The discovery of kinase-specific inhibitors is intensely pursued within the pharmaceutical industry. In spite of multitude of X-ray structures for this protein family, it is increasingly apparent that a number of critical challenges must be overcome to make structure-based drug design (SBDD) a completely reliable tool for the discovery process, especially in the later phases of a drug discovery (lead optimization) that deal with balancing of potency with selectivity, in vivo target engagement, and pharmacokinetic profile. Even when multiple X-ray structures for a selected target are available, our knowledge of all the relevant conformations relevant for ligand binding remains limited, thus hindering the full potential of SBDD. Lack in inhibitor selectivity often leads to undesired side effects caused by off-target binding. But while thermodynamic binding equilibrium considerations are critical, it is also important to go beyond to explain and predict the association/dissociation kinetic rates and the residence time of inhibitors. The latter can strongly affect many aspects of in vivo pharmacokinetics. All these issues become especially important when trying to rationally design and optimize specific covalent inhibitors whose mode of action is sensitive to both thermodynamic and kinetic factors. There is an urgent need to begin to systematically overcome these challenges for technologies like SBDD to play an increasing role in the development of targeted therapies based on kinase-specific inhibitors. A research plan comprising of four specific aims is proposed to develop a comprehensive computational/theoretical framework in order to systematically overcome and address these challenges. Our computational framework will integrate the information from explicit-solvent molecular dynamics simulations, adaptive enhanced sampling strategies, transition pathways from the string method, de novo structure prediction, and Markov state models. Specifically, we will develop, test and validate an integrated computational approach to accurately predict and rank-order all the accessible conformational variants of a target protein; then expand this approach to de-novo predict and rank-order all the accessible binding poses of a ligand in a given kinase of interest. We will also develop, test and validate an integrated computational approach to quantitatively determine and predict the associate/dissociation rates kon and koff of ligand binding, then expand this approach to simulate the formation of covalent linkage (reversible and irreversible) between a ligand and a target kinase, accounting for binding mode and reactivity. Finally, the computational framework will be used to investigate the molecular determinants for the specificity of a novel family of pan-kinase probes and test whether they are compatible with genome-wide profiling. The entire computational framework will be automated and streamlined user- friendly tools will be freely distributed to the general community through our web site. All results of this work will be made publically accessible to the whole community through our web site.
Proteins kinases are critical therapeutic targets for the treatment of cancer as well as other diseases. While the discovery of kinase-specific drugs is pursued by the pharmaceutical industry, overcoming important challenges arising from the complexity of structural flexibility of these proteins would greatly accelerate discovery of inhibitors. The development and validation of computational approaches is proposed to overcome these challenges and open new avenues for structure based design.
Showing the most recent 10 out of 39 publications
