Intrinsically Multifunctional Energy Landscapes: A New Paradigm for Molecular Design

This project aims to advance theory and computer simulation to understand and design molecules capable of functioning as nanoscale devices. The inspiration comes from a recent study of an "intrinsically disordered" protein, which suggests new design principles for systems that can be switched in a controlled fashion between alternative configurations.

The underlying theoretical framework is based on analysis of the potential energy landscape, which defines the variation of potential energy with particle positions for any molecular or condensed matter system. In particular, we formulate observable properties in terms of local minima on the energy landscape, and the transition states and pathways that connect them. Within a well-defined set of approximations, this view reduces the corresponding computational framework largely to geometry optimisation. The results are translated into experimental observables using the tools of statistical mechanics and unimolecular rate theory. The applications will address two Priority Areas: nanoscale design of functional materials, and understanding of biological processes.

In previous work, we have established that systems with self-organising properties are associated with funnelled potential energy landscapes, where configurations are guided downhill towards a target morphology. This paradigm establishes a universality class, which includes magic number clusters (such as buckminsterfullerene), crystallisation, self-assembly, and protein folding. The realisation that intrinsically disordered proteins define an alternative class of behaviour leads us to consider a new paradigm for multifunctional systems. The research hypothesis addressed in the present proposal is that multifunctional molecules are associated with multifunnel energy landscapes. Understanding how naturally occurring systems exploit this capability, for example to bind different ligands, will provide design principles for artificial nanodevices that are switchable between alternative structures.

Project goals will be achieved through a series of work packages:

(1) Recent advances in methodology will be exploited to access experimental time and length scales. Implementing the corresponding computer programs on graphics processing units can provide efficiency gains exceeding two orders of magnitude. A variety of new ideas to further transform the sampling will be implemented and tested.

(2) Intrinsically disordered proteins can perform multiple cellular functions by binding different partners. We aim to test the hypothesis that multiple functions are associated with an intrinsically multifunnel potential energy landscape. The focussing effect of binding partners on the structure of the landscape will be examined for two particular proteins.

(3) The evolution of specificity for antibodies in the presence of antigens will be analysed in terms of the underlying landscape. Structure prediction and the effect of antigen binding and successive mutation will be related to changes in dynamics.

(4) Multifunnel landscapes will be investigated for nucleic acids. Competition between G-quadruplex structures is predicted to result in alternative morphologies separated by high barriers, which may represent important targets for drug discovery. Design principles for ultraresponsive DNA-based devices will be deduced for structures that incorporate fast-folding segments.

(5) The insight gained in the above projects will be used to design artificial nanodevices. Here we will consider switching via both external conditions, such as applied fields, and internal degrees of freedom that are accessible experimentally. For example, devices based upon helix inversion have the potential to couple linear and rotatory motion. To exploit this possibility we will design a photoswitchable chiral ligand. Transitions between the B and Z forms of DNA can also provide a route to nanoscale switches.

This project focuses on new theory and computer simulation techniques to understand the appearance of multifunctional behaviour and specificity in natural systems, and the exploitation of this insight to design artificial nanodevices. The framework involved is rather general, and applications could therefore be exploited directly wherever molecular simulation is used. Hence the immediate (non-academic) impact of this research would include two distinct communities. First we have industrial research teams specifically concerned with construction of nanodevices or exploitation of multifunctional systems. The next-generation antibody therapeutics of interest to Janssen R&D falls into this category. In addition, new methodology is potentially useful for any industrial research that employs computer simulation to predict structure, dynamics, or thermodynamic properties of matter. Examples of specific consultancy arrangements and current discussions with industry illustrate the diversity of applications, which range from:

* computer simulation of antibody binding - this is the new project with Johnson and Johnson. The company has funded an internship to fund a PhD student from the group for a six month collaborative research visit to Philadelphia. This connection provides direct evidence of the importance of the proposed project for industry, and will facilitate rapid knowledge transfer and impact.
* prediction of binding affinities for potential drugs with biomolecules, particularly proteins, which requires us to treat both the structure and thermodynamics (previous project with Evotec OAI - new discussions with companies such as Eli Lilly and Astex)
* structure prediction for metal clusters deposited on surfaces (consultancy with Exxon)
* simulation of phase behaviour for gas hydrates relevant to the petrochemical industry (projects with InfoChem)

An important impact is to provide advantages in terms of research capacity. Transforming the accuracy and speed of predictions from computer simulation has the potential to provide a competitive edge for UK industry, with corresponding benefits for the economy. New opportunities for molecular computer simulation would attract additional investment in R&D programs, with corresponding benefits for the economy. Long term impacts could derive from the discovery of new drugs, providing the possibility of societal benefits within sectors concerned with human health. As a specific example, insight into loop structure and dynamics will have important applications in antibody therapeutic development. More accurate predictions for the properties of materials could help to inform policy making, for example, in terms of how alternative energy reserves based on gas hydrate deposits might usefully be exploited.

Much of this impact for non-academic beneficiaries depends on knowledge transfer, either to industry, or to policy makers. For specific applications in industry, knowledge transfer is greatly facilitated by exchange of students, and explicit consultancy agreements, as for the projects with Evotec, Exxon, and Johnson and Johnson. The latest Knowledge Transfer Fellowship just starting with Biovia, provides further evidence of the relevance and potential impact of this research. Here the focus is on mechanisms for catalysis.

Some particularly successful projects have been initiated by former group members following career paths in industry, who are aware of the enhanced simulation capabilities provided by the computational side of the potential energy landscapes framework. However, other projects, such as the Exxon collaboration, have resulted from dissemination of new methodology via lectures presented at larger international meetings, backed up by publications and information on the group web site. These activities will therefore be extended selectively, choosing invitations to meetings with a view to the potential opportunities that may result.