<p>Adenosine A<sub>3 </sub>receptor
(A<sub>3</sub>R), is a promising
drug target against cancer cell proliferation. Currently
there is no experimentally determined structure of A<sub>3</sub>R. Here, we have investigate a computational model,
previously applied successfully
for agonists binding to A<sub>3</sub>R, using
molecular dynamic (MD) simulations, Molecular Mechanics-Poisson Boltzmann
Surface Area (MM-PBSA) and Molecular Mechanics-Generalized Born Surface Area
(MM-GBSA) binding free energy calculations. Extensive
computations were performed to explore the binding profile of
O4-{[3-(2,6-dichlorophenyl)-5-methylisoxazol-4-yl]carbonyl}-2-methyl-1,3-thiazole-4-carbohydroximamide
(K18) to A<sub>3</sub>R. K18 is a new specific and competitive antagonist at
the orthosteric binding site of A<sub>3</sub>R, discovered using virtual
screening and characterized pharmacologically in our previous studies. The most plausible binding conformation for the dichlorophenyl
group of K18 inside the A<sub>3</sub>R is oriented
towards trans-membrane helices (TM) 5 and 6, according to the MM-PBSA
and MM-GBSA binding free energy calculations, and by the
previous results obtained by mutating residues of TM5, TM6 to alanine
which reduce antagonist potency. The results from 14 site-directed
mutagenesis experiments were interpreted using MD simulations and
MM-GBSA calculations which show that the relative binding free energies of the
mutant A<sub>3</sub>R - K18 complexes compare to the
WT A<sub>3</sub>R are in agreement with the effect of the mutations,
i.e. the reduction, maintenance or increase of antagonist potency. We show that
when the residues V169<sup>5.30</sup>, M177<sup>5.38</sup>, I249<sup>6.54</sup>
involved in direct interactions with K18 are mutated to alanine, the mutant A<sub>3</sub>R
- K18 complexes reduce potency, increase the RMSD value of K18 inside the
binding area and the MM-GBSA binding free energy compared to the WT A<sub>3</sub>R
complex. Our computational model shows that other mutant A<sub>3</sub>R
complexes with K18, including directly interacting residues, i.e. F168<sup>5.29</sup>A,
L246<sup>6.51</sup>A, N250<sup>6.55</sup>A complexes with K18 are not stable.
In these complexes of A<sub>3</sub>R mutated in directly interacting residues
one or more of the interactions between K18 and these residues are lost. In
agreement with the experiments, the computations show that, M174<sup>5.35</sup> a residue which does not make direct
interactions with K18 is critical for K18 binding. A
striking results is that the mutation of residue V169<sup>5.30</sup> to
glutamic acid maintained antagonistic potency. This effect is in agreement with
the binding free energy calculations and it is suggested that is due to K18
re-orientation but also to the plasticity of A<sub>3</sub>R binding area. The
mutation of direct interacting L90<sup>3.32</sup> in the low region and the
non-directly interacting L264<sup>7.35</sup> to alanine in the middle region
increases the antagonistic potency, suggesting that chemical modifications of
K18 can be applied to augment antagonistic potency. The
calculated binding energies Δ<i>G</i><sub>eff</sub>
values of K18 against mutant A<sub>3</sub>Rs displayed
very good correlation with experimental potencies (pA<sub>2</sub>
values). These results further approve the computational model for the
description of K18 binding with critical residues of the orthosteric binding
area which can have implications for the design of more effective antagonists
based on the structure of K18.</p>
Prediction of Binding Free Energy of Protein–Ligand Complexes with a Hybrid Molecular Mechanics/Generalized Born Surface Area and Machine Learning Method
