Binding-affinity predictions of HSP90 in the D3R Grand Challenge 2015 with docking, MM/GBSA, QM/MM, and free-energy simulations
J Comput Aided Mol Des
Binding-affinity predictions of HSP90 in the D3R Grand Challenge 2015 with docking, MM/GBSA, QM/MM, and free-energy simulations
Majda Misini Ignjatovic´ 0
Octav Caldararu 0
Geng Dong 0
Camila Mun˜ oz-Gutierrez 0
Francisco Adasme-Carren˜ o 0
Ulf Ryde 0
0 Centro de Bioinforma ́tica y Simulacio ́n Molecular, Facultad de Ingenier ́ıa, Universidad de Talca , 2 Norte 685, Talca , Chile
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
Ligand-binding affinity; Induced-fit docking; MM/GBSA; QM/MM; Big-QM; Free-energy perturbation; Continuum solvation; Bennett acceptance ratio; D3R grand challenge; Blind-test competition
-
Majda Misini Ignjatovic´, Octav Caldararu, Geng Dong, Camila
Mun˜oz-Gutierrez and Francisco Adasme-Carren˜o have contributed
approximately equal to the investigation: MMI performed the FES
simulations of sets 1 and 3, as well as the GCMC calculations; OC
performed the FES calculations on set 2; GD performed the QM/MM
calculations; CMG and FAD performed the docking and MM/GBSA
calculations.
Department of Theoretical Chemistry, Lund University,
Chemical Centre, P. O. Box 124, 221 00 Lund, Sweden
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.
Introduction
One of the prime challenges of computational chemistry is
to predict the free energy for the binding of small
molecules to biomacromolecules. Many biological functions are
exerted by the binding of substrates or inhibitors to
enzymes or effectors to receptors, and the prime aim of
drug development is to find small molecules that bind
strongly to the target receptor, but with a small effect on
other biosystems. Consequently, much effort has been
spent to develop methods with this aim, ranging from
simple docking and scoring approaches, via end-point
methods, such as MM/GBSA (molecular mechanics
combined with generalised Born and solvent-accessible surface
area solvation) and linear interaction energies (LIE), to
strict free-energy simulation (FES) methods [
1–4
].
Numerous studies have evaluated the performance of
various binding-affinity methods, e.g. docking [
5, 6
], MM/
GBSA [
7, 8
], and FES methods [
9–11
]. The conclusion has
typically been that docking methods can rapidly find the
correct binding pose among several other poses, but that
they have problems to correctly rank the affinities of a set
of ligands to the same protein. MM/GBSA calculations
typically give a better ranking of the ligands and an
understanding of energy terms involved in the binding, but
often vastly overestimate energy differences and the results
strongly depend on the employed continuum-solvation
model [
2, 12
]. Large-scale tests of FES calculations have
given rather impressive results for relative binding
affinities of similar ligands to the same protein, with mean
absolute deviations (MAD) of 4–6 kJ/mol [
9–11
].
However, the comparisons have been primarily directed to
small changes in the ligands and the performance is
uneven, with very good results for some proteins, but quite
poor performance for other proteins, occasionally with
errors of over 20 kJ/mol.
Comparisons of different approaches for the same test
case are less common and often half-hearted in the
meaning that the authors are experts or developers of one
approach (...truncated)