Drug DesignV. Zoete and O. Michielin
Development of EADock: docking of small molecules into protein active sites with a multiobjective evolutionary optimizationA. Grosdidier, V. Zoete and O. Michielin
In recent years, protein-ligand docking has become a powerful tool for drug development. Although several approaches suitable for high throughput screening are available, there is a need for methods able to identify binding modes with high accuracy. This accuracy is essential to reliably compute the binding free energy of the ligand. Such methods are needed when the binding mode of lead compounds is not determined experimentally but is needed for structure-based lead optimization. We present here anew docking software, called EADock, that aims at this goal. It uses an hybrid evolutionary algorithm with two fitness functions, in combination with a sophisticated management of the diversity. EADock is interfaced with the CHARMM package for energy calculations and coordinate handling. A validation was carried out on 37 crystallized protein-ligand complexes featuring 11 different proteins. The search space was defined as a sphere of 15 Å around the center of mass of the ligand position in the crystal structure, and conversely to other benchmarks, our algorithms was fed with optimized ligand positions up to 10 Å root mean square deviation (RMSD) from the crystal structure. This validation illustrates the efficiency of our sampling strategy, as correct binding modes, defined by a RMSD to the crystal structure lower than 2 Å, were identified and ranked first for 68% of the complexes. The success rate increases to 78% when considering the five best ranked clusters, and 92% when all clusters present in the last generation are taken into account. Most failures could be explained by the presence of crystal contacts in the experimental structure. Finally, the ability of EADock to accurately predict binding modes on a real application was illustrated by the successful docking of the RGD cyclic pentapeptide on the pocket (Grosdidier A, Zoete V and Michielin O, Proteins, 67, 1010).
Development of fast methods for ligand-protein binding free energy estimationV. Zoete, A. Grosdidier and O. Michielin
The binding free energy is directly related to the association constant between ligand and protein. It is thus a of major interest in a drug design process where it allows the determination of the most promising ligand candidates in terms of affinity for the targeted macromolecule, and also drives the lead optimization process. However, fine estimates of binding free energy can be time consuming because it requires an accurate determination of the interactions and a reliable treatment of solvent effects. Therefore, finding methods for the evaluation of ligand binding affinity that are fast enough to treat thousands of candidate ligands in a realistic time, while keeping a reasonable accuracy, is a major challenge in computational drug design (Zoete V, Michielin O and Karplus M, J Comput Aided Mol Des. 2003, 17, 861).
A variety of methods have been proposed for estimating the binding affinity in drug design studies. They include the very fast QSAR type approach, as well as regression-based and PMF scoring functions. With the exception of QSAR, these methods are claimed to be universal and applicable per se to all types of proteins. However, recent assessment studies and review articles have shown that existing methods exhibit only very limited efficiency in estimating absolute binding free energies for ligand-protein association. Thus, there is still a need for a universal, fast and reliable scoring function. We are actually developing a new and universal method for estimating absolute ligand-protein binding free energies. This approach will be built using the Ligand Protein Database (LPDB) that contains about 260 different ligand protein complexes, for which experimental structures and binding free energies are available. The ability of the method to rank different possible binding modes of a given ligand will also be assessed, allowing its use as a scoring function in docking approaches.
QM/MM dockingU. F. Röhrig, V. Zoete, A. Grosdider and O. Michielin
We are currently developing schemes to include quantum-mechanically computed binding energies into the scoring function of our docking algorithm EADock. This approach should, for example, improve the performance of EADock when treating metalloproteins. Metalloproteins play an important role in physiological processes, in the binding of potential drugs, and in drug metabolism. In drug design, a description of ligand interactions with transition metals poses a challenge due to the partly covalent nature of the metal-ligand bonds and due to the multiple coordination geometries that may be adapted. These phenomena are most appropriately treated at the quantum mechanical level.
We have carried out tests using both the SCC-DFTB code included in the CHARMM molecular modelling package, or the QM/MM interface of the CP2K code. Promising preliminary results are obtained for the case of indoleamine 2,3-dioxygenase (IDO), where the best predicted binding mode of 4-phenylimidazole improves from having a RMSD of 0.84 A with respect to the X-ray structure to a RMSD of 0.45 A upon inclusion of a quantum correction.
Development of fragment-based computer-aided drug design approachesV. Zoete, U. F. Röhrig, A. Grosdidier and O. Michielin
The fragment-based approach to structure-based computer-aided ligand design consists mainly of three steps. In the first, positions and orientations for a series of molecular fragments are determined in the known structure of the macromolecular target. The EADock program created in our laboratory has also been developed for this purpose. The fragments may correspond to molecular functions and frameworks that can be found frequently in organic molecules, or peptides. In the second step, selected molecular fragments are connected to form putative ligands. The fragments may be connected directly, or through the use of an additional linker. In a first approximation, the connection of the fragments may be done based on simple geometric rules. However, we are also developing a new method that will take account of other physical parameters during the fragment connection, like solvation, and will also allow a dynamic modification of the linker nature as a function of its environment. At this stage, hundreds to many thousands candidate ligands can be obtained. In the final step, the ligand candidates are examined for synthetic feasibility and their free energies of binding are estimated to determine which of them are likely to have the strongest affinities for the target. As mentioned above, new approaches for ligand-protein binding free energy approaches will be derived.
Rational design of integrin peptide inhibitorsV. Zoete, A. Grosdidier and O. Michielin, in collaboration with G. Alghisi and C. Ruegg
One of the major causes of mortality in cancer patients comes from the invasive and metastatic behavior of malignant cells. The migration of cancer cells is regulated by their physical adhesion to other cells and to the extracellular matrix (ECM), and by transmission of signals from the extracellular environment. Integrin receptors are membrane proteins that play an important role in the attachment of the cell to the ECM, and for the signal transduction from the ECM to the cell in relation with cell survival and apoptosis. RGD cyclic peptides, such as cilengitide, that target and block the avb3 receptor induce apoptosis in endothelial cells, inhibit angiogenesis, and block tumor growth. In collaboration with the CePO and ISREC, our group is involved in the computer-aided structure-based rational design of new ligands of different subtypes of integrin for use in treatment against cancer.
Docking study of Indoleamine 2,3-Dioxygenase ligandsU. F. Röhrig, V. Zoete, A. Grosdider and O. Michielin
Inhibition of the enzyme indoleamine 2,3-dioxygenase (IDO), which catalyzes the rate-limiting step in the tryptophan catabolism, has been shown to enhance the effectiveness of cancer immunotherapy. We use the evolutionary docking algorithm EADock in conjunction with the recently resolved crystal structure of human IDO in order to elucidate the binding mode of tryptophan and known inhibitors. The validity of our docking algorithm is tested on the co-crystallized inhibitor 4-phenylimidazole, for which an excellent agreement with the X-ray structure is found (rmsd 0.4 A). For tryptophan as well as different classes of known inhibitors, a consistent binding mode is found inside the active site pocket, opening the way to rational structural modifications in order to obtain ligands with higher binding affinities.