Accurate calculation of the absolute free energy of binding for drug molecules

Abstract

Accurate prediction of binding affinities has been a central goal of computational chemistry for decades, yet remains elusive. Despite good progress, the required accuracy for use in a drug-discovery context has not been consistently achieved for drug-like molecules. Here, we perform absolute free energy calculations based on a thermodynamic cycle for a set of diverse inhibitors binding to bromodomain-containing protein 4 (BRD4) and demonstrate that a mean absolute error of 0.6 kcal mol^-1 can be achieved. We also show a similar level of accuracy (1.0 kcal mol^-1) can be achieved in pseudo prospective approach. Bromodomains are epigenetic mark readers that recognize acetylation motifs and regulate gene transcription, and are currently being investigated as therapeutic targets for cancer and inflammation. The unprecedented accuracy offers the exciting prospect that the binding free energy of drug-like compounds can be predicted for pharmacologically relevant targets.

Fig. 1. Bromodomain fold and acetyl-lysine binding pocket. (a) Cartoon representation of the structure of BRD4(1) bromodomain in complex with an acetylated peptide. Crystallographically observed water molecules are represented as red spheres. (b) BRD4(1) acetyl-lysine binding site with key interacting residues labeled.

Fig. 2. Non-physical thermodynamic cycle. Scheme of the alchemical thermodynamic cycle used to obtain the absolute binding free energies. The fully interacting ligand (orange) in solution at the top left (A) is transformed into a non-interacting solute (B, white) during a series of equilibrium simulations where its electrostatic and van der Waals interactions are scaled to zero, providing the term ΔGsolvelec+vdw. The ligand is then restrained while still non-interacting with the environment (C). This step (ΔGsolvrestr) is computed analytically in accordance with the protocol described by Boresch et al. This state is equivalent to having the non-interacting ligand restrained within the protein cavity (D). The restrained and non-interacting ligand in complex with the protein has its electrostatic and VdW interactions turned back on again (E), giving ΔGprotelec+vdw. The restraints between ligand and protein are then removed (ΔGprotrestr), closing the cycle, and the final state is the unrestrained and fully interacting ligand in complex with the protein (F).

Fig. 3. Chemical structure of the ligands. The structures of the compounds analyzed in this study are shown and are labeled with Arabic numerals in descending order of affinity.

Fig. 4. Binding poses suggested by docking. In red are the crystallographic structures, and in green are the docked ligands. The ligand number and cluster letter are reported on each pose.

Fig. 5. Scatter and correlation plots of the results. Correlation plots for (a) the free energy calculations starting from the X-ray structures, (b) the docking free energy scores and (c) the free energy calculations starting from the docked structures.

Fig. 6. Multiple potential binding conformations of ligand 3. (a) Overlay of the crystal structures for ligand 3 and ligand 7 (4MR3 and ; 4MR4 respectively). (b) Overlay of docking poses 3-a and 3-c.