Computational Drug Design and Synthesis - Biophysical Chemistry - Computational Organic Chemistry

A. Virtual Screening: Discovery of Novel Adenosine Receptor Antagonists through a Combined Structure- and Ligand-Based Approach

An intense effort is made by pharmaceutical and academic research laboratories to identify and develop selective antagonists for each adenosine receptor (AR) subtype as potential clinical candidates for "soft" treatment of various diseases. Crystal structures of subtypes A_2A and A₁ARs offer exciting opportunities for structure-based drug design.

• A virtual screening on Maybridge HitFinder^TM library of 14400 compounds was tested (Scheme 1). A combination of structure-based against the crystal structure of A_2AAR and ligand-based methodologies were applied. The docking poses were obtained after GOLD/ChemPLP and rescoring with ROCS. The resulting docking poses were further re-scored by CHARMM energy minimization and calculation of the desolvation energy using Poisson-Boltzmann equation electrostatics (Scheme 2). Out of the eight selected and tested compounds, five were found positive hits (63% success) based on their binding constants measured using radio-labelled assays (in collaboration, Professor Karl-Norbert Klotz, Wurtzburg)  Based on these results, 19 compounds characterized by novel chemotypes were purchased and tested. Sixteen of them were identified as AR antagonists with affinity towards combinations of the AR family isoforms (A_2A/A_3, A₁/A₃, A₁/A_2A/A₃ and A₃). Of particular interest is that the 2-amino-thiophene-3-carboxamides, 3-acylamino-5-aryl-thiophene-2-carboxamides and carbonyloxycarbo-ximidamide derivatives were found to be selective and possess a micromolar to low micromolar affinity for A₃ receptor.

Scheme 1. Workflow used for the discovery of new hits.

• The performance of hundreds of molecular dynamics (MD) simulations of complexes between the ARs and the total of 27 ligands to resolve the binding interactions of the active compounds, which were not achieved by docking calculations alone. This computational work allowed the prediction of stable and unstable complexes which agree with the experimental results of potent and inactive compounds respectively (Figure 1).

• Carbonyloxycarboximidamide derivatives were found to be selective and possess a micromolar to low micromolar affinity for A₃ receptor (Figure 2) (with Professor K.-N. Klotz, Medicinal Chemistry and Pharmacology Wurzburg).

Scheme 2. The free energy for the formation of ligand A-receptor R complex can be calculated using the end-points of this thermodynamic cycle including the bound and unbound states of the ligand according to equation 1. The MM-PBSA model was applied. For the calculation of the free energies of solvation ΔG_solv ie, for transferring ligand, protein receptor and complex from the gas to the solution phase, the electrostatic component was calculated using the Poisson-Boltzmann equation and the non-polar component was calculated based on the surface of the cave formed by solute inside the solvent. ΔΕ_ΜΜ is the molecular mechanics energy change between bound and unbound ligand. Entropy change is taken to be approximately zero.

• Currently we are performing virtual screening with small chemical libraries, consisting of compounds synthesized from researchers in our Section in NKUA, and we have identified micromolar hits of novel structure as A₁R and A₃R antagonists (with Prof. P. Marakos, Prof. N. Pouli, Assist. Prof. N. Lougiakis, Pharmacy, NKUA and Dr G. Ladds, Pharmacology, Cambridge).

Figure 1. Predicted binding modes before and after 100 ns MD simulations for ligand 17 which binds A, A2A and A3 in the orthosteric binding site of (a) A₁R, (b) A_2AR and (c) A₃R, resulting in stable complexes. Compound 17 can adopt a binding orientation inside A_2AAR in which the amido group of thiophene ring is hydrogen-bonded to N(6.55) and E(5.30), and van der Waals interactions stabilize the ligand inside the binding cavity. The replacement of E(5.30) with V in A₃R orthosteric cavity (panel c) retained the binding of compound 17, as a result of hydrogen-bonding to N(6.55) and additional favourable van der Waals interactions of its bulky lipophilic group in the vicinity of V(5.30). In A₁AR, which has a broader binding cavity, the 5-aryl-thiophene ring is inclined and the 3-NHCOR substituent is directed towards TM2.  Binding orientation of the ligand after the MD simulation is shown in yellow sticks and sidechains of some amino acids involved in ligand binding are displayed as gray sticks while the starting ligand and Ν(6.55)/Ε(5.30) side chain positions are shown in green wires. Hydrogen atoms are omitted except for those involved in hydrogen bond interactions, highlighted as black dashed lines.

 

Figure 2. (a) Τhe 3-phenyloxazole 5b is a  low micromolar A₃R selective antagonist, and an analogue (but more active) of hit 5a. (b) MD screenshot of 5b inside A₃AR orthosteric binding area.

B. Computational Biochemistry: Structural Characterization of Agonists and Antagonists Binding to A3 Adenosine Receptor

A₃AR is over-expressed in various tumor cells, compared to normal cells where it was found having low or no expression. Thus, A₃AR and its signaling pathway is a promising drug target against cancer cell proliferation and for a number of other conditions like inflammatory diseases, including asthma and rheumatoid arthritis, and ischemic injury. Structures of subtypes A_2A and A₁ARs obtained using X-ray crystallography and the structure of A_2A in complex with G-protein obtained using cryo-electron microscopy offer exciting opportunities for structure-based drug design. Currently there is no experimental structure for A₃AR.

• We investigate the orthosteric binding site of A₃AR in complex with two agonists of adenosine (Ado (1)), the selective IB-MECA (4) and the non-selective NECA (3) (Scheme 3) and a recently discovered potent antagonist (5b) (Figure 3) with a novel structure by applying an array of approaches including: (a) the performance of MD simulations (Figure 4) and Molecular Mechanics-Generalized Born Surface Area (MM-GBSA) free energy calculations of WT and mutant A₃ARs in complex with agonists IB-MECA (4), NECA (3)(Scheme 4) (b) theantagonist 5b (Figure 5), (c) in combination with several site-directed mutagenesis studies, and biological data from functional assays to validate the in silico predictions.

• The mutagenesis analysis show mutations with alanine in critical residues for agonist and antagonist activity at A₃AR, included L90^3.32, T94^3.36, W185^5.46, S271^7.42, H272^7.43, L90^3.32, F168^5.29, M174^5.35_, M177^5.38, L246^6.51, I249^6.54, N250^6.55, L264^7.35, I268^7.39, V169^5.30, I253^6.58 V169^5.30/W185^5.46 (with Dr G. Ladds, Pharmacology, Cambridge).

Scheme 3. The structures of non-selective A₃R agonists 1, 2 and of the selective A₃R agonist 3.

Figure 3. (A) Snapshot of IB-MECA (3) (left) and NECA (2) (right) inside the orthosteric binding area of WT A₃R at 150 ns of the MD simulation, (B) 2D interaction diagram and (C) receptor-ligand interaction histogram for IB-MECA (3) (left) and NECA (2) (right) inside the orthosteric binding area of WT A₃R for 0-150 ns of MD simulations. Bars are plotted for residues with interaction frequencies ≥ 0.2 for either IB-MECA (3) or NECA (2). Color scheme (A): Ligand=cyan sticks, receptor=white ribbon and sticks, H-bond interactions=yellow dashes, pi-pi interactions=pink dashes; Color scheme (B): polar surface/residues=blue, hydrophobic residues=green; Color scheme (C): H-bond interactions=dark blue, hydrophobic interactions=grey, pi-pi interactions=green, water-mediated interactions=light blue.