Binding-affinity predictions of HSP90 in the D3R Grand Challenge 2015 with docking, MM/GBSA, QM/MM, and free-energy simulations

Abstract

We have estimated the binding affinity of three sets of ligands of the heat-shock protein 90 in the D3R grand challenge blind test competition. We have employed four different methods, based on five different crystal structures: first, we docked the ligands to the proteins with induced-fit docking with the Glide software and calculated binding affinities with three energy functions. Second, the docked structures were minimised in a continuum solvent and binding affinities were calculated with the MM/GBSA method (molecular mechanics combined with generalised Born and solvent-accessible surface area solvation). Third, the docked structures were re-optimised by combined quantum mechanics and molecular mechanics (QM/MM) calculations. Then, interaction energies were calculated with quantum mechanical calculations employing 970-1160 atoms in a continuum solvent, combined with energy corrections for dispersion, zero-point energy and entropy, ligand distortion, ligand solvation, and an increase of the basis set to quadruple-zeta quality. Fourth, relative binding affinities were estimated by free-energy simulations, using the multi-state Bennett acceptance-ratio approach. Unfortunately, the results were varying and rather poor, with only one calculation giving a correlation to the experimental affinities larger than 0.7, and with no consistent difference in the quality of the predictions from the various methods. For one set of ligands, the results could be strongly improved (after experimental data were revealed) if it was recognised that one of the ligands displaced one or two water molecules. For the other two sets, the problem is probably that the ligands bind in different modes than in the crystal structures employed or that the conformation of the ligand-binding site or the whole protein changes.

Keywords: Bennett acceptance ratio; Big-QM; Blind-test competition; Continuum solvation; D3R grand challenge; Free-energy perturbation; Induced-fit docking; Ligand-binding affinity; MM/GBSA; QM/MM.

Fig. 1

Structures of all ligands from sets 1, 2, and 3, considered in this study. The additional reference ligands that were employed for sets 1 and 3 are also shown. The numbering of ligands is the same as in the HSP90 D3R grand challenge data set. Ligands of sets 1 and 3 are shown in conformation 1

Fig. 2

The QM systems used in the QM/MM optimisations for sets 1 and 2 (a), and set 3 (b), as well as in the big-QM calculations (c, d). The ligand is shown in ball-and-sticks representation

Fig. 3

Binding modes for the three series of HSP90 inhibitor from the docking calculations: a set 1, b original docking for set 2, based on the 3VHA crystal structure (submitted), c set 2 in the 2WI7 crystal structure, keeping all water molecules, and d set 3. Carbon atoms of the residues are shown in light grey tubes, showing some movements as result of the induced-fit docking protocol. Carbon atoms of the ligands are shown as green tubes. Water molecules that interact with the ligands are displayed in thick tube representation and labelled as WAT. Reference crystal structures (3VHA, 2WI7, and 3OW6 [36, 37, 39]) are coloured in cyan for comparison (both ligands and protein). Nitrogen and oxygen atoms are blue and red, respectively. Hydrogen bonds are represented as yellow dashed lines (purple if the acceptor is a halogen atom). Cation-π and π-stacking interactions are represented as dark green and dark cyan dashed lines, respectively

Fig. 4

Correlation between the experimental [23] and calculated binding affinities. Sets 1–3 are marked with squares, triangles, and circles, respectively. For GScore, the original score is shown, whereas for E_model, IFDScore, and MM/GBSA, the mean signed error is subtracted (to give a similar scale of all the calculated results). The line shows the perfect correlation. Ligand 61 was experimentally found to be a non-binder, i.e. with a K _i > 50 µM, which corresponds to ∆G _bind > −25 kJ/mol

Fig. 5

Binding modes in the FES calculations. a ligand 80 (set 1; all the other ligands in this set bind in a similar mode), b set 2 ligands, based on the 2WI7 crystal structure, c ligands 101, 105, and 106 (set 2) with three water molecules in different colours (the one in magenta corresponds to Wat2 and that in orange corresponds to Wat3), d ligand 100 (set 2), based on the 3FT5 crystal structure, and e ligand 10 (set 3; all the other ligands in this set bind in a similar mode). Hydrogen bonds are indicated by green dotted lines

Fig. 6

Water clusters obtained by GCMC method for the a 2WI7 and b 3FT5 structures with set 2 ligands. In both figures, ligands and the corresponding water molecules are presented in different colours: ligand 100—blue, ligand 101—red, ligand 105—yellow, and ligand 106—green