inhibitors, structure- and computer-based approaches were applied to analyze targetligand interactions and to rationalize the development of small molecule modulators
[3]. This chapter will illustrate various structure-based approaches which have been
successfully implemented to identify and design small molecule inhibitors of
histone-modifying enzymes.
2 Computational Methods to Predict the Target Interaction
of Small Molecules and Their Binding Affinities
Designing small molecules with desired biological activity has long been a major
goal of structure-based drug design. Different approaches have been tested to predict
the biological activity of an inhibitor from its calculated interaction with the target.
Molecular docking tools are able to predict the interactions in a putative binding site
and can easily screen large numbers of molecules and rank them according to their
calculated affinities [4, 5]. However, using a single frozen protein structure in most
of the cases hampers the true conformational and orientational sampling of an
inhibitor in its putative binding pocket [6]. Moreover, currently available scoring
functions that estimate the binding affinity of calculated poses for a ligand are still far
away from ranking the biologically active molecules in a reasonable agreement with
the experimental data [7]. To overcome this challenge, more sophisticated methods
that take into account conformational changes, desolvation effects, as well as
entropic effects on binding are needed to predict the binding strength of molecules
of interest.
In contrast to simple docking methods, free energy calculations were shown to
give results that are in better agreement with experimental data [8]. Two different
approaches have been widely studied for their applications in estimating the binding
affinity of small molecule inhibitors. The first approach includes the so-called
pathway methods such as free energy perturbation (FEP) and thermodynamic
integration (TI) which consider all small changes that occur between the initial and
final states [9]. These methods are more accurate and robust in predicting reliable
affinities than docking methods but are expensive and time-consuming, thus, making
them inapplicable for large datasets of compounds. Meanwhile, the second group of
approaches (endpoint methods) including linear interaction energy (LIE), molecular
mechanics Poisson-Boltzmann surface area (MM-PBSA), and molecular mechanics
generalized Born surface area (MM-GBSA) methods calculate only the difference
between the starting and last step of these two states leading to much faster
calculations than FEP and TI [10]. MM-PBSA and MM-GBSA methods have
found various applications in drug discovery studies such as pose selection, discrimination of the true actives from decoys or inactive molecules in virtual screening
(VS) studies, and prediction of the relative binding free energies in the lead optimization stage [11, 12]. The applicability of such approximations has been tested and
shown to be valuable as post-docking filter to rescore docking results and to increase
458
D. Robaa et al.
[3]. This chapter will illustrate various structure-based approaches which have been
successfully implemented to identify and design small molecule inhibitors of
histone-modifying enzymes.
2 Computational Methods to Predict the Target Interaction
of Small Molecules and Their Binding Affinities
Designing small molecules with desired biological activity has long been a major
goal of structure-based drug design. Different approaches have been tested to predict
the biological activity of an inhibitor from its calculated interaction with the target.
Molecular docking tools are able to predict the interactions in a putative binding site
and can easily screen large numbers of molecules and rank them according to their
calculated affinities [4, 5]. However, using a single frozen protein structure in most
of the cases hampers the true conformational and orientational sampling of an
inhibitor in its putative binding pocket [6]. Moreover, currently available scoring
functions that estimate the binding affinity of calculated poses for a ligand are still far
away from ranking the biologically active molecules in a reasonable agreement with
the experimental data [7]. To overcome this challenge, more sophisticated methods
that take into account conformational changes, desolvation effects, as well as
entropic effects on binding are needed to predict the binding strength of molecules
of interest.
In contrast to simple docking methods, free energy calculations were shown to
give results that are in better agreement with experimental data [8]. Two different
approaches have been widely studied for their applications in estimating the binding
affinity of small molecule inhibitors. The first approach includes the so-called
pathway methods such as free energy perturbation (FEP) and thermodynamic
integration (TI) which consider all small changes that occur between the initial and
final states [9]. These methods are more accurate and robust in predicting reliable
affinities than docking methods but are expensive and time-consuming, thus, making
them inapplicable for large datasets of compounds. Meanwhile, the second group of
approaches (endpoint methods) including linear interaction energy (LIE), molecular
mechanics Poisson-Boltzmann surface area (MM-PBSA), and molecular mechanics
generalized Born surface area (MM-GBSA) methods calculate only the difference
between the starting and last step of these two states leading to much faster
calculations than FEP and TI [10]. MM-PBSA and MM-GBSA methods have
found various applications in drug discovery studies such as pose selection, discrimination of the true actives from decoys or inactive molecules in virtual screening
(VS) studies, and prediction of the relative binding free energies in the lead optimization stage [11, 12]. The applicability of such approximations has been tested and
shown to be valuable as post-docking filter to rescore docking results and to increase
458
D. Robaa et al.