Molecular simulations of the coupled binding and folding dynamics of intrinsically unstructured proteins

The classical view of protein function was that after protein synthesis in the cell, the protein chain needed to 'fold' up into a highly specific three-dimensional structure in order to perform its function. For example, enzymes fold into a structure that allows them to selectively bind to their substrate molecules and to catalyze the conversion into products by means of precisely positioned amino acid residues in the 'active site'. Interest has recently centred on a large class of proteins which do not form a single well-defined three-dimensional structure. These so-called 'intrinsically unstructured proteins' (IUPs) account for about 30 % of the coding regions of the human genome. They frequently play a role in transcription or signalling and often occur as modules in larger proteins that facilitate the assembly of protein-protein complexes. Many IUPs do form a structure when they bind to their target molecule in a coupled folding-binding process, which allows high specificity via a large interaction surface, but low affinity because of the entropic cost of binding. This project aims to study the mechanism of coupled folding and binding for several intrinsically unstructured proteins using computer simulations of atomistic models of the protein - that is, where all of the atoms of the protein are represented. We will begin by developing the energy functions to be used in the simulations, building on work already done by the PI, since an accurate energy function is critical for obtaining the correct mechanism. Next, we will develop simulation methods suitable for studying coupled folding and binding. The major hurdle to this application is the gap between the time scales which can be accessed in the simulations (0.1 - 1 microseconds) and those which can be accessed experimentally (milliseconds to seconds or even longer). We will do this by means of a combination of enhanced sampling methods to be developed as part of the project. We will apply the novel methods developed to study the coupled binding and folding of two systems. Firstly, we study a system which has become a paradigm for binding-induced folding, and which has been very well characterized experimentally. This involves the binding of a transcription factor (a protein which triggers initiation of the transcription of genes by the cellular machinery) to its coactivator (a protein which enhances the effect of the transcription factor). Binding of the intrinsically unstructured pKID domain of the transcription factor CREB to the KIX domain of its coactivator CBP causes pKID to assume a folded two-helix structure. We will characterize the thermodynamics and kinetics of binding by simulation and compare these with experiment. The main purpose of this system is as a test of the simulation methods, however it will also provide an atomistic picture of the binding mechanism and may suggest further experiments to probe this mechanism. Our second application is to a transcription factor-coactivator pair that is implicated in tumour growth. The response of cells to low oxygen conditions inside a tumour is triggered by the binding of the CAD domain of the transcription factor Hif-1alpha to the TAZ-1 domain of the coactivator CBP. In the bound structure, the intrinsically unstructured CAD domain wraps almost completely around TAZ-1, forming three short helices. We will determine the mechanism and association rate of this pair of proteins and compare them with experimental data. In this case, we also intend to use the simulation results to suggest ways in which the binding mechanism could be interfered with for the development of novel anti-tumour drugs. We hope to deduce some general principles for the mechanism of coupled folding and binding based on the examples studied. Furthermore, the methods developed as part of this proposal should be generally applicable to other instances of coupled folding and binding.

This proposal aims to develop novel methods for studying the mechanism of coupled folding and binding of intrinsically unstructured proteins with all-atom molecular simulations, and to apply these to two well-characterised experimental examples. The first system is the coupled folding-binding of the phosphorylated KID domain (pKID) of the transcription factor CREB to the KIX domain of its co-activator CBP. We will characterize the binding equilibrium using umbrella sampling along optimized reaction coordinates, where we will employ the Bayesian methods developed by the PI to find the optimal coordinates. We can compare the results with the experimentally determined binding affinities and free energy changes upon mutation. We will also determine the binding mechanism by sampling transition paths for binding via a transition-path sampling approach. From this we will be able to determine transition states and compute association rates for coupled folding-binding. A recent relaxation-dispersion NMR study provides detailed experimental information on the binding kinetics which can be compared with the simulations. The second system we will study is the binding of the CAD domain of the hypoxia-inducible transcription factor Hif-1 to the TAZ1 domain of its co-activator CBP: this interaction is important in the response to hypoxia in tumours. Despite a complex binding mode in which it wraps completely around TAZ1, the CAD domain associates with a rate that appears to be diffusion limited. We will apply the same simulation methods as for the binding of pKID to KIX in order to study the binding mechanism and association rate, in order to understand the very rapid binding in this case. In addition, we will use the mechanistic information gained from the simulations to suggest novel strategies for rational design of inhibitors of the CAD/TAZ1 interaction.

PRINCIPAL NON-ACADEMIC BENEFICIARIES The outcomes of the proposed research that could be of use to non-academic beneficiaries are: (i) Better force fields for molecular simulation (ii) Advanced sampling methods for studying coupled folding-binding and related phenomena (iii) Specific insights into the coupled folding-binding mechanism of the binding partners studied: pKID/KIX and CAD/TAZ1 The principal non-academic beneficiaries of these outcomes are: * Private Sector 1. Software companies selling software for molecular simulations will benefit from both the force fields and methods developed, (i) and (ii) above. 2. The pharmaceutical industry will benefit from all three outcomes, partially via software bought from the molecular simulations industry, and via any new insights into the binding of the CAD domain of Hif-1a to the TAZ1 domain of CBP. * Public/Third Sector 1. State-funded or Charity-funded medical research labs may benefit from the same outcomes as the pharmaceutical industry, (i)-(iii). 2. Ministry of Defence (MoD) Defence Science and Technology Laboratory (DSTL): biological research labs may also benefit from (i)-(ii). * Wider Public There will probably be no direct benefits to the wider public, but they may benefit indirectly from any outcomes that are useful to the pharmaceutical or medical research industry, for example. COMMUNICATIONS AND ENGAGEMENT The PI's Royal Society URF is funded by the MoD DSTL. As such, he has already engaged with the DSTL research laboratories in presenting his research to senior DSTL scientists. Although it is not clear yet whether this specific proposal will be relevant to them, he will be visiting the DSTL laboratories later in 2009 in order to meet more directly with the researchers there. He will then be able to gauge which aspects of the research will be most useful to them and how these benefits can be realized. The results of the research will be published in academic journals as early as possible and presented at academic conferences, where they can also be accessed by the non-academic beneficiaries. Similarly, the force fields and other data generated by the project will be made available as supporting information alongside publications, as well as online on the research group's website. If any non-academic beneficiaries are interested in further data or details on published research, the PI will provide these upon request. When results are published, the relevant software companies in the UK will be contacted in order to make them aware of this; a similar approach will be taken with pharmaceutical companies. If the results are of interest, they will be engaged to find out what future directions would be most useful to them, and whether a joint venture would be appropriate (e.g. through industry-funded studentship). COLLABORATION Should any of the potential beneficiaries become interested in the outcomes of the research, an informal collaboration would be started, possibly leading to a more formal relationship (e.g. through an industrial partnership). EXPLOITATION AND APPLICATION The maximum impact of the research will be achieved through open access to the results, which will be achieved principally through rapid and open publication: in this way the outcomes will be accessible to the largest possible number of beneficiaries. Although the expected benefits are outlined in the first section, the actual benefits can only be determined by engaging with potential beneficiaries. CAPABILITY The main engagement with the non-academic world will probably be done by the PI, with the PDRA involved if he is interested. RESOURCES FOR THE ACTIVITY Funds have been requested to cover the PI's travel costs involved with engaging with the MoD and other possible beneficiaries; funds for open access and presenting the results at academic conferences are also included.