A more complicated example is the discovery of a hidden trench
in HIV integrase, which is not accessible within the available crystal
structure [40]. Other examples include the identification of
uncharacterized binding sites on the Cruzain, a therapeutic target
for Chagas’ disease [41] and the identification of novel transient
allosteric binding sites in Ras proteins, a hallmark of diverse
cancers [42].
All these examples emphasize the importance of introducing
protein flexibility in the search process for a binding site. In this
context, MD simulations can be used to explore the dynamicity of
the target protein [43, 44]. The method became very popular in
studying protein structures and in extracting conformational
ensembles of these targets. These simulations can be carried out
in the context of the solvent, ions, and various physiological para-
meters. In this way, one will not only able to identify and under-
stand the flexibility of a given binding site, but will be also able to
explore its interaction with water molecules and ions. Moreover, by
analyzing the MD trajectory, one can also measure the persistence
of the binding site, duration of the hydrogen bonds, and varying
depth and width of the binding site.2.4 Generation
of Conformation
Ensemble
A conformational ensemble is a group of structures from the same
target with different conformations, reflecting the flexibility of the
backbone and side chains. In the early 1990s, Pang and Kozikowski
made the first attempt to extract multiple conformations of the
acetycholinesterase enzyme from MD trajectories. Since this
moment, MD simulations become the most common way to gen-
erate such structural ensembles [45]. The depth and the shape of a
given protein pocket can be assessed during the simulation and
transient cavities can be identified and explored [44].Fig. 4Structure of Bcl-xL in (a) unbound (orange surface, PDB ID: 2LP8) and (b) bound (blue surface)
conformation. The Bak peptide (green ribbon, side chains show) is detected with the Bcl-xL by using NMR
(PDB ID: 2LPC) and superimposed onto the unbound conformation. Based on the geometric feature, the
binding site can be recognized as high druggability in bound conformation (a). However, the same binding site
of unbound Bcl-xL is identified as low druggability in (b). Generated by Maestro
Prediction of Druggable Binding Sites 93