Binding site dynamics and solvation

Background.

Binding site flexibility, as well as solvation and desolvation, strongly affect the ability of proteins to bind to substrates and inhibitors. These two aspects hamper traditional computer-aided drug design methods and are also still major challenges for more elaborate free-energy calculations. In the past we have contributed to the field by addressing the (dis)appearance of water molecules in the active site explicitly during free-energy calculations and by developing methods to either enhance or focus the conformational freedom of molecular species. (1-3)
Particularly challenging are processes for which larger conformational changes are expected. For example, when a loop needs to change its conformation to accommodate a larger ligand, or when a conformational change takes place at a distant location of the protein. Typical examples are allosteric effects of modulators on protein activity, or mutations that occur at a distant site and affect the binding of inhibitors. We should not think of such events as a single conformational change, but rather as a shift of the conformational ensemble of the protein.

Free-energy calculations are increasingly often applied to predict (relative) binding free energies. Typically, an alchemical modification is performed gradually changing a small binding molecule into another variant, or to non-interacting dummy particles. This process involves simulation at different intermediate states. The challenge in free-energy calculations arises from the fact that this conformational ensemble, together with such subtle shifts, should be sampled in all intermediates of the alchemical process. This requires long simulation times or the use of well-tuned replica exchange perturbations. Alternatively, one can use methods that are based on individual, long simulations of a (potentially unphysical) reference state, such as in the OSP method or in EDS. The advantage is that the conformational equilibrium needs to be captured once and can potentially be enforced by the use of enhanced sampling algorithms. The free-energy calculation can subsequently correct for any unphysical biases and yields the free-energy differences between different ligands or amino acids in the context of the conformational change.

Aims and methods.

In this project, we will use enhanced sampling methods to describe the conformational flexibility of the RAS and CYP superfamilies of proteins. Active site dynamics and solvation effects of the bound and unbound ligands will be focus points. For the CYP superfamily, insight from site-of-metabolism prediction by our collaborators will be used to guide the docking of substrates into the active site. The generated conformational ensembles can be used for hot spot analysis, or for the generation of dynamic pharmacophore models. Furthermore, we will explore the effect of large conformational flexibility on the accuracy of free-energy calculations based on single simulations. The RAS proteins again offer an excellent playground for these effects: mutations may affect the direct binding affinity, or the conformational ensemble of the protein itself, affecting the occurrence of states that are particularly relevant for specific interactions. Simultaneously, differences between the active and inactive states can be captured and relevant thermodynamic cycles capturing the different states alchemically or through enhanced sampling can be designed.

Anticipated outcomes and innovation.

A thorough characterization of the conformational ensembles of selected members of the RAS and CYP superfamilies of enzymes will be obtained. We will explore the use of novel grand-canonical methodologies to sample the absence and presence of water molecules to enhance the sampling of different solvation states and combine these with enhanced sampling techniques like accelerated MD to observe conformational changes. The ensembles are of interest by themselves to understand the conformational landscape of these challenging proteins, and will also be used directly to analyze hot spots for active site solvation and molecular interactions. Furthermore, once broad ensembles can be generated, the methodologies will be extended to obtain binding free-energy estimates of flexible and dynamically solvated active sites.

1. A. de Ruiter and C. Oostenbrink, Efficient and accurate free energy calculations on trypsin inhibitors. J. Chem. Theory Comp. 8, 3686 - 3695 (2012)
2. A. M. Maurer, S.B.A. de Beer and C. Oostenbrink, Calculation of relative binding free energy in the water-filled active site of oligopeptide-binding protein. Molecules 21, 499 (2016)
3. M. Maurer, N. Hansen and C. Oostenbrink, Comparison of free-energy methods using a tripeptide-water model system. J. Comput. Chem. 39, 2226 - 2242 (2018)