Kinetic of protein kinases inhibitors and Affinity by flexible Docking – ChADock

The in silico method widely used to predict protein-ligand interactions is molecular docking. However, despite the large improvements of these technics in the last decades, several limitations remain that lead to inaccurate predictions. One major limitation of molecular docking is its inability to take into account the flexibility of the protein receptor .
Molecular dynamics simulation is an in silico method that allows to estimate the motion of a system taking into account its environment. By combining different numerical simulation methods ( at different scales of representation , with or without constraints, etc ...), we will be able to overcome the shortfalls of molecular docking to refine predictions and guide experimentalists to design innovative active compounds.

Development of the predictive tool :
-----------------------------------------------
The development of this tool consists in three main steps:
1) development of a multiconformational docking protocol (the aim is to enrich the number of selected compounds by limiting the number of false negatives)
2) simulation of the ligands' entry paths (the aim is to limit the number of false positive compounds in the set of potentially active ligands )
3) prediction of the affinity constants and kinetic rate constants .
Currently , we achieved and validated the step 1) of this process.

Application to the design of LIMKs inhibitors :
------------------------------------------------------------
To perform this step of the project, we built homology models of LIMK-1 and LIMK-2 proteins in their active and inactive forms.

Development of the predictive tool :
-----------------------------------------------
The development of the prediction tool continues with the second step, which is to simulate the ligand entry into the active site of the receptor. This entry is divided in two distinct parts:
a) the diffusion of the ligand to a favourite gate on the surface of the protein
b) the passing the ligand from the solvent toward the binding site of the protein
These two steps will be performed using different numerical simulation methods. The whole process of association (diffusion of the ligand and its passage through the protein to bind the active site) will be then used to predict the kinetic constant of association (kon).

Application to the design of LIMKs inhibitors :
------------------------------------------------------------
The homology models will be used in conventional docking approaches to guide the chemists in the design of new LMIKs' inhibitors by using their own scaffolds. These models will also be used as starting conformations for a conformational sampling and mult-docking conformations from virtual chemical libraries . The objective will be to identify new lead compounds that will be optimized by our collaborators.

This project have been the subject of three posters presentations at the following congress:
- Journées de la section régionale centre-ouest - Société Chimique de France, Orléans (France), 19-20 février 2015
- 4ème congrès du GT Enzymes et 19ème congrès du GGMM, Sète (France), 25-28 mai 2015
- 2nd Novalix conference Biophysics in Drug Discovery, Strasbourg (France), 9-12 juin 2015

Protein-ligand interactions are essential for a majority of biological processes. In silico drug design, which aims at characterizing those interactions, plays a fundamental role in the development of medicinal chemistry expertise. The docking methods were especially developed in the rational design field since they aim to predict the binding affinity and the molecular interactions of a molecule for a target.
Several enhancements have been made in the past few years to allow docking methods to treat ligand flexibility or to improve scoring functions that evaluate the binding affinity between the ligand and the receptor. However, none of the current methods is able to properly consider the receptor flexibility during the ligand binding process in the active site. It is nevertheless well accepted that the ligand flexibility but also the intrinsic receptor flexibility are crucial for the formation of protein-ligand complex.
Moreover, the existent docking methods consist in positioning the ligand straight into the active site of the receptor, meaning to focus only on the evaluation of the ligand-receptor interactions and to discard the dynamic and kinetic aspects of the general binding process. The prediction of the ligand entry pathway and the evaluation of the kinetic rate constants (kd, kon, koff) could provide insights on the complex stability and on the ligand true efficiency.
This project aims to develop a new predictive method that allows the selection of a subset of potentially active compounds for a given target and then to evaluate i) their binding affinity calculated by free energy of binding ii) the entry pathway taken by the ligand and iii) the association rate constant (kon) of the biological mechanism. The most promising compounds will be synthetsized and tested onto the target of interest in order to compare the predicted and experimental data.
The methodology will be first developed and validated using receptor-ligand datasets extracted from the DUD (a Directory of Useful Decoys) database. We focus our development on protein kinase targets, which are highly flexible proteins due to several loops in the close vicinity of the ATP substrate active site. The validated method will then be applied on current internal drug design project to search novel efficient inhibitors for promising targets such as LIMK protein kinase family.
Two serine/threonine and tyrosine kinases, LIMK1 and LIMK2, constitute this protein family. They are structurally close (72% of kinase domain identity) but have different expression levels, subcellular localizations and functions. They are involved in many cellular functions such as migration, cell cycle and neuronal differentiation and in pathological processes such as tumor invasion and neurodevelopmental disorders. The known substrate of LIMKs is cofilin, a depolymerisation factor of actin.
By phosphorylating cofilin and inactivating it, these two kinases modulate the structure and the activity of the actin cytoskeleton, which is involved in cellular morphogenesis, motility, division, differentiation and apoptosis. LIMK are also involved in microtubules dynamics during mitosis, although the molecular mechanism is unknown. Recent data have shown that LIMK could be considered as emerging targets for cancer treatment. Therefore, our project to develop new LIMK inhibitors might be of great interest in developing new anti-cancer therapeutics.
An interesting perspective of this work is to explore further the prediction of kinetic data with calculating the dissociation rate constant (koff) of the protein-ligand process. Indeed, it was recently proposed that koff could be a key indicator of the in vivo compound duration efficacy. Moreover, this constant could provide insights into the in vivo drug selectivity. The prediction of such biochemical data could be very useful for preclinical compound selection and optimization. The outcome of the project results will be fully adapted to other biological systems.