Chemistry Reference
In-Depth Information
unambiguous solution since the ligand decomposition may have subjective errors, there may be many variants of
adjustment of the fragments to each other and it may be necessary to estimate the conformational flexibility of the
whole system. Some authors propose that the reasons for the low predictivity of the docking, with rms deviations
above 2 Å may be attributed to the experimental data and the weaknesses of the docking methods [303].
A new methodology to describe the interactions in 'receptor-ligand' complexes was introduced in order to address
some of the above-mentioned problems, i.e the 3D/4D QSAR algorithm BiS/Multiconformational (BiS/MC) and
CoCon algorithms [98]. The first algorithm performs the restricted docking of compounds to receptor pockets. The
other algorithm determines the relationships between the bioactivity and the parameters of interactions in the
'receptor-ligand' complexes, including a new formalism for estimating hydrogen bond energies. 3D structures of the
ligand-receptor complexes are used and the receptor geometry in the complexes is prepared for the interaction with a
ligand changing, to a lesser degree, as it receives different ligands.
The method aligns molecules onto each other by considering their Coulomb and van der Waals potentials at points
on the molecular surface. The algorithm reconstructs the model receptor as being complementary to the field of the
generalized molecular set whereas the receptor is represented as a set of pseudoatoms whose parameters can be
calculated from the complementarity formalism. The molecules are oriented in the complementary receptor model
and the maximum total probability of the interactions of a molecule with the model receptor is optimized. The use of
probability increases the influence of the most important `receptor atom-ligand atom` interactions on the molecular
orientation. The model receptor is flexible and every ligand is represented as a set of conformers, found with the
algorithm, interfaced with the MM3 force field which allows taking into account the flexibilities of both the receptor
and ligands. The reconstruction of the complementary model receptor is performed by means of an algorithm which
gets additional information on the molecular orientation in the receptor cavity.
The BiS/MC method uses the MERA (Model of Effective Radii of atoms) force field which is a nonparametric
model for simulating the effective atomic radii whereas each atom fills the space allowed by other atoms such that
the atom is represented by an expanding balloon. Other expanding spheres limit the expansion of the atomic spheres
such that the expansion of the balloon ceases when the external pressure provided by the other atoms is equal to the
internal pressure of the balloon. The internal density of the atom
i
is proportional to its inverse volume such that the
external density equals the sum of the inverse volumes of spheres with radii equal to the distances between atoms
i
and
j.
It is possible to calculate the zero-order approximation volume of each atom whereas the sum of the atomic
volumes must be equal to the molecular volume allowing the calculation of the density of the compound [309].
A new approach (CoCon) for determination of mechanisms of biological action of compounds as well as search for
active centers of ligands and receptors was reported [98]
.
The basic idea of the CoCon algorithm is that biological
activity of a compound depends mostly on the interaction ability of the active centers of the receptor and ligand sites
whereas relationships are created between the biological activity and the parameters of interactions in the 'receptor-
ligand' complexes. The interaction depends on the atom-atomic potentials of van der Waals and Coulomb
interactions which permit the decomposition of the interaction energies in search of the best linear relationship
between the biological activity and the parameters of the interaction. In CoCon the parameters that determine
biological activity belong to the receptor and ligand atoms that are supposed to be active centers. Forward and
backward stepwise procedures of the regression analysis can be used for the decomposition with this algorithm. The
leave-one-out cross validation technique can be used to estimate the quality of the relationships [98].
There is still considerable debate regarding whether the conformation of the protein in the ligand-bound state was
induced by the ligand or whether this state already existed before binding by the ligand and was selected from an
ensemble of different conformations. Certainly, for the determination of the conformation of the protein to which it
is bound, the ligand plays a central role. For half a century the existence of ligand-induced flexibility was considered
inconsequential or rare whereas today it is known that a large number of proteins contain as much as three flexible
residues in the active site and thus single rigid dockings predict incorrect binding pose for most ligands. Major
domain rearrangements, simpler backbone-loop and side-chain movements are often observed in flexible residues
leading in general to an unclear picture of protein changes induced by ligands. Simultaneous ligand effects on
protein conformation, separating the amount of protein effects on ligand conformation, inadequate understanding of
receptor flexibility, due to paucity of experimental data, are some of the issues that need to be addressed in
theoretical-computational techniques [336].
