now form a "cast" of the shape of the binding site. These can then be used in either bespoke design or high-throughput in silico screening processes to find novel high-affinity ligands.
Applied to protein kinases, ASP can detect differences between apparently very similar ATP binding sites that stems from the fact that though their structures are similar, their deformabilities are different. This opens up new opportunities for the design of selective inhibitors.
Top
Solvent Effects in Protein-Ligand Recognition
The hydrophobic effect plays a major role in drug-target recognition, yet it remains far from fully understood. In collaborations with the Homans group and others at Leeds, we are studying the role of solvation in the binding of hydrophobic ligands to the Major Urinary Protein (MUP). The Homans group have shown that, contrary to expectation, ligand binding is enthaly, not entropy, driven. Using long timescale molecular dynamics simulations we have shown that the ligand binding cavity in MUP is so intensely hydrophobic that, in the absence of a ligand, water molecules avoid it, despite the fact that this creates an effective partial vacuum within the cavity {Barratt, 2005 #37}{Malham, 2005 #33}. The generality of this "dewetting" phenomenon, which explains the enthalpy-driven recognition, is now being further examined, using MUP mutants (collaboration with Stephanie Allen, LBSA).
Top
Understanding DNA Flexibility and Recognition
The cellular processing of DNA requires it to be extensively remodelled. The ability of DNA to be bent, unwound, denatured or reassembled into structures very different to the classic regular double helix is critically dependent on the base sequence, and indeed is now regarded as a "fourth dimension" to the genetic code. However, that code remains largely undeciphered. We have used molecular dynamics simulations to study the role of DNA flexibility in ligand recognition {Harris, 2001 #58}, how it deforms and breaks when stretched in AFM experiments {Harris, 2005 #40}, and how unusual DNA structures may be identified {Laughton, 2004 #41}.
Current work includes studies of how DNA sequence affects its compressibility again a simulation of AFM-type experiments; how DNA damage can signal to repair proteins through altered flexibility, studies on molecular motors that remodel DNA, and sequence effects in DNA supercoiling (collaboration with Jonathan Wattis, Maths, and the Harris group, Leeds). We have produced a web-accessible database of simulation data on DNA structures deposited in the PDB. We are also involved in a major international collaboration for the development of more accurate forcefields for the simulation of nucleic acids, and are a member of the Ascona B-DNA Consortium (ABC), a joint European/US initiative developing a very large database analysing sequence-dependent DNA flexibility.
Top
Promoting and Enhancing Biomolecular Simulation Techniques
Biomolecular simulation techniques are rapidly evolving, but the validation of new methods is frequently very difficult, since the key parameters which have to be validated are emergent properties of the system. We have developed a number of statistical analytical techniques to help with this process (collaboration with Ian Dryden, Maths) {Murdock, 2006 #26; Meyer, 2006 #30; Jha, 2005 #36; Sands, 2004 #43}. We are the current "headquarters" of the Collaborative Computational Project for Biomolecular Simulation (CCPB), established in 2006 with BBSRC support to "increase the UK�s capacity and capability in Biomolecular Simulation". CCPB organises networking activities, specialist training courses and an annual conference. It also supports ventures to develop standards, enhance the interoperability of specialist software, interface with other communities (e.g. NMR spectroscopists through CCPN), and ease access to high-performance computational resources for non-computer specialists.
Top
Computational Systems Biology
A newer direction for the division involves the development and application of analytical and stochastic simulation techniques to non-atomistic descriptions of systems as part of integrative biology studies. One project involves simulation of DNA repair, examining the "continuity testing" hypothesis for rapid lesion detection via the injection of mobile electrons into the DNA chain. A second project is concerned with the development of a "virtual cancer cell" that simulates cell cycle progression, multiplication, senescence and death. The model is being used to integrate and cross-validate experimental data (growth inhibition, cell cycle distributions, senescence and apoptosis) resulting from the treatment of cell lines with selected antitumour agents, and to look for agents with novel mechanisms of action (collaboration with Helen Byrne, Maths).
Top